{
  "best_metric": 0.3083859980106354,
  "best_model_checkpoint": "./w2v-bert-2.0-chichewa_34_34h/checkpoint-9000",
  "epoch": 25.831024930747922,
  "eval_steps": 1000,
  "global_step": 14000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0018467220683287165,
      "grad_norm": 22.621971130371094,
      "learning_rate": 3.0000000000000004e-09,
      "loss": 9.3625,
      "step": 1
    },
    {
      "epoch": 0.003693444136657433,
      "grad_norm": 20.207006454467773,
      "learning_rate": 6.000000000000001e-09,
      "loss": 8.455,
      "step": 2
    },
    {
      "epoch": 0.00554016620498615,
      "grad_norm": 19.7971134185791,
      "learning_rate": 9e-09,
      "loss": 8.2344,
      "step": 3
    },
    {
      "epoch": 0.007386888273314866,
      "grad_norm": 19.891284942626953,
      "learning_rate": 1.2000000000000002e-08,
      "loss": 8.3048,
      "step": 4
    },
    {
      "epoch": 0.009233610341643583,
      "grad_norm": 19.400951385498047,
      "learning_rate": 1.5000000000000002e-08,
      "loss": 8.0988,
      "step": 5
    },
    {
      "epoch": 0.0110803324099723,
      "grad_norm": 20.50615882873535,
      "learning_rate": 1.8e-08,
      "loss": 8.3828,
      "step": 6
    },
    {
      "epoch": 0.012927054478301015,
      "grad_norm": 20.289794921875,
      "learning_rate": 2.1e-08,
      "loss": 8.2395,
      "step": 7
    },
    {
      "epoch": 0.014773776546629732,
      "grad_norm": 20.422374725341797,
      "learning_rate": 2.4000000000000003e-08,
      "loss": 8.3791,
      "step": 8
    },
    {
      "epoch": 0.01662049861495845,
      "grad_norm": 20.068317413330078,
      "learning_rate": 2.7e-08,
      "loss": 8.223,
      "step": 9
    },
    {
      "epoch": 0.018467220683287166,
      "grad_norm": 20.406330108642578,
      "learning_rate": 3.0000000000000004e-08,
      "loss": 8.2373,
      "step": 10
    },
    {
      "epoch": 0.020313942751615882,
      "grad_norm": 20.23192024230957,
      "learning_rate": 3.3000000000000004e-08,
      "loss": 8.1534,
      "step": 11
    },
    {
      "epoch": 0.0221606648199446,
      "grad_norm": 20.09835433959961,
      "learning_rate": 3.6e-08,
      "loss": 8.0391,
      "step": 12
    },
    {
      "epoch": 0.024007386888273315,
      "grad_norm": 21.759641647338867,
      "learning_rate": 3.9e-08,
      "loss": 8.7457,
      "step": 13
    },
    {
      "epoch": 0.02585410895660203,
      "grad_norm": 21.865198135375977,
      "learning_rate": 4.2e-08,
      "loss": 8.8412,
      "step": 14
    },
    {
      "epoch": 0.027700831024930747,
      "grad_norm": 22.47687530517578,
      "learning_rate": 4.5e-08,
      "loss": 8.9815,
      "step": 15
    },
    {
      "epoch": 0.029547553093259463,
      "grad_norm": 24.19575309753418,
      "learning_rate": 4.8000000000000006e-08,
      "loss": 9.5453,
      "step": 16
    },
    {
      "epoch": 0.03139427516158818,
      "grad_norm": 23.769922256469727,
      "learning_rate": 5.1e-08,
      "loss": 9.4247,
      "step": 17
    },
    {
      "epoch": 0.0332409972299169,
      "grad_norm": 26.34709358215332,
      "learning_rate": 5.4e-08,
      "loss": 10.171,
      "step": 18
    },
    {
      "epoch": 0.03508771929824561,
      "grad_norm": 27.064607620239258,
      "learning_rate": 5.7e-08,
      "loss": 10.2378,
      "step": 19
    },
    {
      "epoch": 0.03693444136657433,
      "grad_norm": 25.482282638549805,
      "learning_rate": 6.000000000000001e-08,
      "loss": 9.9684,
      "step": 20
    },
    {
      "epoch": 0.038781163434903045,
      "grad_norm": 26.783565521240234,
      "learning_rate": 6.3e-08,
      "loss": 10.2665,
      "step": 21
    },
    {
      "epoch": 0.040627885503231764,
      "grad_norm": 25.946687698364258,
      "learning_rate": 6.600000000000001e-08,
      "loss": 9.8199,
      "step": 22
    },
    {
      "epoch": 0.04247460757156048,
      "grad_norm": 26.15570831298828,
      "learning_rate": 6.9e-08,
      "loss": 10.0497,
      "step": 23
    },
    {
      "epoch": 0.0443213296398892,
      "grad_norm": 26.88849639892578,
      "learning_rate": 7.2e-08,
      "loss": 10.1436,
      "step": 24
    },
    {
      "epoch": 0.046168051708217916,
      "grad_norm": 25.654088973999023,
      "learning_rate": 7.500000000000001e-08,
      "loss": 9.8762,
      "step": 25
    },
    {
      "epoch": 0.04801477377654663,
      "grad_norm": 26.350845336914062,
      "learning_rate": 7.8e-08,
      "loss": 9.9597,
      "step": 26
    },
    {
      "epoch": 0.04986149584487535,
      "grad_norm": 26.18523597717285,
      "learning_rate": 8.100000000000001e-08,
      "loss": 9.9127,
      "step": 27
    },
    {
      "epoch": 0.05170821791320406,
      "grad_norm": 25.764511108398438,
      "learning_rate": 8.4e-08,
      "loss": 9.9067,
      "step": 28
    },
    {
      "epoch": 0.05355493998153278,
      "grad_norm": 26.577726364135742,
      "learning_rate": 8.7e-08,
      "loss": 9.9984,
      "step": 29
    },
    {
      "epoch": 0.055401662049861494,
      "grad_norm": 26.31147575378418,
      "learning_rate": 9e-08,
      "loss": 9.9988,
      "step": 30
    },
    {
      "epoch": 0.057248384118190214,
      "grad_norm": 26.770029067993164,
      "learning_rate": 9.3e-08,
      "loss": 10.0926,
      "step": 31
    },
    {
      "epoch": 0.05909510618651893,
      "grad_norm": 26.448558807373047,
      "learning_rate": 9.600000000000001e-08,
      "loss": 9.9067,
      "step": 32
    },
    {
      "epoch": 0.060941828254847646,
      "grad_norm": 26.373918533325195,
      "learning_rate": 9.9e-08,
      "loss": 9.9779,
      "step": 33
    },
    {
      "epoch": 0.06278855032317636,
      "grad_norm": 27.763418197631836,
      "learning_rate": 1.02e-07,
      "loss": 10.1244,
      "step": 34
    },
    {
      "epoch": 0.06463527239150507,
      "grad_norm": 25.96488380432129,
      "learning_rate": 1.05e-07,
      "loss": 9.7475,
      "step": 35
    },
    {
      "epoch": 0.0664819944598338,
      "grad_norm": 25.816864013671875,
      "learning_rate": 1.08e-07,
      "loss": 9.7923,
      "step": 36
    },
    {
      "epoch": 0.06832871652816251,
      "grad_norm": 26.328279495239258,
      "learning_rate": 1.11e-07,
      "loss": 9.8985,
      "step": 37
    },
    {
      "epoch": 0.07017543859649122,
      "grad_norm": 25.940914154052734,
      "learning_rate": 1.14e-07,
      "loss": 9.7801,
      "step": 38
    },
    {
      "epoch": 0.07202216066481995,
      "grad_norm": 25.886962890625,
      "learning_rate": 1.17e-07,
      "loss": 9.7807,
      "step": 39
    },
    {
      "epoch": 0.07386888273314866,
      "grad_norm": 27.217947006225586,
      "learning_rate": 1.2000000000000002e-07,
      "loss": 9.8574,
      "step": 40
    },
    {
      "epoch": 0.07571560480147738,
      "grad_norm": 26.309337615966797,
      "learning_rate": 1.23e-07,
      "loss": 9.5134,
      "step": 41
    },
    {
      "epoch": 0.07756232686980609,
      "grad_norm": 25.59661102294922,
      "learning_rate": 1.26e-07,
      "loss": 9.513,
      "step": 42
    },
    {
      "epoch": 0.07940904893813482,
      "grad_norm": 27.14220428466797,
      "learning_rate": 1.29e-07,
      "loss": 9.6874,
      "step": 43
    },
    {
      "epoch": 0.08125577100646353,
      "grad_norm": 24.923891067504883,
      "learning_rate": 1.3200000000000002e-07,
      "loss": 9.3105,
      "step": 44
    },
    {
      "epoch": 0.08310249307479224,
      "grad_norm": 25.352060317993164,
      "learning_rate": 1.35e-07,
      "loss": 9.3556,
      "step": 45
    },
    {
      "epoch": 0.08494921514312095,
      "grad_norm": 25.561046600341797,
      "learning_rate": 1.38e-07,
      "loss": 9.2043,
      "step": 46
    },
    {
      "epoch": 0.08679593721144968,
      "grad_norm": 25.8244686126709,
      "learning_rate": 1.41e-07,
      "loss": 9.4057,
      "step": 47
    },
    {
      "epoch": 0.0886426592797784,
      "grad_norm": 26.3791561126709,
      "learning_rate": 1.44e-07,
      "loss": 9.2032,
      "step": 48
    },
    {
      "epoch": 0.0904893813481071,
      "grad_norm": 25.30979347229004,
      "learning_rate": 1.47e-07,
      "loss": 9.1291,
      "step": 49
    },
    {
      "epoch": 0.09233610341643583,
      "grad_norm": 23.93961524963379,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 8.591,
      "step": 50
    },
    {
      "epoch": 0.09418282548476455,
      "grad_norm": 23.533157348632812,
      "learning_rate": 1.53e-07,
      "loss": 9.1986,
      "step": 51
    },
    {
      "epoch": 0.09602954755309326,
      "grad_norm": 19.902755737304688,
      "learning_rate": 1.56e-07,
      "loss": 7.9289,
      "step": 52
    },
    {
      "epoch": 0.09787626962142197,
      "grad_norm": 21.54668426513672,
      "learning_rate": 1.59e-07,
      "loss": 8.4068,
      "step": 53
    },
    {
      "epoch": 0.0997229916897507,
      "grad_norm": 20.574657440185547,
      "learning_rate": 1.6200000000000002e-07,
      "loss": 8.0642,
      "step": 54
    },
    {
      "epoch": 0.10156971375807941,
      "grad_norm": 20.305767059326172,
      "learning_rate": 1.6499999999999998e-07,
      "loss": 7.8933,
      "step": 55
    },
    {
      "epoch": 0.10341643582640812,
      "grad_norm": 21.524248123168945,
      "learning_rate": 1.68e-07,
      "loss": 8.2149,
      "step": 56
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 20.279159545898438,
      "learning_rate": 1.71e-07,
      "loss": 7.8225,
      "step": 57
    },
    {
      "epoch": 0.10710987996306556,
      "grad_norm": 20.45210838317871,
      "learning_rate": 1.74e-07,
      "loss": 7.8799,
      "step": 58
    },
    {
      "epoch": 0.10895660203139428,
      "grad_norm": 21.43587303161621,
      "learning_rate": 1.77e-07,
      "loss": 7.9268,
      "step": 59
    },
    {
      "epoch": 0.11080332409972299,
      "grad_norm": 20.783554077148438,
      "learning_rate": 1.8e-07,
      "loss": 7.9966,
      "step": 60
    },
    {
      "epoch": 0.11265004616805172,
      "grad_norm": 20.195615768432617,
      "learning_rate": 1.83e-07,
      "loss": 7.7444,
      "step": 61
    },
    {
      "epoch": 0.11449676823638043,
      "grad_norm": 22.26091766357422,
      "learning_rate": 1.86e-07,
      "loss": 8.2015,
      "step": 62
    },
    {
      "epoch": 0.11634349030470914,
      "grad_norm": 22.69646453857422,
      "learning_rate": 1.89e-07,
      "loss": 8.3655,
      "step": 63
    },
    {
      "epoch": 0.11819021237303785,
      "grad_norm": 22.68636131286621,
      "learning_rate": 1.9200000000000003e-07,
      "loss": 8.1947,
      "step": 64
    },
    {
      "epoch": 0.12003693444136658,
      "grad_norm": 25.018808364868164,
      "learning_rate": 1.9499999999999999e-07,
      "loss": 8.832,
      "step": 65
    },
    {
      "epoch": 0.12188365650969529,
      "grad_norm": 27.025386810302734,
      "learning_rate": 1.98e-07,
      "loss": 9.3476,
      "step": 66
    },
    {
      "epoch": 0.123730378578024,
      "grad_norm": 27.390701293945312,
      "learning_rate": 2.01e-07,
      "loss": 9.5083,
      "step": 67
    },
    {
      "epoch": 0.12557710064635272,
      "grad_norm": 28.93164825439453,
      "learning_rate": 2.04e-07,
      "loss": 9.9266,
      "step": 68
    },
    {
      "epoch": 0.12742382271468145,
      "grad_norm": 29.568239212036133,
      "learning_rate": 2.0700000000000001e-07,
      "loss": 9.9968,
      "step": 69
    },
    {
      "epoch": 0.12927054478301014,
      "grad_norm": 31.310028076171875,
      "learning_rate": 2.1e-07,
      "loss": 10.2074,
      "step": 70
    },
    {
      "epoch": 0.13111726685133887,
      "grad_norm": 29.128494262695312,
      "learning_rate": 2.1300000000000001e-07,
      "loss": 9.6556,
      "step": 71
    },
    {
      "epoch": 0.1329639889196676,
      "grad_norm": 29.401884078979492,
      "learning_rate": 2.16e-07,
      "loss": 9.7587,
      "step": 72
    },
    {
      "epoch": 0.1348107109879963,
      "grad_norm": 29.164548873901367,
      "learning_rate": 2.1900000000000002e-07,
      "loss": 9.6016,
      "step": 73
    },
    {
      "epoch": 0.13665743305632502,
      "grad_norm": 29.890323638916016,
      "learning_rate": 2.22e-07,
      "loss": 9.6611,
      "step": 74
    },
    {
      "epoch": 0.13850415512465375,
      "grad_norm": 30.824186325073242,
      "learning_rate": 2.25e-07,
      "loss": 9.8628,
      "step": 75
    },
    {
      "epoch": 0.14035087719298245,
      "grad_norm": 30.39980125427246,
      "learning_rate": 2.28e-07,
      "loss": 9.7659,
      "step": 76
    },
    {
      "epoch": 0.14219759926131118,
      "grad_norm": 28.811660766601562,
      "learning_rate": 2.3100000000000002e-07,
      "loss": 9.4846,
      "step": 77
    },
    {
      "epoch": 0.1440443213296399,
      "grad_norm": 30.074052810668945,
      "learning_rate": 2.34e-07,
      "loss": 9.6802,
      "step": 78
    },
    {
      "epoch": 0.1458910433979686,
      "grad_norm": 30.820892333984375,
      "learning_rate": 2.3700000000000002e-07,
      "loss": 9.5901,
      "step": 79
    },
    {
      "epoch": 0.14773776546629733,
      "grad_norm": 28.861011505126953,
      "learning_rate": 2.4000000000000003e-07,
      "loss": 9.3774,
      "step": 80
    },
    {
      "epoch": 0.14958448753462603,
      "grad_norm": 29.80718231201172,
      "learning_rate": 2.43e-07,
      "loss": 9.5449,
      "step": 81
    },
    {
      "epoch": 0.15143120960295475,
      "grad_norm": 29.73357391357422,
      "learning_rate": 2.46e-07,
      "loss": 9.2557,
      "step": 82
    },
    {
      "epoch": 0.15327793167128348,
      "grad_norm": 29.006399154663086,
      "learning_rate": 2.49e-07,
      "loss": 9.2959,
      "step": 83
    },
    {
      "epoch": 0.15512465373961218,
      "grad_norm": 31.09878921508789,
      "learning_rate": 2.52e-07,
      "loss": 9.5172,
      "step": 84
    },
    {
      "epoch": 0.1569713758079409,
      "grad_norm": 30.797834396362305,
      "learning_rate": 2.5500000000000005e-07,
      "loss": 9.4964,
      "step": 85
    },
    {
      "epoch": 0.15881809787626963,
      "grad_norm": 30.867095947265625,
      "learning_rate": 2.58e-07,
      "loss": 9.4663,
      "step": 86
    },
    {
      "epoch": 0.16066481994459833,
      "grad_norm": 30.52632713317871,
      "learning_rate": 2.6099999999999997e-07,
      "loss": 9.1204,
      "step": 87
    },
    {
      "epoch": 0.16251154201292706,
      "grad_norm": 30.950654983520508,
      "learning_rate": 2.6400000000000003e-07,
      "loss": 9.2421,
      "step": 88
    },
    {
      "epoch": 0.16435826408125578,
      "grad_norm": 31.74310302734375,
      "learning_rate": 2.67e-07,
      "loss": 9.2844,
      "step": 89
    },
    {
      "epoch": 0.16620498614958448,
      "grad_norm": 31.408519744873047,
      "learning_rate": 2.7e-07,
      "loss": 9.2126,
      "step": 90
    },
    {
      "epoch": 0.1680517082179132,
      "grad_norm": 31.419960021972656,
      "learning_rate": 2.73e-07,
      "loss": 9.2066,
      "step": 91
    },
    {
      "epoch": 0.1698984302862419,
      "grad_norm": 30.93468475341797,
      "learning_rate": 2.76e-07,
      "loss": 9.0334,
      "step": 92
    },
    {
      "epoch": 0.17174515235457063,
      "grad_norm": 31.12645721435547,
      "learning_rate": 2.79e-07,
      "loss": 8.8968,
      "step": 93
    },
    {
      "epoch": 0.17359187442289936,
      "grad_norm": 30.42648696899414,
      "learning_rate": 2.82e-07,
      "loss": 8.8951,
      "step": 94
    },
    {
      "epoch": 0.17543859649122806,
      "grad_norm": 29.975467681884766,
      "learning_rate": 2.85e-07,
      "loss": 8.8357,
      "step": 95
    },
    {
      "epoch": 0.1772853185595568,
      "grad_norm": 30.835067749023438,
      "learning_rate": 2.88e-07,
      "loss": 8.7768,
      "step": 96
    },
    {
      "epoch": 0.1791320406278855,
      "grad_norm": 30.060455322265625,
      "learning_rate": 2.91e-07,
      "loss": 8.7634,
      "step": 97
    },
    {
      "epoch": 0.1809787626962142,
      "grad_norm": 30.031410217285156,
      "learning_rate": 2.94e-07,
      "loss": 8.5461,
      "step": 98
    },
    {
      "epoch": 0.18282548476454294,
      "grad_norm": 31.30252456665039,
      "learning_rate": 2.97e-07,
      "loss": 8.5539,
      "step": 99
    },
    {
      "epoch": 0.18467220683287167,
      "grad_norm": 29.008281707763672,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 8.2293,
      "step": 100
    },
    {
      "epoch": 0.18651892890120036,
      "grad_norm": 32.37430191040039,
      "learning_rate": 3.03e-07,
      "loss": 9.5889,
      "step": 101
    },
    {
      "epoch": 0.1883656509695291,
      "grad_norm": 25.16545295715332,
      "learning_rate": 3.06e-07,
      "loss": 7.8944,
      "step": 102
    },
    {
      "epoch": 0.1902123730378578,
      "grad_norm": 23.66377067565918,
      "learning_rate": 3.0900000000000003e-07,
      "loss": 7.4857,
      "step": 103
    },
    {
      "epoch": 0.19205909510618652,
      "grad_norm": 24.09159278869629,
      "learning_rate": 3.12e-07,
      "loss": 7.511,
      "step": 104
    },
    {
      "epoch": 0.19390581717451524,
      "grad_norm": 23.55128288269043,
      "learning_rate": 3.15e-07,
      "loss": 7.3832,
      "step": 105
    },
    {
      "epoch": 0.19575253924284394,
      "grad_norm": 24.18902587890625,
      "learning_rate": 3.18e-07,
      "loss": 7.4328,
      "step": 106
    },
    {
      "epoch": 0.19759926131117267,
      "grad_norm": 23.721189498901367,
      "learning_rate": 3.21e-07,
      "loss": 7.3789,
      "step": 107
    },
    {
      "epoch": 0.1994459833795014,
      "grad_norm": 25.13196563720703,
      "learning_rate": 3.2400000000000004e-07,
      "loss": 7.4585,
      "step": 108
    },
    {
      "epoch": 0.2012927054478301,
      "grad_norm": 23.71413230895996,
      "learning_rate": 3.27e-07,
      "loss": 7.1487,
      "step": 109
    },
    {
      "epoch": 0.20313942751615882,
      "grad_norm": 24.34363555908203,
      "learning_rate": 3.2999999999999996e-07,
      "loss": 7.1516,
      "step": 110
    },
    {
      "epoch": 0.20498614958448755,
      "grad_norm": 24.962244033813477,
      "learning_rate": 3.3300000000000003e-07,
      "loss": 7.2355,
      "step": 111
    },
    {
      "epoch": 0.20683287165281625,
      "grad_norm": 25.698650360107422,
      "learning_rate": 3.36e-07,
      "loss": 7.3704,
      "step": 112
    },
    {
      "epoch": 0.20867959372114497,
      "grad_norm": 29.467025756835938,
      "learning_rate": 3.39e-07,
      "loss": 7.9443,
      "step": 113
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 28.43014907836914,
      "learning_rate": 3.42e-07,
      "loss": 7.7096,
      "step": 114
    },
    {
      "epoch": 0.2123730378578024,
      "grad_norm": 33.474647521972656,
      "learning_rate": 3.45e-07,
      "loss": 8.204,
      "step": 115
    },
    {
      "epoch": 0.21421975992613113,
      "grad_norm": Infinity,
      "learning_rate": 3.45e-07,
      "loss": 8.3935,
      "step": 116
    },
    {
      "epoch": 0.21606648199445982,
      "grad_norm": 33.81159591674805,
      "learning_rate": 3.48e-07,
      "loss": 8.3841,
      "step": 117
    },
    {
      "epoch": 0.21791320406278855,
      "grad_norm": 35.76697540283203,
      "learning_rate": 3.51e-07,
      "loss": 8.526,
      "step": 118
    },
    {
      "epoch": 0.21975992613111728,
      "grad_norm": 37.332489013671875,
      "learning_rate": 3.54e-07,
      "loss": 8.5882,
      "step": 119
    },
    {
      "epoch": 0.22160664819944598,
      "grad_norm": 38.213417053222656,
      "learning_rate": 3.5700000000000003e-07,
      "loss": 8.5769,
      "step": 120
    },
    {
      "epoch": 0.2234533702677747,
      "grad_norm": 37.23543167114258,
      "learning_rate": 3.6e-07,
      "loss": 8.5438,
      "step": 121
    },
    {
      "epoch": 0.22530009233610343,
      "grad_norm": 41.91793441772461,
      "learning_rate": 3.63e-07,
      "loss": 8.8844,
      "step": 122
    },
    {
      "epoch": 0.22714681440443213,
      "grad_norm": 38.34842300415039,
      "learning_rate": 3.66e-07,
      "loss": 8.4606,
      "step": 123
    },
    {
      "epoch": 0.22899353647276086,
      "grad_norm": 43.13532638549805,
      "learning_rate": 3.6900000000000004e-07,
      "loss": 8.8698,
      "step": 124
    },
    {
      "epoch": 0.23084025854108955,
      "grad_norm": 39.65250015258789,
      "learning_rate": 3.72e-07,
      "loss": 8.4624,
      "step": 125
    },
    {
      "epoch": 0.23268698060941828,
      "grad_norm": 39.97956848144531,
      "learning_rate": 3.75e-07,
      "loss": 8.4699,
      "step": 126
    },
    {
      "epoch": 0.234533702677747,
      "grad_norm": 39.858314514160156,
      "learning_rate": 3.78e-07,
      "loss": 8.4046,
      "step": 127
    },
    {
      "epoch": 0.2363804247460757,
      "grad_norm": 40.01725387573242,
      "learning_rate": 3.81e-07,
      "loss": 8.3207,
      "step": 128
    },
    {
      "epoch": 0.23822714681440443,
      "grad_norm": 38.98045349121094,
      "learning_rate": 3.8400000000000005e-07,
      "loss": 8.0651,
      "step": 129
    },
    {
      "epoch": 0.24007386888273316,
      "grad_norm": 41.58202362060547,
      "learning_rate": 3.87e-07,
      "loss": 8.1929,
      "step": 130
    },
    {
      "epoch": 0.24192059095106186,
      "grad_norm": 41.836204528808594,
      "learning_rate": 3.8999999999999997e-07,
      "loss": 8.0894,
      "step": 131
    },
    {
      "epoch": 0.24376731301939059,
      "grad_norm": 40.90363693237305,
      "learning_rate": 3.9300000000000004e-07,
      "loss": 7.9278,
      "step": 132
    },
    {
      "epoch": 0.24561403508771928,
      "grad_norm": 44.515872955322266,
      "learning_rate": 3.96e-07,
      "loss": 8.1413,
      "step": 133
    },
    {
      "epoch": 0.247460757156048,
      "grad_norm": 42.47338104248047,
      "learning_rate": 3.99e-07,
      "loss": 8.0039,
      "step": 134
    },
    {
      "epoch": 0.24930747922437674,
      "grad_norm": 44.11112594604492,
      "learning_rate": 4.02e-07,
      "loss": 8.0041,
      "step": 135
    },
    {
      "epoch": 0.25115420129270544,
      "grad_norm": 46.43265914916992,
      "learning_rate": 4.05e-07,
      "loss": 8.0428,
      "step": 136
    },
    {
      "epoch": 0.2530009233610342,
      "grad_norm": 45.253013610839844,
      "learning_rate": 4.08e-07,
      "loss": 8.0573,
      "step": 137
    },
    {
      "epoch": 0.2548476454293629,
      "grad_norm": 43.769744873046875,
      "learning_rate": 4.11e-07,
      "loss": 7.8205,
      "step": 138
    },
    {
      "epoch": 0.2566943674976916,
      "grad_norm": 46.40185546875,
      "learning_rate": 4.1400000000000003e-07,
      "loss": 7.7984,
      "step": 139
    },
    {
      "epoch": 0.2585410895660203,
      "grad_norm": 44.123138427734375,
      "learning_rate": 4.17e-07,
      "loss": 7.5095,
      "step": 140
    },
    {
      "epoch": 0.26038781163434904,
      "grad_norm": 47.95862579345703,
      "learning_rate": 4.2e-07,
      "loss": 7.7006,
      "step": 141
    },
    {
      "epoch": 0.26223453370267774,
      "grad_norm": 43.312129974365234,
      "learning_rate": 4.23e-07,
      "loss": 7.2459,
      "step": 142
    },
    {
      "epoch": 0.26408125577100644,
      "grad_norm": 45.633419036865234,
      "learning_rate": 4.2600000000000003e-07,
      "loss": 7.3567,
      "step": 143
    },
    {
      "epoch": 0.2659279778393352,
      "grad_norm": 43.2322998046875,
      "learning_rate": 4.2900000000000004e-07,
      "loss": 7.2379,
      "step": 144
    },
    {
      "epoch": 0.2677746999076639,
      "grad_norm": 41.97365188598633,
      "learning_rate": 4.32e-07,
      "loss": 6.94,
      "step": 145
    },
    {
      "epoch": 0.2696214219759926,
      "grad_norm": 45.6364860534668,
      "learning_rate": 4.35e-07,
      "loss": 7.1133,
      "step": 146
    },
    {
      "epoch": 0.27146814404432135,
      "grad_norm": 44.61563491821289,
      "learning_rate": 4.3800000000000003e-07,
      "loss": 6.9913,
      "step": 147
    },
    {
      "epoch": 0.27331486611265005,
      "grad_norm": 43.38426208496094,
      "learning_rate": 4.41e-07,
      "loss": 6.8122,
      "step": 148
    },
    {
      "epoch": 0.27516158818097874,
      "grad_norm": 44.04775619506836,
      "learning_rate": 4.44e-07,
      "loss": 6.7077,
      "step": 149
    },
    {
      "epoch": 0.2770083102493075,
      "grad_norm": 41.649017333984375,
      "learning_rate": 4.47e-07,
      "loss": 6.4878,
      "step": 150
    },
    {
      "epoch": 0.2788550323176362,
      "grad_norm": 38.133544921875,
      "learning_rate": 4.5e-07,
      "loss": 6.4685,
      "step": 151
    },
    {
      "epoch": 0.2807017543859649,
      "grad_norm": 39.247467041015625,
      "learning_rate": 4.5300000000000005e-07,
      "loss": 6.4852,
      "step": 152
    },
    {
      "epoch": 0.28254847645429365,
      "grad_norm": 36.58827590942383,
      "learning_rate": 4.56e-07,
      "loss": 6.2618,
      "step": 153
    },
    {
      "epoch": 0.28439519852262235,
      "grad_norm": 33.99684524536133,
      "learning_rate": 4.5899999999999997e-07,
      "loss": 6.0097,
      "step": 154
    },
    {
      "epoch": 0.28624192059095105,
      "grad_norm": 37.434478759765625,
      "learning_rate": 4.6200000000000003e-07,
      "loss": 6.1671,
      "step": 155
    },
    {
      "epoch": 0.2880886426592798,
      "grad_norm": 35.457984924316406,
      "learning_rate": 4.65e-07,
      "loss": 5.9534,
      "step": 156
    },
    {
      "epoch": 0.2899353647276085,
      "grad_norm": 37.76537322998047,
      "learning_rate": 4.68e-07,
      "loss": 6.0054,
      "step": 157
    },
    {
      "epoch": 0.2917820867959372,
      "grad_norm": 34.144412994384766,
      "learning_rate": 4.7099999999999997e-07,
      "loss": 5.7492,
      "step": 158
    },
    {
      "epoch": 0.29362880886426596,
      "grad_norm": 35.62531280517578,
      "learning_rate": 4.7400000000000004e-07,
      "loss": 5.7278,
      "step": 159
    },
    {
      "epoch": 0.29547553093259465,
      "grad_norm": 36.27285385131836,
      "learning_rate": 4.77e-07,
      "loss": 5.7173,
      "step": 160
    },
    {
      "epoch": 0.29732225300092335,
      "grad_norm": 36.02254104614258,
      "learning_rate": 4.800000000000001e-07,
      "loss": 5.5855,
      "step": 161
    },
    {
      "epoch": 0.29916897506925205,
      "grad_norm": 37.09486770629883,
      "learning_rate": 4.83e-07,
      "loss": 5.663,
      "step": 162
    },
    {
      "epoch": 0.3010156971375808,
      "grad_norm": 37.96799087524414,
      "learning_rate": 4.86e-07,
      "loss": 5.6351,
      "step": 163
    },
    {
      "epoch": 0.3028624192059095,
      "grad_norm": 40.50882339477539,
      "learning_rate": 4.89e-07,
      "loss": 5.6276,
      "step": 164
    },
    {
      "epoch": 0.3047091412742382,
      "grad_norm": 41.79549789428711,
      "learning_rate": 4.92e-07,
      "loss": 5.6743,
      "step": 165
    },
    {
      "epoch": 0.30655586334256696,
      "grad_norm": 43.414180755615234,
      "learning_rate": 4.95e-07,
      "loss": 5.7052,
      "step": 166
    },
    {
      "epoch": 0.30840258541089566,
      "grad_norm": 43.84006118774414,
      "learning_rate": 4.98e-07,
      "loss": 5.6869,
      "step": 167
    },
    {
      "epoch": 0.31024930747922436,
      "grad_norm": 46.30332565307617,
      "learning_rate": 5.01e-07,
      "loss": 5.7381,
      "step": 168
    },
    {
      "epoch": 0.3120960295475531,
      "grad_norm": 44.577213287353516,
      "learning_rate": 5.04e-07,
      "loss": 5.6472,
      "step": 169
    },
    {
      "epoch": 0.3139427516158818,
      "grad_norm": 44.917484283447266,
      "learning_rate": 5.07e-07,
      "loss": 5.6299,
      "step": 170
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 47.360355377197266,
      "learning_rate": 5.100000000000001e-07,
      "loss": 5.6951,
      "step": 171
    },
    {
      "epoch": 0.31763619575253926,
      "grad_norm": 41.15222930908203,
      "learning_rate": 5.13e-07,
      "loss": 5.4645,
      "step": 172
    },
    {
      "epoch": 0.31948291782086796,
      "grad_norm": 42.47678756713867,
      "learning_rate": 5.16e-07,
      "loss": 5.4574,
      "step": 173
    },
    {
      "epoch": 0.32132963988919666,
      "grad_norm": 40.974754333496094,
      "learning_rate": 5.19e-07,
      "loss": 5.3821,
      "step": 174
    },
    {
      "epoch": 0.3231763619575254,
      "grad_norm": 37.154075622558594,
      "learning_rate": 5.219999999999999e-07,
      "loss": 5.2527,
      "step": 175
    },
    {
      "epoch": 0.3250230840258541,
      "grad_norm": 34.92766571044922,
      "learning_rate": 5.250000000000001e-07,
      "loss": 5.1485,
      "step": 176
    },
    {
      "epoch": 0.3268698060941828,
      "grad_norm": 33.639408111572266,
      "learning_rate": 5.280000000000001e-07,
      "loss": 5.1358,
      "step": 177
    },
    {
      "epoch": 0.32871652816251157,
      "grad_norm": 31.7966365814209,
      "learning_rate": 5.31e-07,
      "loss": 5.0301,
      "step": 178
    },
    {
      "epoch": 0.33056325023084027,
      "grad_norm": 32.888492584228516,
      "learning_rate": 5.34e-07,
      "loss": 5.0694,
      "step": 179
    },
    {
      "epoch": 0.33240997229916897,
      "grad_norm": 28.766515731811523,
      "learning_rate": 5.37e-07,
      "loss": 4.9529,
      "step": 180
    },
    {
      "epoch": 0.33425669436749766,
      "grad_norm": 27.88300323486328,
      "learning_rate": 5.4e-07,
      "loss": 4.928,
      "step": 181
    },
    {
      "epoch": 0.3361034164358264,
      "grad_norm": 26.481048583984375,
      "learning_rate": 5.43e-07,
      "loss": 4.9091,
      "step": 182
    },
    {
      "epoch": 0.3379501385041551,
      "grad_norm": 23.27083969116211,
      "learning_rate": 5.46e-07,
      "loss": 4.86,
      "step": 183
    },
    {
      "epoch": 0.3397968605724838,
      "grad_norm": 21.233184814453125,
      "learning_rate": 5.490000000000001e-07,
      "loss": 4.7144,
      "step": 184
    },
    {
      "epoch": 0.34164358264081257,
      "grad_norm": 18.95135498046875,
      "learning_rate": 5.52e-07,
      "loss": 4.7257,
      "step": 185
    },
    {
      "epoch": 0.34349030470914127,
      "grad_norm": 16.528169631958008,
      "learning_rate": 5.55e-07,
      "loss": 4.7432,
      "step": 186
    },
    {
      "epoch": 0.34533702677746997,
      "grad_norm": 14.118398666381836,
      "learning_rate": 5.58e-07,
      "loss": 4.6586,
      "step": 187
    },
    {
      "epoch": 0.3471837488457987,
      "grad_norm": 11.961546897888184,
      "learning_rate": 5.61e-07,
      "loss": 4.5834,
      "step": 188
    },
    {
      "epoch": 0.3490304709141274,
      "grad_norm": 10.75606918334961,
      "learning_rate": 5.64e-07,
      "loss": 4.6245,
      "step": 189
    },
    {
      "epoch": 0.3508771929824561,
      "grad_norm": 8.708870887756348,
      "learning_rate": 5.67e-07,
      "loss": 4.5646,
      "step": 190
    },
    {
      "epoch": 0.3527239150507849,
      "grad_norm": 7.9037346839904785,
      "learning_rate": 5.7e-07,
      "loss": 4.4985,
      "step": 191
    },
    {
      "epoch": 0.3545706371191136,
      "grad_norm": 7.6336259841918945,
      "learning_rate": 5.73e-07,
      "loss": 4.549,
      "step": 192
    },
    {
      "epoch": 0.3564173591874423,
      "grad_norm": 8.644391059875488,
      "learning_rate": 5.76e-07,
      "loss": 4.5836,
      "step": 193
    },
    {
      "epoch": 0.358264081255771,
      "grad_norm": 8.554842948913574,
      "learning_rate": 5.790000000000001e-07,
      "loss": 4.5207,
      "step": 194
    },
    {
      "epoch": 0.3601108033240997,
      "grad_norm": 9.063313484191895,
      "learning_rate": 5.82e-07,
      "loss": 4.4461,
      "step": 195
    },
    {
      "epoch": 0.3619575253924284,
      "grad_norm": 10.217832565307617,
      "learning_rate": 5.85e-07,
      "loss": 4.5165,
      "step": 196
    },
    {
      "epoch": 0.3638042474607572,
      "grad_norm": 10.64183235168457,
      "learning_rate": 5.88e-07,
      "loss": 4.4659,
      "step": 197
    },
    {
      "epoch": 0.3656509695290859,
      "grad_norm": 11.097018241882324,
      "learning_rate": 5.909999999999999e-07,
      "loss": 4.4518,
      "step": 198
    },
    {
      "epoch": 0.3674976915974146,
      "grad_norm": 11.132061958312988,
      "learning_rate": 5.94e-07,
      "loss": 4.4149,
      "step": 199
    },
    {
      "epoch": 0.36934441366574333,
      "grad_norm": 12.801709175109863,
      "learning_rate": 5.970000000000001e-07,
      "loss": 4.4589,
      "step": 200
    },
    {
      "epoch": 0.37119113573407203,
      "grad_norm": 7.300925254821777,
      "learning_rate": 6.000000000000001e-07,
      "loss": 4.3065,
      "step": 201
    },
    {
      "epoch": 0.37303785780240073,
      "grad_norm": 6.761147975921631,
      "learning_rate": 6.03e-07,
      "loss": 4.2383,
      "step": 202
    },
    {
      "epoch": 0.37488457987072943,
      "grad_norm": 7.358983516693115,
      "learning_rate": 6.06e-07,
      "loss": 4.1914,
      "step": 203
    },
    {
      "epoch": 0.3767313019390582,
      "grad_norm": 6.69858455657959,
      "learning_rate": 6.09e-07,
      "loss": 4.2535,
      "step": 204
    },
    {
      "epoch": 0.3785780240073869,
      "grad_norm": 7.087483882904053,
      "learning_rate": 6.12e-07,
      "loss": 4.1431,
      "step": 205
    },
    {
      "epoch": 0.3804247460757156,
      "grad_norm": 7.197114944458008,
      "learning_rate": 6.15e-07,
      "loss": 4.1343,
      "step": 206
    },
    {
      "epoch": 0.38227146814404434,
      "grad_norm": 7.602322101593018,
      "learning_rate": 6.180000000000001e-07,
      "loss": 4.0637,
      "step": 207
    },
    {
      "epoch": 0.38411819021237303,
      "grad_norm": 7.007232666015625,
      "learning_rate": 6.21e-07,
      "loss": 4.0418,
      "step": 208
    },
    {
      "epoch": 0.38596491228070173,
      "grad_norm": 6.680031776428223,
      "learning_rate": 6.24e-07,
      "loss": 4.0599,
      "step": 209
    },
    {
      "epoch": 0.3878116343490305,
      "grad_norm": 6.525915145874023,
      "learning_rate": 6.27e-07,
      "loss": 4.004,
      "step": 210
    },
    {
      "epoch": 0.3896583564173592,
      "grad_norm": 6.130704879760742,
      "learning_rate": 6.3e-07,
      "loss": 4.0534,
      "step": 211
    },
    {
      "epoch": 0.3915050784856879,
      "grad_norm": 5.882370471954346,
      "learning_rate": 6.33e-07,
      "loss": 4.0821,
      "step": 212
    },
    {
      "epoch": 0.39335180055401664,
      "grad_norm": 5.746700763702393,
      "learning_rate": 6.36e-07,
      "loss": 4.1033,
      "step": 213
    },
    {
      "epoch": 0.39519852262234534,
      "grad_norm": 5.587550640106201,
      "learning_rate": 6.39e-07,
      "loss": 4.0979,
      "step": 214
    },
    {
      "epoch": 0.39704524469067404,
      "grad_norm": 5.65580415725708,
      "learning_rate": 6.42e-07,
      "loss": 4.1246,
      "step": 215
    },
    {
      "epoch": 0.3988919667590028,
      "grad_norm": 5.67427921295166,
      "learning_rate": 6.45e-07,
      "loss": 4.1217,
      "step": 216
    },
    {
      "epoch": 0.4007386888273315,
      "grad_norm": 6.284714698791504,
      "learning_rate": 6.480000000000001e-07,
      "loss": 4.2012,
      "step": 217
    },
    {
      "epoch": 0.4025854108956602,
      "grad_norm": 5.949926376342773,
      "learning_rate": 6.51e-07,
      "loss": 4.1859,
      "step": 218
    },
    {
      "epoch": 0.40443213296398894,
      "grad_norm": 6.066436767578125,
      "learning_rate": 6.54e-07,
      "loss": 4.1414,
      "step": 219
    },
    {
      "epoch": 0.40627885503231764,
      "grad_norm": 6.133418083190918,
      "learning_rate": 6.57e-07,
      "loss": 4.1737,
      "step": 220
    },
    {
      "epoch": 0.40812557710064634,
      "grad_norm": 5.733587265014648,
      "learning_rate": 6.599999999999999e-07,
      "loss": 4.1053,
      "step": 221
    },
    {
      "epoch": 0.4099722991689751,
      "grad_norm": 5.346182823181152,
      "learning_rate": 6.63e-07,
      "loss": 4.0954,
      "step": 222
    },
    {
      "epoch": 0.4118190212373038,
      "grad_norm": 5.801249980926514,
      "learning_rate": 6.660000000000001e-07,
      "loss": 4.1524,
      "step": 223
    },
    {
      "epoch": 0.4136657433056325,
      "grad_norm": 5.427605628967285,
      "learning_rate": 6.690000000000001e-07,
      "loss": 4.1132,
      "step": 224
    },
    {
      "epoch": 0.4155124653739612,
      "grad_norm": 5.082065582275391,
      "learning_rate": 6.72e-07,
      "loss": 4.0983,
      "step": 225
    },
    {
      "epoch": 0.41735918744228995,
      "grad_norm": 5.642475605010986,
      "learning_rate": 6.75e-07,
      "loss": 4.089,
      "step": 226
    },
    {
      "epoch": 0.41920590951061865,
      "grad_norm": 5.176789283752441,
      "learning_rate": 6.78e-07,
      "loss": 4.0512,
      "step": 227
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 5.5746541023254395,
      "learning_rate": 6.81e-07,
      "loss": 4.0565,
      "step": 228
    },
    {
      "epoch": 0.4228993536472761,
      "grad_norm": 5.4985032081604,
      "learning_rate": 6.84e-07,
      "loss": 4.0774,
      "step": 229
    },
    {
      "epoch": 0.4247460757156048,
      "grad_norm": 5.544588565826416,
      "learning_rate": 6.87e-07,
      "loss": 4.0619,
      "step": 230
    },
    {
      "epoch": 0.4265927977839335,
      "grad_norm": 5.085567474365234,
      "learning_rate": 6.9e-07,
      "loss": 3.9928,
      "step": 231
    },
    {
      "epoch": 0.42843951985226225,
      "grad_norm": 5.949645519256592,
      "learning_rate": 6.93e-07,
      "loss": 4.0557,
      "step": 232
    },
    {
      "epoch": 0.43028624192059095,
      "grad_norm": 5.037247657775879,
      "learning_rate": 6.96e-07,
      "loss": 4.005,
      "step": 233
    },
    {
      "epoch": 0.43213296398891965,
      "grad_norm": 5.97141695022583,
      "learning_rate": 6.990000000000001e-07,
      "loss": 4.0016,
      "step": 234
    },
    {
      "epoch": 0.4339796860572484,
      "grad_norm": 4.68813943862915,
      "learning_rate": 7.02e-07,
      "loss": 3.9743,
      "step": 235
    },
    {
      "epoch": 0.4358264081255771,
      "grad_norm": 14.047794342041016,
      "learning_rate": 7.05e-07,
      "loss": 3.9354,
      "step": 236
    },
    {
      "epoch": 0.4376731301939058,
      "grad_norm": 5.274556636810303,
      "learning_rate": 7.08e-07,
      "loss": 4.0027,
      "step": 237
    },
    {
      "epoch": 0.43951985226223456,
      "grad_norm": 4.5572733879089355,
      "learning_rate": 7.11e-07,
      "loss": 3.9243,
      "step": 238
    },
    {
      "epoch": 0.44136657433056325,
      "grad_norm": 4.318416118621826,
      "learning_rate": 7.140000000000001e-07,
      "loss": 3.889,
      "step": 239
    },
    {
      "epoch": 0.44321329639889195,
      "grad_norm": 4.351585388183594,
      "learning_rate": 7.170000000000001e-07,
      "loss": 3.9835,
      "step": 240
    },
    {
      "epoch": 0.4450600184672207,
      "grad_norm": 4.199488639831543,
      "learning_rate": 7.2e-07,
      "loss": 3.9077,
      "step": 241
    },
    {
      "epoch": 0.4469067405355494,
      "grad_norm": 4.0089898109436035,
      "learning_rate": 7.23e-07,
      "loss": 3.8684,
      "step": 242
    },
    {
      "epoch": 0.4487534626038781,
      "grad_norm": 4.100970268249512,
      "learning_rate": 7.26e-07,
      "loss": 3.8637,
      "step": 243
    },
    {
      "epoch": 0.45060018467220686,
      "grad_norm": 4.072782516479492,
      "learning_rate": 7.29e-07,
      "loss": 3.8156,
      "step": 244
    },
    {
      "epoch": 0.45244690674053556,
      "grad_norm": 4.012927532196045,
      "learning_rate": 7.32e-07,
      "loss": 3.8037,
      "step": 245
    },
    {
      "epoch": 0.45429362880886426,
      "grad_norm": 6.060505390167236,
      "learning_rate": 7.350000000000001e-07,
      "loss": 3.9028,
      "step": 246
    },
    {
      "epoch": 0.45614035087719296,
      "grad_norm": 4.179002285003662,
      "learning_rate": 7.380000000000001e-07,
      "loss": 3.8943,
      "step": 247
    },
    {
      "epoch": 0.4579870729455217,
      "grad_norm": 5.213497638702393,
      "learning_rate": 7.41e-07,
      "loss": 3.8375,
      "step": 248
    },
    {
      "epoch": 0.4598337950138504,
      "grad_norm": 4.26724910736084,
      "learning_rate": 7.44e-07,
      "loss": 3.7996,
      "step": 249
    },
    {
      "epoch": 0.4616805170821791,
      "grad_norm": 12.426194190979004,
      "learning_rate": 7.47e-07,
      "loss": 3.8624,
      "step": 250
    },
    {
      "epoch": 0.46352723915050786,
      "grad_norm": 4.382394790649414,
      "learning_rate": 7.5e-07,
      "loss": 3.7192,
      "step": 251
    },
    {
      "epoch": 0.46537396121883656,
      "grad_norm": 4.744300365447998,
      "learning_rate": 7.53e-07,
      "loss": 3.6259,
      "step": 252
    },
    {
      "epoch": 0.46722068328716526,
      "grad_norm": 4.654346942901611,
      "learning_rate": 7.56e-07,
      "loss": 3.5965,
      "step": 253
    },
    {
      "epoch": 0.469067405355494,
      "grad_norm": 3.6834309101104736,
      "learning_rate": 7.59e-07,
      "loss": 3.5887,
      "step": 254
    },
    {
      "epoch": 0.4709141274238227,
      "grad_norm": 4.092491149902344,
      "learning_rate": 7.62e-07,
      "loss": 3.5426,
      "step": 255
    },
    {
      "epoch": 0.4727608494921514,
      "grad_norm": 4.001455783843994,
      "learning_rate": 7.65e-07,
      "loss": 3.5544,
      "step": 256
    },
    {
      "epoch": 0.47460757156048017,
      "grad_norm": 5.414680480957031,
      "learning_rate": 7.680000000000001e-07,
      "loss": 3.4742,
      "step": 257
    },
    {
      "epoch": 0.47645429362880887,
      "grad_norm": 3.3546714782714844,
      "learning_rate": 7.71e-07,
      "loss": 3.4772,
      "step": 258
    },
    {
      "epoch": 0.47830101569713757,
      "grad_norm": 3.5425257682800293,
      "learning_rate": 7.74e-07,
      "loss": 3.4811,
      "step": 259
    },
    {
      "epoch": 0.4801477377654663,
      "grad_norm": 3.1755106449127197,
      "learning_rate": 7.77e-07,
      "loss": 3.4188,
      "step": 260
    },
    {
      "epoch": 0.481994459833795,
      "grad_norm": 4.707540988922119,
      "learning_rate": 7.799999999999999e-07,
      "loss": 3.3899,
      "step": 261
    },
    {
      "epoch": 0.4838411819021237,
      "grad_norm": 3.285004138946533,
      "learning_rate": 7.830000000000001e-07,
      "loss": 3.4876,
      "step": 262
    },
    {
      "epoch": 0.4856879039704525,
      "grad_norm": 4.9399638175964355,
      "learning_rate": 7.860000000000001e-07,
      "loss": 3.4514,
      "step": 263
    },
    {
      "epoch": 0.48753462603878117,
      "grad_norm": 3.9122636318206787,
      "learning_rate": 7.89e-07,
      "loss": 3.476,
      "step": 264
    },
    {
      "epoch": 0.48938134810710987,
      "grad_norm": 5.649412155151367,
      "learning_rate": 7.92e-07,
      "loss": 3.5428,
      "step": 265
    },
    {
      "epoch": 0.49122807017543857,
      "grad_norm": 14.08123779296875,
      "learning_rate": 7.95e-07,
      "loss": 3.5073,
      "step": 266
    },
    {
      "epoch": 0.4930747922437673,
      "grad_norm": 4.978379726409912,
      "learning_rate": 7.98e-07,
      "loss": 3.5416,
      "step": 267
    },
    {
      "epoch": 0.494921514312096,
      "grad_norm": 4.730838298797607,
      "learning_rate": 8.01e-07,
      "loss": 3.5272,
      "step": 268
    },
    {
      "epoch": 0.4967682363804247,
      "grad_norm": 4.087198257446289,
      "learning_rate": 8.04e-07,
      "loss": 3.4973,
      "step": 269
    },
    {
      "epoch": 0.4986149584487535,
      "grad_norm": 4.161842346191406,
      "learning_rate": 8.070000000000001e-07,
      "loss": 3.5052,
      "step": 270
    },
    {
      "epoch": 0.5004616805170822,
      "grad_norm": 5.521448135375977,
      "learning_rate": 8.1e-07,
      "loss": 3.5112,
      "step": 271
    },
    {
      "epoch": 0.5023084025854109,
      "grad_norm": 3.795762062072754,
      "learning_rate": 8.13e-07,
      "loss": 3.4638,
      "step": 272
    },
    {
      "epoch": 0.5041551246537396,
      "grad_norm": 4.849123001098633,
      "learning_rate": 8.16e-07,
      "loss": 3.4319,
      "step": 273
    },
    {
      "epoch": 0.5060018467220684,
      "grad_norm": 4.427425861358643,
      "learning_rate": 8.19e-07,
      "loss": 3.4609,
      "step": 274
    },
    {
      "epoch": 0.5078485687903971,
      "grad_norm": 4.559118747711182,
      "learning_rate": 8.22e-07,
      "loss": 3.4332,
      "step": 275
    },
    {
      "epoch": 0.5096952908587258,
      "grad_norm": 5.236905574798584,
      "learning_rate": 8.25e-07,
      "loss": 3.4064,
      "step": 276
    },
    {
      "epoch": 0.5115420129270545,
      "grad_norm": 6.23512077331543,
      "learning_rate": 8.280000000000001e-07,
      "loss": 3.4709,
      "step": 277
    },
    {
      "epoch": 0.5133887349953832,
      "grad_norm": 5.505606174468994,
      "learning_rate": 8.31e-07,
      "loss": 3.4046,
      "step": 278
    },
    {
      "epoch": 0.5152354570637119,
      "grad_norm": 15.238096237182617,
      "learning_rate": 8.34e-07,
      "loss": 3.3852,
      "step": 279
    },
    {
      "epoch": 0.5170821791320406,
      "grad_norm": 7.080745697021484,
      "learning_rate": 8.370000000000001e-07,
      "loss": 3.3836,
      "step": 280
    },
    {
      "epoch": 0.5189289012003694,
      "grad_norm": 4.308071136474609,
      "learning_rate": 8.4e-07,
      "loss": 3.3356,
      "step": 281
    },
    {
      "epoch": 0.5207756232686981,
      "grad_norm": 5.031737327575684,
      "learning_rate": 8.43e-07,
      "loss": 3.3544,
      "step": 282
    },
    {
      "epoch": 0.5226223453370268,
      "grad_norm": 4.0981268882751465,
      "learning_rate": 8.46e-07,
      "loss": 3.3315,
      "step": 283
    },
    {
      "epoch": 0.5244690674053555,
      "grad_norm": 8.215164184570312,
      "learning_rate": 8.489999999999999e-07,
      "loss": 3.3009,
      "step": 284
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 4.3822245597839355,
      "learning_rate": 8.520000000000001e-07,
      "loss": 3.3268,
      "step": 285
    },
    {
      "epoch": 0.5281625115420129,
      "grad_norm": 3.282374382019043,
      "learning_rate": 8.550000000000001e-07,
      "loss": 3.2991,
      "step": 286
    },
    {
      "epoch": 0.5300092336103417,
      "grad_norm": 5.968320846557617,
      "learning_rate": 8.580000000000001e-07,
      "loss": 3.2981,
      "step": 287
    },
    {
      "epoch": 0.5318559556786704,
      "grad_norm": 10.67861270904541,
      "learning_rate": 8.61e-07,
      "loss": 3.2595,
      "step": 288
    },
    {
      "epoch": 0.5337026777469991,
      "grad_norm": 3.8899168968200684,
      "learning_rate": 8.64e-07,
      "loss": 3.2271,
      "step": 289
    },
    {
      "epoch": 0.5355493998153278,
      "grad_norm": 4.796634674072266,
      "learning_rate": 8.67e-07,
      "loss": 3.3094,
      "step": 290
    },
    {
      "epoch": 0.5373961218836565,
      "grad_norm": 4.089239120483398,
      "learning_rate": 8.7e-07,
      "loss": 3.2095,
      "step": 291
    },
    {
      "epoch": 0.5392428439519852,
      "grad_norm": 5.146942615509033,
      "learning_rate": 8.73e-07,
      "loss": 3.2382,
      "step": 292
    },
    {
      "epoch": 0.541089566020314,
      "grad_norm": 5.682231903076172,
      "learning_rate": 8.760000000000001e-07,
      "loss": 3.2271,
      "step": 293
    },
    {
      "epoch": 0.5429362880886427,
      "grad_norm": 4.716771602630615,
      "learning_rate": 8.79e-07,
      "loss": 3.2294,
      "step": 294
    },
    {
      "epoch": 0.5447830101569714,
      "grad_norm": 4.983368873596191,
      "learning_rate": 8.82e-07,
      "loss": 3.2358,
      "step": 295
    },
    {
      "epoch": 0.5466297322253001,
      "grad_norm": 7.215044975280762,
      "learning_rate": 8.85e-07,
      "loss": 3.2556,
      "step": 296
    },
    {
      "epoch": 0.5484764542936288,
      "grad_norm": 4.684432506561279,
      "learning_rate": 8.88e-07,
      "loss": 3.2566,
      "step": 297
    },
    {
      "epoch": 0.5503231763619575,
      "grad_norm": 4.458353519439697,
      "learning_rate": 8.91e-07,
      "loss": 3.2236,
      "step": 298
    },
    {
      "epoch": 0.5521698984302862,
      "grad_norm": 8.782550811767578,
      "learning_rate": 8.94e-07,
      "loss": 3.2065,
      "step": 299
    },
    {
      "epoch": 0.554016620498615,
      "grad_norm": 12.405173301696777,
      "learning_rate": 8.97e-07,
      "loss": 3.1937,
      "step": 300
    },
    {
      "epoch": 0.5558633425669437,
      "grad_norm": 12.274544715881348,
      "learning_rate": 9e-07,
      "loss": 3.3293,
      "step": 301
    },
    {
      "epoch": 0.5577100646352724,
      "grad_norm": 10.548261642456055,
      "learning_rate": 9.03e-07,
      "loss": 3.2509,
      "step": 302
    },
    {
      "epoch": 0.5595567867036011,
      "grad_norm": 7.260283946990967,
      "learning_rate": 9.060000000000001e-07,
      "loss": 3.1922,
      "step": 303
    },
    {
      "epoch": 0.5614035087719298,
      "grad_norm": 6.222684860229492,
      "learning_rate": 9.09e-07,
      "loss": 3.2229,
      "step": 304
    },
    {
      "epoch": 0.5632502308402585,
      "grad_norm": 3.192690849304199,
      "learning_rate": 9.12e-07,
      "loss": 3.2094,
      "step": 305
    },
    {
      "epoch": 0.5650969529085873,
      "grad_norm": 4.268246173858643,
      "learning_rate": 9.15e-07,
      "loss": 3.1659,
      "step": 306
    },
    {
      "epoch": 0.566943674976916,
      "grad_norm": 5.622379779815674,
      "learning_rate": 9.179999999999999e-07,
      "loss": 3.1662,
      "step": 307
    },
    {
      "epoch": 0.5687903970452447,
      "grad_norm": 5.32727575302124,
      "learning_rate": 9.210000000000001e-07,
      "loss": 3.1983,
      "step": 308
    },
    {
      "epoch": 0.5706371191135734,
      "grad_norm": 4.53847599029541,
      "learning_rate": 9.240000000000001e-07,
      "loss": 3.2035,
      "step": 309
    },
    {
      "epoch": 0.5724838411819021,
      "grad_norm": 5.334409713745117,
      "learning_rate": 9.270000000000001e-07,
      "loss": 3.1746,
      "step": 310
    },
    {
      "epoch": 0.5743305632502308,
      "grad_norm": 4.618269920349121,
      "learning_rate": 9.3e-07,
      "loss": 3.1904,
      "step": 311
    },
    {
      "epoch": 0.5761772853185596,
      "grad_norm": 2.947032928466797,
      "learning_rate": 9.33e-07,
      "loss": 3.1811,
      "step": 312
    },
    {
      "epoch": 0.5780240073868883,
      "grad_norm": 9.83170223236084,
      "learning_rate": 9.36e-07,
      "loss": 3.2034,
      "step": 313
    },
    {
      "epoch": 0.579870729455217,
      "grad_norm": 4.297440052032471,
      "learning_rate": 9.39e-07,
      "loss": 3.1925,
      "step": 314
    },
    {
      "epoch": 0.5817174515235457,
      "grad_norm": 3.8568122386932373,
      "learning_rate": 9.419999999999999e-07,
      "loss": 3.1584,
      "step": 315
    },
    {
      "epoch": 0.5835641735918744,
      "grad_norm": 3.8877062797546387,
      "learning_rate": 9.450000000000001e-07,
      "loss": 3.1015,
      "step": 316
    },
    {
      "epoch": 0.5854108956602031,
      "grad_norm": 3.579789400100708,
      "learning_rate": 9.480000000000001e-07,
      "loss": 3.1743,
      "step": 317
    },
    {
      "epoch": 0.5872576177285319,
      "grad_norm": 3.8187687397003174,
      "learning_rate": 9.51e-07,
      "loss": 3.2222,
      "step": 318
    },
    {
      "epoch": 0.5891043397968606,
      "grad_norm": 5.050365447998047,
      "learning_rate": 9.54e-07,
      "loss": 3.1866,
      "step": 319
    },
    {
      "epoch": 0.5909510618651893,
      "grad_norm": 4.368477821350098,
      "learning_rate": 9.57e-07,
      "loss": 3.1362,
      "step": 320
    },
    {
      "epoch": 0.592797783933518,
      "grad_norm": 7.682777404785156,
      "learning_rate": 9.600000000000001e-07,
      "loss": 3.1577,
      "step": 321
    },
    {
      "epoch": 0.5946445060018467,
      "grad_norm": 4.716755390167236,
      "learning_rate": 9.63e-07,
      "loss": 3.1668,
      "step": 322
    },
    {
      "epoch": 0.5964912280701754,
      "grad_norm": 7.155860424041748,
      "learning_rate": 9.66e-07,
      "loss": 3.1626,
      "step": 323
    },
    {
      "epoch": 0.5983379501385041,
      "grad_norm": 3.2187845706939697,
      "learning_rate": 9.690000000000002e-07,
      "loss": 3.0961,
      "step": 324
    },
    {
      "epoch": 0.6001846722068329,
      "grad_norm": 7.184601783752441,
      "learning_rate": 9.72e-07,
      "loss": 3.1912,
      "step": 325
    },
    {
      "epoch": 0.6020313942751616,
      "grad_norm": 6.903328895568848,
      "learning_rate": 9.75e-07,
      "loss": 3.1594,
      "step": 326
    },
    {
      "epoch": 0.6038781163434903,
      "grad_norm": 6.267014980316162,
      "learning_rate": 9.78e-07,
      "loss": 3.1753,
      "step": 327
    },
    {
      "epoch": 0.605724838411819,
      "grad_norm": 6.188835620880127,
      "learning_rate": 9.81e-07,
      "loss": 3.1726,
      "step": 328
    },
    {
      "epoch": 0.6075715604801477,
      "grad_norm": 4.90877103805542,
      "learning_rate": 9.84e-07,
      "loss": 3.1101,
      "step": 329
    },
    {
      "epoch": 0.6094182825484764,
      "grad_norm": 5.561330318450928,
      "learning_rate": 9.87e-07,
      "loss": 3.1213,
      "step": 330
    },
    {
      "epoch": 0.6112650046168052,
      "grad_norm": 4.409107208251953,
      "learning_rate": 9.9e-07,
      "loss": 3.1224,
      "step": 331
    },
    {
      "epoch": 0.6131117266851339,
      "grad_norm": 2.835259437561035,
      "learning_rate": 9.929999999999999e-07,
      "loss": 3.1419,
      "step": 332
    },
    {
      "epoch": 0.6149584487534626,
      "grad_norm": 5.65529203414917,
      "learning_rate": 9.96e-07,
      "loss": 3.1314,
      "step": 333
    },
    {
      "epoch": 0.6168051708217913,
      "grad_norm": 5.926425457000732,
      "learning_rate": 9.99e-07,
      "loss": 3.1238,
      "step": 334
    },
    {
      "epoch": 0.61865189289012,
      "grad_norm": 4.398102760314941,
      "learning_rate": 1.002e-06,
      "loss": 3.1094,
      "step": 335
    },
    {
      "epoch": 0.6204986149584487,
      "grad_norm": 5.749934673309326,
      "learning_rate": 1.0050000000000001e-06,
      "loss": 3.158,
      "step": 336
    },
    {
      "epoch": 0.6223453370267775,
      "grad_norm": 4.760705947875977,
      "learning_rate": 1.008e-06,
      "loss": 3.1159,
      "step": 337
    },
    {
      "epoch": 0.6241920590951062,
      "grad_norm": 7.050318241119385,
      "learning_rate": 1.0110000000000001e-06,
      "loss": 3.1001,
      "step": 338
    },
    {
      "epoch": 0.6260387811634349,
      "grad_norm": 7.832237720489502,
      "learning_rate": 1.014e-06,
      "loss": 3.1326,
      "step": 339
    },
    {
      "epoch": 0.6278855032317636,
      "grad_norm": 4.517617225646973,
      "learning_rate": 1.017e-06,
      "loss": 3.1487,
      "step": 340
    },
    {
      "epoch": 0.6297322253000923,
      "grad_norm": 6.685205936431885,
      "learning_rate": 1.0200000000000002e-06,
      "loss": 3.0753,
      "step": 341
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 6.85542106628418,
      "learning_rate": 1.023e-06,
      "loss": 3.1213,
      "step": 342
    },
    {
      "epoch": 0.6334256694367497,
      "grad_norm": 6.824644565582275,
      "learning_rate": 1.026e-06,
      "loss": 3.0636,
      "step": 343
    },
    {
      "epoch": 0.6352723915050785,
      "grad_norm": 8.230368614196777,
      "learning_rate": 1.029e-06,
      "loss": 3.0875,
      "step": 344
    },
    {
      "epoch": 0.6371191135734072,
      "grad_norm": 3.8798701763153076,
      "learning_rate": 1.032e-06,
      "loss": 3.1108,
      "step": 345
    },
    {
      "epoch": 0.6389658356417359,
      "grad_norm": 4.092694282531738,
      "learning_rate": 1.035e-06,
      "loss": 3.1304,
      "step": 346
    },
    {
      "epoch": 0.6408125577100646,
      "grad_norm": 4.441507339477539,
      "learning_rate": 1.038e-06,
      "loss": 3.1214,
      "step": 347
    },
    {
      "epoch": 0.6426592797783933,
      "grad_norm": 8.2803955078125,
      "learning_rate": 1.041e-06,
      "loss": 3.0863,
      "step": 348
    },
    {
      "epoch": 0.644506001846722,
      "grad_norm": 5.707834720611572,
      "learning_rate": 1.0439999999999999e-06,
      "loss": 3.1261,
      "step": 349
    },
    {
      "epoch": 0.6463527239150508,
      "grad_norm": 6.7514238357543945,
      "learning_rate": 1.047e-06,
      "loss": 3.0936,
      "step": 350
    },
    {
      "epoch": 0.6481994459833795,
      "grad_norm": 9.191317558288574,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 3.1867,
      "step": 351
    },
    {
      "epoch": 0.6500461680517082,
      "grad_norm": 8.901002883911133,
      "learning_rate": 1.053e-06,
      "loss": 3.1831,
      "step": 352
    },
    {
      "epoch": 0.6518928901200369,
      "grad_norm": 5.3682169914245605,
      "learning_rate": 1.0560000000000001e-06,
      "loss": 3.115,
      "step": 353
    },
    {
      "epoch": 0.6537396121883656,
      "grad_norm": 3.2384531497955322,
      "learning_rate": 1.059e-06,
      "loss": 3.1332,
      "step": 354
    },
    {
      "epoch": 0.6555863342566943,
      "grad_norm": 2.113199472427368,
      "learning_rate": 1.062e-06,
      "loss": 3.1003,
      "step": 355
    },
    {
      "epoch": 0.6574330563250231,
      "grad_norm": 3.5084481239318848,
      "learning_rate": 1.065e-06,
      "loss": 3.0896,
      "step": 356
    },
    {
      "epoch": 0.6592797783933518,
      "grad_norm": 4.078393936157227,
      "learning_rate": 1.068e-06,
      "loss": 3.0751,
      "step": 357
    },
    {
      "epoch": 0.6611265004616805,
      "grad_norm": 6.067633628845215,
      "learning_rate": 1.0710000000000002e-06,
      "loss": 3.0848,
      "step": 358
    },
    {
      "epoch": 0.6629732225300092,
      "grad_norm": 5.45023775100708,
      "learning_rate": 1.074e-06,
      "loss": 3.0941,
      "step": 359
    },
    {
      "epoch": 0.6648199445983379,
      "grad_norm": 4.068233966827393,
      "learning_rate": 1.077e-06,
      "loss": 3.0869,
      "step": 360
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 3.762965202331543,
      "learning_rate": 1.08e-06,
      "loss": 3.1117,
      "step": 361
    },
    {
      "epoch": 0.6685133887349953,
      "grad_norm": 2.311380624771118,
      "learning_rate": 1.083e-06,
      "loss": 3.0524,
      "step": 362
    },
    {
      "epoch": 0.6703601108033241,
      "grad_norm": 4.134860038757324,
      "learning_rate": 1.086e-06,
      "loss": 3.0798,
      "step": 363
    },
    {
      "epoch": 0.6722068328716528,
      "grad_norm": 3.7544875144958496,
      "learning_rate": 1.089e-06,
      "loss": 3.1159,
      "step": 364
    },
    {
      "epoch": 0.6740535549399815,
      "grad_norm": 3.5989551544189453,
      "learning_rate": 1.092e-06,
      "loss": 3.0632,
      "step": 365
    },
    {
      "epoch": 0.6759002770083102,
      "grad_norm": 4.7507853507995605,
      "learning_rate": 1.0949999999999999e-06,
      "loss": 3.0727,
      "step": 366
    },
    {
      "epoch": 0.6777469990766389,
      "grad_norm": 7.671990871429443,
      "learning_rate": 1.0980000000000001e-06,
      "loss": 3.0851,
      "step": 367
    },
    {
      "epoch": 0.6795937211449676,
      "grad_norm": 6.9083380699157715,
      "learning_rate": 1.1010000000000001e-06,
      "loss": 3.0607,
      "step": 368
    },
    {
      "epoch": 0.6814404432132964,
      "grad_norm": 4.966256618499756,
      "learning_rate": 1.104e-06,
      "loss": 3.0911,
      "step": 369
    },
    {
      "epoch": 0.6832871652816251,
      "grad_norm": 12.021642684936523,
      "learning_rate": 1.1070000000000002e-06,
      "loss": 3.0722,
      "step": 370
    },
    {
      "epoch": 0.6851338873499538,
      "grad_norm": 3.737931728363037,
      "learning_rate": 1.11e-06,
      "loss": 3.0455,
      "step": 371
    },
    {
      "epoch": 0.6869806094182825,
      "grad_norm": 4.485469341278076,
      "learning_rate": 1.113e-06,
      "loss": 3.0536,
      "step": 372
    },
    {
      "epoch": 0.6888273314866112,
      "grad_norm": 6.518353462219238,
      "learning_rate": 1.116e-06,
      "loss": 3.049,
      "step": 373
    },
    {
      "epoch": 0.6906740535549399,
      "grad_norm": 4.276355743408203,
      "learning_rate": 1.119e-06,
      "loss": 3.0453,
      "step": 374
    },
    {
      "epoch": 0.6925207756232687,
      "grad_norm": 7.017218112945557,
      "learning_rate": 1.122e-06,
      "loss": 3.0717,
      "step": 375
    },
    {
      "epoch": 0.6943674976915974,
      "grad_norm": 4.840349197387695,
      "learning_rate": 1.125e-06,
      "loss": 3.0627,
      "step": 376
    },
    {
      "epoch": 0.6962142197599261,
      "grad_norm": 6.178562164306641,
      "learning_rate": 1.128e-06,
      "loss": 3.053,
      "step": 377
    },
    {
      "epoch": 0.6980609418282548,
      "grad_norm": 7.595609664916992,
      "learning_rate": 1.131e-06,
      "loss": 3.1017,
      "step": 378
    },
    {
      "epoch": 0.6999076638965835,
      "grad_norm": 4.348419666290283,
      "learning_rate": 1.134e-06,
      "loss": 3.0428,
      "step": 379
    },
    {
      "epoch": 0.7017543859649122,
      "grad_norm": 4.139669418334961,
      "learning_rate": 1.137e-06,
      "loss": 3.0727,
      "step": 380
    },
    {
      "epoch": 0.703601108033241,
      "grad_norm": 5.153532981872559,
      "learning_rate": 1.14e-06,
      "loss": 3.0405,
      "step": 381
    },
    {
      "epoch": 0.7054478301015698,
      "grad_norm": 4.004812240600586,
      "learning_rate": 1.1430000000000001e-06,
      "loss": 3.0353,
      "step": 382
    },
    {
      "epoch": 0.7072945521698984,
      "grad_norm": 5.635801792144775,
      "learning_rate": 1.146e-06,
      "loss": 3.0672,
      "step": 383
    },
    {
      "epoch": 0.7091412742382271,
      "grad_norm": 3.6818764209747314,
      "learning_rate": 1.1490000000000001e-06,
      "loss": 3.0607,
      "step": 384
    },
    {
      "epoch": 0.7109879963065558,
      "grad_norm": 8.386887550354004,
      "learning_rate": 1.152e-06,
      "loss": 3.0002,
      "step": 385
    },
    {
      "epoch": 0.7128347183748845,
      "grad_norm": 7.178035259246826,
      "learning_rate": 1.155e-06,
      "loss": 3.0347,
      "step": 386
    },
    {
      "epoch": 0.7146814404432132,
      "grad_norm": 3.6302337646484375,
      "learning_rate": 1.1580000000000002e-06,
      "loss": 3.0274,
      "step": 387
    },
    {
      "epoch": 0.716528162511542,
      "grad_norm": 3.4200942516326904,
      "learning_rate": 1.161e-06,
      "loss": 3.0377,
      "step": 388
    },
    {
      "epoch": 0.7183748845798708,
      "grad_norm": 3.180461883544922,
      "learning_rate": 1.164e-06,
      "loss": 3.0216,
      "step": 389
    },
    {
      "epoch": 0.7202216066481995,
      "grad_norm": 3.9970366954803467,
      "learning_rate": 1.167e-06,
      "loss": 3.0542,
      "step": 390
    },
    {
      "epoch": 0.7220683287165282,
      "grad_norm": 8.964760780334473,
      "learning_rate": 1.17e-06,
      "loss": 3.0152,
      "step": 391
    },
    {
      "epoch": 0.7239150507848569,
      "grad_norm": 4.677817344665527,
      "learning_rate": 1.173e-06,
      "loss": 3.0372,
      "step": 392
    },
    {
      "epoch": 0.7257617728531855,
      "grad_norm": 5.287402153015137,
      "learning_rate": 1.176e-06,
      "loss": 2.9933,
      "step": 393
    },
    {
      "epoch": 0.7276084949215144,
      "grad_norm": 4.688521385192871,
      "learning_rate": 1.179e-06,
      "loss": 3.0533,
      "step": 394
    },
    {
      "epoch": 0.7294552169898431,
      "grad_norm": 5.032068729400635,
      "learning_rate": 1.1819999999999999e-06,
      "loss": 3.0147,
      "step": 395
    },
    {
      "epoch": 0.7313019390581718,
      "grad_norm": 6.2361626625061035,
      "learning_rate": 1.185e-06,
      "loss": 3.0879,
      "step": 396
    },
    {
      "epoch": 0.7331486611265005,
      "grad_norm": 7.5859270095825195,
      "learning_rate": 1.188e-06,
      "loss": 3.0227,
      "step": 397
    },
    {
      "epoch": 0.7349953831948292,
      "grad_norm": 6.370418071746826,
      "learning_rate": 1.191e-06,
      "loss": 2.9768,
      "step": 398
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 4.1537675857543945,
      "learning_rate": 1.1940000000000001e-06,
      "loss": 3.0001,
      "step": 399
    },
    {
      "epoch": 0.7386888273314867,
      "grad_norm": 4.340860843658447,
      "learning_rate": 1.197e-06,
      "loss": 3.0386,
      "step": 400
    },
    {
      "epoch": 0.7405355493998154,
      "grad_norm": 7.044209003448486,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 3.187,
      "step": 401
    },
    {
      "epoch": 0.7423822714681441,
      "grad_norm": 9.477621078491211,
      "learning_rate": 1.203e-06,
      "loss": 3.0968,
      "step": 402
    },
    {
      "epoch": 0.7442289935364728,
      "grad_norm": 4.682133197784424,
      "learning_rate": 1.206e-06,
      "loss": 3.0578,
      "step": 403
    },
    {
      "epoch": 0.7460757156048015,
      "grad_norm": 3.7250590324401855,
      "learning_rate": 1.2090000000000002e-06,
      "loss": 3.0519,
      "step": 404
    },
    {
      "epoch": 0.7479224376731302,
      "grad_norm": 2.828423500061035,
      "learning_rate": 1.212e-06,
      "loss": 3.0602,
      "step": 405
    },
    {
      "epoch": 0.7497691597414589,
      "grad_norm": 4.87825870513916,
      "learning_rate": 1.215e-06,
      "loss": 3.0154,
      "step": 406
    },
    {
      "epoch": 0.7516158818097877,
      "grad_norm": 5.9660234451293945,
      "learning_rate": 1.218e-06,
      "loss": 3.0498,
      "step": 407
    },
    {
      "epoch": 0.7534626038781164,
      "grad_norm": 6.8960862159729,
      "learning_rate": 1.221e-06,
      "loss": 3.014,
      "step": 408
    },
    {
      "epoch": 0.7553093259464451,
      "grad_norm": 3.969679117202759,
      "learning_rate": 1.224e-06,
      "loss": 3.0294,
      "step": 409
    },
    {
      "epoch": 0.7571560480147738,
      "grad_norm": 4.408971786499023,
      "learning_rate": 1.227e-06,
      "loss": 2.9686,
      "step": 410
    },
    {
      "epoch": 0.7590027700831025,
      "grad_norm": 4.6965107917785645,
      "learning_rate": 1.23e-06,
      "loss": 3.0472,
      "step": 411
    },
    {
      "epoch": 0.7608494921514312,
      "grad_norm": 3.0788402557373047,
      "learning_rate": 1.2329999999999999e-06,
      "loss": 2.9867,
      "step": 412
    },
    {
      "epoch": 0.76269621421976,
      "grad_norm": 5.122383117675781,
      "learning_rate": 1.2360000000000001e-06,
      "loss": 3.0021,
      "step": 413
    },
    {
      "epoch": 0.7645429362880887,
      "grad_norm": 3.920349359512329,
      "learning_rate": 1.2390000000000001e-06,
      "loss": 2.9976,
      "step": 414
    },
    {
      "epoch": 0.7663896583564174,
      "grad_norm": 8.338603019714355,
      "learning_rate": 1.242e-06,
      "loss": 2.9776,
      "step": 415
    },
    {
      "epoch": 0.7682363804247461,
      "grad_norm": 3.2974908351898193,
      "learning_rate": 1.2450000000000002e-06,
      "loss": 3.0257,
      "step": 416
    },
    {
      "epoch": 0.7700831024930748,
      "grad_norm": 4.498806953430176,
      "learning_rate": 1.248e-06,
      "loss": 2.9794,
      "step": 417
    },
    {
      "epoch": 0.7719298245614035,
      "grad_norm": 3.317884683609009,
      "learning_rate": 1.251e-06,
      "loss": 3.0031,
      "step": 418
    },
    {
      "epoch": 0.7737765466297323,
      "grad_norm": 5.48812198638916,
      "learning_rate": 1.254e-06,
      "loss": 3.0484,
      "step": 419
    },
    {
      "epoch": 0.775623268698061,
      "grad_norm": 5.9689154624938965,
      "learning_rate": 1.257e-06,
      "loss": 3.0451,
      "step": 420
    },
    {
      "epoch": 0.7774699907663897,
      "grad_norm": 4.714944839477539,
      "learning_rate": 1.26e-06,
      "loss": 3.0069,
      "step": 421
    },
    {
      "epoch": 0.7793167128347184,
      "grad_norm": 4.204563617706299,
      "learning_rate": 1.263e-06,
      "loss": 3.0274,
      "step": 422
    },
    {
      "epoch": 0.7811634349030471,
      "grad_norm": 4.774872779846191,
      "learning_rate": 1.266e-06,
      "loss": 2.9794,
      "step": 423
    },
    {
      "epoch": 0.7830101569713758,
      "grad_norm": 5.415470600128174,
      "learning_rate": 1.269e-06,
      "loss": 3.0081,
      "step": 424
    },
    {
      "epoch": 0.7848568790397045,
      "grad_norm": 4.5735554695129395,
      "learning_rate": 1.272e-06,
      "loss": 3.0158,
      "step": 425
    },
    {
      "epoch": 0.7867036011080333,
      "grad_norm": 4.952023506164551,
      "learning_rate": 1.275e-06,
      "loss": 3.0651,
      "step": 426
    },
    {
      "epoch": 0.788550323176362,
      "grad_norm": 4.689613342285156,
      "learning_rate": 1.278e-06,
      "loss": 2.9854,
      "step": 427
    },
    {
      "epoch": 0.7903970452446907,
      "grad_norm": 5.4375762939453125,
      "learning_rate": 1.281e-06,
      "loss": 2.9901,
      "step": 428
    },
    {
      "epoch": 0.7922437673130194,
      "grad_norm": 4.584848880767822,
      "learning_rate": 1.284e-06,
      "loss": 3.0129,
      "step": 429
    },
    {
      "epoch": 0.7940904893813481,
      "grad_norm": 5.961929798126221,
      "learning_rate": 1.2870000000000001e-06,
      "loss": 3.017,
      "step": 430
    },
    {
      "epoch": 0.7959372114496768,
      "grad_norm": 6.332926273345947,
      "learning_rate": 1.29e-06,
      "loss": 3.0211,
      "step": 431
    },
    {
      "epoch": 0.7977839335180056,
      "grad_norm": 4.257788181304932,
      "learning_rate": 1.293e-06,
      "loss": 3.0044,
      "step": 432
    },
    {
      "epoch": 0.7996306555863343,
      "grad_norm": 4.114194393157959,
      "learning_rate": 1.2960000000000002e-06,
      "loss": 2.9987,
      "step": 433
    },
    {
      "epoch": 0.801477377654663,
      "grad_norm": 12.284896850585938,
      "learning_rate": 1.299e-06,
      "loss": 2.9784,
      "step": 434
    },
    {
      "epoch": 0.8033240997229917,
      "grad_norm": 7.53204870223999,
      "learning_rate": 1.302e-06,
      "loss": 3.0016,
      "step": 435
    },
    {
      "epoch": 0.8051708217913204,
      "grad_norm": 3.8289263248443604,
      "learning_rate": 1.305e-06,
      "loss": 3.0036,
      "step": 436
    },
    {
      "epoch": 0.8070175438596491,
      "grad_norm": 6.599183559417725,
      "learning_rate": 1.308e-06,
      "loss": 2.9835,
      "step": 437
    },
    {
      "epoch": 0.8088642659279779,
      "grad_norm": 4.600238800048828,
      "learning_rate": 1.311e-06,
      "loss": 3.0261,
      "step": 438
    },
    {
      "epoch": 0.8107109879963066,
      "grad_norm": 4.442101955413818,
      "learning_rate": 1.314e-06,
      "loss": 2.9914,
      "step": 439
    },
    {
      "epoch": 0.8125577100646353,
      "grad_norm": 4.377070903778076,
      "learning_rate": 1.317e-06,
      "loss": 2.9763,
      "step": 440
    },
    {
      "epoch": 0.814404432132964,
      "grad_norm": 5.045103549957275,
      "learning_rate": 1.3199999999999999e-06,
      "loss": 2.9904,
      "step": 441
    },
    {
      "epoch": 0.8162511542012927,
      "grad_norm": 6.373979568481445,
      "learning_rate": 1.323e-06,
      "loss": 3.0087,
      "step": 442
    },
    {
      "epoch": 0.8180978762696214,
      "grad_norm": 4.301845073699951,
      "learning_rate": 1.326e-06,
      "loss": 2.9853,
      "step": 443
    },
    {
      "epoch": 0.8199445983379502,
      "grad_norm": 15.533903121948242,
      "learning_rate": 1.3290000000000001e-06,
      "loss": 3.0014,
      "step": 444
    },
    {
      "epoch": 0.8217913204062789,
      "grad_norm": 4.522268772125244,
      "learning_rate": 1.3320000000000001e-06,
      "loss": 2.9524,
      "step": 445
    },
    {
      "epoch": 0.8236380424746076,
      "grad_norm": 4.2080559730529785,
      "learning_rate": 1.335e-06,
      "loss": 3.0163,
      "step": 446
    },
    {
      "epoch": 0.8254847645429363,
      "grad_norm": 4.199611663818359,
      "learning_rate": 1.3380000000000001e-06,
      "loss": 2.996,
      "step": 447
    },
    {
      "epoch": 0.827331486611265,
      "grad_norm": 4.514741897583008,
      "learning_rate": 1.341e-06,
      "loss": 2.9532,
      "step": 448
    },
    {
      "epoch": 0.8291782086795937,
      "grad_norm": 4.151061058044434,
      "learning_rate": 1.344e-06,
      "loss": 2.973,
      "step": 449
    },
    {
      "epoch": 0.8310249307479224,
      "grad_norm": 8.964682579040527,
      "learning_rate": 1.3470000000000002e-06,
      "loss": 2.998,
      "step": 450
    },
    {
      "epoch": 0.8328716528162512,
      "grad_norm": 7.244773864746094,
      "learning_rate": 1.35e-06,
      "loss": 3.1545,
      "step": 451
    },
    {
      "epoch": 0.8347183748845799,
      "grad_norm": 5.385263919830322,
      "learning_rate": 1.353e-06,
      "loss": 3.0276,
      "step": 452
    },
    {
      "epoch": 0.8365650969529086,
      "grad_norm": 2.983064889907837,
      "learning_rate": 1.356e-06,
      "loss": 2.9877,
      "step": 453
    },
    {
      "epoch": 0.8384118190212373,
      "grad_norm": 4.306495666503906,
      "learning_rate": 1.359e-06,
      "loss": 2.9808,
      "step": 454
    },
    {
      "epoch": 0.840258541089566,
      "grad_norm": 5.624070167541504,
      "learning_rate": 1.362e-06,
      "loss": 3.0441,
      "step": 455
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 5.876847267150879,
      "learning_rate": 1.365e-06,
      "loss": 3.014,
      "step": 456
    },
    {
      "epoch": 0.8439519852262235,
      "grad_norm": 5.433878421783447,
      "learning_rate": 1.368e-06,
      "loss": 3.0038,
      "step": 457
    },
    {
      "epoch": 0.8457987072945522,
      "grad_norm": 2.7806458473205566,
      "learning_rate": 1.3709999999999999e-06,
      "loss": 2.9971,
      "step": 458
    },
    {
      "epoch": 0.8476454293628809,
      "grad_norm": 3.286996603012085,
      "learning_rate": 1.374e-06,
      "loss": 2.9475,
      "step": 459
    },
    {
      "epoch": 0.8494921514312096,
      "grad_norm": 4.864804267883301,
      "learning_rate": 1.3770000000000001e-06,
      "loss": 2.9524,
      "step": 460
    },
    {
      "epoch": 0.8513388734995383,
      "grad_norm": 5.1426615715026855,
      "learning_rate": 1.38e-06,
      "loss": 2.9754,
      "step": 461
    },
    {
      "epoch": 0.853185595567867,
      "grad_norm": 6.4956955909729,
      "learning_rate": 1.3830000000000001e-06,
      "loss": 2.9736,
      "step": 462
    },
    {
      "epoch": 0.8550323176361958,
      "grad_norm": 5.773088455200195,
      "learning_rate": 1.386e-06,
      "loss": 2.989,
      "step": 463
    },
    {
      "epoch": 0.8568790397045245,
      "grad_norm": 4.204294204711914,
      "learning_rate": 1.389e-06,
      "loss": 2.9945,
      "step": 464
    },
    {
      "epoch": 0.8587257617728532,
      "grad_norm": 4.098605632781982,
      "learning_rate": 1.392e-06,
      "loss": 2.9446,
      "step": 465
    },
    {
      "epoch": 0.8605724838411819,
      "grad_norm": 5.481067180633545,
      "learning_rate": 1.395e-06,
      "loss": 2.9319,
      "step": 466
    },
    {
      "epoch": 0.8624192059095106,
      "grad_norm": 6.614928245544434,
      "learning_rate": 1.3980000000000002e-06,
      "loss": 2.9674,
      "step": 467
    },
    {
      "epoch": 0.8642659279778393,
      "grad_norm": 7.987035751342773,
      "learning_rate": 1.401e-06,
      "loss": 2.9716,
      "step": 468
    },
    {
      "epoch": 0.866112650046168,
      "grad_norm": 5.396458625793457,
      "learning_rate": 1.404e-06,
      "loss": 2.9757,
      "step": 469
    },
    {
      "epoch": 0.8679593721144968,
      "grad_norm": 7.34463357925415,
      "learning_rate": 1.407e-06,
      "loss": 2.9759,
      "step": 470
    },
    {
      "epoch": 0.8698060941828255,
      "grad_norm": 4.453892707824707,
      "learning_rate": 1.41e-06,
      "loss": 2.9274,
      "step": 471
    },
    {
      "epoch": 0.8716528162511542,
      "grad_norm": 6.051670551300049,
      "learning_rate": 1.413e-06,
      "loss": 2.9369,
      "step": 472
    },
    {
      "epoch": 0.8734995383194829,
      "grad_norm": 5.094610214233398,
      "learning_rate": 1.416e-06,
      "loss": 2.9687,
      "step": 473
    },
    {
      "epoch": 0.8753462603878116,
      "grad_norm": 6.582310199737549,
      "learning_rate": 1.419e-06,
      "loss": 2.9616,
      "step": 474
    },
    {
      "epoch": 0.8771929824561403,
      "grad_norm": 6.141293525695801,
      "learning_rate": 1.422e-06,
      "loss": 2.9422,
      "step": 475
    },
    {
      "epoch": 0.8790397045244691,
      "grad_norm": 11.331812858581543,
      "learning_rate": 1.4250000000000001e-06,
      "loss": 3.0109,
      "step": 476
    },
    {
      "epoch": 0.8808864265927978,
      "grad_norm": 3.9797582626342773,
      "learning_rate": 1.4280000000000001e-06,
      "loss": 2.9676,
      "step": 477
    },
    {
      "epoch": 0.8827331486611265,
      "grad_norm": 10.876532554626465,
      "learning_rate": 1.431e-06,
      "loss": 2.9751,
      "step": 478
    },
    {
      "epoch": 0.8845798707294552,
      "grad_norm": 6.678962707519531,
      "learning_rate": 1.4340000000000002e-06,
      "loss": 2.9888,
      "step": 479
    },
    {
      "epoch": 0.8864265927977839,
      "grad_norm": 5.212357044219971,
      "learning_rate": 1.437e-06,
      "loss": 2.9691,
      "step": 480
    },
    {
      "epoch": 0.8882733148661126,
      "grad_norm": 10.343843460083008,
      "learning_rate": 1.44e-06,
      "loss": 2.9692,
      "step": 481
    },
    {
      "epoch": 0.8901200369344414,
      "grad_norm": 5.704240798950195,
      "learning_rate": 1.443e-06,
      "loss": 2.9718,
      "step": 482
    },
    {
      "epoch": 0.8919667590027701,
      "grad_norm": 5.897530555725098,
      "learning_rate": 1.446e-06,
      "loss": 2.9566,
      "step": 483
    },
    {
      "epoch": 0.8938134810710988,
      "grad_norm": 6.475763320922852,
      "learning_rate": 1.449e-06,
      "loss": 2.9642,
      "step": 484
    },
    {
      "epoch": 0.8956602031394275,
      "grad_norm": 12.50463581085205,
      "learning_rate": 1.452e-06,
      "loss": 2.9554,
      "step": 485
    },
    {
      "epoch": 0.8975069252077562,
      "grad_norm": 9.410432815551758,
      "learning_rate": 1.455e-06,
      "loss": 2.9559,
      "step": 486
    },
    {
      "epoch": 0.8993536472760849,
      "grad_norm": 5.922210216522217,
      "learning_rate": 1.458e-06,
      "loss": 3.0047,
      "step": 487
    },
    {
      "epoch": 0.9012003693444137,
      "grad_norm": 4.550955295562744,
      "learning_rate": 1.461e-06,
      "loss": 2.9478,
      "step": 488
    },
    {
      "epoch": 0.9030470914127424,
      "grad_norm": 4.072491645812988,
      "learning_rate": 1.464e-06,
      "loss": 2.9659,
      "step": 489
    },
    {
      "epoch": 0.9048938134810711,
      "grad_norm": 5.787655353546143,
      "learning_rate": 1.467e-06,
      "loss": 2.9295,
      "step": 490
    },
    {
      "epoch": 0.9067405355493998,
      "grad_norm": 6.225001335144043,
      "learning_rate": 1.4700000000000001e-06,
      "loss": 2.9969,
      "step": 491
    },
    {
      "epoch": 0.9085872576177285,
      "grad_norm": 5.337482452392578,
      "learning_rate": 1.473e-06,
      "loss": 2.9654,
      "step": 492
    },
    {
      "epoch": 0.9104339796860572,
      "grad_norm": 5.667120456695557,
      "learning_rate": 1.4760000000000001e-06,
      "loss": 2.9851,
      "step": 493
    },
    {
      "epoch": 0.9122807017543859,
      "grad_norm": 5.20329475402832,
      "learning_rate": 1.479e-06,
      "loss": 2.9161,
      "step": 494
    },
    {
      "epoch": 0.9141274238227147,
      "grad_norm": 8.450933456420898,
      "learning_rate": 1.482e-06,
      "loss": 2.94,
      "step": 495
    },
    {
      "epoch": 0.9159741458910434,
      "grad_norm": 4.999294281005859,
      "learning_rate": 1.4850000000000002e-06,
      "loss": 2.9509,
      "step": 496
    },
    {
      "epoch": 0.9178208679593721,
      "grad_norm": 12.855695724487305,
      "learning_rate": 1.488e-06,
      "loss": 2.9833,
      "step": 497
    },
    {
      "epoch": 0.9196675900277008,
      "grad_norm": 3.0434823036193848,
      "learning_rate": 1.491e-06,
      "loss": 2.9338,
      "step": 498
    },
    {
      "epoch": 0.9215143120960295,
      "grad_norm": 7.846760272979736,
      "learning_rate": 1.494e-06,
      "loss": 3.0113,
      "step": 499
    },
    {
      "epoch": 0.9233610341643582,
      "grad_norm": 7.970994472503662,
      "learning_rate": 1.497e-06,
      "loss": 3.0131,
      "step": 500
    },
    {
      "epoch": 0.925207756232687,
      "grad_norm": 5.931797981262207,
      "learning_rate": 1.5e-06,
      "loss": 3.1065,
      "step": 501
    },
    {
      "epoch": 0.9270544783010157,
      "grad_norm": 5.381325721740723,
      "learning_rate": 1.503e-06,
      "loss": 2.9859,
      "step": 502
    },
    {
      "epoch": 0.9289012003693444,
      "grad_norm": 4.340394020080566,
      "learning_rate": 1.506e-06,
      "loss": 2.964,
      "step": 503
    },
    {
      "epoch": 0.9307479224376731,
      "grad_norm": 10.615586280822754,
      "learning_rate": 1.5089999999999999e-06,
      "loss": 2.9981,
      "step": 504
    },
    {
      "epoch": 0.9325946445060018,
      "grad_norm": 4.364864826202393,
      "learning_rate": 1.512e-06,
      "loss": 2.9368,
      "step": 505
    },
    {
      "epoch": 0.9344413665743305,
      "grad_norm": 4.9499359130859375,
      "learning_rate": 1.5150000000000001e-06,
      "loss": 2.9529,
      "step": 506
    },
    {
      "epoch": 0.9362880886426593,
      "grad_norm": 3.906594753265381,
      "learning_rate": 1.518e-06,
      "loss": 2.9482,
      "step": 507
    },
    {
      "epoch": 0.938134810710988,
      "grad_norm": 4.437647819519043,
      "learning_rate": 1.5210000000000001e-06,
      "loss": 2.9554,
      "step": 508
    },
    {
      "epoch": 0.9399815327793167,
      "grad_norm": 3.466791868209839,
      "learning_rate": 1.524e-06,
      "loss": 2.9123,
      "step": 509
    },
    {
      "epoch": 0.9418282548476454,
      "grad_norm": 4.363769054412842,
      "learning_rate": 1.5270000000000002e-06,
      "loss": 2.9242,
      "step": 510
    },
    {
      "epoch": 0.9436749769159741,
      "grad_norm": 5.1405863761901855,
      "learning_rate": 1.53e-06,
      "loss": 2.9258,
      "step": 511
    },
    {
      "epoch": 0.9455216989843028,
      "grad_norm": 4.913668632507324,
      "learning_rate": 1.533e-06,
      "loss": 2.9457,
      "step": 512
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 6.878814220428467,
      "learning_rate": 1.5360000000000002e-06,
      "loss": 2.9271,
      "step": 513
    },
    {
      "epoch": 0.9492151431209603,
      "grad_norm": 5.439577102661133,
      "learning_rate": 1.539e-06,
      "loss": 2.9196,
      "step": 514
    },
    {
      "epoch": 0.951061865189289,
      "grad_norm": 3.7509171962738037,
      "learning_rate": 1.542e-06,
      "loss": 2.9413,
      "step": 515
    },
    {
      "epoch": 0.9529085872576177,
      "grad_norm": 6.960578918457031,
      "learning_rate": 1.545e-06,
      "loss": 2.9481,
      "step": 516
    },
    {
      "epoch": 0.9547553093259464,
      "grad_norm": 7.0822930335998535,
      "learning_rate": 1.548e-06,
      "loss": 2.9559,
      "step": 517
    },
    {
      "epoch": 0.9566020313942751,
      "grad_norm": 6.05672550201416,
      "learning_rate": 1.551e-06,
      "loss": 2.9432,
      "step": 518
    },
    {
      "epoch": 0.9584487534626038,
      "grad_norm": 5.092657566070557,
      "learning_rate": 1.554e-06,
      "loss": 2.9519,
      "step": 519
    },
    {
      "epoch": 0.9602954755309326,
      "grad_norm": 8.910967826843262,
      "learning_rate": 1.557e-06,
      "loss": 2.9207,
      "step": 520
    },
    {
      "epoch": 0.9621421975992613,
      "grad_norm": 6.184726238250732,
      "learning_rate": 1.5599999999999999e-06,
      "loss": 2.9386,
      "step": 521
    },
    {
      "epoch": 0.96398891966759,
      "grad_norm": 7.006811141967773,
      "learning_rate": 1.5630000000000001e-06,
      "loss": 2.9322,
      "step": 522
    },
    {
      "epoch": 0.9658356417359187,
      "grad_norm": 6.83420467376709,
      "learning_rate": 1.5660000000000001e-06,
      "loss": 2.9562,
      "step": 523
    },
    {
      "epoch": 0.9676823638042474,
      "grad_norm": 7.393522262573242,
      "learning_rate": 1.569e-06,
      "loss": 2.9554,
      "step": 524
    },
    {
      "epoch": 0.9695290858725761,
      "grad_norm": 5.073296070098877,
      "learning_rate": 1.5720000000000002e-06,
      "loss": 2.9534,
      "step": 525
    },
    {
      "epoch": 0.971375807940905,
      "grad_norm": 6.644114017486572,
      "learning_rate": 1.575e-06,
      "loss": 2.9545,
      "step": 526
    },
    {
      "epoch": 0.9732225300092336,
      "grad_norm": 5.469161510467529,
      "learning_rate": 1.578e-06,
      "loss": 2.934,
      "step": 527
    },
    {
      "epoch": 0.9750692520775623,
      "grad_norm": 8.649992942810059,
      "learning_rate": 1.581e-06,
      "loss": 2.9644,
      "step": 528
    },
    {
      "epoch": 0.976915974145891,
      "grad_norm": 5.227014064788818,
      "learning_rate": 1.584e-06,
      "loss": 2.8967,
      "step": 529
    },
    {
      "epoch": 0.9787626962142197,
      "grad_norm": 6.67928409576416,
      "learning_rate": 1.5870000000000002e-06,
      "loss": 2.9356,
      "step": 530
    },
    {
      "epoch": 0.9806094182825484,
      "grad_norm": 6.659364700317383,
      "learning_rate": 1.59e-06,
      "loss": 2.939,
      "step": 531
    },
    {
      "epoch": 0.9824561403508771,
      "grad_norm": 9.442721366882324,
      "learning_rate": 1.593e-06,
      "loss": 2.9603,
      "step": 532
    },
    {
      "epoch": 0.984302862419206,
      "grad_norm": 6.252249240875244,
      "learning_rate": 1.596e-06,
      "loss": 2.9177,
      "step": 533
    },
    {
      "epoch": 0.9861495844875346,
      "grad_norm": 6.107838153839111,
      "learning_rate": 1.599e-06,
      "loss": 2.9185,
      "step": 534
    },
    {
      "epoch": 0.9879963065558633,
      "grad_norm": 8.275815963745117,
      "learning_rate": 1.602e-06,
      "loss": 2.9011,
      "step": 535
    },
    {
      "epoch": 0.989843028624192,
      "grad_norm": 5.931856155395508,
      "learning_rate": 1.605e-06,
      "loss": 2.9344,
      "step": 536
    },
    {
      "epoch": 0.9916897506925207,
      "grad_norm": 7.1943511962890625,
      "learning_rate": 1.608e-06,
      "loss": 2.8943,
      "step": 537
    },
    {
      "epoch": 0.9935364727608494,
      "grad_norm": 12.691488265991211,
      "learning_rate": 1.611e-06,
      "loss": 2.9317,
      "step": 538
    },
    {
      "epoch": 0.9953831948291783,
      "grad_norm": 4.4045233726501465,
      "learning_rate": 1.6140000000000001e-06,
      "loss": 2.8875,
      "step": 539
    },
    {
      "epoch": 0.997229916897507,
      "grad_norm": 6.350393772125244,
      "learning_rate": 1.6170000000000001e-06,
      "loss": 2.9094,
      "step": 540
    },
    {
      "epoch": 0.9990766389658357,
      "grad_norm": 6.049736499786377,
      "learning_rate": 1.62e-06,
      "loss": 2.944,
      "step": 541
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.06985330581665,
      "learning_rate": 1.6230000000000002e-06,
      "loss": 1.4558,
      "step": 542
    },
    {
      "epoch": 1.0018467220683287,
      "grad_norm": 12.09255599975586,
      "learning_rate": 1.626e-06,
      "loss": 3.0718,
      "step": 543
    },
    {
      "epoch": 1.0036934441366574,
      "grad_norm": 9.780098915100098,
      "learning_rate": 1.629e-06,
      "loss": 3.0656,
      "step": 544
    },
    {
      "epoch": 1.005540166204986,
      "grad_norm": 5.1886467933654785,
      "learning_rate": 1.632e-06,
      "loss": 3.0008,
      "step": 545
    },
    {
      "epoch": 1.0073868882733148,
      "grad_norm": 4.111461162567139,
      "learning_rate": 1.635e-06,
      "loss": 2.9359,
      "step": 546
    },
    {
      "epoch": 1.0092336103416435,
      "grad_norm": 5.650781154632568,
      "learning_rate": 1.638e-06,
      "loss": 2.9783,
      "step": 547
    },
    {
      "epoch": 1.0110803324099722,
      "grad_norm": 7.087272644042969,
      "learning_rate": 1.641e-06,
      "loss": 3.0137,
      "step": 548
    },
    {
      "epoch": 1.0129270544783011,
      "grad_norm": 6.621068477630615,
      "learning_rate": 1.644e-06,
      "loss": 2.9767,
      "step": 549
    },
    {
      "epoch": 1.0147737765466298,
      "grad_norm": 4.337469100952148,
      "learning_rate": 1.6469999999999999e-06,
      "loss": 2.948,
      "step": 550
    },
    {
      "epoch": 1.0166204986149585,
      "grad_norm": 3.3184478282928467,
      "learning_rate": 1.65e-06,
      "loss": 2.9006,
      "step": 551
    },
    {
      "epoch": 1.0184672206832872,
      "grad_norm": 3.802716016769409,
      "learning_rate": 1.653e-06,
      "loss": 2.9339,
      "step": 552
    },
    {
      "epoch": 1.020313942751616,
      "grad_norm": 5.8885297775268555,
      "learning_rate": 1.6560000000000001e-06,
      "loss": 2.9551,
      "step": 553
    },
    {
      "epoch": 1.0221606648199446,
      "grad_norm": 10.987329483032227,
      "learning_rate": 1.6590000000000001e-06,
      "loss": 2.9743,
      "step": 554
    },
    {
      "epoch": 1.0240073868882733,
      "grad_norm": 8.780963897705078,
      "learning_rate": 1.662e-06,
      "loss": 2.9448,
      "step": 555
    },
    {
      "epoch": 1.025854108956602,
      "grad_norm": 10.191997528076172,
      "learning_rate": 1.6650000000000002e-06,
      "loss": 2.971,
      "step": 556
    },
    {
      "epoch": 1.0277008310249307,
      "grad_norm": 6.121068954467773,
      "learning_rate": 1.668e-06,
      "loss": 2.8992,
      "step": 557
    },
    {
      "epoch": 1.0295475530932594,
      "grad_norm": 5.866390228271484,
      "learning_rate": 1.671e-06,
      "loss": 2.915,
      "step": 558
    },
    {
      "epoch": 1.031394275161588,
      "grad_norm": 4.83311653137207,
      "learning_rate": 1.6740000000000002e-06,
      "loss": 2.9023,
      "step": 559
    },
    {
      "epoch": 1.0332409972299168,
      "grad_norm": 12.177200317382812,
      "learning_rate": 1.677e-06,
      "loss": 2.938,
      "step": 560
    },
    {
      "epoch": 1.0350877192982457,
      "grad_norm": 11.86186695098877,
      "learning_rate": 1.68e-06,
      "loss": 2.9344,
      "step": 561
    },
    {
      "epoch": 1.0369344413665744,
      "grad_norm": 8.23930835723877,
      "learning_rate": 1.683e-06,
      "loss": 2.9468,
      "step": 562
    },
    {
      "epoch": 1.0387811634349031,
      "grad_norm": 5.7900166511535645,
      "learning_rate": 1.686e-06,
      "loss": 2.8912,
      "step": 563
    },
    {
      "epoch": 1.0406278855032318,
      "grad_norm": 9.679719924926758,
      "learning_rate": 1.689e-06,
      "loss": 2.8847,
      "step": 564
    },
    {
      "epoch": 1.0424746075715605,
      "grad_norm": 4.417483806610107,
      "learning_rate": 1.692e-06,
      "loss": 2.9201,
      "step": 565
    },
    {
      "epoch": 1.0443213296398892,
      "grad_norm": 11.755255699157715,
      "learning_rate": 1.695e-06,
      "loss": 2.9109,
      "step": 566
    },
    {
      "epoch": 1.046168051708218,
      "grad_norm": 6.306344985961914,
      "learning_rate": 1.6979999999999999e-06,
      "loss": 2.8868,
      "step": 567
    },
    {
      "epoch": 1.0480147737765466,
      "grad_norm": 7.374333381652832,
      "learning_rate": 1.701e-06,
      "loss": 2.9801,
      "step": 568
    },
    {
      "epoch": 1.0498614958448753,
      "grad_norm": 9.103034973144531,
      "learning_rate": 1.7040000000000001e-06,
      "loss": 2.9041,
      "step": 569
    },
    {
      "epoch": 1.051708217913204,
      "grad_norm": 6.525115013122559,
      "learning_rate": 1.707e-06,
      "loss": 2.9199,
      "step": 570
    },
    {
      "epoch": 1.0535549399815327,
      "grad_norm": 5.839273452758789,
      "learning_rate": 1.7100000000000001e-06,
      "loss": 2.9292,
      "step": 571
    },
    {
      "epoch": 1.0554016620498614,
      "grad_norm": 7.118470668792725,
      "learning_rate": 1.713e-06,
      "loss": 2.92,
      "step": 572
    },
    {
      "epoch": 1.05724838411819,
      "grad_norm": 7.868495464324951,
      "learning_rate": 1.7160000000000002e-06,
      "loss": 2.8916,
      "step": 573
    },
    {
      "epoch": 1.059095106186519,
      "grad_norm": 14.98963737487793,
      "learning_rate": 1.719e-06,
      "loss": 2.9395,
      "step": 574
    },
    {
      "epoch": 1.0609418282548477,
      "grad_norm": 6.7287774085998535,
      "learning_rate": 1.722e-06,
      "loss": 2.9224,
      "step": 575
    },
    {
      "epoch": 1.0627885503231764,
      "grad_norm": 6.143113613128662,
      "learning_rate": 1.7250000000000002e-06,
      "loss": 2.932,
      "step": 576
    },
    {
      "epoch": 1.0646352723915051,
      "grad_norm": 5.883518695831299,
      "learning_rate": 1.728e-06,
      "loss": 2.8912,
      "step": 577
    },
    {
      "epoch": 1.0664819944598338,
      "grad_norm": 6.572492599487305,
      "learning_rate": 1.731e-06,
      "loss": 2.9304,
      "step": 578
    },
    {
      "epoch": 1.0683287165281625,
      "grad_norm": 8.820541381835938,
      "learning_rate": 1.734e-06,
      "loss": 2.8974,
      "step": 579
    },
    {
      "epoch": 1.0701754385964912,
      "grad_norm": 8.882281303405762,
      "learning_rate": 1.737e-06,
      "loss": 2.9745,
      "step": 580
    },
    {
      "epoch": 1.07202216066482,
      "grad_norm": 14.256232261657715,
      "learning_rate": 1.74e-06,
      "loss": 2.9035,
      "step": 581
    },
    {
      "epoch": 1.0738688827331486,
      "grad_norm": 7.182823181152344,
      "learning_rate": 1.743e-06,
      "loss": 2.8644,
      "step": 582
    },
    {
      "epoch": 1.0757156048014773,
      "grad_norm": 7.0797600746154785,
      "learning_rate": 1.746e-06,
      "loss": 2.8771,
      "step": 583
    },
    {
      "epoch": 1.077562326869806,
      "grad_norm": 11.359004974365234,
      "learning_rate": 1.749e-06,
      "loss": 2.9435,
      "step": 584
    },
    {
      "epoch": 1.0794090489381347,
      "grad_norm": 21.062517166137695,
      "learning_rate": 1.7520000000000001e-06,
      "loss": 2.9306,
      "step": 585
    },
    {
      "epoch": 1.0812557710064636,
      "grad_norm": 10.017592430114746,
      "learning_rate": 1.7550000000000001e-06,
      "loss": 2.9346,
      "step": 586
    },
    {
      "epoch": 1.0831024930747923,
      "grad_norm": NaN,
      "learning_rate": 1.7550000000000001e-06,
      "loss": 2.8635,
      "step": 587
    },
    {
      "epoch": 1.084949215143121,
      "grad_norm": 8.263057708740234,
      "learning_rate": 1.758e-06,
      "loss": 2.9339,
      "step": 588
    },
    {
      "epoch": 1.0867959372114497,
      "grad_norm": 6.318991661071777,
      "learning_rate": 1.7610000000000002e-06,
      "loss": 2.8913,
      "step": 589
    },
    {
      "epoch": 1.0886426592797784,
      "grad_norm": 5.331015110015869,
      "learning_rate": 1.764e-06,
      "loss": 2.8813,
      "step": 590
    },
    {
      "epoch": 1.0904893813481071,
      "grad_norm": 7.7847089767456055,
      "learning_rate": 1.767e-06,
      "loss": 2.9171,
      "step": 591
    },
    {
      "epoch": 1.0923361034164358,
      "grad_norm": 4.977515697479248,
      "learning_rate": 1.77e-06,
      "loss": 2.8977,
      "step": 592
    },
    {
      "epoch": 1.0941828254847645,
      "grad_norm": 5.907567024230957,
      "learning_rate": 1.773e-06,
      "loss": 3.0373,
      "step": 593
    },
    {
      "epoch": 1.0960295475530932,
      "grad_norm": 4.246371269226074,
      "learning_rate": 1.776e-06,
      "loss": 2.9802,
      "step": 594
    },
    {
      "epoch": 1.097876269621422,
      "grad_norm": 3.133854389190674,
      "learning_rate": 1.779e-06,
      "loss": 2.9497,
      "step": 595
    },
    {
      "epoch": 1.0997229916897506,
      "grad_norm": 3.6142194271087646,
      "learning_rate": 1.782e-06,
      "loss": 2.9732,
      "step": 596
    },
    {
      "epoch": 1.1015697137580793,
      "grad_norm": 3.5081558227539062,
      "learning_rate": 1.785e-06,
      "loss": 2.934,
      "step": 597
    },
    {
      "epoch": 1.103416435826408,
      "grad_norm": 5.0536956787109375,
      "learning_rate": 1.788e-06,
      "loss": 2.9035,
      "step": 598
    },
    {
      "epoch": 1.1052631578947367,
      "grad_norm": 6.321654796600342,
      "learning_rate": 1.791e-06,
      "loss": 2.9239,
      "step": 599
    },
    {
      "epoch": 1.1071098799630656,
      "grad_norm": 6.200789451599121,
      "learning_rate": 1.794e-06,
      "loss": 2.9125,
      "step": 600
    },
    {
      "epoch": 1.1089566020313943,
      "grad_norm": 12.870471954345703,
      "learning_rate": 1.7970000000000001e-06,
      "loss": 2.9131,
      "step": 601
    },
    {
      "epoch": 1.110803324099723,
      "grad_norm": 4.161349296569824,
      "learning_rate": 1.8e-06,
      "loss": 2.9016,
      "step": 602
    },
    {
      "epoch": 1.1126500461680517,
      "grad_norm": 7.384085655212402,
      "learning_rate": 1.8030000000000001e-06,
      "loss": 2.9012,
      "step": 603
    },
    {
      "epoch": 1.1144967682363804,
      "grad_norm": 5.1679840087890625,
      "learning_rate": 1.806e-06,
      "loss": 2.9031,
      "step": 604
    },
    {
      "epoch": 1.1163434903047091,
      "grad_norm": 5.276459693908691,
      "learning_rate": 1.809e-06,
      "loss": 2.8899,
      "step": 605
    },
    {
      "epoch": 1.1181902123730378,
      "grad_norm": 5.313704967498779,
      "learning_rate": 1.8120000000000002e-06,
      "loss": 2.8708,
      "step": 606
    },
    {
      "epoch": 1.1200369344413665,
      "grad_norm": 5.521195888519287,
      "learning_rate": 1.815e-06,
      "loss": 2.9006,
      "step": 607
    },
    {
      "epoch": 1.1218836565096952,
      "grad_norm": 8.239352226257324,
      "learning_rate": 1.818e-06,
      "loss": 2.905,
      "step": 608
    },
    {
      "epoch": 1.123730378578024,
      "grad_norm": 5.124113082885742,
      "learning_rate": 1.821e-06,
      "loss": 2.8871,
      "step": 609
    },
    {
      "epoch": 1.1255771006463526,
      "grad_norm": 4.876293659210205,
      "learning_rate": 1.824e-06,
      "loss": 2.8934,
      "step": 610
    },
    {
      "epoch": 1.1274238227146816,
      "grad_norm": 10.052160263061523,
      "learning_rate": 1.827e-06,
      "loss": 2.9043,
      "step": 611
    },
    {
      "epoch": 1.1292705447830103,
      "grad_norm": 8.201723098754883,
      "learning_rate": 1.83e-06,
      "loss": 2.9086,
      "step": 612
    },
    {
      "epoch": 1.131117266851339,
      "grad_norm": 4.239204406738281,
      "learning_rate": 1.833e-06,
      "loss": 2.8758,
      "step": 613
    },
    {
      "epoch": 1.1329639889196677,
      "grad_norm": 5.192922592163086,
      "learning_rate": 1.8359999999999999e-06,
      "loss": 2.8823,
      "step": 614
    },
    {
      "epoch": 1.1348107109879964,
      "grad_norm": 8.6290864944458,
      "learning_rate": 1.839e-06,
      "loss": 2.8688,
      "step": 615
    },
    {
      "epoch": 1.136657433056325,
      "grad_norm": 6.887111663818359,
      "learning_rate": 1.8420000000000001e-06,
      "loss": 2.8978,
      "step": 616
    },
    {
      "epoch": 1.1385041551246537,
      "grad_norm": 4.787206172943115,
      "learning_rate": 1.8450000000000001e-06,
      "loss": 2.8983,
      "step": 617
    },
    {
      "epoch": 1.1403508771929824,
      "grad_norm": 10.718205451965332,
      "learning_rate": 1.8480000000000001e-06,
      "loss": 2.9018,
      "step": 618
    },
    {
      "epoch": 1.1421975992613111,
      "grad_norm": 6.42782735824585,
      "learning_rate": 1.851e-06,
      "loss": 2.8909,
      "step": 619
    },
    {
      "epoch": 1.1440443213296398,
      "grad_norm": 7.556795120239258,
      "learning_rate": 1.8540000000000002e-06,
      "loss": 2.9127,
      "step": 620
    },
    {
      "epoch": 1.1458910433979685,
      "grad_norm": 9.342985153198242,
      "learning_rate": 1.857e-06,
      "loss": 2.885,
      "step": 621
    },
    {
      "epoch": 1.1477377654662972,
      "grad_norm": 9.603569984436035,
      "learning_rate": 1.86e-06,
      "loss": 2.8918,
      "step": 622
    },
    {
      "epoch": 1.149584487534626,
      "grad_norm": 15.569653511047363,
      "learning_rate": 1.8630000000000002e-06,
      "loss": 2.8943,
      "step": 623
    },
    {
      "epoch": 1.1514312096029546,
      "grad_norm": 8.585500717163086,
      "learning_rate": 1.866e-06,
      "loss": 2.889,
      "step": 624
    },
    {
      "epoch": 1.1532779316712836,
      "grad_norm": 4.9472150802612305,
      "learning_rate": 1.869e-06,
      "loss": 2.8983,
      "step": 625
    },
    {
      "epoch": 1.1551246537396123,
      "grad_norm": 14.335213661193848,
      "learning_rate": 1.872e-06,
      "loss": 2.8734,
      "step": 626
    },
    {
      "epoch": 1.156971375807941,
      "grad_norm": 10.756952285766602,
      "learning_rate": 1.875e-06,
      "loss": 2.9226,
      "step": 627
    },
    {
      "epoch": 1.1588180978762697,
      "grad_norm": 5.745858192443848,
      "learning_rate": 1.878e-06,
      "loss": 2.8554,
      "step": 628
    },
    {
      "epoch": 1.1606648199445984,
      "grad_norm": 8.257650375366211,
      "learning_rate": 1.8810000000000003e-06,
      "loss": 2.9119,
      "step": 629
    },
    {
      "epoch": 1.162511542012927,
      "grad_norm": 13.96142292022705,
      "learning_rate": 1.8839999999999999e-06,
      "loss": 2.8432,
      "step": 630
    },
    {
      "epoch": 1.1643582640812558,
      "grad_norm": 8.9722900390625,
      "learning_rate": 1.8869999999999999e-06,
      "loss": 2.9012,
      "step": 631
    },
    {
      "epoch": 1.1662049861495845,
      "grad_norm": 5.260127544403076,
      "learning_rate": 1.8900000000000001e-06,
      "loss": 2.852,
      "step": 632
    },
    {
      "epoch": 1.1680517082179132,
      "grad_norm": 12.734224319458008,
      "learning_rate": 1.8930000000000001e-06,
      "loss": 2.8762,
      "step": 633
    },
    {
      "epoch": 1.1698984302862419,
      "grad_norm": 10.812736511230469,
      "learning_rate": 1.8960000000000001e-06,
      "loss": 2.8813,
      "step": 634
    },
    {
      "epoch": 1.1717451523545706,
      "grad_norm": 7.148223876953125,
      "learning_rate": 1.899e-06,
      "loss": 2.8701,
      "step": 635
    },
    {
      "epoch": 1.1735918744228995,
      "grad_norm": 10.687859535217285,
      "learning_rate": 1.902e-06,
      "loss": 2.8745,
      "step": 636
    },
    {
      "epoch": 1.1754385964912282,
      "grad_norm": 14.53646469116211,
      "learning_rate": 1.905e-06,
      "loss": 2.881,
      "step": 637
    },
    {
      "epoch": 1.1772853185595569,
      "grad_norm": 9.041234970092773,
      "learning_rate": 1.908e-06,
      "loss": 2.8867,
      "step": 638
    },
    {
      "epoch": 1.1791320406278856,
      "grad_norm": 6.838634967803955,
      "learning_rate": 1.9110000000000004e-06,
      "loss": 2.8243,
      "step": 639
    },
    {
      "epoch": 1.1809787626962143,
      "grad_norm": 6.505585193634033,
      "learning_rate": 1.914e-06,
      "loss": 2.8369,
      "step": 640
    },
    {
      "epoch": 1.182825484764543,
      "grad_norm": 4.176347732543945,
      "learning_rate": 1.917e-06,
      "loss": 2.8408,
      "step": 641
    },
    {
      "epoch": 1.1846722068328717,
      "grad_norm": 9.723844528198242,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 2.8581,
      "step": 642
    },
    {
      "epoch": 1.1865189289012004,
      "grad_norm": 3.1684012413024902,
      "learning_rate": 1.923e-06,
      "loss": 2.9751,
      "step": 643
    },
    {
      "epoch": 1.188365650969529,
      "grad_norm": 3.2966151237487793,
      "learning_rate": 1.926e-06,
      "loss": 2.9243,
      "step": 644
    },
    {
      "epoch": 1.1902123730378578,
      "grad_norm": 3.4464704990386963,
      "learning_rate": 1.929e-06,
      "loss": 2.9082,
      "step": 645
    },
    {
      "epoch": 1.1920590951061865,
      "grad_norm": 2.4331507682800293,
      "learning_rate": 1.932e-06,
      "loss": 2.905,
      "step": 646
    },
    {
      "epoch": 1.1939058171745152,
      "grad_norm": 4.029451370239258,
      "learning_rate": 1.935e-06,
      "loss": 2.9147,
      "step": 647
    },
    {
      "epoch": 1.1957525392428439,
      "grad_norm": 15.655887603759766,
      "learning_rate": 1.9380000000000003e-06,
      "loss": 2.8587,
      "step": 648
    },
    {
      "epoch": 1.1975992613111726,
      "grad_norm": 4.774689197540283,
      "learning_rate": 1.9409999999999997e-06,
      "loss": 2.8423,
      "step": 649
    },
    {
      "epoch": 1.1994459833795015,
      "grad_norm": 3.9174821376800537,
      "learning_rate": 1.944e-06,
      "loss": 2.8417,
      "step": 650
    },
    {
      "epoch": 1.2012927054478302,
      "grad_norm": 2.8984720706939697,
      "learning_rate": 1.947e-06,
      "loss": 2.8605,
      "step": 651
    },
    {
      "epoch": 1.2031394275161589,
      "grad_norm": 5.679083824157715,
      "learning_rate": 1.95e-06,
      "loss": 2.8173,
      "step": 652
    },
    {
      "epoch": 1.2049861495844876,
      "grad_norm": 3.6044998168945312,
      "learning_rate": 1.953e-06,
      "loss": 2.8322,
      "step": 653
    },
    {
      "epoch": 1.2068328716528163,
      "grad_norm": 5.281006813049316,
      "learning_rate": 1.956e-06,
      "loss": 2.8368,
      "step": 654
    },
    {
      "epoch": 1.208679593721145,
      "grad_norm": 4.353389263153076,
      "learning_rate": 1.959e-06,
      "loss": 2.822,
      "step": 655
    },
    {
      "epoch": 1.2105263157894737,
      "grad_norm": 5.798543930053711,
      "learning_rate": 1.962e-06,
      "loss": 2.8404,
      "step": 656
    },
    {
      "epoch": 1.2123730378578024,
      "grad_norm": 5.485929489135742,
      "learning_rate": 1.9650000000000002e-06,
      "loss": 2.8379,
      "step": 657
    },
    {
      "epoch": 1.214219759926131,
      "grad_norm": 4.834137916564941,
      "learning_rate": 1.968e-06,
      "loss": 2.8366,
      "step": 658
    },
    {
      "epoch": 1.2160664819944598,
      "grad_norm": 5.665460586547852,
      "learning_rate": 1.971e-06,
      "loss": 2.8385,
      "step": 659
    },
    {
      "epoch": 1.2179132040627885,
      "grad_norm": 7.043797492980957,
      "learning_rate": 1.974e-06,
      "loss": 2.8221,
      "step": 660
    },
    {
      "epoch": 1.2197599261311174,
      "grad_norm": 5.808165073394775,
      "learning_rate": 1.977e-06,
      "loss": 2.8227,
      "step": 661
    },
    {
      "epoch": 1.221606648199446,
      "grad_norm": 21.553096771240234,
      "learning_rate": 1.98e-06,
      "loss": 2.8435,
      "step": 662
    },
    {
      "epoch": 1.2234533702677748,
      "grad_norm": 4.092555046081543,
      "learning_rate": 1.9830000000000003e-06,
      "loss": 2.8029,
      "step": 663
    },
    {
      "epoch": 1.2253000923361035,
      "grad_norm": 6.962827682495117,
      "learning_rate": 1.9859999999999997e-06,
      "loss": 2.8122,
      "step": 664
    },
    {
      "epoch": 1.2271468144044322,
      "grad_norm": 4.3983659744262695,
      "learning_rate": 1.989e-06,
      "loss": 2.833,
      "step": 665
    },
    {
      "epoch": 1.2289935364727609,
      "grad_norm": 4.870649814605713,
      "learning_rate": 1.992e-06,
      "loss": 2.8313,
      "step": 666
    },
    {
      "epoch": 1.2308402585410896,
      "grad_norm": 8.451695442199707,
      "learning_rate": 1.995e-06,
      "loss": 2.8211,
      "step": 667
    },
    {
      "epoch": 1.2326869806094183,
      "grad_norm": 9.19526481628418,
      "learning_rate": 1.998e-06,
      "loss": 2.8427,
      "step": 668
    },
    {
      "epoch": 1.234533702677747,
      "grad_norm": 7.379989147186279,
      "learning_rate": 2.001e-06,
      "loss": 2.8373,
      "step": 669
    },
    {
      "epoch": 1.2363804247460757,
      "grad_norm": 10.438669204711914,
      "learning_rate": 2.004e-06,
      "loss": 2.8293,
      "step": 670
    },
    {
      "epoch": 1.2382271468144044,
      "grad_norm": 3.9440643787384033,
      "learning_rate": 2.007e-06,
      "loss": 2.8231,
      "step": 671
    },
    {
      "epoch": 1.240073868882733,
      "grad_norm": 7.063518524169922,
      "learning_rate": 2.0100000000000002e-06,
      "loss": 2.846,
      "step": 672
    },
    {
      "epoch": 1.2419205909510618,
      "grad_norm": 5.866770267486572,
      "learning_rate": 2.0130000000000005e-06,
      "loss": 2.8227,
      "step": 673
    },
    {
      "epoch": 1.2437673130193905,
      "grad_norm": 4.861609935760498,
      "learning_rate": 2.016e-06,
      "loss": 2.8078,
      "step": 674
    },
    {
      "epoch": 1.2456140350877192,
      "grad_norm": 4.8793463706970215,
      "learning_rate": 2.019e-06,
      "loss": 2.8705,
      "step": 675
    },
    {
      "epoch": 1.247460757156048,
      "grad_norm": 5.221220016479492,
      "learning_rate": 2.0220000000000003e-06,
      "loss": 2.8138,
      "step": 676
    },
    {
      "epoch": 1.2493074792243768,
      "grad_norm": 6.798542022705078,
      "learning_rate": 2.025e-06,
      "loss": 2.795,
      "step": 677
    },
    {
      "epoch": 1.2511542012927055,
      "grad_norm": 6.260668754577637,
      "learning_rate": 2.028e-06,
      "loss": 2.8433,
      "step": 678
    },
    {
      "epoch": 1.2530009233610342,
      "grad_norm": 4.344798564910889,
      "learning_rate": 2.031e-06,
      "loss": 2.7893,
      "step": 679
    },
    {
      "epoch": 1.254847645429363,
      "grad_norm": 3.513256549835205,
      "learning_rate": 2.034e-06,
      "loss": 2.8129,
      "step": 680
    },
    {
      "epoch": 1.2566943674976916,
      "grad_norm": 5.4455718994140625,
      "learning_rate": 2.037e-06,
      "loss": 2.7959,
      "step": 681
    },
    {
      "epoch": 1.2585410895660203,
      "grad_norm": 9.517064094543457,
      "learning_rate": 2.0400000000000004e-06,
      "loss": 2.8195,
      "step": 682
    },
    {
      "epoch": 1.260387811634349,
      "grad_norm": 6.500871658325195,
      "learning_rate": 2.0429999999999998e-06,
      "loss": 2.8049,
      "step": 683
    },
    {
      "epoch": 1.2622345337026777,
      "grad_norm": 4.144886016845703,
      "learning_rate": 2.046e-06,
      "loss": 2.7864,
      "step": 684
    },
    {
      "epoch": 1.2640812557710064,
      "grad_norm": 5.43059778213501,
      "learning_rate": 2.049e-06,
      "loss": 2.7869,
      "step": 685
    },
    {
      "epoch": 1.2659279778393353,
      "grad_norm": 7.421367168426514,
      "learning_rate": 2.052e-06,
      "loss": 2.8101,
      "step": 686
    },
    {
      "epoch": 1.267774699907664,
      "grad_norm": 5.0669941902160645,
      "learning_rate": 2.0550000000000002e-06,
      "loss": 2.7979,
      "step": 687
    },
    {
      "epoch": 1.2696214219759927,
      "grad_norm": 5.999461650848389,
      "learning_rate": 2.058e-06,
      "loss": 2.7967,
      "step": 688
    },
    {
      "epoch": 1.2714681440443214,
      "grad_norm": 5.63998556137085,
      "learning_rate": 2.061e-06,
      "loss": 2.8188,
      "step": 689
    },
    {
      "epoch": 1.27331486611265,
      "grad_norm": 10.207146644592285,
      "learning_rate": 2.064e-06,
      "loss": 2.819,
      "step": 690
    },
    {
      "epoch": 1.2751615881809788,
      "grad_norm": 2.9560375213623047,
      "learning_rate": 2.0670000000000003e-06,
      "loss": 2.7793,
      "step": 691
    },
    {
      "epoch": 1.2770083102493075,
      "grad_norm": 3.9872069358825684,
      "learning_rate": 2.07e-06,
      "loss": 2.7723,
      "step": 692
    },
    {
      "epoch": 1.2788550323176362,
      "grad_norm": 9.619598388671875,
      "learning_rate": 2.073e-06,
      "loss": 2.927,
      "step": 693
    },
    {
      "epoch": 1.280701754385965,
      "grad_norm": 10.317651748657227,
      "learning_rate": 2.076e-06,
      "loss": 2.8825,
      "step": 694
    },
    {
      "epoch": 1.2825484764542936,
      "grad_norm": 4.086780071258545,
      "learning_rate": 2.079e-06,
      "loss": 2.8653,
      "step": 695
    },
    {
      "epoch": 1.2843951985226223,
      "grad_norm": 6.283045768737793,
      "learning_rate": 2.082e-06,
      "loss": 2.8542,
      "step": 696
    },
    {
      "epoch": 1.286241920590951,
      "grad_norm": 9.914119720458984,
      "learning_rate": 2.0850000000000004e-06,
      "loss": 2.8521,
      "step": 697
    },
    {
      "epoch": 1.2880886426592797,
      "grad_norm": 9.009072303771973,
      "learning_rate": 2.0879999999999997e-06,
      "loss": 2.861,
      "step": 698
    },
    {
      "epoch": 1.2899353647276084,
      "grad_norm": 5.214295864105225,
      "learning_rate": 2.091e-06,
      "loss": 2.8336,
      "step": 699
    },
    {
      "epoch": 1.291782086795937,
      "grad_norm": 4.523592472076416,
      "learning_rate": 2.094e-06,
      "loss": 2.7949,
      "step": 700
    },
    {
      "epoch": 1.293628808864266,
      "grad_norm": 5.53308629989624,
      "learning_rate": 2.097e-06,
      "loss": 2.8029,
      "step": 701
    },
    {
      "epoch": 1.2954755309325947,
      "grad_norm": 5.987646579742432,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 2.7991,
      "step": 702
    },
    {
      "epoch": 1.2973222530009234,
      "grad_norm": 6.248629570007324,
      "learning_rate": 2.103e-06,
      "loss": 2.7892,
      "step": 703
    },
    {
      "epoch": 1.299168975069252,
      "grad_norm": 2.0059845447540283,
      "learning_rate": 2.106e-06,
      "loss": 2.7865,
      "step": 704
    },
    {
      "epoch": 1.3010156971375808,
      "grad_norm": 4.6807098388671875,
      "learning_rate": 2.109e-06,
      "loss": 2.7815,
      "step": 705
    },
    {
      "epoch": 1.3028624192059095,
      "grad_norm": 3.9518628120422363,
      "learning_rate": 2.1120000000000003e-06,
      "loss": 2.8086,
      "step": 706
    },
    {
      "epoch": 1.3047091412742382,
      "grad_norm": 4.064188003540039,
      "learning_rate": 2.1149999999999997e-06,
      "loss": 2.7868,
      "step": 707
    },
    {
      "epoch": 1.306555863342567,
      "grad_norm": 3.1247639656066895,
      "learning_rate": 2.118e-06,
      "loss": 2.7812,
      "step": 708
    },
    {
      "epoch": 1.3084025854108956,
      "grad_norm": 4.630458831787109,
      "learning_rate": 2.121e-06,
      "loss": 2.7804,
      "step": 709
    },
    {
      "epoch": 1.3102493074792243,
      "grad_norm": 22.884502410888672,
      "learning_rate": 2.124e-06,
      "loss": 2.748,
      "step": 710
    },
    {
      "epoch": 1.3120960295475532,
      "grad_norm": 4.427460670471191,
      "learning_rate": 2.127e-06,
      "loss": 2.8104,
      "step": 711
    },
    {
      "epoch": 1.313942751615882,
      "grad_norm": 4.474238872528076,
      "learning_rate": 2.13e-06,
      "loss": 2.7915,
      "step": 712
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 2.491396903991699,
      "learning_rate": 2.133e-06,
      "loss": 2.7305,
      "step": 713
    },
    {
      "epoch": 1.3176361957525393,
      "grad_norm": 4.121035099029541,
      "learning_rate": 2.136e-06,
      "loss": 2.7733,
      "step": 714
    },
    {
      "epoch": 1.319482917820868,
      "grad_norm": 4.982003211975098,
      "learning_rate": 2.139e-06,
      "loss": 2.7607,
      "step": 715
    },
    {
      "epoch": 1.3213296398891967,
      "grad_norm": 2.3569273948669434,
      "learning_rate": 2.1420000000000004e-06,
      "loss": 2.7735,
      "step": 716
    },
    {
      "epoch": 1.3231763619575254,
      "grad_norm": 4.553219318389893,
      "learning_rate": 2.145e-06,
      "loss": 2.7653,
      "step": 717
    },
    {
      "epoch": 1.3250230840258541,
      "grad_norm": 3.5110154151916504,
      "learning_rate": 2.148e-06,
      "loss": 2.7853,
      "step": 718
    },
    {
      "epoch": 1.3268698060941828,
      "grad_norm": 3.3325610160827637,
      "learning_rate": 2.1510000000000002e-06,
      "loss": 2.7661,
      "step": 719
    },
    {
      "epoch": 1.3287165281625115,
      "grad_norm": 5.185281753540039,
      "learning_rate": 2.154e-06,
      "loss": 2.7715,
      "step": 720
    },
    {
      "epoch": 1.3305632502308402,
      "grad_norm": 5.594532489776611,
      "learning_rate": 2.1570000000000003e-06,
      "loss": 2.7784,
      "step": 721
    },
    {
      "epoch": 1.332409972299169,
      "grad_norm": 5.433917999267578,
      "learning_rate": 2.16e-06,
      "loss": 2.7305,
      "step": 722
    },
    {
      "epoch": 1.3342566943674976,
      "grad_norm": 4.988367080688477,
      "learning_rate": 2.163e-06,
      "loss": 2.7773,
      "step": 723
    },
    {
      "epoch": 1.3361034164358263,
      "grad_norm": 3.9520010948181152,
      "learning_rate": 2.166e-06,
      "loss": 2.7538,
      "step": 724
    },
    {
      "epoch": 1.337950138504155,
      "grad_norm": 6.544180393218994,
      "learning_rate": 2.1690000000000003e-06,
      "loss": 2.7799,
      "step": 725
    },
    {
      "epoch": 1.3397968605724837,
      "grad_norm": 7.745989799499512,
      "learning_rate": 2.172e-06,
      "loss": 2.7307,
      "step": 726
    },
    {
      "epoch": 1.3416435826408126,
      "grad_norm": 4.38186502456665,
      "learning_rate": 2.175e-06,
      "loss": 2.8229,
      "step": 727
    },
    {
      "epoch": 1.3434903047091413,
      "grad_norm": 3.6738603115081787,
      "learning_rate": 2.178e-06,
      "loss": 2.7445,
      "step": 728
    },
    {
      "epoch": 1.34533702677747,
      "grad_norm": 4.38754940032959,
      "learning_rate": 2.181e-06,
      "loss": 2.7487,
      "step": 729
    },
    {
      "epoch": 1.3471837488457987,
      "grad_norm": 3.183619260787964,
      "learning_rate": 2.184e-06,
      "loss": 2.75,
      "step": 730
    },
    {
      "epoch": 1.3490304709141274,
      "grad_norm": 6.966119289398193,
      "learning_rate": 2.1870000000000004e-06,
      "loss": 2.765,
      "step": 731
    },
    {
      "epoch": 1.3508771929824561,
      "grad_norm": 4.263189792633057,
      "learning_rate": 2.1899999999999998e-06,
      "loss": 2.7622,
      "step": 732
    },
    {
      "epoch": 1.3527239150507848,
      "grad_norm": 2.978217124938965,
      "learning_rate": 2.193e-06,
      "loss": 2.7429,
      "step": 733
    },
    {
      "epoch": 1.3545706371191135,
      "grad_norm": 4.708694934844971,
      "learning_rate": 2.1960000000000002e-06,
      "loss": 2.7563,
      "step": 734
    },
    {
      "epoch": 1.3564173591874422,
      "grad_norm": 4.529090881347656,
      "learning_rate": 2.199e-06,
      "loss": 2.7436,
      "step": 735
    },
    {
      "epoch": 1.3582640812557711,
      "grad_norm": 4.785562992095947,
      "learning_rate": 2.2020000000000003e-06,
      "loss": 2.7406,
      "step": 736
    },
    {
      "epoch": 1.3601108033240998,
      "grad_norm": 10.316986083984375,
      "learning_rate": 2.205e-06,
      "loss": 2.7357,
      "step": 737
    },
    {
      "epoch": 1.3619575253924285,
      "grad_norm": 4.90836238861084,
      "learning_rate": 2.208e-06,
      "loss": 2.7566,
      "step": 738
    },
    {
      "epoch": 1.3638042474607572,
      "grad_norm": 8.626869201660156,
      "learning_rate": 2.211e-06,
      "loss": 2.7523,
      "step": 739
    },
    {
      "epoch": 1.365650969529086,
      "grad_norm": 11.956260681152344,
      "learning_rate": 2.2140000000000003e-06,
      "loss": 2.7543,
      "step": 740
    },
    {
      "epoch": 1.3674976915974146,
      "grad_norm": 5.312864303588867,
      "learning_rate": 2.2169999999999997e-06,
      "loss": 2.7477,
      "step": 741
    },
    {
      "epoch": 1.3693444136657433,
      "grad_norm": 4.989658355712891,
      "learning_rate": 2.22e-06,
      "loss": 2.7476,
      "step": 742
    },
    {
      "epoch": 1.371191135734072,
      "grad_norm": 6.6182637214660645,
      "learning_rate": 2.223e-06,
      "loss": 2.8991,
      "step": 743
    },
    {
      "epoch": 1.3730378578024007,
      "grad_norm": 6.964872360229492,
      "learning_rate": 2.226e-06,
      "loss": 2.8556,
      "step": 744
    },
    {
      "epoch": 1.3748845798707294,
      "grad_norm": 3.2420098781585693,
      "learning_rate": 2.229e-06,
      "loss": 2.7974,
      "step": 745
    },
    {
      "epoch": 1.3767313019390581,
      "grad_norm": 9.735184669494629,
      "learning_rate": 2.232e-06,
      "loss": 2.8501,
      "step": 746
    },
    {
      "epoch": 1.3785780240073868,
      "grad_norm": 10.228830337524414,
      "learning_rate": 2.2349999999999998e-06,
      "loss": 2.8392,
      "step": 747
    },
    {
      "epoch": 1.3804247460757155,
      "grad_norm": 10.16733455657959,
      "learning_rate": 2.238e-06,
      "loss": 2.8219,
      "step": 748
    },
    {
      "epoch": 1.3822714681440442,
      "grad_norm": 8.562270164489746,
      "learning_rate": 2.2410000000000002e-06,
      "loss": 2.7874,
      "step": 749
    },
    {
      "epoch": 1.384118190212373,
      "grad_norm": 7.701913833618164,
      "learning_rate": 2.244e-06,
      "loss": 2.7978,
      "step": 750
    },
    {
      "epoch": 1.3859649122807016,
      "grad_norm": 8.914648056030273,
      "learning_rate": 2.247e-06,
      "loss": 2.7869,
      "step": 751
    },
    {
      "epoch": 1.3878116343490305,
      "grad_norm": 11.16434383392334,
      "learning_rate": 2.25e-06,
      "loss": 2.7883,
      "step": 752
    },
    {
      "epoch": 1.3896583564173592,
      "grad_norm": 10.750763893127441,
      "learning_rate": 2.253e-06,
      "loss": 2.7952,
      "step": 753
    },
    {
      "epoch": 1.391505078485688,
      "grad_norm": 7.983006954193115,
      "learning_rate": 2.256e-06,
      "loss": 2.77,
      "step": 754
    },
    {
      "epoch": 1.3933518005540166,
      "grad_norm": 4.1849188804626465,
      "learning_rate": 2.2590000000000003e-06,
      "loss": 2.736,
      "step": 755
    },
    {
      "epoch": 1.3951985226223453,
      "grad_norm": 5.593008995056152,
      "learning_rate": 2.262e-06,
      "loss": 2.7679,
      "step": 756
    },
    {
      "epoch": 1.397045244690674,
      "grad_norm": 8.314484596252441,
      "learning_rate": 2.265e-06,
      "loss": 2.7588,
      "step": 757
    },
    {
      "epoch": 1.3988919667590027,
      "grad_norm": 10.190176010131836,
      "learning_rate": 2.268e-06,
      "loss": 2.7147,
      "step": 758
    },
    {
      "epoch": 1.4007386888273314,
      "grad_norm": 6.350856304168701,
      "learning_rate": 2.2710000000000004e-06,
      "loss": 2.726,
      "step": 759
    },
    {
      "epoch": 1.4025854108956601,
      "grad_norm": 2.3066318035125732,
      "learning_rate": 2.274e-06,
      "loss": 2.6919,
      "step": 760
    },
    {
      "epoch": 1.404432132963989,
      "grad_norm": 3.782069683074951,
      "learning_rate": 2.277e-06,
      "loss": 2.7338,
      "step": 761
    },
    {
      "epoch": 1.4062788550323178,
      "grad_norm": 4.446532726287842,
      "learning_rate": 2.28e-06,
      "loss": 2.7093,
      "step": 762
    },
    {
      "epoch": 1.4081255771006465,
      "grad_norm": 6.186867713928223,
      "learning_rate": 2.283e-06,
      "loss": 2.7388,
      "step": 763
    },
    {
      "epoch": 1.4099722991689752,
      "grad_norm": 5.769529819488525,
      "learning_rate": 2.2860000000000002e-06,
      "loss": 2.6958,
      "step": 764
    },
    {
      "epoch": 1.4118190212373039,
      "grad_norm": 4.115484237670898,
      "learning_rate": 2.2890000000000004e-06,
      "loss": 2.719,
      "step": 765
    },
    {
      "epoch": 1.4136657433056325,
      "grad_norm": 5.355208873748779,
      "learning_rate": 2.292e-06,
      "loss": 2.7121,
      "step": 766
    },
    {
      "epoch": 1.4155124653739612,
      "grad_norm": 4.1041035652160645,
      "learning_rate": 2.295e-06,
      "loss": 2.6666,
      "step": 767
    },
    {
      "epoch": 1.41735918744229,
      "grad_norm": 6.766019344329834,
      "learning_rate": 2.2980000000000003e-06,
      "loss": 2.7147,
      "step": 768
    },
    {
      "epoch": 1.4192059095106186,
      "grad_norm": 8.364627838134766,
      "learning_rate": 2.301e-06,
      "loss": 2.7387,
      "step": 769
    },
    {
      "epoch": 1.4210526315789473,
      "grad_norm": 5.883131504058838,
      "learning_rate": 2.304e-06,
      "loss": 2.7544,
      "step": 770
    },
    {
      "epoch": 1.422899353647276,
      "grad_norm": 5.160414695739746,
      "learning_rate": 2.307e-06,
      "loss": 2.7187,
      "step": 771
    },
    {
      "epoch": 1.4247460757156047,
      "grad_norm": 6.408952236175537,
      "learning_rate": 2.31e-06,
      "loss": 2.7144,
      "step": 772
    },
    {
      "epoch": 1.4265927977839334,
      "grad_norm": 4.58326530456543,
      "learning_rate": 2.313e-06,
      "loss": 2.6918,
      "step": 773
    },
    {
      "epoch": 1.4284395198522621,
      "grad_norm": 4.147642135620117,
      "learning_rate": 2.3160000000000004e-06,
      "loss": 2.7173,
      "step": 774
    },
    {
      "epoch": 1.4302862419205908,
      "grad_norm": 5.40259313583374,
      "learning_rate": 2.3189999999999997e-06,
      "loss": 2.7323,
      "step": 775
    },
    {
      "epoch": 1.4321329639889195,
      "grad_norm": 9.753005981445312,
      "learning_rate": 2.322e-06,
      "loss": 2.7337,
      "step": 776
    },
    {
      "epoch": 1.4339796860572485,
      "grad_norm": 4.117685794830322,
      "learning_rate": 2.325e-06,
      "loss": 2.7285,
      "step": 777
    },
    {
      "epoch": 1.4358264081255772,
      "grad_norm": 6.1139421463012695,
      "learning_rate": 2.328e-06,
      "loss": 2.7262,
      "step": 778
    },
    {
      "epoch": 1.4376731301939059,
      "grad_norm": 4.455411434173584,
      "learning_rate": 2.3310000000000002e-06,
      "loss": 2.7265,
      "step": 779
    },
    {
      "epoch": 1.4395198522622346,
      "grad_norm": 4.577296733856201,
      "learning_rate": 2.334e-06,
      "loss": 2.68,
      "step": 780
    },
    {
      "epoch": 1.4413665743305633,
      "grad_norm": 3.3163914680480957,
      "learning_rate": 2.337e-06,
      "loss": 2.7446,
      "step": 781
    },
    {
      "epoch": 1.443213296398892,
      "grad_norm": 9.669931411743164,
      "learning_rate": 2.34e-06,
      "loss": 2.7558,
      "step": 782
    },
    {
      "epoch": 1.4450600184672207,
      "grad_norm": 5.4542555809021,
      "learning_rate": 2.3430000000000003e-06,
      "loss": 2.7302,
      "step": 783
    },
    {
      "epoch": 1.4469067405355494,
      "grad_norm": 9.963101387023926,
      "learning_rate": 2.346e-06,
      "loss": 2.7001,
      "step": 784
    },
    {
      "epoch": 1.448753462603878,
      "grad_norm": 11.758763313293457,
      "learning_rate": 2.349e-06,
      "loss": 2.6938,
      "step": 785
    },
    {
      "epoch": 1.450600184672207,
      "grad_norm": 5.225876808166504,
      "learning_rate": 2.352e-06,
      "loss": 2.7307,
      "step": 786
    },
    {
      "epoch": 1.4524469067405357,
      "grad_norm": 14.65092945098877,
      "learning_rate": 2.355e-06,
      "loss": 2.6993,
      "step": 787
    },
    {
      "epoch": 1.4542936288088644,
      "grad_norm": 10.20727825164795,
      "learning_rate": 2.358e-06,
      "loss": 2.7313,
      "step": 788
    },
    {
      "epoch": 1.456140350877193,
      "grad_norm": 5.7672505378723145,
      "learning_rate": 2.3610000000000003e-06,
      "loss": 2.7177,
      "step": 789
    },
    {
      "epoch": 1.4579870729455218,
      "grad_norm": 9.435592651367188,
      "learning_rate": 2.3639999999999997e-06,
      "loss": 2.7246,
      "step": 790
    },
    {
      "epoch": 1.4598337950138505,
      "grad_norm": 26.1789608001709,
      "learning_rate": 2.367e-06,
      "loss": 2.745,
      "step": 791
    },
    {
      "epoch": 1.4616805170821792,
      "grad_norm": 3.997400999069214,
      "learning_rate": 2.37e-06,
      "loss": 2.7087,
      "step": 792
    },
    {
      "epoch": 1.4635272391505079,
      "grad_norm": 3.2992331981658936,
      "learning_rate": 2.373e-06,
      "loss": 2.8327,
      "step": 793
    },
    {
      "epoch": 1.4653739612188366,
      "grad_norm": 4.294439792633057,
      "learning_rate": 2.376e-06,
      "loss": 2.8169,
      "step": 794
    },
    {
      "epoch": 1.4672206832871653,
      "grad_norm": 4.262382507324219,
      "learning_rate": 2.379e-06,
      "loss": 2.7942,
      "step": 795
    },
    {
      "epoch": 1.469067405355494,
      "grad_norm": 6.730347633361816,
      "learning_rate": 2.382e-06,
      "loss": 2.8166,
      "step": 796
    },
    {
      "epoch": 1.4709141274238227,
      "grad_norm": 4.085928440093994,
      "learning_rate": 2.385e-06,
      "loss": 2.788,
      "step": 797
    },
    {
      "epoch": 1.4727608494921514,
      "grad_norm": 3.401494264602661,
      "learning_rate": 2.3880000000000003e-06,
      "loss": 2.7739,
      "step": 798
    },
    {
      "epoch": 1.47460757156048,
      "grad_norm": 6.391687870025635,
      "learning_rate": 2.391e-06,
      "loss": 2.771,
      "step": 799
    },
    {
      "epoch": 1.4764542936288088,
      "grad_norm": 9.127158164978027,
      "learning_rate": 2.394e-06,
      "loss": 2.765,
      "step": 800
    },
    {
      "epoch": 1.4783010156971375,
      "grad_norm": 7.535577774047852,
      "learning_rate": 2.397e-06,
      "loss": 2.755,
      "step": 801
    },
    {
      "epoch": 1.4801477377654664,
      "grad_norm": 2.549172878265381,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 2.7157,
      "step": 802
    },
    {
      "epoch": 1.481994459833795,
      "grad_norm": 4.9487128257751465,
      "learning_rate": 2.403e-06,
      "loss": 2.7071,
      "step": 803
    },
    {
      "epoch": 1.4838411819021238,
      "grad_norm": 4.3724799156188965,
      "learning_rate": 2.406e-06,
      "loss": 2.7335,
      "step": 804
    },
    {
      "epoch": 1.4856879039704525,
      "grad_norm": 5.787751197814941,
      "learning_rate": 2.409e-06,
      "loss": 2.7254,
      "step": 805
    },
    {
      "epoch": 1.4875346260387812,
      "grad_norm": 3.1072328090667725,
      "learning_rate": 2.412e-06,
      "loss": 2.723,
      "step": 806
    },
    {
      "epoch": 1.4893813481071099,
      "grad_norm": 4.013876438140869,
      "learning_rate": 2.415e-06,
      "loss": 2.6765,
      "step": 807
    },
    {
      "epoch": 1.4912280701754386,
      "grad_norm": 5.162932872772217,
      "learning_rate": 2.4180000000000004e-06,
      "loss": 2.6791,
      "step": 808
    },
    {
      "epoch": 1.4930747922437673,
      "grad_norm": 3.466583728790283,
      "learning_rate": 2.4209999999999998e-06,
      "loss": 2.6731,
      "step": 809
    },
    {
      "epoch": 1.494921514312096,
      "grad_norm": 5.3154191970825195,
      "learning_rate": 2.424e-06,
      "loss": 2.7359,
      "step": 810
    },
    {
      "epoch": 1.4967682363804247,
      "grad_norm": 6.798288822174072,
      "learning_rate": 2.4270000000000002e-06,
      "loss": 2.7224,
      "step": 811
    },
    {
      "epoch": 1.4986149584487536,
      "grad_norm": 7.94334077835083,
      "learning_rate": 2.43e-06,
      "loss": 2.7205,
      "step": 812
    },
    {
      "epoch": 1.5004616805170823,
      "grad_norm": 6.905054092407227,
      "learning_rate": 2.4330000000000003e-06,
      "loss": 2.6951,
      "step": 813
    },
    {
      "epoch": 1.502308402585411,
      "grad_norm": 5.6237688064575195,
      "learning_rate": 2.436e-06,
      "loss": 2.6836,
      "step": 814
    },
    {
      "epoch": 1.5041551246537397,
      "grad_norm": 6.948147773742676,
      "learning_rate": 2.439e-06,
      "loss": 2.6905,
      "step": 815
    },
    {
      "epoch": 1.5060018467220684,
      "grad_norm": 7.7601847648620605,
      "learning_rate": 2.442e-06,
      "loss": 2.7105,
      "step": 816
    },
    {
      "epoch": 1.507848568790397,
      "grad_norm": 6.125829696655273,
      "learning_rate": 2.4450000000000003e-06,
      "loss": 2.6873,
      "step": 817
    },
    {
      "epoch": 1.5096952908587258,
      "grad_norm": 9.361186981201172,
      "learning_rate": 2.448e-06,
      "loss": 2.6749,
      "step": 818
    },
    {
      "epoch": 1.5115420129270545,
      "grad_norm": 5.572545528411865,
      "learning_rate": 2.451e-06,
      "loss": 2.6779,
      "step": 819
    },
    {
      "epoch": 1.5133887349953832,
      "grad_norm": 9.183470726013184,
      "learning_rate": 2.454e-06,
      "loss": 2.7265,
      "step": 820
    },
    {
      "epoch": 1.5152354570637119,
      "grad_norm": 5.680612087249756,
      "learning_rate": 2.457e-06,
      "loss": 2.6752,
      "step": 821
    },
    {
      "epoch": 1.5170821791320406,
      "grad_norm": 4.805209159851074,
      "learning_rate": 2.46e-06,
      "loss": 2.6951,
      "step": 822
    },
    {
      "epoch": 1.5189289012003693,
      "grad_norm": 7.211028099060059,
      "learning_rate": 2.4630000000000004e-06,
      "loss": 2.6981,
      "step": 823
    },
    {
      "epoch": 1.520775623268698,
      "grad_norm": 6.059957504272461,
      "learning_rate": 2.4659999999999998e-06,
      "loss": 2.6957,
      "step": 824
    },
    {
      "epoch": 1.5226223453370267,
      "grad_norm": 4.662355899810791,
      "learning_rate": 2.469e-06,
      "loss": 2.6932,
      "step": 825
    },
    {
      "epoch": 1.5244690674053554,
      "grad_norm": 9.695551872253418,
      "learning_rate": 2.4720000000000002e-06,
      "loss": 2.698,
      "step": 826
    },
    {
      "epoch": 1.526315789473684,
      "grad_norm": 6.460937976837158,
      "learning_rate": 2.475e-06,
      "loss": 2.6918,
      "step": 827
    },
    {
      "epoch": 1.5281625115420128,
      "grad_norm": 4.860535144805908,
      "learning_rate": 2.4780000000000002e-06,
      "loss": 2.6999,
      "step": 828
    },
    {
      "epoch": 1.5300092336103417,
      "grad_norm": 3.691718578338623,
      "learning_rate": 2.481e-06,
      "loss": 2.6793,
      "step": 829
    },
    {
      "epoch": 1.5318559556786704,
      "grad_norm": 6.444655895233154,
      "learning_rate": 2.484e-06,
      "loss": 2.6883,
      "step": 830
    },
    {
      "epoch": 1.533702677746999,
      "grad_norm": 4.39552116394043,
      "learning_rate": 2.487e-06,
      "loss": 2.6928,
      "step": 831
    },
    {
      "epoch": 1.5355493998153278,
      "grad_norm": 4.5506696701049805,
      "learning_rate": 2.4900000000000003e-06,
      "loss": 2.7049,
      "step": 832
    },
    {
      "epoch": 1.5373961218836565,
      "grad_norm": 3.6617109775543213,
      "learning_rate": 2.4929999999999997e-06,
      "loss": 2.698,
      "step": 833
    },
    {
      "epoch": 1.5392428439519852,
      "grad_norm": 2.4751598834991455,
      "learning_rate": 2.496e-06,
      "loss": 2.6257,
      "step": 834
    },
    {
      "epoch": 1.541089566020314,
      "grad_norm": 3.382323980331421,
      "learning_rate": 2.499e-06,
      "loss": 2.7211,
      "step": 835
    },
    {
      "epoch": 1.5429362880886428,
      "grad_norm": 5.438704967498779,
      "learning_rate": 2.502e-06,
      "loss": 2.6868,
      "step": 836
    },
    {
      "epoch": 1.5447830101569715,
      "grad_norm": 7.3822784423828125,
      "learning_rate": 2.505e-06,
      "loss": 2.6893,
      "step": 837
    },
    {
      "epoch": 1.5466297322253002,
      "grad_norm": 3.2348599433898926,
      "learning_rate": 2.508e-06,
      "loss": 2.6683,
      "step": 838
    },
    {
      "epoch": 1.548476454293629,
      "grad_norm": 4.4207377433776855,
      "learning_rate": 2.5109999999999998e-06,
      "loss": 2.6667,
      "step": 839
    },
    {
      "epoch": 1.5503231763619576,
      "grad_norm": 2.893880844116211,
      "learning_rate": 2.514e-06,
      "loss": 2.6793,
      "step": 840
    },
    {
      "epoch": 1.5521698984302863,
      "grad_norm": 3.3121488094329834,
      "learning_rate": 2.517e-06,
      "loss": 2.6794,
      "step": 841
    },
    {
      "epoch": 1.554016620498615,
      "grad_norm": 4.944551944732666,
      "learning_rate": 2.52e-06,
      "loss": 2.6549,
      "step": 842
    },
    {
      "epoch": 1.5558633425669437,
      "grad_norm": 4.086531162261963,
      "learning_rate": 2.523e-06,
      "loss": 2.8968,
      "step": 843
    },
    {
      "epoch": 1.5577100646352724,
      "grad_norm": 2.7039895057678223,
      "learning_rate": 2.526e-06,
      "loss": 2.8208,
      "step": 844
    },
    {
      "epoch": 1.559556786703601,
      "grad_norm": 2.362365245819092,
      "learning_rate": 2.5290000000000003e-06,
      "loss": 2.7717,
      "step": 845
    },
    {
      "epoch": 1.5614035087719298,
      "grad_norm": 5.044862747192383,
      "learning_rate": 2.532e-06,
      "loss": 2.7733,
      "step": 846
    },
    {
      "epoch": 1.5632502308402585,
      "grad_norm": 5.624716281890869,
      "learning_rate": 2.5350000000000003e-06,
      "loss": 2.7638,
      "step": 847
    },
    {
      "epoch": 1.5650969529085872,
      "grad_norm": 6.3323163986206055,
      "learning_rate": 2.538e-06,
      "loss": 2.7417,
      "step": 848
    },
    {
      "epoch": 1.566943674976916,
      "grad_norm": 4.66796350479126,
      "learning_rate": 2.541e-06,
      "loss": 2.707,
      "step": 849
    },
    {
      "epoch": 1.5687903970452446,
      "grad_norm": 6.403055191040039,
      "learning_rate": 2.544e-06,
      "loss": 2.7194,
      "step": 850
    },
    {
      "epoch": 1.5706371191135733,
      "grad_norm": 5.171673774719238,
      "learning_rate": 2.5470000000000003e-06,
      "loss": 2.7007,
      "step": 851
    },
    {
      "epoch": 1.572483841181902,
      "grad_norm": 3.812385320663452,
      "learning_rate": 2.55e-06,
      "loss": 2.7044,
      "step": 852
    },
    {
      "epoch": 1.5743305632502307,
      "grad_norm": 8.64167594909668,
      "learning_rate": 2.553e-06,
      "loss": 2.7368,
      "step": 853
    },
    {
      "epoch": 1.5761772853185596,
      "grad_norm": 6.886471271514893,
      "learning_rate": 2.556e-06,
      "loss": 2.6865,
      "step": 854
    },
    {
      "epoch": 1.5780240073868883,
      "grad_norm": 4.122204303741455,
      "learning_rate": 2.559e-06,
      "loss": 2.6864,
      "step": 855
    },
    {
      "epoch": 1.579870729455217,
      "grad_norm": 5.5636067390441895,
      "learning_rate": 2.562e-06,
      "loss": 2.6743,
      "step": 856
    },
    {
      "epoch": 1.5817174515235457,
      "grad_norm": 4.138462543487549,
      "learning_rate": 2.5650000000000004e-06,
      "loss": 2.6896,
      "step": 857
    },
    {
      "epoch": 1.5835641735918744,
      "grad_norm": 3.9760003089904785,
      "learning_rate": 2.568e-06,
      "loss": 2.6389,
      "step": 858
    },
    {
      "epoch": 1.585410895660203,
      "grad_norm": 3.1889688968658447,
      "learning_rate": 2.571e-06,
      "loss": 2.6655,
      "step": 859
    },
    {
      "epoch": 1.587257617728532,
      "grad_norm": 5.70900821685791,
      "learning_rate": 2.5740000000000003e-06,
      "loss": 2.6764,
      "step": 860
    },
    {
      "epoch": 1.5891043397968607,
      "grad_norm": 4.828238487243652,
      "learning_rate": 2.577e-06,
      "loss": 2.6499,
      "step": 861
    },
    {
      "epoch": 1.5909510618651894,
      "grad_norm": 3.9876034259796143,
      "learning_rate": 2.58e-06,
      "loss": 2.6698,
      "step": 862
    },
    {
      "epoch": 1.5927977839335181,
      "grad_norm": 3.536850690841675,
      "learning_rate": 2.583e-06,
      "loss": 2.6383,
      "step": 863
    },
    {
      "epoch": 1.5946445060018468,
      "grad_norm": 3.826817035675049,
      "learning_rate": 2.586e-06,
      "loss": 2.6255,
      "step": 864
    },
    {
      "epoch": 1.5964912280701755,
      "grad_norm": 6.136444091796875,
      "learning_rate": 2.589e-06,
      "loss": 2.6355,
      "step": 865
    },
    {
      "epoch": 1.5983379501385042,
      "grad_norm": 3.2077252864837646,
      "learning_rate": 2.5920000000000003e-06,
      "loss": 2.677,
      "step": 866
    },
    {
      "epoch": 1.600184672206833,
      "grad_norm": 3.540379762649536,
      "learning_rate": 2.5949999999999997e-06,
      "loss": 2.6584,
      "step": 867
    },
    {
      "epoch": 1.6020313942751616,
      "grad_norm": 3.182039737701416,
      "learning_rate": 2.598e-06,
      "loss": 2.6294,
      "step": 868
    },
    {
      "epoch": 1.6038781163434903,
      "grad_norm": 2.667015314102173,
      "learning_rate": 2.601e-06,
      "loss": 2.6402,
      "step": 869
    },
    {
      "epoch": 1.605724838411819,
      "grad_norm": 4.8609938621521,
      "learning_rate": 2.604e-06,
      "loss": 2.6315,
      "step": 870
    },
    {
      "epoch": 1.6075715604801477,
      "grad_norm": 3.460028648376465,
      "learning_rate": 2.607e-06,
      "loss": 2.5968,
      "step": 871
    },
    {
      "epoch": 1.6094182825484764,
      "grad_norm": 2.8984220027923584,
      "learning_rate": 2.61e-06,
      "loss": 2.6517,
      "step": 872
    },
    {
      "epoch": 1.611265004616805,
      "grad_norm": 5.12603759765625,
      "learning_rate": 2.613e-06,
      "loss": 2.5993,
      "step": 873
    },
    {
      "epoch": 1.6131117266851338,
      "grad_norm": 4.4763641357421875,
      "learning_rate": 2.616e-06,
      "loss": 2.6411,
      "step": 874
    },
    {
      "epoch": 1.6149584487534625,
      "grad_norm": 3.163438558578491,
      "learning_rate": 2.6190000000000003e-06,
      "loss": 2.619,
      "step": 875
    },
    {
      "epoch": 1.6168051708217912,
      "grad_norm": 5.125920295715332,
      "learning_rate": 2.622e-06,
      "loss": 2.609,
      "step": 876
    },
    {
      "epoch": 1.61865189289012,
      "grad_norm": 3.5490097999572754,
      "learning_rate": 2.625e-06,
      "loss": 2.6037,
      "step": 877
    },
    {
      "epoch": 1.6204986149584486,
      "grad_norm": 6.430298805236816,
      "learning_rate": 2.628e-06,
      "loss": 2.6346,
      "step": 878
    },
    {
      "epoch": 1.6223453370267775,
      "grad_norm": 3.71742844581604,
      "learning_rate": 2.631e-06,
      "loss": 2.6379,
      "step": 879
    },
    {
      "epoch": 1.6241920590951062,
      "grad_norm": 3.395528554916382,
      "learning_rate": 2.634e-06,
      "loss": 2.6065,
      "step": 880
    },
    {
      "epoch": 1.626038781163435,
      "grad_norm": 5.155135631561279,
      "learning_rate": 2.6370000000000003e-06,
      "loss": 2.6013,
      "step": 881
    },
    {
      "epoch": 1.6278855032317636,
      "grad_norm": 4.800289630889893,
      "learning_rate": 2.6399999999999997e-06,
      "loss": 2.6259,
      "step": 882
    },
    {
      "epoch": 1.6297322253000923,
      "grad_norm": 5.167157173156738,
      "learning_rate": 2.643e-06,
      "loss": 2.6065,
      "step": 883
    },
    {
      "epoch": 1.631578947368421,
      "grad_norm": 5.212462902069092,
      "learning_rate": 2.646e-06,
      "loss": 2.6042,
      "step": 884
    },
    {
      "epoch": 1.6334256694367497,
      "grad_norm": 3.7768590450286865,
      "learning_rate": 2.649e-06,
      "loss": 2.5636,
      "step": 885
    },
    {
      "epoch": 1.6352723915050786,
      "grad_norm": 3.708564281463623,
      "learning_rate": 2.652e-06,
      "loss": 2.624,
      "step": 886
    },
    {
      "epoch": 1.6371191135734073,
      "grad_norm": 3.8434557914733887,
      "learning_rate": 2.655e-06,
      "loss": 2.5539,
      "step": 887
    },
    {
      "epoch": 1.638965835641736,
      "grad_norm": 5.406363487243652,
      "learning_rate": 2.6580000000000002e-06,
      "loss": 2.5992,
      "step": 888
    },
    {
      "epoch": 1.6408125577100647,
      "grad_norm": 3.292588710784912,
      "learning_rate": 2.661e-06,
      "loss": 2.6162,
      "step": 889
    },
    {
      "epoch": 1.6426592797783934,
      "grad_norm": 6.086481094360352,
      "learning_rate": 2.6640000000000002e-06,
      "loss": 2.6225,
      "step": 890
    },
    {
      "epoch": 1.6445060018467221,
      "grad_norm": 5.598721504211426,
      "learning_rate": 2.6670000000000005e-06,
      "loss": 2.6216,
      "step": 891
    },
    {
      "epoch": 1.6463527239150508,
      "grad_norm": 4.04868221282959,
      "learning_rate": 2.67e-06,
      "loss": 2.5437,
      "step": 892
    },
    {
      "epoch": 1.6481994459833795,
      "grad_norm": 4.556248188018799,
      "learning_rate": 2.673e-06,
      "loss": 2.7949,
      "step": 893
    },
    {
      "epoch": 1.6500461680517082,
      "grad_norm": 3.4863927364349365,
      "learning_rate": 2.6760000000000003e-06,
      "loss": 2.7995,
      "step": 894
    },
    {
      "epoch": 1.651892890120037,
      "grad_norm": 2.3505067825317383,
      "learning_rate": 2.679e-06,
      "loss": 2.7359,
      "step": 895
    },
    {
      "epoch": 1.6537396121883656,
      "grad_norm": 2.904036521911621,
      "learning_rate": 2.682e-06,
      "loss": 2.7324,
      "step": 896
    },
    {
      "epoch": 1.6555863342566943,
      "grad_norm": 2.6287262439727783,
      "learning_rate": 2.685e-06,
      "loss": 2.7101,
      "step": 897
    },
    {
      "epoch": 1.657433056325023,
      "grad_norm": 4.635859966278076,
      "learning_rate": 2.688e-06,
      "loss": 2.6604,
      "step": 898
    },
    {
      "epoch": 1.6592797783933517,
      "grad_norm": 3.0158092975616455,
      "learning_rate": 2.691e-06,
      "loss": 2.6693,
      "step": 899
    },
    {
      "epoch": 1.6611265004616804,
      "grad_norm": 2.7617454528808594,
      "learning_rate": 2.6940000000000004e-06,
      "loss": 2.6603,
      "step": 900
    },
    {
      "epoch": 1.6629732225300091,
      "grad_norm": 4.1142449378967285,
      "learning_rate": 2.6969999999999998e-06,
      "loss": 2.6057,
      "step": 901
    },
    {
      "epoch": 1.6648199445983378,
      "grad_norm": 4.510326385498047,
      "learning_rate": 2.7e-06,
      "loss": 2.6286,
      "step": 902
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 3.2970147132873535,
      "learning_rate": 2.703e-06,
      "loss": 2.6388,
      "step": 903
    },
    {
      "epoch": 1.6685133887349952,
      "grad_norm": 4.057903289794922,
      "learning_rate": 2.706e-06,
      "loss": 2.6044,
      "step": 904
    },
    {
      "epoch": 1.6703601108033241,
      "grad_norm": 2.707749605178833,
      "learning_rate": 2.7090000000000002e-06,
      "loss": 2.5706,
      "step": 905
    },
    {
      "epoch": 1.6722068328716528,
      "grad_norm": 3.057931423187256,
      "learning_rate": 2.712e-06,
      "loss": 2.5373,
      "step": 906
    },
    {
      "epoch": 1.6740535549399815,
      "grad_norm": 3.6897811889648438,
      "learning_rate": 2.715e-06,
      "loss": 2.5179,
      "step": 907
    },
    {
      "epoch": 1.6759002770083102,
      "grad_norm": 3.4612832069396973,
      "learning_rate": 2.718e-06,
      "loss": 2.5037,
      "step": 908
    },
    {
      "epoch": 1.677746999076639,
      "grad_norm": 4.006030559539795,
      "learning_rate": 2.7210000000000003e-06,
      "loss": 2.5421,
      "step": 909
    },
    {
      "epoch": 1.6795937211449676,
      "grad_norm": 4.664487838745117,
      "learning_rate": 2.724e-06,
      "loss": 2.5214,
      "step": 910
    },
    {
      "epoch": 1.6814404432132966,
      "grad_norm": 4.485028266906738,
      "learning_rate": 2.727e-06,
      "loss": 2.4488,
      "step": 911
    },
    {
      "epoch": 1.6832871652816253,
      "grad_norm": 3.4627418518066406,
      "learning_rate": 2.73e-06,
      "loss": 2.4525,
      "step": 912
    },
    {
      "epoch": 1.685133887349954,
      "grad_norm": 3.839110851287842,
      "learning_rate": 2.733e-06,
      "loss": 2.4505,
      "step": 913
    },
    {
      "epoch": 1.6869806094182827,
      "grad_norm": 3.6342220306396484,
      "learning_rate": 2.736e-06,
      "loss": 2.4108,
      "step": 914
    },
    {
      "epoch": 1.6888273314866113,
      "grad_norm": 4.183078765869141,
      "learning_rate": 2.7390000000000004e-06,
      "loss": 2.4424,
      "step": 915
    },
    {
      "epoch": 1.69067405355494,
      "grad_norm": 4.1377854347229,
      "learning_rate": 2.7419999999999998e-06,
      "loss": 2.3767,
      "step": 916
    },
    {
      "epoch": 1.6925207756232687,
      "grad_norm": 3.869332790374756,
      "learning_rate": 2.745e-06,
      "loss": 2.3823,
      "step": 917
    },
    {
      "epoch": 1.6943674976915974,
      "grad_norm": 3.835867166519165,
      "learning_rate": 2.748e-06,
      "loss": 2.3853,
      "step": 918
    },
    {
      "epoch": 1.6962142197599261,
      "grad_norm": 5.005284309387207,
      "learning_rate": 2.751e-06,
      "loss": 2.3881,
      "step": 919
    },
    {
      "epoch": 1.6980609418282548,
      "grad_norm": 4.473355770111084,
      "learning_rate": 2.7540000000000002e-06,
      "loss": 2.3436,
      "step": 920
    },
    {
      "epoch": 1.6999076638965835,
      "grad_norm": 3.1353838443756104,
      "learning_rate": 2.757e-06,
      "loss": 2.3374,
      "step": 921
    },
    {
      "epoch": 1.7017543859649122,
      "grad_norm": 3.464097499847412,
      "learning_rate": 2.76e-06,
      "loss": 2.2889,
      "step": 922
    },
    {
      "epoch": 1.703601108033241,
      "grad_norm": 3.300783634185791,
      "learning_rate": 2.763e-06,
      "loss": 2.2898,
      "step": 923
    },
    {
      "epoch": 1.7054478301015696,
      "grad_norm": 4.972532272338867,
      "learning_rate": 2.7660000000000003e-06,
      "loss": 2.2912,
      "step": 924
    },
    {
      "epoch": 1.7072945521698983,
      "grad_norm": 5.833555698394775,
      "learning_rate": 2.7689999999999997e-06,
      "loss": 2.2812,
      "step": 925
    },
    {
      "epoch": 1.709141274238227,
      "grad_norm": 4.416055202484131,
      "learning_rate": 2.772e-06,
      "loss": 2.2297,
      "step": 926
    },
    {
      "epoch": 1.7109879963065557,
      "grad_norm": 8.705913543701172,
      "learning_rate": 2.775e-06,
      "loss": 2.243,
      "step": 927
    },
    {
      "epoch": 1.7128347183748844,
      "grad_norm": 3.3832077980041504,
      "learning_rate": 2.778e-06,
      "loss": 2.2515,
      "step": 928
    },
    {
      "epoch": 1.7146814404432131,
      "grad_norm": 4.412527561187744,
      "learning_rate": 2.781e-06,
      "loss": 2.193,
      "step": 929
    },
    {
      "epoch": 1.716528162511542,
      "grad_norm": 4.890414237976074,
      "learning_rate": 2.784e-06,
      "loss": 2.2242,
      "step": 930
    },
    {
      "epoch": 1.7183748845798708,
      "grad_norm": 4.201949596405029,
      "learning_rate": 2.787e-06,
      "loss": 2.1902,
      "step": 931
    },
    {
      "epoch": 1.7202216066481995,
      "grad_norm": 3.7861242294311523,
      "learning_rate": 2.79e-06,
      "loss": 2.188,
      "step": 932
    },
    {
      "epoch": 1.7220683287165282,
      "grad_norm": 4.323360443115234,
      "learning_rate": 2.793e-06,
      "loss": 2.1455,
      "step": 933
    },
    {
      "epoch": 1.7239150507848569,
      "grad_norm": 2.85317325592041,
      "learning_rate": 2.7960000000000004e-06,
      "loss": 2.099,
      "step": 934
    },
    {
      "epoch": 1.7257617728531855,
      "grad_norm": 3.6395180225372314,
      "learning_rate": 2.799e-06,
      "loss": 2.0754,
      "step": 935
    },
    {
      "epoch": 1.7276084949215145,
      "grad_norm": 4.379446983337402,
      "learning_rate": 2.802e-06,
      "loss": 2.0919,
      "step": 936
    },
    {
      "epoch": 1.7294552169898432,
      "grad_norm": 3.4269096851348877,
      "learning_rate": 2.8050000000000002e-06,
      "loss": 2.0443,
      "step": 937
    },
    {
      "epoch": 1.7313019390581719,
      "grad_norm": 4.84335994720459,
      "learning_rate": 2.808e-06,
      "loss": 2.0577,
      "step": 938
    },
    {
      "epoch": 1.7331486611265006,
      "grad_norm": 7.587337493896484,
      "learning_rate": 2.8110000000000003e-06,
      "loss": 2.0765,
      "step": 939
    },
    {
      "epoch": 1.7349953831948293,
      "grad_norm": 6.143586158752441,
      "learning_rate": 2.814e-06,
      "loss": 2.0361,
      "step": 940
    },
    {
      "epoch": 1.736842105263158,
      "grad_norm": 4.004847526550293,
      "learning_rate": 2.817e-06,
      "loss": 1.9983,
      "step": 941
    },
    {
      "epoch": 1.7386888273314867,
      "grad_norm": 6.8099260330200195,
      "learning_rate": 2.82e-06,
      "loss": 2.0316,
      "step": 942
    },
    {
      "epoch": 1.7405355493998154,
      "grad_norm": 9.880424499511719,
      "learning_rate": 2.8230000000000003e-06,
      "loss": 2.4311,
      "step": 943
    },
    {
      "epoch": 1.742382271468144,
      "grad_norm": 6.309079170227051,
      "learning_rate": 2.826e-06,
      "loss": 2.3447,
      "step": 944
    },
    {
      "epoch": 1.7442289935364728,
      "grad_norm": 4.839874744415283,
      "learning_rate": 2.829e-06,
      "loss": 2.2527,
      "step": 945
    },
    {
      "epoch": 1.7460757156048015,
      "grad_norm": 3.779041051864624,
      "learning_rate": 2.832e-06,
      "loss": 2.1876,
      "step": 946
    },
    {
      "epoch": 1.7479224376731302,
      "grad_norm": 6.362606048583984,
      "learning_rate": 2.835e-06,
      "loss": 2.1817,
      "step": 947
    },
    {
      "epoch": 1.7497691597414589,
      "grad_norm": 8.506185531616211,
      "learning_rate": 2.838e-06,
      "loss": 2.1166,
      "step": 948
    },
    {
      "epoch": 1.7516158818097876,
      "grad_norm": 5.72320032119751,
      "learning_rate": 2.8410000000000004e-06,
      "loss": 2.1144,
      "step": 949
    },
    {
      "epoch": 1.7534626038781163,
      "grad_norm": 4.872931957244873,
      "learning_rate": 2.844e-06,
      "loss": 2.0904,
      "step": 950
    },
    {
      "epoch": 1.755309325946445,
      "grad_norm": 5.8115153312683105,
      "learning_rate": 2.847e-06,
      "loss": 2.0425,
      "step": 951
    },
    {
      "epoch": 1.7571560480147737,
      "grad_norm": 5.148961067199707,
      "learning_rate": 2.8500000000000002e-06,
      "loss": 1.9555,
      "step": 952
    },
    {
      "epoch": 1.7590027700831024,
      "grad_norm": 4.77873420715332,
      "learning_rate": 2.853e-06,
      "loss": 1.8868,
      "step": 953
    },
    {
      "epoch": 1.760849492151431,
      "grad_norm": 6.63554573059082,
      "learning_rate": 2.8560000000000003e-06,
      "loss": 1.9266,
      "step": 954
    },
    {
      "epoch": 1.76269621421976,
      "grad_norm": 5.592360019683838,
      "learning_rate": 2.859e-06,
      "loss": 1.9111,
      "step": 955
    },
    {
      "epoch": 1.7645429362880887,
      "grad_norm": 6.387895107269287,
      "learning_rate": 2.862e-06,
      "loss": 1.8245,
      "step": 956
    },
    {
      "epoch": 1.7663896583564174,
      "grad_norm": 5.357057571411133,
      "learning_rate": 2.865e-06,
      "loss": 1.748,
      "step": 957
    },
    {
      "epoch": 1.768236380424746,
      "grad_norm": 13.47465991973877,
      "learning_rate": 2.8680000000000003e-06,
      "loss": 1.7151,
      "step": 958
    },
    {
      "epoch": 1.7700831024930748,
      "grad_norm": 3.3738362789154053,
      "learning_rate": 2.8709999999999997e-06,
      "loss": 1.673,
      "step": 959
    },
    {
      "epoch": 1.7719298245614035,
      "grad_norm": 4.385458946228027,
      "learning_rate": 2.874e-06,
      "loss": 1.7046,
      "step": 960
    },
    {
      "epoch": 1.7737765466297324,
      "grad_norm": 2.9951484203338623,
      "learning_rate": 2.877e-06,
      "loss": 1.6725,
      "step": 961
    },
    {
      "epoch": 1.775623268698061,
      "grad_norm": 5.280602931976318,
      "learning_rate": 2.88e-06,
      "loss": 1.6102,
      "step": 962
    },
    {
      "epoch": 1.7774699907663898,
      "grad_norm": 3.529313802719116,
      "learning_rate": 2.883e-06,
      "loss": 1.5605,
      "step": 963
    },
    {
      "epoch": 1.7793167128347185,
      "grad_norm": 8.768348693847656,
      "learning_rate": 2.886e-06,
      "loss": 1.5792,
      "step": 964
    },
    {
      "epoch": 1.7811634349030472,
      "grad_norm": 4.621875286102295,
      "learning_rate": 2.8889999999999998e-06,
      "loss": 1.5769,
      "step": 965
    },
    {
      "epoch": 1.7830101569713759,
      "grad_norm": 2.7766876220703125,
      "learning_rate": 2.892e-06,
      "loss": 1.522,
      "step": 966
    },
    {
      "epoch": 1.7848568790397046,
      "grad_norm": 3.9150314331054688,
      "learning_rate": 2.8950000000000002e-06,
      "loss": 1.539,
      "step": 967
    },
    {
      "epoch": 1.7867036011080333,
      "grad_norm": 7.411330699920654,
      "learning_rate": 2.898e-06,
      "loss": 1.4637,
      "step": 968
    },
    {
      "epoch": 1.788550323176362,
      "grad_norm": 4.648345470428467,
      "learning_rate": 2.901e-06,
      "loss": 1.4333,
      "step": 969
    },
    {
      "epoch": 1.7903970452446907,
      "grad_norm": 4.080045223236084,
      "learning_rate": 2.904e-06,
      "loss": 1.4157,
      "step": 970
    },
    {
      "epoch": 1.7922437673130194,
      "grad_norm": 15.073267936706543,
      "learning_rate": 2.907e-06,
      "loss": 1.3991,
      "step": 971
    },
    {
      "epoch": 1.794090489381348,
      "grad_norm": 5.176401138305664,
      "learning_rate": 2.91e-06,
      "loss": 1.3885,
      "step": 972
    },
    {
      "epoch": 1.7959372114496768,
      "grad_norm": 3.8407390117645264,
      "learning_rate": 2.9130000000000003e-06,
      "loss": 1.3605,
      "step": 973
    },
    {
      "epoch": 1.7977839335180055,
      "grad_norm": 3.0771899223327637,
      "learning_rate": 2.916e-06,
      "loss": 1.3694,
      "step": 974
    },
    {
      "epoch": 1.7996306555863342,
      "grad_norm": 4.342832088470459,
      "learning_rate": 2.919e-06,
      "loss": 1.2812,
      "step": 975
    },
    {
      "epoch": 1.8014773776546629,
      "grad_norm": 3.102047920227051,
      "learning_rate": 2.922e-06,
      "loss": 1.2985,
      "step": 976
    },
    {
      "epoch": 1.8033240997229916,
      "grad_norm": 2.795923948287964,
      "learning_rate": 2.9250000000000004e-06,
      "loss": 1.2971,
      "step": 977
    },
    {
      "epoch": 1.8051708217913203,
      "grad_norm": 6.470213413238525,
      "learning_rate": 2.928e-06,
      "loss": 1.1874,
      "step": 978
    },
    {
      "epoch": 1.807017543859649,
      "grad_norm": 9.83168888092041,
      "learning_rate": 2.931e-06,
      "loss": 1.2674,
      "step": 979
    },
    {
      "epoch": 1.8088642659279779,
      "grad_norm": 5.468801498413086,
      "learning_rate": 2.934e-06,
      "loss": 1.2251,
      "step": 980
    },
    {
      "epoch": 1.8107109879963066,
      "grad_norm": 4.2570109367370605,
      "learning_rate": 2.937e-06,
      "loss": 1.1727,
      "step": 981
    },
    {
      "epoch": 1.8125577100646353,
      "grad_norm": 7.420529842376709,
      "learning_rate": 2.9400000000000002e-06,
      "loss": 1.1752,
      "step": 982
    },
    {
      "epoch": 1.814404432132964,
      "grad_norm": NaN,
      "learning_rate": 2.9400000000000002e-06,
      "loss": 1.185,
      "step": 983
    },
    {
      "epoch": 1.8162511542012927,
      "grad_norm": 4.976733207702637,
      "learning_rate": 2.9430000000000005e-06,
      "loss": 1.178,
      "step": 984
    },
    {
      "epoch": 1.8180978762696214,
      "grad_norm": 4.1213765144348145,
      "learning_rate": 2.946e-06,
      "loss": 1.1264,
      "step": 985
    },
    {
      "epoch": 1.8199445983379503,
      "grad_norm": 3.392833948135376,
      "learning_rate": 2.949e-06,
      "loss": 1.1685,
      "step": 986
    },
    {
      "epoch": 1.821791320406279,
      "grad_norm": 21.531436920166016,
      "learning_rate": 2.9520000000000003e-06,
      "loss": 1.1395,
      "step": 987
    },
    {
      "epoch": 1.8236380424746077,
      "grad_norm": 4.77427339553833,
      "learning_rate": 2.955e-06,
      "loss": 1.1045,
      "step": 988
    },
    {
      "epoch": 1.8254847645429364,
      "grad_norm": 5.12446928024292,
      "learning_rate": 2.958e-06,
      "loss": 1.1141,
      "step": 989
    },
    {
      "epoch": 1.827331486611265,
      "grad_norm": 4.657517910003662,
      "learning_rate": 2.961e-06,
      "loss": 1.0613,
      "step": 990
    },
    {
      "epoch": 1.8291782086795938,
      "grad_norm": 9.723724365234375,
      "learning_rate": 2.964e-06,
      "loss": 1.0681,
      "step": 991
    },
    {
      "epoch": 1.8310249307479225,
      "grad_norm": 3.2556118965148926,
      "learning_rate": 2.967e-06,
      "loss": 1.1434,
      "step": 992
    },
    {
      "epoch": 1.8328716528162512,
      "grad_norm": 10.147623062133789,
      "learning_rate": 2.9700000000000004e-06,
      "loss": 1.8507,
      "step": 993
    },
    {
      "epoch": 1.83471837488458,
      "grad_norm": 5.80134391784668,
      "learning_rate": 2.9729999999999997e-06,
      "loss": 1.5426,
      "step": 994
    },
    {
      "epoch": 1.8365650969529086,
      "grad_norm": 4.6190876960754395,
      "learning_rate": 2.976e-06,
      "loss": 1.486,
      "step": 995
    },
    {
      "epoch": 1.8384118190212373,
      "grad_norm": 3.1225485801696777,
      "learning_rate": 2.979e-06,
      "loss": 1.4601,
      "step": 996
    },
    {
      "epoch": 1.840258541089566,
      "grad_norm": 3.720115900039673,
      "learning_rate": 2.982e-06,
      "loss": 1.4278,
      "step": 997
    },
    {
      "epoch": 1.8421052631578947,
      "grad_norm": 10.240564346313477,
      "learning_rate": 2.9850000000000002e-06,
      "loss": 1.2362,
      "step": 998
    },
    {
      "epoch": 1.8439519852262234,
      "grad_norm": 5.202827453613281,
      "learning_rate": 2.988e-06,
      "loss": 1.2611,
      "step": 999
    },
    {
      "epoch": 1.845798707294552,
      "grad_norm": 4.68275785446167,
      "learning_rate": 2.991e-06,
      "loss": 1.2884,
      "step": 1000
    },
    {
      "epoch": 1.845798707294552,
      "eval_cer": 0.3808227700443269,
      "eval_loss": 1.387242317199707,
      "eval_runtime": 16.2128,
      "eval_samples_per_second": 18.751,
      "eval_steps_per_second": 0.617,
      "eval_wer": 1.0113200306983883,
      "step": 1000
    },
    {
      "epoch": 1.8476454293628808,
      "grad_norm": 5.817086219787598,
      "learning_rate": 2.994e-06,
      "loss": 1.2335,
      "step": 1001
    },
    {
      "epoch": 1.8494921514312095,
      "grad_norm": 2.8305552005767822,
      "learning_rate": 2.9970000000000003e-06,
      "loss": 1.1889,
      "step": 1002
    },
    {
      "epoch": 1.8513388734995382,
      "grad_norm": 5.681527614593506,
      "learning_rate": 3e-06,
      "loss": 1.1404,
      "step": 1003
    },
    {
      "epoch": 1.8531855955678669,
      "grad_norm": 4.868209362030029,
      "learning_rate": 3.003e-06,
      "loss": 1.1088,
      "step": 1004
    },
    {
      "epoch": 1.8550323176361958,
      "grad_norm": 4.205574989318848,
      "learning_rate": 3.006e-06,
      "loss": 1.0609,
      "step": 1005
    },
    {
      "epoch": 1.8568790397045245,
      "grad_norm": 9.881863594055176,
      "learning_rate": 3.009e-06,
      "loss": 0.9499,
      "step": 1006
    },
    {
      "epoch": 1.8587257617728532,
      "grad_norm": 3.2579362392425537,
      "learning_rate": 3.012e-06,
      "loss": 0.8826,
      "step": 1007
    },
    {
      "epoch": 1.860572483841182,
      "grad_norm": 5.675721645355225,
      "learning_rate": 3.0150000000000004e-06,
      "loss": 0.873,
      "step": 1008
    },
    {
      "epoch": 1.8624192059095106,
      "grad_norm": 4.980535507202148,
      "learning_rate": 3.0179999999999997e-06,
      "loss": 0.8653,
      "step": 1009
    },
    {
      "epoch": 1.8642659279778393,
      "grad_norm": 11.08716869354248,
      "learning_rate": 3.021e-06,
      "loss": 0.903,
      "step": 1010
    },
    {
      "epoch": 1.866112650046168,
      "grad_norm": 4.356354236602783,
      "learning_rate": 3.024e-06,
      "loss": 0.939,
      "step": 1011
    },
    {
      "epoch": 1.867959372114497,
      "grad_norm": 6.092064380645752,
      "learning_rate": 3.027e-06,
      "loss": 0.8441,
      "step": 1012
    },
    {
      "epoch": 1.8698060941828256,
      "grad_norm": 10.622693061828613,
      "learning_rate": 3.0300000000000002e-06,
      "loss": 0.7909,
      "step": 1013
    },
    {
      "epoch": 1.8716528162511543,
      "grad_norm": 5.019896030426025,
      "learning_rate": 3.033e-06,
      "loss": 0.8149,
      "step": 1014
    },
    {
      "epoch": 1.873499538319483,
      "grad_norm": 3.3237733840942383,
      "learning_rate": 3.036e-06,
      "loss": 0.741,
      "step": 1015
    },
    {
      "epoch": 1.8753462603878117,
      "grad_norm": 4.057783603668213,
      "learning_rate": 3.039e-06,
      "loss": 0.7615,
      "step": 1016
    },
    {
      "epoch": 1.8771929824561404,
      "grad_norm": 3.0382447242736816,
      "learning_rate": 3.0420000000000003e-06,
      "loss": 0.7755,
      "step": 1017
    },
    {
      "epoch": 1.8790397045244691,
      "grad_norm": 3.909538745880127,
      "learning_rate": 3.0450000000000005e-06,
      "loss": 0.7565,
      "step": 1018
    },
    {
      "epoch": 1.8808864265927978,
      "grad_norm": 7.082792282104492,
      "learning_rate": 3.048e-06,
      "loss": 0.8341,
      "step": 1019
    },
    {
      "epoch": 1.8827331486611265,
      "grad_norm": 3.4757003784179688,
      "learning_rate": 3.051e-06,
      "loss": 0.7641,
      "step": 1020
    },
    {
      "epoch": 1.8845798707294552,
      "grad_norm": 4.288313388824463,
      "learning_rate": 3.0540000000000003e-06,
      "loss": 0.7306,
      "step": 1021
    },
    {
      "epoch": 1.886426592797784,
      "grad_norm": 12.425573348999023,
      "learning_rate": 3.057e-06,
      "loss": 0.6782,
      "step": 1022
    },
    {
      "epoch": 1.8882733148661126,
      "grad_norm": 3.1311159133911133,
      "learning_rate": 3.06e-06,
      "loss": 0.7367,
      "step": 1023
    },
    {
      "epoch": 1.8901200369344413,
      "grad_norm": 3.636770248413086,
      "learning_rate": 3.063e-06,
      "loss": 0.7476,
      "step": 1024
    },
    {
      "epoch": 1.89196675900277,
      "grad_norm": 3.5470006465911865,
      "learning_rate": 3.066e-06,
      "loss": 0.7078,
      "step": 1025
    },
    {
      "epoch": 1.8938134810710987,
      "grad_norm": 4.514321804046631,
      "learning_rate": 3.069e-06,
      "loss": 0.6822,
      "step": 1026
    },
    {
      "epoch": 1.8956602031394274,
      "grad_norm": 4.672080993652344,
      "learning_rate": 3.0720000000000004e-06,
      "loss": 0.6728,
      "step": 1027
    },
    {
      "epoch": 1.897506925207756,
      "grad_norm": 20.278226852416992,
      "learning_rate": 3.0749999999999998e-06,
      "loss": 0.6897,
      "step": 1028
    },
    {
      "epoch": 1.8993536472760848,
      "grad_norm": 7.047516345977783,
      "learning_rate": 3.078e-06,
      "loss": 0.6794,
      "step": 1029
    },
    {
      "epoch": 1.9012003693444137,
      "grad_norm": 2.9382872581481934,
      "learning_rate": 3.0810000000000002e-06,
      "loss": 0.738,
      "step": 1030
    },
    {
      "epoch": 1.9030470914127424,
      "grad_norm": 3.0094921588897705,
      "learning_rate": 3.084e-06,
      "loss": 0.7073,
      "step": 1031
    },
    {
      "epoch": 1.9048938134810711,
      "grad_norm": 20.426923751831055,
      "learning_rate": 3.0870000000000003e-06,
      "loss": 0.6801,
      "step": 1032
    },
    {
      "epoch": 1.9067405355493998,
      "grad_norm": 5.870213508605957,
      "learning_rate": 3.09e-06,
      "loss": 0.6999,
      "step": 1033
    },
    {
      "epoch": 1.9085872576177285,
      "grad_norm": 4.475285530090332,
      "learning_rate": 3.093e-06,
      "loss": 0.6844,
      "step": 1034
    },
    {
      "epoch": 1.9104339796860572,
      "grad_norm": 24.2203311920166,
      "learning_rate": 3.096e-06,
      "loss": 0.6208,
      "step": 1035
    },
    {
      "epoch": 1.912280701754386,
      "grad_norm": 4.748508453369141,
      "learning_rate": 3.0990000000000003e-06,
      "loss": 0.6121,
      "step": 1036
    },
    {
      "epoch": 1.9141274238227148,
      "grad_norm": 3.6154093742370605,
      "learning_rate": 3.102e-06,
      "loss": 0.6886,
      "step": 1037
    },
    {
      "epoch": 1.9159741458910435,
      "grad_norm": 3.9529449939727783,
      "learning_rate": 3.105e-06,
      "loss": 0.6453,
      "step": 1038
    },
    {
      "epoch": 1.9178208679593722,
      "grad_norm": 18.980297088623047,
      "learning_rate": 3.108e-06,
      "loss": 0.6588,
      "step": 1039
    },
    {
      "epoch": 1.919667590027701,
      "grad_norm": 4.728965759277344,
      "learning_rate": 3.111e-06,
      "loss": 0.6229,
      "step": 1040
    },
    {
      "epoch": 1.9215143120960296,
      "grad_norm": 8.780813217163086,
      "learning_rate": 3.114e-06,
      "loss": 0.7132,
      "step": 1041
    },
    {
      "epoch": 1.9233610341643583,
      "grad_norm": 7.412342071533203,
      "learning_rate": 3.1170000000000004e-06,
      "loss": 0.8253,
      "step": 1042
    },
    {
      "epoch": 1.925207756232687,
      "grad_norm": 8.90256118774414,
      "learning_rate": 3.1199999999999998e-06,
      "loss": 1.3585,
      "step": 1043
    },
    {
      "epoch": 1.9270544783010157,
      "grad_norm": 4.4571428298950195,
      "learning_rate": 3.123e-06,
      "loss": 1.1706,
      "step": 1044
    },
    {
      "epoch": 1.9289012003693444,
      "grad_norm": 4.3071184158325195,
      "learning_rate": 3.1260000000000002e-06,
      "loss": 1.0833,
      "step": 1045
    },
    {
      "epoch": 1.9307479224376731,
      "grad_norm": 9.08230972290039,
      "learning_rate": 3.129e-06,
      "loss": 0.9173,
      "step": 1046
    },
    {
      "epoch": 1.9325946445060018,
      "grad_norm": 5.986129283905029,
      "learning_rate": 3.1320000000000003e-06,
      "loss": 0.886,
      "step": 1047
    },
    {
      "epoch": 1.9344413665743305,
      "grad_norm": 4.313248157501221,
      "learning_rate": 3.135e-06,
      "loss": 0.9184,
      "step": 1048
    },
    {
      "epoch": 1.9362880886426592,
      "grad_norm": 3.475137948989868,
      "learning_rate": 3.138e-06,
      "loss": 0.8592,
      "step": 1049
    },
    {
      "epoch": 1.938134810710988,
      "grad_norm": 3.4487123489379883,
      "learning_rate": 3.141e-06,
      "loss": 0.8229,
      "step": 1050
    },
    {
      "epoch": 1.9399815327793166,
      "grad_norm": 5.085602760314941,
      "learning_rate": 3.1440000000000003e-06,
      "loss": 0.7537,
      "step": 1051
    },
    {
      "epoch": 1.9418282548476453,
      "grad_norm": 3.04370379447937,
      "learning_rate": 3.1469999999999997e-06,
      "loss": 0.778,
      "step": 1052
    },
    {
      "epoch": 1.943674976915974,
      "grad_norm": 3.9167068004608154,
      "learning_rate": 3.15e-06,
      "loss": 0.6315,
      "step": 1053
    },
    {
      "epoch": 1.9455216989843027,
      "grad_norm": 2.40921950340271,
      "learning_rate": 3.153e-06,
      "loss": 0.6394,
      "step": 1054
    },
    {
      "epoch": 1.9473684210526314,
      "grad_norm": 9.868019104003906,
      "learning_rate": 3.156e-06,
      "loss": 0.6353,
      "step": 1055
    },
    {
      "epoch": 1.9492151431209603,
      "grad_norm": 2.789872646331787,
      "learning_rate": 3.159e-06,
      "loss": 0.5673,
      "step": 1056
    },
    {
      "epoch": 1.951061865189289,
      "grad_norm": 2.2566208839416504,
      "learning_rate": 3.162e-06,
      "loss": 0.6103,
      "step": 1057
    },
    {
      "epoch": 1.9529085872576177,
      "grad_norm": 3.410893678665161,
      "learning_rate": 3.1649999999999998e-06,
      "loss": 0.5088,
      "step": 1058
    },
    {
      "epoch": 1.9547553093259464,
      "grad_norm": 2.58414626121521,
      "learning_rate": 3.168e-06,
      "loss": 0.5453,
      "step": 1059
    },
    {
      "epoch": 1.9566020313942751,
      "grad_norm": 6.431880950927734,
      "learning_rate": 3.1710000000000002e-06,
      "loss": 0.4953,
      "step": 1060
    },
    {
      "epoch": 1.9584487534626038,
      "grad_norm": 2.7993009090423584,
      "learning_rate": 3.1740000000000004e-06,
      "loss": 0.5452,
      "step": 1061
    },
    {
      "epoch": 1.9602954755309328,
      "grad_norm": 5.0864386558532715,
      "learning_rate": 3.177e-06,
      "loss": 0.5177,
      "step": 1062
    },
    {
      "epoch": 1.9621421975992615,
      "grad_norm": 2.416808605194092,
      "learning_rate": 3.18e-06,
      "loss": 0.5564,
      "step": 1063
    },
    {
      "epoch": 1.9639889196675901,
      "grad_norm": 3.0072779655456543,
      "learning_rate": 3.1830000000000003e-06,
      "loss": 0.4664,
      "step": 1064
    },
    {
      "epoch": 1.9658356417359188,
      "grad_norm": 3.39021635055542,
      "learning_rate": 3.186e-06,
      "loss": 0.5069,
      "step": 1065
    },
    {
      "epoch": 1.9676823638042475,
      "grad_norm": 2.3784401416778564,
      "learning_rate": 3.1890000000000003e-06,
      "loss": 0.5095,
      "step": 1066
    },
    {
      "epoch": 1.9695290858725762,
      "grad_norm": 2.980907440185547,
      "learning_rate": 3.192e-06,
      "loss": 0.581,
      "step": 1067
    },
    {
      "epoch": 1.971375807940905,
      "grad_norm": 3.834425687789917,
      "learning_rate": 3.195e-06,
      "loss": 0.5368,
      "step": 1068
    },
    {
      "epoch": 1.9732225300092336,
      "grad_norm": 4.101207733154297,
      "learning_rate": 3.198e-06,
      "loss": 0.5162,
      "step": 1069
    },
    {
      "epoch": 1.9750692520775623,
      "grad_norm": 2.8639001846313477,
      "learning_rate": 3.2010000000000004e-06,
      "loss": 0.4687,
      "step": 1070
    },
    {
      "epoch": 1.976915974145891,
      "grad_norm": 3.660947561264038,
      "learning_rate": 3.204e-06,
      "loss": 0.5063,
      "step": 1071
    },
    {
      "epoch": 1.9787626962142197,
      "grad_norm": 4.399283409118652,
      "learning_rate": 3.207e-06,
      "loss": 0.5655,
      "step": 1072
    },
    {
      "epoch": 1.9806094182825484,
      "grad_norm": 5.054996967315674,
      "learning_rate": 3.21e-06,
      "loss": 0.5052,
      "step": 1073
    },
    {
      "epoch": 1.9824561403508771,
      "grad_norm": 3.4301655292510986,
      "learning_rate": 3.213e-06,
      "loss": 0.484,
      "step": 1074
    },
    {
      "epoch": 1.9843028624192058,
      "grad_norm": 3.875305652618408,
      "learning_rate": 3.216e-06,
      "loss": 0.5369,
      "step": 1075
    },
    {
      "epoch": 1.9861495844875345,
      "grad_norm": 4.462026119232178,
      "learning_rate": 3.2190000000000004e-06,
      "loss": 0.5141,
      "step": 1076
    },
    {
      "epoch": 1.9879963065558632,
      "grad_norm": 4.290552139282227,
      "learning_rate": 3.222e-06,
      "loss": 0.5084,
      "step": 1077
    },
    {
      "epoch": 1.989843028624192,
      "grad_norm": 2.693206787109375,
      "learning_rate": 3.225e-06,
      "loss": 0.5143,
      "step": 1078
    },
    {
      "epoch": 1.9916897506925206,
      "grad_norm": 13.908217430114746,
      "learning_rate": 3.2280000000000003e-06,
      "loss": 0.5154,
      "step": 1079
    },
    {
      "epoch": 1.9935364727608493,
      "grad_norm": 4.1457600593566895,
      "learning_rate": 3.231e-06,
      "loss": 0.5337,
      "step": 1080
    },
    {
      "epoch": 1.9953831948291783,
      "grad_norm": 4.087045669555664,
      "learning_rate": 3.2340000000000003e-06,
      "loss": 0.5313,
      "step": 1081
    },
    {
      "epoch": 1.997229916897507,
      "grad_norm": 6.225906848907471,
      "learning_rate": 3.237e-06,
      "loss": 0.5336,
      "step": 1082
    },
    {
      "epoch": 1.9990766389658357,
      "grad_norm": 4.523451328277588,
      "learning_rate": 3.24e-06,
      "loss": 0.5495,
      "step": 1083
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.176987648010254,
      "learning_rate": 3.243e-06,
      "loss": 0.3116,
      "step": 1084
    },
    {
      "epoch": 2.0018467220683287,
      "grad_norm": 5.960220813751221,
      "learning_rate": 3.2460000000000003e-06,
      "loss": 1.2542,
      "step": 1085
    },
    {
      "epoch": 2.0036934441366574,
      "grad_norm": 5.376857757568359,
      "learning_rate": 3.2489999999999997e-06,
      "loss": 1.1583,
      "step": 1086
    },
    {
      "epoch": 2.005540166204986,
      "grad_norm": 3.4985902309417725,
      "learning_rate": 3.252e-06,
      "loss": 0.9727,
      "step": 1087
    },
    {
      "epoch": 2.007386888273315,
      "grad_norm": 3.4755091667175293,
      "learning_rate": 3.255e-06,
      "loss": 1.0657,
      "step": 1088
    },
    {
      "epoch": 2.0092336103416435,
      "grad_norm": 2.7386229038238525,
      "learning_rate": 3.258e-06,
      "loss": 0.8495,
      "step": 1089
    },
    {
      "epoch": 2.011080332409972,
      "grad_norm": 4.1003737449646,
      "learning_rate": 3.261e-06,
      "loss": 0.8668,
      "step": 1090
    },
    {
      "epoch": 2.012927054478301,
      "grad_norm": 6.044220447540283,
      "learning_rate": 3.264e-06,
      "loss": 0.8225,
      "step": 1091
    },
    {
      "epoch": 2.0147737765466296,
      "grad_norm": 3.062387704849243,
      "learning_rate": 3.267e-06,
      "loss": 0.6906,
      "step": 1092
    },
    {
      "epoch": 2.0166204986149583,
      "grad_norm": 5.289513111114502,
      "learning_rate": 3.27e-06,
      "loss": 0.6469,
      "step": 1093
    },
    {
      "epoch": 2.018467220683287,
      "grad_norm": 2.689310312271118,
      "learning_rate": 3.2730000000000003e-06,
      "loss": 0.7619,
      "step": 1094
    },
    {
      "epoch": 2.0203139427516157,
      "grad_norm": 6.174566745758057,
      "learning_rate": 3.276e-06,
      "loss": 0.6249,
      "step": 1095
    },
    {
      "epoch": 2.0221606648199444,
      "grad_norm": 6.018412113189697,
      "learning_rate": 3.279e-06,
      "loss": 0.5918,
      "step": 1096
    },
    {
      "epoch": 2.0240073868882735,
      "grad_norm": 2.6255674362182617,
      "learning_rate": 3.282e-06,
      "loss": 0.578,
      "step": 1097
    },
    {
      "epoch": 2.0258541089566022,
      "grad_norm": 2.5936782360076904,
      "learning_rate": 3.285e-06,
      "loss": 0.6471,
      "step": 1098
    },
    {
      "epoch": 2.027700831024931,
      "grad_norm": 2.8347206115722656,
      "learning_rate": 3.288e-06,
      "loss": 0.5161,
      "step": 1099
    },
    {
      "epoch": 2.0295475530932596,
      "grad_norm": 3.6363515853881836,
      "learning_rate": 3.2910000000000003e-06,
      "loss": 0.4732,
      "step": 1100
    },
    {
      "epoch": 2.0313942751615883,
      "grad_norm": 2.8886845111846924,
      "learning_rate": 3.2939999999999997e-06,
      "loss": 0.4759,
      "step": 1101
    },
    {
      "epoch": 2.033240997229917,
      "grad_norm": 5.658815383911133,
      "learning_rate": 3.297e-06,
      "loss": 0.4244,
      "step": 1102
    },
    {
      "epoch": 2.0350877192982457,
      "grad_norm": 3.8610455989837646,
      "learning_rate": 3.3e-06,
      "loss": 0.5042,
      "step": 1103
    },
    {
      "epoch": 2.0369344413665744,
      "grad_norm": 3.7909250259399414,
      "learning_rate": 3.3030000000000004e-06,
      "loss": 0.4401,
      "step": 1104
    },
    {
      "epoch": 2.038781163434903,
      "grad_norm": 3.0346169471740723,
      "learning_rate": 3.306e-06,
      "loss": 0.4517,
      "step": 1105
    },
    {
      "epoch": 2.040627885503232,
      "grad_norm": 2.9240636825561523,
      "learning_rate": 3.309e-06,
      "loss": 0.512,
      "step": 1106
    },
    {
      "epoch": 2.0424746075715605,
      "grad_norm": 3.549503803253174,
      "learning_rate": 3.3120000000000002e-06,
      "loss": 0.4491,
      "step": 1107
    },
    {
      "epoch": 2.044321329639889,
      "grad_norm": 2.4286088943481445,
      "learning_rate": 3.315e-06,
      "loss": 0.4317,
      "step": 1108
    },
    {
      "epoch": 2.046168051708218,
      "grad_norm": 4.359600067138672,
      "learning_rate": 3.3180000000000003e-06,
      "loss": 0.507,
      "step": 1109
    },
    {
      "epoch": 2.0480147737765466,
      "grad_norm": 2.9661221504211426,
      "learning_rate": 3.3210000000000005e-06,
      "loss": 0.4248,
      "step": 1110
    },
    {
      "epoch": 2.0498614958448753,
      "grad_norm": 18.271678924560547,
      "learning_rate": 3.324e-06,
      "loss": 0.4179,
      "step": 1111
    },
    {
      "epoch": 2.051708217913204,
      "grad_norm": 3.5711114406585693,
      "learning_rate": 3.327e-06,
      "loss": 0.4596,
      "step": 1112
    },
    {
      "epoch": 2.0535549399815327,
      "grad_norm": 3.6461987495422363,
      "learning_rate": 3.3300000000000003e-06,
      "loss": 0.4043,
      "step": 1113
    },
    {
      "epoch": 2.0554016620498614,
      "grad_norm": 3.886255979537964,
      "learning_rate": 3.333e-06,
      "loss": 0.4284,
      "step": 1114
    },
    {
      "epoch": 2.05724838411819,
      "grad_norm": 3.654627799987793,
      "learning_rate": 3.336e-06,
      "loss": 0.4275,
      "step": 1115
    },
    {
      "epoch": 2.059095106186519,
      "grad_norm": 2.4884819984436035,
      "learning_rate": 3.339e-06,
      "loss": 0.3802,
      "step": 1116
    },
    {
      "epoch": 2.0609418282548475,
      "grad_norm": 10.350143432617188,
      "learning_rate": 3.342e-06,
      "loss": 0.4153,
      "step": 1117
    },
    {
      "epoch": 2.062788550323176,
      "grad_norm": 4.766338348388672,
      "learning_rate": 3.345e-06,
      "loss": 0.4242,
      "step": 1118
    },
    {
      "epoch": 2.064635272391505,
      "grad_norm": 2.0844154357910156,
      "learning_rate": 3.3480000000000004e-06,
      "loss": 0.3617,
      "step": 1119
    },
    {
      "epoch": 2.0664819944598336,
      "grad_norm": 5.149382591247559,
      "learning_rate": 3.3509999999999998e-06,
      "loss": 0.4822,
      "step": 1120
    },
    {
      "epoch": 2.0683287165281623,
      "grad_norm": 8.087337493896484,
      "learning_rate": 3.354e-06,
      "loss": 0.4138,
      "step": 1121
    },
    {
      "epoch": 2.0701754385964914,
      "grad_norm": 3.428081750869751,
      "learning_rate": 3.3570000000000002e-06,
      "loss": 0.3851,
      "step": 1122
    },
    {
      "epoch": 2.07202216066482,
      "grad_norm": 2.2771971225738525,
      "learning_rate": 3.36e-06,
      "loss": 0.4297,
      "step": 1123
    },
    {
      "epoch": 2.073868882733149,
      "grad_norm": 3.1540608406066895,
      "learning_rate": 3.3630000000000002e-06,
      "loss": 0.3744,
      "step": 1124
    },
    {
      "epoch": 2.0757156048014775,
      "grad_norm": 4.224101543426514,
      "learning_rate": 3.366e-06,
      "loss": 0.4036,
      "step": 1125
    },
    {
      "epoch": 2.0775623268698062,
      "grad_norm": 3.432368278503418,
      "learning_rate": 3.369e-06,
      "loss": 0.3774,
      "step": 1126
    },
    {
      "epoch": 2.079409048938135,
      "grad_norm": 4.711674690246582,
      "learning_rate": 3.372e-06,
      "loss": 0.3861,
      "step": 1127
    },
    {
      "epoch": 2.0812557710064636,
      "grad_norm": 4.5501909255981445,
      "learning_rate": 3.3750000000000003e-06,
      "loss": 0.4297,
      "step": 1128
    },
    {
      "epoch": 2.0831024930747923,
      "grad_norm": 8.730445861816406,
      "learning_rate": 3.378e-06,
      "loss": 0.3753,
      "step": 1129
    },
    {
      "epoch": 2.084949215143121,
      "grad_norm": 5.603017330169678,
      "learning_rate": 3.381e-06,
      "loss": 0.4272,
      "step": 1130
    },
    {
      "epoch": 2.0867959372114497,
      "grad_norm": 3.613166332244873,
      "learning_rate": 3.384e-06,
      "loss": 0.432,
      "step": 1131
    },
    {
      "epoch": 2.0886426592797784,
      "grad_norm": 4.397871494293213,
      "learning_rate": 3.387e-06,
      "loss": 0.4113,
      "step": 1132
    },
    {
      "epoch": 2.090489381348107,
      "grad_norm": 3.521243095397949,
      "learning_rate": 3.39e-06,
      "loss": 0.433,
      "step": 1133
    },
    {
      "epoch": 2.092336103416436,
      "grad_norm": 4.172756671905518,
      "learning_rate": 3.3930000000000004e-06,
      "loss": 0.4833,
      "step": 1134
    },
    {
      "epoch": 2.0941828254847645,
      "grad_norm": 6.483964920043945,
      "learning_rate": 3.3959999999999998e-06,
      "loss": 1.2138,
      "step": 1135
    },
    {
      "epoch": 2.0960295475530932,
      "grad_norm": 3.9152333736419678,
      "learning_rate": 3.399e-06,
      "loss": 0.9175,
      "step": 1136
    },
    {
      "epoch": 2.097876269621422,
      "grad_norm": 4.725359916687012,
      "learning_rate": 3.402e-06,
      "loss": 0.9543,
      "step": 1137
    },
    {
      "epoch": 2.0997229916897506,
      "grad_norm": 2.669206142425537,
      "learning_rate": 3.405e-06,
      "loss": 0.8489,
      "step": 1138
    },
    {
      "epoch": 2.1015697137580793,
      "grad_norm": 7.135278224945068,
      "learning_rate": 3.4080000000000002e-06,
      "loss": 0.8881,
      "step": 1139
    },
    {
      "epoch": 2.103416435826408,
      "grad_norm": 4.369724273681641,
      "learning_rate": 3.411e-06,
      "loss": 0.7659,
      "step": 1140
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 6.003192901611328,
      "learning_rate": 3.414e-06,
      "loss": 0.8159,
      "step": 1141
    },
    {
      "epoch": 2.1071098799630654,
      "grad_norm": 4.358618259429932,
      "learning_rate": 3.417e-06,
      "loss": 0.6835,
      "step": 1142
    },
    {
      "epoch": 2.108956602031394,
      "grad_norm": 5.704505920410156,
      "learning_rate": 3.4200000000000003e-06,
      "loss": 0.6636,
      "step": 1143
    },
    {
      "epoch": 2.110803324099723,
      "grad_norm": 5.818912506103516,
      "learning_rate": 3.4229999999999997e-06,
      "loss": 0.6394,
      "step": 1144
    },
    {
      "epoch": 2.1126500461680515,
      "grad_norm": 4.240537166595459,
      "learning_rate": 3.426e-06,
      "loss": 0.5845,
      "step": 1145
    },
    {
      "epoch": 2.11449676823638,
      "grad_norm": 4.361617088317871,
      "learning_rate": 3.429e-06,
      "loss": 0.5257,
      "step": 1146
    },
    {
      "epoch": 2.1163434903047094,
      "grad_norm": 5.02565860748291,
      "learning_rate": 3.4320000000000003e-06,
      "loss": 0.5765,
      "step": 1147
    },
    {
      "epoch": 2.118190212373038,
      "grad_norm": 2.9467175006866455,
      "learning_rate": 3.435e-06,
      "loss": 0.6254,
      "step": 1148
    },
    {
      "epoch": 2.1200369344413668,
      "grad_norm": 2.518874168395996,
      "learning_rate": 3.438e-06,
      "loss": 0.4371,
      "step": 1149
    },
    {
      "epoch": 2.1218836565096955,
      "grad_norm": 3.232793092727661,
      "learning_rate": 3.441e-06,
      "loss": 0.386,
      "step": 1150
    },
    {
      "epoch": 2.123730378578024,
      "grad_norm": 2.345193862915039,
      "learning_rate": 3.444e-06,
      "loss": 0.3993,
      "step": 1151
    },
    {
      "epoch": 2.125577100646353,
      "grad_norm": 4.542880535125732,
      "learning_rate": 3.447e-06,
      "loss": 0.4761,
      "step": 1152
    },
    {
      "epoch": 2.1274238227146816,
      "grad_norm": 5.272704124450684,
      "learning_rate": 3.4500000000000004e-06,
      "loss": 0.4417,
      "step": 1153
    },
    {
      "epoch": 2.1292705447830103,
      "grad_norm": 2.8965117931365967,
      "learning_rate": 3.453e-06,
      "loss": 0.409,
      "step": 1154
    },
    {
      "epoch": 2.131117266851339,
      "grad_norm": 15.167119026184082,
      "learning_rate": 3.456e-06,
      "loss": 0.3611,
      "step": 1155
    },
    {
      "epoch": 2.1329639889196677,
      "grad_norm": 5.197674751281738,
      "learning_rate": 3.4590000000000003e-06,
      "loss": 0.3397,
      "step": 1156
    },
    {
      "epoch": 2.1348107109879964,
      "grad_norm": 2.644599199295044,
      "learning_rate": 3.462e-06,
      "loss": 0.3809,
      "step": 1157
    },
    {
      "epoch": 2.136657433056325,
      "grad_norm": 14.982160568237305,
      "learning_rate": 3.4650000000000003e-06,
      "loss": 0.3453,
      "step": 1158
    },
    {
      "epoch": 2.1385041551246537,
      "grad_norm": 2.48482346534729,
      "learning_rate": 3.468e-06,
      "loss": 0.351,
      "step": 1159
    },
    {
      "epoch": 2.1403508771929824,
      "grad_norm": 2.524677038192749,
      "learning_rate": 3.471e-06,
      "loss": 0.306,
      "step": 1160
    },
    {
      "epoch": 2.142197599261311,
      "grad_norm": 4.211189270019531,
      "learning_rate": 3.474e-06,
      "loss": 0.3509,
      "step": 1161
    },
    {
      "epoch": 2.14404432132964,
      "grad_norm": 2.2646098136901855,
      "learning_rate": 3.4770000000000003e-06,
      "loss": 0.3004,
      "step": 1162
    },
    {
      "epoch": 2.1458910433979685,
      "grad_norm": 2.701345443725586,
      "learning_rate": 3.48e-06,
      "loss": 0.3862,
      "step": 1163
    },
    {
      "epoch": 2.1477377654662972,
      "grad_norm": 2.695441961288452,
      "learning_rate": 3.483e-06,
      "loss": 0.441,
      "step": 1164
    },
    {
      "epoch": 2.149584487534626,
      "grad_norm": 3.7059075832366943,
      "learning_rate": 3.486e-06,
      "loss": 0.3228,
      "step": 1165
    },
    {
      "epoch": 2.1514312096029546,
      "grad_norm": 5.615117073059082,
      "learning_rate": 3.489e-06,
      "loss": 0.4142,
      "step": 1166
    },
    {
      "epoch": 2.1532779316712833,
      "grad_norm": 3.990037441253662,
      "learning_rate": 3.492e-06,
      "loss": 0.3822,
      "step": 1167
    },
    {
      "epoch": 2.155124653739612,
      "grad_norm": 2.899252414703369,
      "learning_rate": 3.4950000000000004e-06,
      "loss": 0.3355,
      "step": 1168
    },
    {
      "epoch": 2.1569713758079407,
      "grad_norm": 3.141970634460449,
      "learning_rate": 3.498e-06,
      "loss": 0.3535,
      "step": 1169
    },
    {
      "epoch": 2.1588180978762694,
      "grad_norm": 2.840134620666504,
      "learning_rate": 3.501e-06,
      "loss": 0.3351,
      "step": 1170
    },
    {
      "epoch": 2.160664819944598,
      "grad_norm": 2.2682883739471436,
      "learning_rate": 3.5040000000000002e-06,
      "loss": 0.3443,
      "step": 1171
    },
    {
      "epoch": 2.1625115420129273,
      "grad_norm": 3.5608057975769043,
      "learning_rate": 3.507e-06,
      "loss": 0.3911,
      "step": 1172
    },
    {
      "epoch": 2.164358264081256,
      "grad_norm": 3.4463419914245605,
      "learning_rate": 3.5100000000000003e-06,
      "loss": 0.3563,
      "step": 1173
    },
    {
      "epoch": 2.1662049861495847,
      "grad_norm": 3.2165534496307373,
      "learning_rate": 3.513e-06,
      "loss": 0.3671,
      "step": 1174
    },
    {
      "epoch": 2.1680517082179134,
      "grad_norm": 4.781904220581055,
      "learning_rate": 3.516e-06,
      "loss": 0.3282,
      "step": 1175
    },
    {
      "epoch": 2.169898430286242,
      "grad_norm": 22.736854553222656,
      "learning_rate": 3.519e-06,
      "loss": 0.375,
      "step": 1176
    },
    {
      "epoch": 2.1717451523545708,
      "grad_norm": 3.0178444385528564,
      "learning_rate": 3.5220000000000003e-06,
      "loss": 0.3934,
      "step": 1177
    },
    {
      "epoch": 2.1735918744228995,
      "grad_norm": 7.673983573913574,
      "learning_rate": 3.5249999999999997e-06,
      "loss": 0.3651,
      "step": 1178
    },
    {
      "epoch": 2.175438596491228,
      "grad_norm": 2.905626058578491,
      "learning_rate": 3.528e-06,
      "loss": 0.3587,
      "step": 1179
    },
    {
      "epoch": 2.177285318559557,
      "grad_norm": 5.361052989959717,
      "learning_rate": 3.531e-06,
      "loss": 0.3424,
      "step": 1180
    },
    {
      "epoch": 2.1791320406278856,
      "grad_norm": 2.9514048099517822,
      "learning_rate": 3.534e-06,
      "loss": 0.3506,
      "step": 1181
    },
    {
      "epoch": 2.1809787626962143,
      "grad_norm": 6.453454494476318,
      "learning_rate": 3.537e-06,
      "loss": 0.4183,
      "step": 1182
    },
    {
      "epoch": 2.182825484764543,
      "grad_norm": 4.925314903259277,
      "learning_rate": 3.54e-06,
      "loss": 0.4085,
      "step": 1183
    },
    {
      "epoch": 2.1846722068328717,
      "grad_norm": 3.5400233268737793,
      "learning_rate": 3.543e-06,
      "loss": 0.3729,
      "step": 1184
    },
    {
      "epoch": 2.1865189289012004,
      "grad_norm": 7.280455589294434,
      "learning_rate": 3.546e-06,
      "loss": 1.11,
      "step": 1185
    },
    {
      "epoch": 2.188365650969529,
      "grad_norm": 3.9851276874542236,
      "learning_rate": 3.5490000000000002e-06,
      "loss": 0.9891,
      "step": 1186
    },
    {
      "epoch": 2.1902123730378578,
      "grad_norm": 3.3686535358428955,
      "learning_rate": 3.552e-06,
      "loss": 0.907,
      "step": 1187
    },
    {
      "epoch": 2.1920590951061865,
      "grad_norm": 3.4856481552124023,
      "learning_rate": 3.555e-06,
      "loss": 0.7925,
      "step": 1188
    },
    {
      "epoch": 2.193905817174515,
      "grad_norm": 3.7457706928253174,
      "learning_rate": 3.558e-06,
      "loss": 0.6853,
      "step": 1189
    },
    {
      "epoch": 2.195752539242844,
      "grad_norm": 5.772076606750488,
      "learning_rate": 3.5610000000000003e-06,
      "loss": 0.6675,
      "step": 1190
    },
    {
      "epoch": 2.1975992613111726,
      "grad_norm": 3.661681890487671,
      "learning_rate": 3.564e-06,
      "loss": 0.5446,
      "step": 1191
    },
    {
      "epoch": 2.1994459833795013,
      "grad_norm": 3.1582486629486084,
      "learning_rate": 3.5670000000000003e-06,
      "loss": 0.5409,
      "step": 1192
    },
    {
      "epoch": 2.20129270544783,
      "grad_norm": 3.120471954345703,
      "learning_rate": 3.57e-06,
      "loss": 0.5944,
      "step": 1193
    },
    {
      "epoch": 2.2031394275161587,
      "grad_norm": 3.0466837882995605,
      "learning_rate": 3.573e-06,
      "loss": 0.5486,
      "step": 1194
    },
    {
      "epoch": 2.2049861495844874,
      "grad_norm": 2.113574743270874,
      "learning_rate": 3.576e-06,
      "loss": 0.516,
      "step": 1195
    },
    {
      "epoch": 2.206832871652816,
      "grad_norm": 2.616821527481079,
      "learning_rate": 3.5790000000000004e-06,
      "loss": 0.5089,
      "step": 1196
    },
    {
      "epoch": 2.208679593721145,
      "grad_norm": 3.7034990787506104,
      "learning_rate": 3.582e-06,
      "loss": 0.4616,
      "step": 1197
    },
    {
      "epoch": 2.2105263157894735,
      "grad_norm": 3.5968942642211914,
      "learning_rate": 3.585e-06,
      "loss": 0.4608,
      "step": 1198
    },
    {
      "epoch": 2.2123730378578026,
      "grad_norm": 2.87996506690979,
      "learning_rate": 3.588e-06,
      "loss": 0.4004,
      "step": 1199
    },
    {
      "epoch": 2.2142197599261313,
      "grad_norm": 2.8459644317626953,
      "learning_rate": 3.591e-06,
      "loss": 0.4628,
      "step": 1200
    },
    {
      "epoch": 2.21606648199446,
      "grad_norm": 1.9606138467788696,
      "learning_rate": 3.5940000000000002e-06,
      "loss": 0.3203,
      "step": 1201
    },
    {
      "epoch": 2.2179132040627887,
      "grad_norm": 3.2116522789001465,
      "learning_rate": 3.5970000000000005e-06,
      "loss": 0.3311,
      "step": 1202
    },
    {
      "epoch": 2.2197599261311174,
      "grad_norm": 3.8338441848754883,
      "learning_rate": 3.6e-06,
      "loss": 0.3352,
      "step": 1203
    },
    {
      "epoch": 2.221606648199446,
      "grad_norm": 2.1496729850769043,
      "learning_rate": 3.603e-06,
      "loss": 0.3138,
      "step": 1204
    },
    {
      "epoch": 2.223453370267775,
      "grad_norm": 4.74179220199585,
      "learning_rate": 3.6060000000000003e-06,
      "loss": 0.3463,
      "step": 1205
    },
    {
      "epoch": 2.2253000923361035,
      "grad_norm": 2.1020402908325195,
      "learning_rate": 3.609e-06,
      "loss": 0.3382,
      "step": 1206
    },
    {
      "epoch": 2.227146814404432,
      "grad_norm": 3.37388277053833,
      "learning_rate": 3.612e-06,
      "loss": 0.3619,
      "step": 1207
    },
    {
      "epoch": 2.228993536472761,
      "grad_norm": 3.114112377166748,
      "learning_rate": 3.615e-06,
      "loss": 0.2963,
      "step": 1208
    },
    {
      "epoch": 2.2308402585410896,
      "grad_norm": 15.57221794128418,
      "learning_rate": 3.618e-06,
      "loss": 0.3247,
      "step": 1209
    },
    {
      "epoch": 2.2326869806094183,
      "grad_norm": 3.1523525714874268,
      "learning_rate": 3.621e-06,
      "loss": 0.3519,
      "step": 1210
    },
    {
      "epoch": 2.234533702677747,
      "grad_norm": 4.5500688552856445,
      "learning_rate": 3.6240000000000004e-06,
      "loss": 0.3248,
      "step": 1211
    },
    {
      "epoch": 2.2363804247460757,
      "grad_norm": 3.338387966156006,
      "learning_rate": 3.6269999999999997e-06,
      "loss": 0.2886,
      "step": 1212
    },
    {
      "epoch": 2.2382271468144044,
      "grad_norm": 13.98320484161377,
      "learning_rate": 3.63e-06,
      "loss": 0.3965,
      "step": 1213
    },
    {
      "epoch": 2.240073868882733,
      "grad_norm": 2.9857637882232666,
      "learning_rate": 3.633e-06,
      "loss": 0.322,
      "step": 1214
    },
    {
      "epoch": 2.2419205909510618,
      "grad_norm": 2.754495143890381,
      "learning_rate": 3.636e-06,
      "loss": 0.3352,
      "step": 1215
    },
    {
      "epoch": 2.2437673130193905,
      "grad_norm": 3.092918634414673,
      "learning_rate": 3.6390000000000002e-06,
      "loss": 0.2727,
      "step": 1216
    },
    {
      "epoch": 2.245614035087719,
      "grad_norm": 2.6611902713775635,
      "learning_rate": 3.642e-06,
      "loss": 0.3195,
      "step": 1217
    },
    {
      "epoch": 2.247460757156048,
      "grad_norm": 5.46854305267334,
      "learning_rate": 3.645e-06,
      "loss": 0.3389,
      "step": 1218
    },
    {
      "epoch": 2.2493074792243766,
      "grad_norm": 3.306468963623047,
      "learning_rate": 3.648e-06,
      "loss": 0.3301,
      "step": 1219
    },
    {
      "epoch": 2.2511542012927053,
      "grad_norm": 5.192465782165527,
      "learning_rate": 3.6510000000000003e-06,
      "loss": 0.2876,
      "step": 1220
    },
    {
      "epoch": 2.253000923361034,
      "grad_norm": 6.205870628356934,
      "learning_rate": 3.654e-06,
      "loss": 0.3413,
      "step": 1221
    },
    {
      "epoch": 2.254847645429363,
      "grad_norm": 5.976049900054932,
      "learning_rate": 3.657e-06,
      "loss": 0.3459,
      "step": 1222
    },
    {
      "epoch": 2.2566943674976914,
      "grad_norm": 4.196659564971924,
      "learning_rate": 3.66e-06,
      "loss": 0.3013,
      "step": 1223
    },
    {
      "epoch": 2.2585410895660205,
      "grad_norm": 4.1666107177734375,
      "learning_rate": 3.663e-06,
      "loss": 0.3416,
      "step": 1224
    },
    {
      "epoch": 2.260387811634349,
      "grad_norm": 6.647060394287109,
      "learning_rate": 3.666e-06,
      "loss": 0.3489,
      "step": 1225
    },
    {
      "epoch": 2.262234533702678,
      "grad_norm": 10.108695983886719,
      "learning_rate": 3.6690000000000004e-06,
      "loss": 0.3276,
      "step": 1226
    },
    {
      "epoch": 2.2640812557710066,
      "grad_norm": 3.9936745166778564,
      "learning_rate": 3.6719999999999997e-06,
      "loss": 0.3288,
      "step": 1227
    },
    {
      "epoch": 2.2659279778393353,
      "grad_norm": 3.6058974266052246,
      "learning_rate": 3.675e-06,
      "loss": 0.3284,
      "step": 1228
    },
    {
      "epoch": 2.267774699907664,
      "grad_norm": 9.84175968170166,
      "learning_rate": 3.678e-06,
      "loss": 0.3364,
      "step": 1229
    },
    {
      "epoch": 2.2696214219759927,
      "grad_norm": 4.596037864685059,
      "learning_rate": 3.681e-06,
      "loss": 0.348,
      "step": 1230
    },
    {
      "epoch": 2.2714681440443214,
      "grad_norm": 3.893502712249756,
      "learning_rate": 3.6840000000000002e-06,
      "loss": 0.3163,
      "step": 1231
    },
    {
      "epoch": 2.27331486611265,
      "grad_norm": 2.600052833557129,
      "learning_rate": 3.687e-06,
      "loss": 0.3415,
      "step": 1232
    },
    {
      "epoch": 2.275161588180979,
      "grad_norm": 3.8423099517822266,
      "learning_rate": 3.6900000000000002e-06,
      "loss": 0.3369,
      "step": 1233
    },
    {
      "epoch": 2.2770083102493075,
      "grad_norm": 5.047562599182129,
      "learning_rate": 3.693e-06,
      "loss": 0.3862,
      "step": 1234
    },
    {
      "epoch": 2.278855032317636,
      "grad_norm": 6.9906697273254395,
      "learning_rate": 3.6960000000000003e-06,
      "loss": 1.0494,
      "step": 1235
    },
    {
      "epoch": 2.280701754385965,
      "grad_norm": 4.055783748626709,
      "learning_rate": 3.6990000000000005e-06,
      "loss": 0.8679,
      "step": 1236
    },
    {
      "epoch": 2.2825484764542936,
      "grad_norm": 22.55494499206543,
      "learning_rate": 3.702e-06,
      "loss": 0.6641,
      "step": 1237
    },
    {
      "epoch": 2.2843951985226223,
      "grad_norm": 4.827858924865723,
      "learning_rate": 3.705e-06,
      "loss": 0.771,
      "step": 1238
    },
    {
      "epoch": 2.286241920590951,
      "grad_norm": 2.824592113494873,
      "learning_rate": 3.7080000000000003e-06,
      "loss": 0.5988,
      "step": 1239
    },
    {
      "epoch": 2.2880886426592797,
      "grad_norm": 2.9814367294311523,
      "learning_rate": 3.711e-06,
      "loss": 0.6778,
      "step": 1240
    },
    {
      "epoch": 2.2899353647276084,
      "grad_norm": 2.789565324783325,
      "learning_rate": 3.714e-06,
      "loss": 0.6152,
      "step": 1241
    },
    {
      "epoch": 2.291782086795937,
      "grad_norm": 7.15195369720459,
      "learning_rate": 3.717e-06,
      "loss": 0.6609,
      "step": 1242
    },
    {
      "epoch": 2.293628808864266,
      "grad_norm": 2.280348062515259,
      "learning_rate": 3.72e-06,
      "loss": 0.489,
      "step": 1243
    },
    {
      "epoch": 2.2954755309325945,
      "grad_norm": 2.8647119998931885,
      "learning_rate": 3.723e-06,
      "loss": 0.4645,
      "step": 1244
    },
    {
      "epoch": 2.297322253000923,
      "grad_norm": 2.9245121479034424,
      "learning_rate": 3.7260000000000004e-06,
      "loss": 0.5328,
      "step": 1245
    },
    {
      "epoch": 2.299168975069252,
      "grad_norm": 2.771186113357544,
      "learning_rate": 3.7289999999999998e-06,
      "loss": 0.4194,
      "step": 1246
    },
    {
      "epoch": 2.301015697137581,
      "grad_norm": 2.297076463699341,
      "learning_rate": 3.732e-06,
      "loss": 0.4229,
      "step": 1247
    },
    {
      "epoch": 2.3028624192059093,
      "grad_norm": 4.746479511260986,
      "learning_rate": 3.7350000000000002e-06,
      "loss": 0.3892,
      "step": 1248
    },
    {
      "epoch": 2.3047091412742384,
      "grad_norm": 4.564180374145508,
      "learning_rate": 3.738e-06,
      "loss": 0.2941,
      "step": 1249
    },
    {
      "epoch": 2.306555863342567,
      "grad_norm": 3.5963242053985596,
      "learning_rate": 3.7410000000000003e-06,
      "loss": 0.4098,
      "step": 1250
    },
    {
      "epoch": 2.308402585410896,
      "grad_norm": 3.23600435256958,
      "learning_rate": 3.744e-06,
      "loss": 0.3319,
      "step": 1251
    },
    {
      "epoch": 2.3102493074792245,
      "grad_norm": 1.7888182401657104,
      "learning_rate": 3.747e-06,
      "loss": 0.279,
      "step": 1252
    },
    {
      "epoch": 2.312096029547553,
      "grad_norm": 2.327932357788086,
      "learning_rate": 3.75e-06,
      "loss": 0.3066,
      "step": 1253
    },
    {
      "epoch": 2.313942751615882,
      "grad_norm": 2.380089282989502,
      "learning_rate": 3.753e-06,
      "loss": 0.3174,
      "step": 1254
    },
    {
      "epoch": 2.3157894736842106,
      "grad_norm": 6.162848949432373,
      "learning_rate": 3.756e-06,
      "loss": 0.291,
      "step": 1255
    },
    {
      "epoch": 2.3176361957525393,
      "grad_norm": 2.3624391555786133,
      "learning_rate": 3.759e-06,
      "loss": 0.3356,
      "step": 1256
    },
    {
      "epoch": 2.319482917820868,
      "grad_norm": 2.3133296966552734,
      "learning_rate": 3.7620000000000006e-06,
      "loss": 0.2871,
      "step": 1257
    },
    {
      "epoch": 2.3213296398891967,
      "grad_norm": 2.8012380599975586,
      "learning_rate": 3.765e-06,
      "loss": 0.2981,
      "step": 1258
    },
    {
      "epoch": 2.3231763619575254,
      "grad_norm": 6.965078353881836,
      "learning_rate": 3.7679999999999998e-06,
      "loss": 0.2726,
      "step": 1259
    },
    {
      "epoch": 2.325023084025854,
      "grad_norm": 2.270150899887085,
      "learning_rate": 3.7710000000000004e-06,
      "loss": 0.3289,
      "step": 1260
    },
    {
      "epoch": 2.326869806094183,
      "grad_norm": 12.200182914733887,
      "learning_rate": 3.7739999999999998e-06,
      "loss": 0.308,
      "step": 1261
    },
    {
      "epoch": 2.3287165281625115,
      "grad_norm": 3.335463285446167,
      "learning_rate": 3.7770000000000004e-06,
      "loss": 0.2571,
      "step": 1262
    },
    {
      "epoch": 2.33056325023084,
      "grad_norm": 3.326897144317627,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 0.275,
      "step": 1263
    },
    {
      "epoch": 2.332409972299169,
      "grad_norm": 2.531522274017334,
      "learning_rate": 3.7829999999999996e-06,
      "loss": 0.2572,
      "step": 1264
    },
    {
      "epoch": 2.3342566943674976,
      "grad_norm": 3.5164692401885986,
      "learning_rate": 3.7860000000000003e-06,
      "loss": 0.3005,
      "step": 1265
    },
    {
      "epoch": 2.3361034164358263,
      "grad_norm": 2.500251531600952,
      "learning_rate": 3.789e-06,
      "loss": 0.2758,
      "step": 1266
    },
    {
      "epoch": 2.337950138504155,
      "grad_norm": 2.022705078125,
      "learning_rate": 3.7920000000000003e-06,
      "loss": 0.2581,
      "step": 1267
    },
    {
      "epoch": 2.3397968605724837,
      "grad_norm": 3.829115629196167,
      "learning_rate": 3.795e-06,
      "loss": 0.2543,
      "step": 1268
    },
    {
      "epoch": 2.3416435826408124,
      "grad_norm": 2.7531073093414307,
      "learning_rate": 3.798e-06,
      "loss": 0.2756,
      "step": 1269
    },
    {
      "epoch": 2.343490304709141,
      "grad_norm": 3.9547321796417236,
      "learning_rate": 3.801e-06,
      "loss": 0.2785,
      "step": 1270
    },
    {
      "epoch": 2.34533702677747,
      "grad_norm": 2.6992123126983643,
      "learning_rate": 3.804e-06,
      "loss": 0.308,
      "step": 1271
    },
    {
      "epoch": 2.347183748845799,
      "grad_norm": 3.0379865169525146,
      "learning_rate": 3.8070000000000006e-06,
      "loss": 0.2926,
      "step": 1272
    },
    {
      "epoch": 2.349030470914127,
      "grad_norm": 7.398136615753174,
      "learning_rate": 3.81e-06,
      "loss": 0.2887,
      "step": 1273
    },
    {
      "epoch": 2.3508771929824563,
      "grad_norm": 2.398902177810669,
      "learning_rate": 3.8129999999999997e-06,
      "loss": 0.2795,
      "step": 1274
    },
    {
      "epoch": 2.352723915050785,
      "grad_norm": 3.1875367164611816,
      "learning_rate": 3.816e-06,
      "loss": 0.2811,
      "step": 1275
    },
    {
      "epoch": 2.3545706371191137,
      "grad_norm": 5.333486080169678,
      "learning_rate": 3.819e-06,
      "loss": 0.2525,
      "step": 1276
    },
    {
      "epoch": 2.3564173591874424,
      "grad_norm": 3.6999690532684326,
      "learning_rate": 3.822000000000001e-06,
      "loss": 0.3073,
      "step": 1277
    },
    {
      "epoch": 2.358264081255771,
      "grad_norm": 9.801976203918457,
      "learning_rate": 3.825e-06,
      "loss": 0.2757,
      "step": 1278
    },
    {
      "epoch": 2.3601108033241,
      "grad_norm": 3.6513302326202393,
      "learning_rate": 3.828e-06,
      "loss": 0.2975,
      "step": 1279
    },
    {
      "epoch": 2.3619575253924285,
      "grad_norm": 19.606441497802734,
      "learning_rate": 3.831e-06,
      "loss": 0.3085,
      "step": 1280
    },
    {
      "epoch": 2.3638042474607572,
      "grad_norm": 3.0111238956451416,
      "learning_rate": 3.834e-06,
      "loss": 0.4162,
      "step": 1281
    },
    {
      "epoch": 2.365650969529086,
      "grad_norm": 3.498967170715332,
      "learning_rate": 3.837000000000001e-06,
      "loss": 0.2412,
      "step": 1282
    },
    {
      "epoch": 2.3674976915974146,
      "grad_norm": 3.1680524349212646,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 0.321,
      "step": 1283
    },
    {
      "epoch": 2.3693444136657433,
      "grad_norm": 3.2572083473205566,
      "learning_rate": 3.8429999999999995e-06,
      "loss": 0.3806,
      "step": 1284
    },
    {
      "epoch": 2.371191135734072,
      "grad_norm": 8.028724670410156,
      "learning_rate": 3.846e-06,
      "loss": 0.9676,
      "step": 1285
    },
    {
      "epoch": 2.3730378578024007,
      "grad_norm": 3.0075464248657227,
      "learning_rate": 3.849e-06,
      "loss": 0.7188,
      "step": 1286
    },
    {
      "epoch": 2.3748845798707294,
      "grad_norm": 3.325401544570923,
      "learning_rate": 3.852e-06,
      "loss": 0.7748,
      "step": 1287
    },
    {
      "epoch": 2.376731301939058,
      "grad_norm": 3.004544973373413,
      "learning_rate": 3.855e-06,
      "loss": 0.6705,
      "step": 1288
    },
    {
      "epoch": 2.378578024007387,
      "grad_norm": 5.643571853637695,
      "learning_rate": 3.858e-06,
      "loss": 0.648,
      "step": 1289
    },
    {
      "epoch": 2.3804247460757155,
      "grad_norm": 3.7997190952301025,
      "learning_rate": 3.861e-06,
      "loss": 0.6225,
      "step": 1290
    },
    {
      "epoch": 2.3822714681440442,
      "grad_norm": 3.801635265350342,
      "learning_rate": 3.864e-06,
      "loss": 0.5235,
      "step": 1291
    },
    {
      "epoch": 2.384118190212373,
      "grad_norm": 6.873345375061035,
      "learning_rate": 3.8669999999999996e-06,
      "loss": 0.5504,
      "step": 1292
    },
    {
      "epoch": 2.3859649122807016,
      "grad_norm": 30.787782669067383,
      "learning_rate": 3.87e-06,
      "loss": 0.6031,
      "step": 1293
    },
    {
      "epoch": 2.3878116343490303,
      "grad_norm": 2.551400661468506,
      "learning_rate": 3.873e-06,
      "loss": 0.4685,
      "step": 1294
    },
    {
      "epoch": 2.389658356417359,
      "grad_norm": 2.9074532985687256,
      "learning_rate": 3.876000000000001e-06,
      "loss": 0.4373,
      "step": 1295
    },
    {
      "epoch": 2.3915050784856877,
      "grad_norm": 2.8484601974487305,
      "learning_rate": 3.8790000000000005e-06,
      "loss": 0.4662,
      "step": 1296
    },
    {
      "epoch": 2.393351800554017,
      "grad_norm": 1.9514797925949097,
      "learning_rate": 3.8819999999999994e-06,
      "loss": 0.3518,
      "step": 1297
    },
    {
      "epoch": 2.395198522622345,
      "grad_norm": 2.5072360038757324,
      "learning_rate": 3.885e-06,
      "loss": 0.3705,
      "step": 1298
    },
    {
      "epoch": 2.3970452446906743,
      "grad_norm": 3.407132863998413,
      "learning_rate": 3.888e-06,
      "loss": 0.4064,
      "step": 1299
    },
    {
      "epoch": 2.398891966759003,
      "grad_norm": 3.2116904258728027,
      "learning_rate": 3.8910000000000005e-06,
      "loss": 0.279,
      "step": 1300
    },
    {
      "epoch": 2.4007386888273317,
      "grad_norm": 2.4596047401428223,
      "learning_rate": 3.894e-06,
      "loss": 0.3032,
      "step": 1301
    },
    {
      "epoch": 2.4025854108956604,
      "grad_norm": 2.024303913116455,
      "learning_rate": 3.897e-06,
      "loss": 0.2755,
      "step": 1302
    },
    {
      "epoch": 2.404432132963989,
      "grad_norm": 2.66182017326355,
      "learning_rate": 3.9e-06,
      "loss": 0.367,
      "step": 1303
    },
    {
      "epoch": 2.4062788550323178,
      "grad_norm": 2.259300470352173,
      "learning_rate": 3.903e-06,
      "loss": 0.2667,
      "step": 1304
    },
    {
      "epoch": 2.4081255771006465,
      "grad_norm": 1.9543769359588623,
      "learning_rate": 3.906e-06,
      "loss": 0.2952,
      "step": 1305
    },
    {
      "epoch": 2.409972299168975,
      "grad_norm": 2.3687384128570557,
      "learning_rate": 3.909e-06,
      "loss": 0.2995,
      "step": 1306
    },
    {
      "epoch": 2.411819021237304,
      "grad_norm": 2.4152207374572754,
      "learning_rate": 3.912e-06,
      "loss": 0.2687,
      "step": 1307
    },
    {
      "epoch": 2.4136657433056325,
      "grad_norm": 2.8712761402130127,
      "learning_rate": 3.915000000000001e-06,
      "loss": 0.3216,
      "step": 1308
    },
    {
      "epoch": 2.4155124653739612,
      "grad_norm": 4.489912033081055,
      "learning_rate": 3.918e-06,
      "loss": 0.2717,
      "step": 1309
    },
    {
      "epoch": 2.41735918744229,
      "grad_norm": 5.658008575439453,
      "learning_rate": 3.921e-06,
      "loss": 0.2734,
      "step": 1310
    },
    {
      "epoch": 2.4192059095106186,
      "grad_norm": 4.131112575531006,
      "learning_rate": 3.924e-06,
      "loss": 0.2723,
      "step": 1311
    },
    {
      "epoch": 2.4210526315789473,
      "grad_norm": 2.3462915420532227,
      "learning_rate": 3.927e-06,
      "loss": 0.231,
      "step": 1312
    },
    {
      "epoch": 2.422899353647276,
      "grad_norm": 2.1342649459838867,
      "learning_rate": 3.9300000000000005e-06,
      "loss": 0.2303,
      "step": 1313
    },
    {
      "epoch": 2.4247460757156047,
      "grad_norm": 2.8530802726745605,
      "learning_rate": 3.933e-06,
      "loss": 0.2828,
      "step": 1314
    },
    {
      "epoch": 2.4265927977839334,
      "grad_norm": 5.358726978302002,
      "learning_rate": 3.936e-06,
      "loss": 0.2884,
      "step": 1315
    },
    {
      "epoch": 2.428439519852262,
      "grad_norm": 4.366053581237793,
      "learning_rate": 3.939e-06,
      "loss": 0.2985,
      "step": 1316
    },
    {
      "epoch": 2.430286241920591,
      "grad_norm": 2.7431342601776123,
      "learning_rate": 3.942e-06,
      "loss": 0.2599,
      "step": 1317
    },
    {
      "epoch": 2.4321329639889195,
      "grad_norm": 2.0079314708709717,
      "learning_rate": 3.945e-06,
      "loss": 0.2414,
      "step": 1318
    },
    {
      "epoch": 2.4339796860572482,
      "grad_norm": 2.379559278488159,
      "learning_rate": 3.948e-06,
      "loss": 0.2709,
      "step": 1319
    },
    {
      "epoch": 2.435826408125577,
      "grad_norm": 2.566490411758423,
      "learning_rate": 3.951000000000001e-06,
      "loss": 0.2579,
      "step": 1320
    },
    {
      "epoch": 2.4376731301939056,
      "grad_norm": 2.704746723175049,
      "learning_rate": 3.954e-06,
      "loss": 0.2446,
      "step": 1321
    },
    {
      "epoch": 2.439519852262235,
      "grad_norm": 2.497107744216919,
      "learning_rate": 3.9569999999999996e-06,
      "loss": 0.2526,
      "step": 1322
    },
    {
      "epoch": 2.441366574330563,
      "grad_norm": 2.8292629718780518,
      "learning_rate": 3.96e-06,
      "loss": 0.31,
      "step": 1323
    },
    {
      "epoch": 2.443213296398892,
      "grad_norm": 11.356720924377441,
      "learning_rate": 3.963e-06,
      "loss": 0.2827,
      "step": 1324
    },
    {
      "epoch": 2.445060018467221,
      "grad_norm": 2.888221263885498,
      "learning_rate": 3.966000000000001e-06,
      "loss": 0.2188,
      "step": 1325
    },
    {
      "epoch": 2.4469067405355496,
      "grad_norm": 2.1754820346832275,
      "learning_rate": 3.9690000000000005e-06,
      "loss": 0.3131,
      "step": 1326
    },
    {
      "epoch": 2.4487534626038783,
      "grad_norm": 3.7316973209381104,
      "learning_rate": 3.971999999999999e-06,
      "loss": 0.286,
      "step": 1327
    },
    {
      "epoch": 2.450600184672207,
      "grad_norm": 4.8265581130981445,
      "learning_rate": 3.975e-06,
      "loss": 0.3083,
      "step": 1328
    },
    {
      "epoch": 2.4524469067405357,
      "grad_norm": 2.3270061016082764,
      "learning_rate": 3.978e-06,
      "loss": 0.2119,
      "step": 1329
    },
    {
      "epoch": 2.4542936288088644,
      "grad_norm": 2.599588394165039,
      "learning_rate": 3.9810000000000005e-06,
      "loss": 0.2849,
      "step": 1330
    },
    {
      "epoch": 2.456140350877193,
      "grad_norm": 3.5899312496185303,
      "learning_rate": 3.984e-06,
      "loss": 0.3126,
      "step": 1331
    },
    {
      "epoch": 2.4579870729455218,
      "grad_norm": 5.132264137268066,
      "learning_rate": 3.987e-06,
      "loss": 0.3042,
      "step": 1332
    },
    {
      "epoch": 2.4598337950138505,
      "grad_norm": 2.8291397094726562,
      "learning_rate": 3.99e-06,
      "loss": 0.2809,
      "step": 1333
    },
    {
      "epoch": 2.461680517082179,
      "grad_norm": 3.450786828994751,
      "learning_rate": 3.993e-06,
      "loss": 0.3302,
      "step": 1334
    },
    {
      "epoch": 2.463527239150508,
      "grad_norm": 4.955879211425781,
      "learning_rate": 3.996e-06,
      "loss": 0.8375,
      "step": 1335
    },
    {
      "epoch": 2.4653739612188366,
      "grad_norm": 5.112767219543457,
      "learning_rate": 3.999e-06,
      "loss": 0.779,
      "step": 1336
    },
    {
      "epoch": 2.4672206832871653,
      "grad_norm": 3.635031223297119,
      "learning_rate": 4.002e-06,
      "loss": 0.6466,
      "step": 1337
    },
    {
      "epoch": 2.469067405355494,
      "grad_norm": 2.0046777725219727,
      "learning_rate": 4.005000000000001e-06,
      "loss": 0.7017,
      "step": 1338
    },
    {
      "epoch": 2.4709141274238227,
      "grad_norm": 3.8623645305633545,
      "learning_rate": 4.008e-06,
      "loss": 0.5682,
      "step": 1339
    },
    {
      "epoch": 2.4727608494921514,
      "grad_norm": 2.804150342941284,
      "learning_rate": 4.011e-06,
      "loss": 0.5209,
      "step": 1340
    },
    {
      "epoch": 2.47460757156048,
      "grad_norm": 2.1142044067382812,
      "learning_rate": 4.014e-06,
      "loss": 0.4553,
      "step": 1341
    },
    {
      "epoch": 2.4764542936288088,
      "grad_norm": 3.2303378582000732,
      "learning_rate": 4.017e-06,
      "loss": 0.5504,
      "step": 1342
    },
    {
      "epoch": 2.4783010156971375,
      "grad_norm": 2.541952610015869,
      "learning_rate": 4.0200000000000005e-06,
      "loss": 0.4207,
      "step": 1343
    },
    {
      "epoch": 2.480147737765466,
      "grad_norm": 7.729579925537109,
      "learning_rate": 4.023e-06,
      "loss": 0.4172,
      "step": 1344
    },
    {
      "epoch": 2.481994459833795,
      "grad_norm": 2.261575698852539,
      "learning_rate": 4.026000000000001e-06,
      "loss": 0.4869,
      "step": 1345
    },
    {
      "epoch": 2.4838411819021236,
      "grad_norm": 3.2447593212127686,
      "learning_rate": 4.029e-06,
      "loss": 0.4203,
      "step": 1346
    },
    {
      "epoch": 2.4856879039704527,
      "grad_norm": 2.228067398071289,
      "learning_rate": 4.032e-06,
      "loss": 0.36,
      "step": 1347
    },
    {
      "epoch": 2.487534626038781,
      "grad_norm": 2.548985719680786,
      "learning_rate": 4.035e-06,
      "loss": 0.3544,
      "step": 1348
    },
    {
      "epoch": 2.48938134810711,
      "grad_norm": 2.606369733810425,
      "learning_rate": 4.038e-06,
      "loss": 0.3304,
      "step": 1349
    },
    {
      "epoch": 2.4912280701754383,
      "grad_norm": 2.1755919456481934,
      "learning_rate": 4.041e-06,
      "loss": 0.3204,
      "step": 1350
    },
    {
      "epoch": 2.4930747922437675,
      "grad_norm": 2.264122724533081,
      "learning_rate": 4.044000000000001e-06,
      "loss": 0.2592,
      "step": 1351
    },
    {
      "epoch": 2.494921514312096,
      "grad_norm": 2.880892515182495,
      "learning_rate": 4.0469999999999995e-06,
      "loss": 0.2682,
      "step": 1352
    },
    {
      "epoch": 2.496768236380425,
      "grad_norm": 2.182246685028076,
      "learning_rate": 4.05e-06,
      "loss": 0.2418,
      "step": 1353
    },
    {
      "epoch": 2.4986149584487536,
      "grad_norm": 1.7105096578598022,
      "learning_rate": 4.053e-06,
      "loss": 0.2489,
      "step": 1354
    },
    {
      "epoch": 2.5004616805170823,
      "grad_norm": 4.293247222900391,
      "learning_rate": 4.056e-06,
      "loss": 0.2585,
      "step": 1355
    },
    {
      "epoch": 2.502308402585411,
      "grad_norm": 1.896207332611084,
      "learning_rate": 4.0590000000000004e-06,
      "loss": 0.2307,
      "step": 1356
    },
    {
      "epoch": 2.5041551246537397,
      "grad_norm": 3.0287654399871826,
      "learning_rate": 4.062e-06,
      "loss": 0.2302,
      "step": 1357
    },
    {
      "epoch": 2.5060018467220684,
      "grad_norm": 1.682420015335083,
      "learning_rate": 4.065e-06,
      "loss": 0.2106,
      "step": 1358
    },
    {
      "epoch": 2.507848568790397,
      "grad_norm": 2.477034330368042,
      "learning_rate": 4.068e-06,
      "loss": 0.2265,
      "step": 1359
    },
    {
      "epoch": 2.509695290858726,
      "grad_norm": 2.927973508834839,
      "learning_rate": 4.071e-06,
      "loss": 0.2257,
      "step": 1360
    },
    {
      "epoch": 2.5115420129270545,
      "grad_norm": 2.208219051361084,
      "learning_rate": 4.074e-06,
      "loss": 0.2496,
      "step": 1361
    },
    {
      "epoch": 2.513388734995383,
      "grad_norm": 2.2488653659820557,
      "learning_rate": 4.077e-06,
      "loss": 0.2835,
      "step": 1362
    },
    {
      "epoch": 2.515235457063712,
      "grad_norm": 2.399538993835449,
      "learning_rate": 4.080000000000001e-06,
      "loss": 0.2286,
      "step": 1363
    },
    {
      "epoch": 2.5170821791320406,
      "grad_norm": 3.4305288791656494,
      "learning_rate": 4.083e-06,
      "loss": 0.2554,
      "step": 1364
    },
    {
      "epoch": 2.5189289012003693,
      "grad_norm": 2.3506622314453125,
      "learning_rate": 4.0859999999999995e-06,
      "loss": 0.3603,
      "step": 1365
    },
    {
      "epoch": 2.520775623268698,
      "grad_norm": 2.6317431926727295,
      "learning_rate": 4.089e-06,
      "loss": 0.2436,
      "step": 1366
    },
    {
      "epoch": 2.5226223453370267,
      "grad_norm": 3.0093767642974854,
      "learning_rate": 4.092e-06,
      "loss": 0.2334,
      "step": 1367
    },
    {
      "epoch": 2.5244690674053554,
      "grad_norm": 3.368942975997925,
      "learning_rate": 4.095000000000001e-06,
      "loss": 0.208,
      "step": 1368
    },
    {
      "epoch": 2.526315789473684,
      "grad_norm": 2.1555802822113037,
      "learning_rate": 4.098e-06,
      "loss": 0.2505,
      "step": 1369
    },
    {
      "epoch": 2.5281625115420128,
      "grad_norm": 2.5781638622283936,
      "learning_rate": 4.100999999999999e-06,
      "loss": 0.219,
      "step": 1370
    },
    {
      "epoch": 2.5300092336103415,
      "grad_norm": 4.542846202850342,
      "learning_rate": 4.104e-06,
      "loss": 0.2179,
      "step": 1371
    },
    {
      "epoch": 2.5318559556786706,
      "grad_norm": 1.7584036588668823,
      "learning_rate": 4.107e-06,
      "loss": 0.1994,
      "step": 1372
    },
    {
      "epoch": 2.533702677746999,
      "grad_norm": 1.8268120288848877,
      "learning_rate": 4.1100000000000005e-06,
      "loss": 0.2002,
      "step": 1373
    },
    {
      "epoch": 2.535549399815328,
      "grad_norm": 2.2652413845062256,
      "learning_rate": 4.113e-06,
      "loss": 0.2568,
      "step": 1374
    },
    {
      "epoch": 2.5373961218836563,
      "grad_norm": 4.071157932281494,
      "learning_rate": 4.116e-06,
      "loss": 0.2373,
      "step": 1375
    },
    {
      "epoch": 2.5392428439519854,
      "grad_norm": 2.830423355102539,
      "learning_rate": 4.119e-06,
      "loss": 0.1984,
      "step": 1376
    },
    {
      "epoch": 2.541089566020314,
      "grad_norm": 2.961411237716675,
      "learning_rate": 4.122e-06,
      "loss": 0.2728,
      "step": 1377
    },
    {
      "epoch": 2.542936288088643,
      "grad_norm": 2.2902283668518066,
      "learning_rate": 4.125e-06,
      "loss": 0.2201,
      "step": 1378
    },
    {
      "epoch": 2.5447830101569715,
      "grad_norm": 2.1575589179992676,
      "learning_rate": 4.128e-06,
      "loss": 0.206,
      "step": 1379
    },
    {
      "epoch": 2.5466297322253,
      "grad_norm": 2.122411012649536,
      "learning_rate": 4.131e-06,
      "loss": 0.2204,
      "step": 1380
    },
    {
      "epoch": 2.548476454293629,
      "grad_norm": 2.2870776653289795,
      "learning_rate": 4.1340000000000006e-06,
      "loss": 0.267,
      "step": 1381
    },
    {
      "epoch": 2.5503231763619576,
      "grad_norm": 2.6069374084472656,
      "learning_rate": 4.137e-06,
      "loss": 0.2183,
      "step": 1382
    },
    {
      "epoch": 2.5521698984302863,
      "grad_norm": 2.5496888160705566,
      "learning_rate": 4.14e-06,
      "loss": 0.2969,
      "step": 1383
    },
    {
      "epoch": 2.554016620498615,
      "grad_norm": 4.724157333374023,
      "learning_rate": 4.143e-06,
      "loss": 0.253,
      "step": 1384
    },
    {
      "epoch": 2.5558633425669437,
      "grad_norm": 6.4206390380859375,
      "learning_rate": 4.146e-06,
      "loss": 0.8273,
      "step": 1385
    },
    {
      "epoch": 2.5577100646352724,
      "grad_norm": 6.399569511413574,
      "learning_rate": 4.1490000000000004e-06,
      "loss": 0.857,
      "step": 1386
    },
    {
      "epoch": 2.559556786703601,
      "grad_norm": 1.8726593255996704,
      "learning_rate": 4.152e-06,
      "loss": 0.5886,
      "step": 1387
    },
    {
      "epoch": 2.56140350877193,
      "grad_norm": 3.172929525375366,
      "learning_rate": 4.155000000000001e-06,
      "loss": 0.5749,
      "step": 1388
    },
    {
      "epoch": 2.5632502308402585,
      "grad_norm": 2.1683387756347656,
      "learning_rate": 4.158e-06,
      "loss": 0.5789,
      "step": 1389
    },
    {
      "epoch": 2.565096952908587,
      "grad_norm": 2.343235969543457,
      "learning_rate": 4.161e-06,
      "loss": 0.597,
      "step": 1390
    },
    {
      "epoch": 2.566943674976916,
      "grad_norm": 2.232675552368164,
      "learning_rate": 4.164e-06,
      "loss": 0.5261,
      "step": 1391
    },
    {
      "epoch": 2.5687903970452446,
      "grad_norm": 1.9980791807174683,
      "learning_rate": 4.167e-06,
      "loss": 0.4831,
      "step": 1392
    },
    {
      "epoch": 2.5706371191135733,
      "grad_norm": 2.491255521774292,
      "learning_rate": 4.170000000000001e-06,
      "loss": 0.5287,
      "step": 1393
    },
    {
      "epoch": 2.572483841181902,
      "grad_norm": 1.8359003067016602,
      "learning_rate": 4.1730000000000005e-06,
      "loss": 0.384,
      "step": 1394
    },
    {
      "epoch": 2.5743305632502307,
      "grad_norm": 2.0215978622436523,
      "learning_rate": 4.1759999999999995e-06,
      "loss": 0.3815,
      "step": 1395
    },
    {
      "epoch": 2.5761772853185594,
      "grad_norm": 1.9105162620544434,
      "learning_rate": 4.179e-06,
      "loss": 0.4254,
      "step": 1396
    },
    {
      "epoch": 2.5780240073868885,
      "grad_norm": 5.220210552215576,
      "learning_rate": 4.182e-06,
      "loss": 0.4206,
      "step": 1397
    },
    {
      "epoch": 2.579870729455217,
      "grad_norm": 1.760629653930664,
      "learning_rate": 4.185000000000001e-06,
      "loss": 0.2673,
      "step": 1398
    },
    {
      "epoch": 2.581717451523546,
      "grad_norm": 2.2482142448425293,
      "learning_rate": 4.188e-06,
      "loss": 0.271,
      "step": 1399
    },
    {
      "epoch": 2.583564173591874,
      "grad_norm": 1.5858904123306274,
      "learning_rate": 4.191e-06,
      "loss": 0.277,
      "step": 1400
    },
    {
      "epoch": 2.5854108956602033,
      "grad_norm": 2.4095640182495117,
      "learning_rate": 4.194e-06,
      "loss": 0.1968,
      "step": 1401
    },
    {
      "epoch": 2.587257617728532,
      "grad_norm": 1.7928316593170166,
      "learning_rate": 4.197e-06,
      "loss": 0.2765,
      "step": 1402
    },
    {
      "epoch": 2.5891043397968607,
      "grad_norm": 2.0820446014404297,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.2139,
      "step": 1403
    },
    {
      "epoch": 2.5909510618651894,
      "grad_norm": 1.5978951454162598,
      "learning_rate": 4.203e-06,
      "loss": 0.1674,
      "step": 1404
    },
    {
      "epoch": 2.592797783933518,
      "grad_norm": 1.8610577583312988,
      "learning_rate": 4.206e-06,
      "loss": 0.1904,
      "step": 1405
    },
    {
      "epoch": 2.594644506001847,
      "grad_norm": 2.742856740951538,
      "learning_rate": 4.209000000000001e-06,
      "loss": 0.2835,
      "step": 1406
    },
    {
      "epoch": 2.5964912280701755,
      "grad_norm": 1.738468885421753,
      "learning_rate": 4.212e-06,
      "loss": 0.1867,
      "step": 1407
    },
    {
      "epoch": 2.598337950138504,
      "grad_norm": 7.230634689331055,
      "learning_rate": 4.215e-06,
      "loss": 0.231,
      "step": 1408
    },
    {
      "epoch": 2.600184672206833,
      "grad_norm": 2.0607056617736816,
      "learning_rate": 4.218e-06,
      "loss": 0.268,
      "step": 1409
    },
    {
      "epoch": 2.6020313942751616,
      "grad_norm": 2.3839364051818848,
      "learning_rate": 4.221e-06,
      "loss": 0.2136,
      "step": 1410
    },
    {
      "epoch": 2.6038781163434903,
      "grad_norm": 3.0196101665496826,
      "learning_rate": 4.2240000000000006e-06,
      "loss": 0.1609,
      "step": 1411
    },
    {
      "epoch": 2.605724838411819,
      "grad_norm": 2.1394782066345215,
      "learning_rate": 4.227e-06,
      "loss": 0.3409,
      "step": 1412
    },
    {
      "epoch": 2.6075715604801477,
      "grad_norm": 2.8312456607818604,
      "learning_rate": 4.229999999999999e-06,
      "loss": 0.2025,
      "step": 1413
    },
    {
      "epoch": 2.6094182825484764,
      "grad_norm": 2.4542574882507324,
      "learning_rate": 4.233e-06,
      "loss": 0.1856,
      "step": 1414
    },
    {
      "epoch": 2.611265004616805,
      "grad_norm": 3.183842658996582,
      "learning_rate": 4.236e-06,
      "loss": 0.228,
      "step": 1415
    },
    {
      "epoch": 2.613111726685134,
      "grad_norm": 2.6048052310943604,
      "learning_rate": 4.239e-06,
      "loss": 0.2479,
      "step": 1416
    },
    {
      "epoch": 2.6149584487534625,
      "grad_norm": 2.054537773132324,
      "learning_rate": 4.242e-06,
      "loss": 0.2015,
      "step": 1417
    },
    {
      "epoch": 2.616805170821791,
      "grad_norm": 2.2461042404174805,
      "learning_rate": 4.245e-06,
      "loss": 0.2257,
      "step": 1418
    },
    {
      "epoch": 2.61865189289012,
      "grad_norm": 2.653735876083374,
      "learning_rate": 4.248e-06,
      "loss": 0.2595,
      "step": 1419
    },
    {
      "epoch": 2.6204986149584486,
      "grad_norm": 1.8901677131652832,
      "learning_rate": 4.251e-06,
      "loss": 0.1943,
      "step": 1420
    },
    {
      "epoch": 2.6223453370267773,
      "grad_norm": 2.2802734375,
      "learning_rate": 4.254e-06,
      "loss": 0.2195,
      "step": 1421
    },
    {
      "epoch": 2.6241920590951064,
      "grad_norm": 2.666233539581299,
      "learning_rate": 4.257e-06,
      "loss": 0.3164,
      "step": 1422
    },
    {
      "epoch": 2.6260387811634347,
      "grad_norm": 3.170790195465088,
      "learning_rate": 4.26e-06,
      "loss": 0.221,
      "step": 1423
    },
    {
      "epoch": 2.627885503231764,
      "grad_norm": 1.983960747718811,
      "learning_rate": 4.2630000000000005e-06,
      "loss": 0.2229,
      "step": 1424
    },
    {
      "epoch": 2.629732225300092,
      "grad_norm": 2.3078384399414062,
      "learning_rate": 4.266e-06,
      "loss": 0.2237,
      "step": 1425
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 1.8761299848556519,
      "learning_rate": 4.269e-06,
      "loss": 0.2458,
      "step": 1426
    },
    {
      "epoch": 2.6334256694367495,
      "grad_norm": 3.2777647972106934,
      "learning_rate": 4.272e-06,
      "loss": 0.2519,
      "step": 1427
    },
    {
      "epoch": 2.6352723915050786,
      "grad_norm": 2.3433635234832764,
      "learning_rate": 4.275e-06,
      "loss": 0.2276,
      "step": 1428
    },
    {
      "epoch": 2.6371191135734073,
      "grad_norm": 3.0046474933624268,
      "learning_rate": 4.278e-06,
      "loss": 0.215,
      "step": 1429
    },
    {
      "epoch": 2.638965835641736,
      "grad_norm": 2.6063246726989746,
      "learning_rate": 4.281e-06,
      "loss": 0.2128,
      "step": 1430
    },
    {
      "epoch": 2.6408125577100647,
      "grad_norm": 2.668186902999878,
      "learning_rate": 4.284000000000001e-06,
      "loss": 0.2393,
      "step": 1431
    },
    {
      "epoch": 2.6426592797783934,
      "grad_norm": 2.6497902870178223,
      "learning_rate": 4.287e-06,
      "loss": 0.2509,
      "step": 1432
    },
    {
      "epoch": 2.644506001846722,
      "grad_norm": 2.981414556503296,
      "learning_rate": 4.29e-06,
      "loss": 0.2447,
      "step": 1433
    },
    {
      "epoch": 2.646352723915051,
      "grad_norm": 2.751962900161743,
      "learning_rate": 4.293e-06,
      "loss": 0.2624,
      "step": 1434
    },
    {
      "epoch": 2.6481994459833795,
      "grad_norm": 4.448115825653076,
      "learning_rate": 4.296e-06,
      "loss": 0.8055,
      "step": 1435
    },
    {
      "epoch": 2.6500461680517082,
      "grad_norm": 2.5063092708587646,
      "learning_rate": 4.299000000000001e-06,
      "loss": 0.7689,
      "step": 1436
    },
    {
      "epoch": 2.651892890120037,
      "grad_norm": 2.950172185897827,
      "learning_rate": 4.3020000000000005e-06,
      "loss": 0.6302,
      "step": 1437
    },
    {
      "epoch": 2.6537396121883656,
      "grad_norm": 3.074154853820801,
      "learning_rate": 4.3049999999999994e-06,
      "loss": 0.5561,
      "step": 1438
    },
    {
      "epoch": 2.6555863342566943,
      "grad_norm": 2.672722101211548,
      "learning_rate": 4.308e-06,
      "loss": 0.605,
      "step": 1439
    },
    {
      "epoch": 2.657433056325023,
      "grad_norm": 1.9718472957611084,
      "learning_rate": 4.311e-06,
      "loss": 0.4401,
      "step": 1440
    },
    {
      "epoch": 2.6592797783933517,
      "grad_norm": 1.7161527872085571,
      "learning_rate": 4.3140000000000005e-06,
      "loss": 0.4255,
      "step": 1441
    },
    {
      "epoch": 2.6611265004616804,
      "grad_norm": 2.125276803970337,
      "learning_rate": 4.317e-06,
      "loss": 0.4338,
      "step": 1442
    },
    {
      "epoch": 2.662973222530009,
      "grad_norm": 2.82174015045166,
      "learning_rate": 4.32e-06,
      "loss": 0.4755,
      "step": 1443
    },
    {
      "epoch": 2.664819944598338,
      "grad_norm": 1.9432512521743774,
      "learning_rate": 4.323e-06,
      "loss": 0.5066,
      "step": 1444
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 1.7846685647964478,
      "learning_rate": 4.326e-06,
      "loss": 0.3927,
      "step": 1445
    },
    {
      "epoch": 2.668513388734995,
      "grad_norm": 2.2068796157836914,
      "learning_rate": 4.329e-06,
      "loss": 0.3077,
      "step": 1446
    },
    {
      "epoch": 2.6703601108033244,
      "grad_norm": 2.306232213973999,
      "learning_rate": 4.332e-06,
      "loss": 0.3395,
      "step": 1447
    },
    {
      "epoch": 2.6722068328716526,
      "grad_norm": 2.3692593574523926,
      "learning_rate": 4.335e-06,
      "loss": 0.2894,
      "step": 1448
    },
    {
      "epoch": 2.6740535549399818,
      "grad_norm": 1.493941307067871,
      "learning_rate": 4.338000000000001e-06,
      "loss": 0.3051,
      "step": 1449
    },
    {
      "epoch": 2.67590027700831,
      "grad_norm": 2.4704809188842773,
      "learning_rate": 4.341e-06,
      "loss": 0.3822,
      "step": 1450
    },
    {
      "epoch": 2.677746999076639,
      "grad_norm": 2.337467670440674,
      "learning_rate": 4.344e-06,
      "loss": 0.2224,
      "step": 1451
    },
    {
      "epoch": 2.6795937211449674,
      "grad_norm": 1.7623108625411987,
      "learning_rate": 4.347e-06,
      "loss": 0.1888,
      "step": 1452
    },
    {
      "epoch": 2.6814404432132966,
      "grad_norm": 3.122779369354248,
      "learning_rate": 4.35e-06,
      "loss": 0.2508,
      "step": 1453
    },
    {
      "epoch": 2.6832871652816253,
      "grad_norm": 2.7472877502441406,
      "learning_rate": 4.3530000000000005e-06,
      "loss": 0.1959,
      "step": 1454
    },
    {
      "epoch": 2.685133887349954,
      "grad_norm": 1.7916723489761353,
      "learning_rate": 4.356e-06,
      "loss": 0.193,
      "step": 1455
    },
    {
      "epoch": 2.6869806094182827,
      "grad_norm": 1.9161310195922852,
      "learning_rate": 4.359e-06,
      "loss": 0.2094,
      "step": 1456
    },
    {
      "epoch": 2.6888273314866113,
      "grad_norm": 1.7572247982025146,
      "learning_rate": 4.362e-06,
      "loss": 0.1904,
      "step": 1457
    },
    {
      "epoch": 2.69067405355494,
      "grad_norm": 1.8705005645751953,
      "learning_rate": 4.365e-06,
      "loss": 0.1926,
      "step": 1458
    },
    {
      "epoch": 2.6925207756232687,
      "grad_norm": 4.131222724914551,
      "learning_rate": 4.368e-06,
      "loss": 0.1821,
      "step": 1459
    },
    {
      "epoch": 2.6943674976915974,
      "grad_norm": 1.7637590169906616,
      "learning_rate": 4.371e-06,
      "loss": 0.2116,
      "step": 1460
    },
    {
      "epoch": 2.696214219759926,
      "grad_norm": 2.1233274936676025,
      "learning_rate": 4.374000000000001e-06,
      "loss": 0.1705,
      "step": 1461
    },
    {
      "epoch": 2.698060941828255,
      "grad_norm": 2.2717113494873047,
      "learning_rate": 4.377e-06,
      "loss": 0.1745,
      "step": 1462
    },
    {
      "epoch": 2.6999076638965835,
      "grad_norm": 1.8533835411071777,
      "learning_rate": 4.3799999999999996e-06,
      "loss": 0.211,
      "step": 1463
    },
    {
      "epoch": 2.7017543859649122,
      "grad_norm": 1.4384863376617432,
      "learning_rate": 4.383e-06,
      "loss": 0.1767,
      "step": 1464
    },
    {
      "epoch": 2.703601108033241,
      "grad_norm": 4.695356369018555,
      "learning_rate": 4.386e-06,
      "loss": 0.2391,
      "step": 1465
    },
    {
      "epoch": 2.7054478301015696,
      "grad_norm": 1.8919119834899902,
      "learning_rate": 4.389000000000001e-06,
      "loss": 0.2206,
      "step": 1466
    },
    {
      "epoch": 2.7072945521698983,
      "grad_norm": 2.0214920043945312,
      "learning_rate": 4.3920000000000005e-06,
      "loss": 0.2006,
      "step": 1467
    },
    {
      "epoch": 2.709141274238227,
      "grad_norm": 2.4652929306030273,
      "learning_rate": 4.395e-06,
      "loss": 0.1902,
      "step": 1468
    },
    {
      "epoch": 2.7109879963065557,
      "grad_norm": 2.2650187015533447,
      "learning_rate": 4.398e-06,
      "loss": 0.1999,
      "step": 1469
    },
    {
      "epoch": 2.7128347183748844,
      "grad_norm": 2.400144338607788,
      "learning_rate": 4.401e-06,
      "loss": 0.2368,
      "step": 1470
    },
    {
      "epoch": 2.714681440443213,
      "grad_norm": 2.085756540298462,
      "learning_rate": 4.4040000000000005e-06,
      "loss": 0.2032,
      "step": 1471
    },
    {
      "epoch": 2.7165281625115423,
      "grad_norm": 3.604992628097534,
      "learning_rate": 4.407e-06,
      "loss": 0.2296,
      "step": 1472
    },
    {
      "epoch": 2.7183748845798705,
      "grad_norm": 1.75570547580719,
      "learning_rate": 4.41e-06,
      "loss": 0.1673,
      "step": 1473
    },
    {
      "epoch": 2.7202216066481997,
      "grad_norm": 2.396017551422119,
      "learning_rate": 4.413000000000001e-06,
      "loss": 0.1703,
      "step": 1474
    },
    {
      "epoch": 2.722068328716528,
      "grad_norm": 2.6366143226623535,
      "learning_rate": 4.416e-06,
      "loss": 0.2345,
      "step": 1475
    },
    {
      "epoch": 2.723915050784857,
      "grad_norm": 1.9363056421279907,
      "learning_rate": 4.4189999999999995e-06,
      "loss": 0.1987,
      "step": 1476
    },
    {
      "epoch": 2.7257617728531853,
      "grad_norm": 2.850567102432251,
      "learning_rate": 4.422e-06,
      "loss": 0.1939,
      "step": 1477
    },
    {
      "epoch": 2.7276084949215145,
      "grad_norm": 2.088304281234741,
      "learning_rate": 4.425e-06,
      "loss": 0.2129,
      "step": 1478
    },
    {
      "epoch": 2.729455216989843,
      "grad_norm": 2.7488322257995605,
      "learning_rate": 4.428000000000001e-06,
      "loss": 0.2247,
      "step": 1479
    },
    {
      "epoch": 2.731301939058172,
      "grad_norm": 3.1374897956848145,
      "learning_rate": 4.4310000000000004e-06,
      "loss": 0.2246,
      "step": 1480
    },
    {
      "epoch": 2.7331486611265006,
      "grad_norm": 2.563474416732788,
      "learning_rate": 4.433999999999999e-06,
      "loss": 0.2151,
      "step": 1481
    },
    {
      "epoch": 2.7349953831948293,
      "grad_norm": 2.290768623352051,
      "learning_rate": 4.437e-06,
      "loss": 0.2009,
      "step": 1482
    },
    {
      "epoch": 2.736842105263158,
      "grad_norm": 3.419640302658081,
      "learning_rate": 4.44e-06,
      "loss": 0.2311,
      "step": 1483
    },
    {
      "epoch": 2.7386888273314867,
      "grad_norm": 3.651583671569824,
      "learning_rate": 4.4430000000000005e-06,
      "loss": 0.303,
      "step": 1484
    },
    {
      "epoch": 2.7405355493998154,
      "grad_norm": 2.9224209785461426,
      "learning_rate": 4.446e-06,
      "loss": 0.6854,
      "step": 1485
    },
    {
      "epoch": 2.742382271468144,
      "grad_norm": 2.3546254634857178,
      "learning_rate": 4.449e-06,
      "loss": 0.6776,
      "step": 1486
    },
    {
      "epoch": 2.7442289935364728,
      "grad_norm": 2.4229791164398193,
      "learning_rate": 4.452e-06,
      "loss": 0.495,
      "step": 1487
    },
    {
      "epoch": 2.7460757156048015,
      "grad_norm": 2.2081387042999268,
      "learning_rate": 4.455e-06,
      "loss": 0.488,
      "step": 1488
    },
    {
      "epoch": 2.74792243767313,
      "grad_norm": 1.6628825664520264,
      "learning_rate": 4.458e-06,
      "loss": 0.4709,
      "step": 1489
    },
    {
      "epoch": 2.749769159741459,
      "grad_norm": 2.055128335952759,
      "learning_rate": 4.461e-06,
      "loss": 0.4067,
      "step": 1490
    },
    {
      "epoch": 2.7516158818097876,
      "grad_norm": 4.5744194984436035,
      "learning_rate": 4.464e-06,
      "loss": 0.3766,
      "step": 1491
    },
    {
      "epoch": 2.7534626038781163,
      "grad_norm": 2.0984883308410645,
      "learning_rate": 4.467000000000001e-06,
      "loss": 0.4352,
      "step": 1492
    },
    {
      "epoch": 2.755309325946445,
      "grad_norm": 1.9840891361236572,
      "learning_rate": 4.4699999999999996e-06,
      "loss": 0.42,
      "step": 1493
    },
    {
      "epoch": 2.7571560480147737,
      "grad_norm": 1.4575275182724,
      "learning_rate": 4.473e-06,
      "loss": 0.3601,
      "step": 1494
    },
    {
      "epoch": 2.7590027700831024,
      "grad_norm": 2.2698862552642822,
      "learning_rate": 4.476e-06,
      "loss": 0.3062,
      "step": 1495
    },
    {
      "epoch": 2.760849492151431,
      "grad_norm": 2.3743550777435303,
      "learning_rate": 4.479e-06,
      "loss": 0.4122,
      "step": 1496
    },
    {
      "epoch": 2.76269621421976,
      "grad_norm": 2.3653502464294434,
      "learning_rate": 4.4820000000000005e-06,
      "loss": 0.446,
      "step": 1497
    },
    {
      "epoch": 2.7645429362880884,
      "grad_norm": 1.8148143291473389,
      "learning_rate": 4.485e-06,
      "loss": 0.3055,
      "step": 1498
    },
    {
      "epoch": 2.7663896583564176,
      "grad_norm": 1.6854664087295532,
      "learning_rate": 4.488e-06,
      "loss": 0.2572,
      "step": 1499
    },
    {
      "epoch": 2.768236380424746,
      "grad_norm": 2.335649013519287,
      "learning_rate": 4.491e-06,
      "loss": 0.2348,
      "step": 1500
    },
    {
      "epoch": 2.770083102493075,
      "grad_norm": 2.0741183757781982,
      "learning_rate": 4.494e-06,
      "loss": 0.2342,
      "step": 1501
    },
    {
      "epoch": 2.7719298245614032,
      "grad_norm": 1.6792997121810913,
      "learning_rate": 4.497e-06,
      "loss": 0.1631,
      "step": 1502
    },
    {
      "epoch": 2.7737765466297324,
      "grad_norm": 2.1965067386627197,
      "learning_rate": 4.5e-06,
      "loss": 0.2065,
      "step": 1503
    },
    {
      "epoch": 2.775623268698061,
      "grad_norm": 3.105736017227173,
      "learning_rate": 4.503000000000001e-06,
      "loss": 0.1888,
      "step": 1504
    },
    {
      "epoch": 2.77746999076639,
      "grad_norm": 1.6356396675109863,
      "learning_rate": 4.506e-06,
      "loss": 0.2168,
      "step": 1505
    },
    {
      "epoch": 2.7793167128347185,
      "grad_norm": 1.6595216989517212,
      "learning_rate": 4.5089999999999995e-06,
      "loss": 0.1784,
      "step": 1506
    },
    {
      "epoch": 2.781163434903047,
      "grad_norm": 1.616804599761963,
      "learning_rate": 4.512e-06,
      "loss": 0.193,
      "step": 1507
    },
    {
      "epoch": 2.783010156971376,
      "grad_norm": 2.185166597366333,
      "learning_rate": 4.515e-06,
      "loss": 0.2007,
      "step": 1508
    },
    {
      "epoch": 2.7848568790397046,
      "grad_norm": 1.600041151046753,
      "learning_rate": 4.518000000000001e-06,
      "loss": 0.1807,
      "step": 1509
    },
    {
      "epoch": 2.7867036011080333,
      "grad_norm": 2.245534896850586,
      "learning_rate": 4.521e-06,
      "loss": 0.1933,
      "step": 1510
    },
    {
      "epoch": 2.788550323176362,
      "grad_norm": 2.0908350944519043,
      "learning_rate": 4.524e-06,
      "loss": 0.1914,
      "step": 1511
    },
    {
      "epoch": 2.7903970452446907,
      "grad_norm": 1.715218424797058,
      "learning_rate": 4.527e-06,
      "loss": 0.175,
      "step": 1512
    },
    {
      "epoch": 2.7922437673130194,
      "grad_norm": 1.8380025625228882,
      "learning_rate": 4.53e-06,
      "loss": 0.1799,
      "step": 1513
    },
    {
      "epoch": 2.794090489381348,
      "grad_norm": 3.243588447570801,
      "learning_rate": 4.5330000000000005e-06,
      "loss": 0.183,
      "step": 1514
    },
    {
      "epoch": 2.7959372114496768,
      "grad_norm": 2.1115996837615967,
      "learning_rate": 4.536e-06,
      "loss": 0.1827,
      "step": 1515
    },
    {
      "epoch": 2.7977839335180055,
      "grad_norm": 3.3165128231048584,
      "learning_rate": 4.539e-06,
      "loss": 0.1978,
      "step": 1516
    },
    {
      "epoch": 2.799630655586334,
      "grad_norm": 1.9535701274871826,
      "learning_rate": 4.542000000000001e-06,
      "loss": 0.1685,
      "step": 1517
    },
    {
      "epoch": 2.801477377654663,
      "grad_norm": 1.7053006887435913,
      "learning_rate": 4.545e-06,
      "loss": 0.1515,
      "step": 1518
    },
    {
      "epoch": 2.8033240997229916,
      "grad_norm": 2.9324450492858887,
      "learning_rate": 4.548e-06,
      "loss": 0.1939,
      "step": 1519
    },
    {
      "epoch": 2.8051708217913203,
      "grad_norm": 1.7410835027694702,
      "learning_rate": 4.551e-06,
      "loss": 0.1666,
      "step": 1520
    },
    {
      "epoch": 2.807017543859649,
      "grad_norm": 1.9593206644058228,
      "learning_rate": 4.554e-06,
      "loss": 0.1814,
      "step": 1521
    },
    {
      "epoch": 2.808864265927978,
      "grad_norm": 6.398316383361816,
      "learning_rate": 4.557000000000001e-06,
      "loss": 0.2177,
      "step": 1522
    },
    {
      "epoch": 2.8107109879963064,
      "grad_norm": 2.2251827716827393,
      "learning_rate": 4.56e-06,
      "loss": 0.2002,
      "step": 1523
    },
    {
      "epoch": 2.8125577100646355,
      "grad_norm": 2.4644134044647217,
      "learning_rate": 4.563e-06,
      "loss": 0.1882,
      "step": 1524
    },
    {
      "epoch": 2.8144044321329638,
      "grad_norm": 2.2142317295074463,
      "learning_rate": 4.566e-06,
      "loss": 0.1932,
      "step": 1525
    },
    {
      "epoch": 2.816251154201293,
      "grad_norm": 3.522850751876831,
      "learning_rate": 4.569e-06,
      "loss": 0.1872,
      "step": 1526
    },
    {
      "epoch": 2.818097876269621,
      "grad_norm": 1.950206995010376,
      "learning_rate": 4.5720000000000004e-06,
      "loss": 0.1613,
      "step": 1527
    },
    {
      "epoch": 2.8199445983379503,
      "grad_norm": 1.9772347211837769,
      "learning_rate": 4.575e-06,
      "loss": 0.1761,
      "step": 1528
    },
    {
      "epoch": 2.821791320406279,
      "grad_norm": 2.442274570465088,
      "learning_rate": 4.578000000000001e-06,
      "loss": 0.186,
      "step": 1529
    },
    {
      "epoch": 2.8236380424746077,
      "grad_norm": 2.1009490489959717,
      "learning_rate": 4.581e-06,
      "loss": 0.1791,
      "step": 1530
    },
    {
      "epoch": 2.8254847645429364,
      "grad_norm": 3.239732027053833,
      "learning_rate": 4.584e-06,
      "loss": 0.2082,
      "step": 1531
    },
    {
      "epoch": 2.827331486611265,
      "grad_norm": 2.345767021179199,
      "learning_rate": 4.587e-06,
      "loss": 0.1945,
      "step": 1532
    },
    {
      "epoch": 2.829178208679594,
      "grad_norm": 3.2505600452423096,
      "learning_rate": 4.59e-06,
      "loss": 0.2387,
      "step": 1533
    },
    {
      "epoch": 2.8310249307479225,
      "grad_norm": 4.839790344238281,
      "learning_rate": 4.593000000000001e-06,
      "loss": 0.3456,
      "step": 1534
    },
    {
      "epoch": 2.832871652816251,
      "grad_norm": 2.4936726093292236,
      "learning_rate": 4.5960000000000006e-06,
      "loss": 0.7057,
      "step": 1535
    },
    {
      "epoch": 2.83471837488458,
      "grad_norm": 3.3598785400390625,
      "learning_rate": 4.5989999999999995e-06,
      "loss": 0.5805,
      "step": 1536
    },
    {
      "epoch": 2.8365650969529086,
      "grad_norm": 1.8686054944992065,
      "learning_rate": 4.602e-06,
      "loss": 0.5297,
      "step": 1537
    },
    {
      "epoch": 2.8384118190212373,
      "grad_norm": 1.5554466247558594,
      "learning_rate": 4.605e-06,
      "loss": 0.4223,
      "step": 1538
    },
    {
      "epoch": 2.840258541089566,
      "grad_norm": 2.1304216384887695,
      "learning_rate": 4.608e-06,
      "loss": 0.4202,
      "step": 1539
    },
    {
      "epoch": 2.8421052631578947,
      "grad_norm": 2.460719108581543,
      "learning_rate": 4.611e-06,
      "loss": 0.4698,
      "step": 1540
    },
    {
      "epoch": 2.8439519852262234,
      "grad_norm": 1.9906436204910278,
      "learning_rate": 4.614e-06,
      "loss": 0.4742,
      "step": 1541
    },
    {
      "epoch": 2.845798707294552,
      "grad_norm": 2.511833906173706,
      "learning_rate": 4.617e-06,
      "loss": 0.3593,
      "step": 1542
    },
    {
      "epoch": 2.847645429362881,
      "grad_norm": 2.1805901527404785,
      "learning_rate": 4.62e-06,
      "loss": 0.3863,
      "step": 1543
    },
    {
      "epoch": 2.8494921514312095,
      "grad_norm": 1.7413595914840698,
      "learning_rate": 4.623e-06,
      "loss": 0.4269,
      "step": 1544
    },
    {
      "epoch": 2.851338873499538,
      "grad_norm": 2.286431312561035,
      "learning_rate": 4.626e-06,
      "loss": 0.3202,
      "step": 1545
    },
    {
      "epoch": 2.853185595567867,
      "grad_norm": 2.3051440715789795,
      "learning_rate": 4.629e-06,
      "loss": 0.3118,
      "step": 1546
    },
    {
      "epoch": 2.855032317636196,
      "grad_norm": 2.242866277694702,
      "learning_rate": 4.632000000000001e-06,
      "loss": 0.355,
      "step": 1547
    },
    {
      "epoch": 2.8568790397045243,
      "grad_norm": 1.6628878116607666,
      "learning_rate": 4.635e-06,
      "loss": 0.2136,
      "step": 1548
    },
    {
      "epoch": 2.8587257617728534,
      "grad_norm": 1.5874680280685425,
      "learning_rate": 4.6379999999999995e-06,
      "loss": 0.2046,
      "step": 1549
    },
    {
      "epoch": 2.8605724838411817,
      "grad_norm": 2.236630439758301,
      "learning_rate": 4.641e-06,
      "loss": 0.1926,
      "step": 1550
    },
    {
      "epoch": 2.862419205909511,
      "grad_norm": 2.906998872756958,
      "learning_rate": 4.644e-06,
      "loss": 0.1719,
      "step": 1551
    },
    {
      "epoch": 2.864265927977839,
      "grad_norm": 1.578851342201233,
      "learning_rate": 4.6470000000000006e-06,
      "loss": 0.2125,
      "step": 1552
    },
    {
      "epoch": 2.866112650046168,
      "grad_norm": 2.263638496398926,
      "learning_rate": 4.65e-06,
      "loss": 0.1364,
      "step": 1553
    },
    {
      "epoch": 2.867959372114497,
      "grad_norm": 1.3934235572814941,
      "learning_rate": 4.653e-06,
      "loss": 0.1701,
      "step": 1554
    },
    {
      "epoch": 2.8698060941828256,
      "grad_norm": 1.5761624574661255,
      "learning_rate": 4.656e-06,
      "loss": 0.1496,
      "step": 1555
    },
    {
      "epoch": 2.8716528162511543,
      "grad_norm": 2.035221576690674,
      "learning_rate": 4.659e-06,
      "loss": 0.15,
      "step": 1556
    },
    {
      "epoch": 2.873499538319483,
      "grad_norm": 1.6897352933883667,
      "learning_rate": 4.6620000000000004e-06,
      "loss": 0.158,
      "step": 1557
    },
    {
      "epoch": 2.8753462603878117,
      "grad_norm": 2.1436944007873535,
      "learning_rate": 4.665e-06,
      "loss": 0.1266,
      "step": 1558
    },
    {
      "epoch": 2.8771929824561404,
      "grad_norm": 2.5911736488342285,
      "learning_rate": 4.668e-06,
      "loss": 0.2247,
      "step": 1559
    },
    {
      "epoch": 2.879039704524469,
      "grad_norm": 2.169934034347534,
      "learning_rate": 4.671000000000001e-06,
      "loss": 0.2369,
      "step": 1560
    },
    {
      "epoch": 2.880886426592798,
      "grad_norm": 1.4315106868743896,
      "learning_rate": 4.674e-06,
      "loss": 0.1544,
      "step": 1561
    },
    {
      "epoch": 2.8827331486611265,
      "grad_norm": 1.5830488204956055,
      "learning_rate": 4.677e-06,
      "loss": 0.1559,
      "step": 1562
    },
    {
      "epoch": 2.884579870729455,
      "grad_norm": 2.221632480621338,
      "learning_rate": 4.68e-06,
      "loss": 0.1657,
      "step": 1563
    },
    {
      "epoch": 2.886426592797784,
      "grad_norm": 2.146801710128784,
      "learning_rate": 4.683e-06,
      "loss": 0.1606,
      "step": 1564
    },
    {
      "epoch": 2.8882733148661126,
      "grad_norm": 2.3033361434936523,
      "learning_rate": 4.6860000000000005e-06,
      "loss": 0.2723,
      "step": 1565
    },
    {
      "epoch": 2.8901200369344413,
      "grad_norm": 2.868062734603882,
      "learning_rate": 4.689e-06,
      "loss": 0.1503,
      "step": 1566
    },
    {
      "epoch": 2.89196675900277,
      "grad_norm": 1.7747246026992798,
      "learning_rate": 4.692e-06,
      "loss": 0.1728,
      "step": 1567
    },
    {
      "epoch": 2.8938134810710987,
      "grad_norm": 1.6821857690811157,
      "learning_rate": 4.695e-06,
      "loss": 0.141,
      "step": 1568
    },
    {
      "epoch": 2.8956602031394274,
      "grad_norm": 2.7381348609924316,
      "learning_rate": 4.698e-06,
      "loss": 0.2091,
      "step": 1569
    },
    {
      "epoch": 2.897506925207756,
      "grad_norm": 2.3321731090545654,
      "learning_rate": 4.701e-06,
      "loss": 0.2118,
      "step": 1570
    },
    {
      "epoch": 2.899353647276085,
      "grad_norm": 1.6755270957946777,
      "learning_rate": 4.704e-06,
      "loss": 0.1553,
      "step": 1571
    },
    {
      "epoch": 2.901200369344414,
      "grad_norm": 1.7802635431289673,
      "learning_rate": 4.707000000000001e-06,
      "loss": 0.1424,
      "step": 1572
    },
    {
      "epoch": 2.903047091412742,
      "grad_norm": 2.5739917755126953,
      "learning_rate": 4.71e-06,
      "loss": 0.1851,
      "step": 1573
    },
    {
      "epoch": 2.9048938134810713,
      "grad_norm": 2.763282060623169,
      "learning_rate": 4.713e-06,
      "loss": 0.2264,
      "step": 1574
    },
    {
      "epoch": 2.9067405355493996,
      "grad_norm": 1.8808858394622803,
      "learning_rate": 4.716e-06,
      "loss": 0.1417,
      "step": 1575
    },
    {
      "epoch": 2.9085872576177287,
      "grad_norm": 3.0323848724365234,
      "learning_rate": 4.719e-06,
      "loss": 0.1822,
      "step": 1576
    },
    {
      "epoch": 2.910433979686057,
      "grad_norm": 2.023909091949463,
      "learning_rate": 4.722000000000001e-06,
      "loss": 0.1641,
      "step": 1577
    },
    {
      "epoch": 2.912280701754386,
      "grad_norm": 1.6922128200531006,
      "learning_rate": 4.7250000000000005e-06,
      "loss": 0.1566,
      "step": 1578
    },
    {
      "epoch": 2.914127423822715,
      "grad_norm": 2.9664113521575928,
      "learning_rate": 4.7279999999999995e-06,
      "loss": 0.1718,
      "step": 1579
    },
    {
      "epoch": 2.9159741458910435,
      "grad_norm": 2.469337224960327,
      "learning_rate": 4.731e-06,
      "loss": 0.2049,
      "step": 1580
    },
    {
      "epoch": 2.9178208679593722,
      "grad_norm": 1.755744218826294,
      "learning_rate": 4.734e-06,
      "loss": 0.1615,
      "step": 1581
    },
    {
      "epoch": 2.919667590027701,
      "grad_norm": 4.633786678314209,
      "learning_rate": 4.7370000000000006e-06,
      "loss": 0.1993,
      "step": 1582
    },
    {
      "epoch": 2.9215143120960296,
      "grad_norm": 1.6162577867507935,
      "learning_rate": 4.74e-06,
      "loss": 0.1641,
      "step": 1583
    },
    {
      "epoch": 2.9233610341643583,
      "grad_norm": 3.0486338138580322,
      "learning_rate": 4.743e-06,
      "loss": 0.2215,
      "step": 1584
    },
    {
      "epoch": 2.925207756232687,
      "grad_norm": 4.850353717803955,
      "learning_rate": 4.746e-06,
      "loss": 0.6632,
      "step": 1585
    },
    {
      "epoch": 2.9270544783010157,
      "grad_norm": 2.6363439559936523,
      "learning_rate": 4.749e-06,
      "loss": 0.6831,
      "step": 1586
    },
    {
      "epoch": 2.9289012003693444,
      "grad_norm": 2.2633118629455566,
      "learning_rate": 4.752e-06,
      "loss": 0.5065,
      "step": 1587
    },
    {
      "epoch": 2.930747922437673,
      "grad_norm": 2.4349043369293213,
      "learning_rate": 4.755e-06,
      "loss": 0.5256,
      "step": 1588
    },
    {
      "epoch": 2.932594644506002,
      "grad_norm": 14.977548599243164,
      "learning_rate": 4.758e-06,
      "loss": 0.6051,
      "step": 1589
    },
    {
      "epoch": 2.9344413665743305,
      "grad_norm": 1.744310975074768,
      "learning_rate": 4.761000000000001e-06,
      "loss": 0.3641,
      "step": 1590
    },
    {
      "epoch": 2.936288088642659,
      "grad_norm": 2.5987777709960938,
      "learning_rate": 4.764e-06,
      "loss": 0.469,
      "step": 1591
    },
    {
      "epoch": 2.938134810710988,
      "grad_norm": 3.108370304107666,
      "learning_rate": 4.767e-06,
      "loss": 0.2756,
      "step": 1592
    },
    {
      "epoch": 2.9399815327793166,
      "grad_norm": 1.5362529754638672,
      "learning_rate": 4.77e-06,
      "loss": 0.3,
      "step": 1593
    },
    {
      "epoch": 2.9418282548476453,
      "grad_norm": 1.8580305576324463,
      "learning_rate": 4.773e-06,
      "loss": 0.2937,
      "step": 1594
    },
    {
      "epoch": 2.943674976915974,
      "grad_norm": 2.898499011993408,
      "learning_rate": 4.7760000000000005e-06,
      "loss": 0.2125,
      "step": 1595
    },
    {
      "epoch": 2.9455216989843027,
      "grad_norm": 2.2666280269622803,
      "learning_rate": 4.779e-06,
      "loss": 0.3014,
      "step": 1596
    },
    {
      "epoch": 2.9473684210526314,
      "grad_norm": 1.9050452709197998,
      "learning_rate": 4.782e-06,
      "loss": 0.2166,
      "step": 1597
    },
    {
      "epoch": 2.94921514312096,
      "grad_norm": 2.1083133220672607,
      "learning_rate": 4.785e-06,
      "loss": 0.1901,
      "step": 1598
    },
    {
      "epoch": 2.9510618651892893,
      "grad_norm": 2.402987480163574,
      "learning_rate": 4.788e-06,
      "loss": 0.2178,
      "step": 1599
    },
    {
      "epoch": 2.9529085872576175,
      "grad_norm": 1.4459607601165771,
      "learning_rate": 4.791e-06,
      "loss": 0.2102,
      "step": 1600
    },
    {
      "epoch": 2.9547553093259467,
      "grad_norm": 1.7319731712341309,
      "learning_rate": 4.794e-06,
      "loss": 0.2142,
      "step": 1601
    },
    {
      "epoch": 2.956602031394275,
      "grad_norm": 1.5321577787399292,
      "learning_rate": 4.797e-06,
      "loss": 0.1676,
      "step": 1602
    },
    {
      "epoch": 2.958448753462604,
      "grad_norm": 1.6956512928009033,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.1819,
      "step": 1603
    },
    {
      "epoch": 2.9602954755309328,
      "grad_norm": 2.4199330806732178,
      "learning_rate": 4.803e-06,
      "loss": 0.1777,
      "step": 1604
    },
    {
      "epoch": 2.9621421975992615,
      "grad_norm": 1.6201728582382202,
      "learning_rate": 4.806e-06,
      "loss": 0.155,
      "step": 1605
    },
    {
      "epoch": 2.96398891966759,
      "grad_norm": 2.5003530979156494,
      "learning_rate": 4.809e-06,
      "loss": 0.1551,
      "step": 1606
    },
    {
      "epoch": 2.965835641735919,
      "grad_norm": 2.197706699371338,
      "learning_rate": 4.812e-06,
      "loss": 0.1395,
      "step": 1607
    },
    {
      "epoch": 2.9676823638042475,
      "grad_norm": 1.5907740592956543,
      "learning_rate": 4.8150000000000005e-06,
      "loss": 0.1677,
      "step": 1608
    },
    {
      "epoch": 2.9695290858725762,
      "grad_norm": 2.2628791332244873,
      "learning_rate": 4.818e-06,
      "loss": 0.1671,
      "step": 1609
    },
    {
      "epoch": 2.971375807940905,
      "grad_norm": 2.739673376083374,
      "learning_rate": 4.821e-06,
      "loss": 0.1803,
      "step": 1610
    },
    {
      "epoch": 2.9732225300092336,
      "grad_norm": 1.834478497505188,
      "learning_rate": 4.824e-06,
      "loss": 0.1546,
      "step": 1611
    },
    {
      "epoch": 2.9750692520775623,
      "grad_norm": 1.9884629249572754,
      "learning_rate": 4.827e-06,
      "loss": 0.1619,
      "step": 1612
    },
    {
      "epoch": 2.976915974145891,
      "grad_norm": 1.7522231340408325,
      "learning_rate": 4.83e-06,
      "loss": 0.1375,
      "step": 1613
    },
    {
      "epoch": 2.9787626962142197,
      "grad_norm": 1.8641668558120728,
      "learning_rate": 4.833e-06,
      "loss": 0.1763,
      "step": 1614
    },
    {
      "epoch": 2.9806094182825484,
      "grad_norm": 2.795372724533081,
      "learning_rate": 4.836000000000001e-06,
      "loss": 0.1819,
      "step": 1615
    },
    {
      "epoch": 2.982456140350877,
      "grad_norm": 1.361962914466858,
      "learning_rate": 4.839e-06,
      "loss": 0.1448,
      "step": 1616
    },
    {
      "epoch": 2.984302862419206,
      "grad_norm": 2.029982805252075,
      "learning_rate": 4.8419999999999996e-06,
      "loss": 0.1818,
      "step": 1617
    },
    {
      "epoch": 2.9861495844875345,
      "grad_norm": 3.986133575439453,
      "learning_rate": 4.845e-06,
      "loss": 0.1743,
      "step": 1618
    },
    {
      "epoch": 2.9879963065558632,
      "grad_norm": 2.4707727432250977,
      "learning_rate": 4.848e-06,
      "loss": 0.1853,
      "step": 1619
    },
    {
      "epoch": 2.989843028624192,
      "grad_norm": 3.1868081092834473,
      "learning_rate": 4.851000000000001e-06,
      "loss": 0.1881,
      "step": 1620
    },
    {
      "epoch": 2.9916897506925206,
      "grad_norm": 2.083648204803467,
      "learning_rate": 4.8540000000000005e-06,
      "loss": 0.1776,
      "step": 1621
    },
    {
      "epoch": 2.9935364727608493,
      "grad_norm": 2.8003695011138916,
      "learning_rate": 4.856999999999999e-06,
      "loss": 0.1843,
      "step": 1622
    },
    {
      "epoch": 2.995383194829178,
      "grad_norm": 2.0337893962860107,
      "learning_rate": 4.86e-06,
      "loss": 0.1757,
      "step": 1623
    },
    {
      "epoch": 2.997229916897507,
      "grad_norm": 2.571591854095459,
      "learning_rate": 4.863e-06,
      "loss": 0.1648,
      "step": 1624
    },
    {
      "epoch": 2.9990766389658354,
      "grad_norm": 2.681323766708374,
      "learning_rate": 4.8660000000000005e-06,
      "loss": 0.2077,
      "step": 1625
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5731614828109741,
      "learning_rate": 4.869e-06,
      "loss": 0.0914,
      "step": 1626
    },
    {
      "epoch": 3.0018467220683287,
      "grad_norm": 2.272230386734009,
      "learning_rate": 4.872e-06,
      "loss": 0.5649,
      "step": 1627
    },
    {
      "epoch": 3.0036934441366574,
      "grad_norm": 2.1991465091705322,
      "learning_rate": 4.875e-06,
      "loss": 0.624,
      "step": 1628
    },
    {
      "epoch": 3.005540166204986,
      "grad_norm": 1.6478530168533325,
      "learning_rate": 4.878e-06,
      "loss": 0.563,
      "step": 1629
    },
    {
      "epoch": 3.007386888273315,
      "grad_norm": 1.7868117094039917,
      "learning_rate": 4.881e-06,
      "loss": 0.4616,
      "step": 1630
    },
    {
      "epoch": 3.0092336103416435,
      "grad_norm": 1.7761942148208618,
      "learning_rate": 4.884e-06,
      "loss": 0.4051,
      "step": 1631
    },
    {
      "epoch": 3.011080332409972,
      "grad_norm": 1.7362419366836548,
      "learning_rate": 4.887e-06,
      "loss": 0.4763,
      "step": 1632
    },
    {
      "epoch": 3.012927054478301,
      "grad_norm": 2.110557794570923,
      "learning_rate": 4.890000000000001e-06,
      "loss": 0.4974,
      "step": 1633
    },
    {
      "epoch": 3.0147737765466296,
      "grad_norm": 1.771098017692566,
      "learning_rate": 4.8929999999999996e-06,
      "loss": 0.3519,
      "step": 1634
    },
    {
      "epoch": 3.0166204986149583,
      "grad_norm": 1.9019705057144165,
      "learning_rate": 4.896e-06,
      "loss": 0.3971,
      "step": 1635
    },
    {
      "epoch": 3.018467220683287,
      "grad_norm": 1.8817514181137085,
      "learning_rate": 4.899e-06,
      "loss": 0.3578,
      "step": 1636
    },
    {
      "epoch": 3.0203139427516157,
      "grad_norm": 1.5816795825958252,
      "learning_rate": 4.902e-06,
      "loss": 0.3423,
      "step": 1637
    },
    {
      "epoch": 3.0221606648199444,
      "grad_norm": 2.711153030395508,
      "learning_rate": 4.9050000000000005e-06,
      "loss": 0.2532,
      "step": 1638
    },
    {
      "epoch": 3.0240073868882735,
      "grad_norm": 1.6076635122299194,
      "learning_rate": 4.908e-06,
      "loss": 0.2882,
      "step": 1639
    },
    {
      "epoch": 3.0258541089566022,
      "grad_norm": 1.8714416027069092,
      "learning_rate": 4.911e-06,
      "loss": 0.2771,
      "step": 1640
    },
    {
      "epoch": 3.027700831024931,
      "grad_norm": 2.4448885917663574,
      "learning_rate": 4.914e-06,
      "loss": 0.2915,
      "step": 1641
    },
    {
      "epoch": 3.0295475530932596,
      "grad_norm": 1.5730973482131958,
      "learning_rate": 4.917e-06,
      "loss": 0.1761,
      "step": 1642
    },
    {
      "epoch": 3.0313942751615883,
      "grad_norm": 1.8119219541549683,
      "learning_rate": 4.92e-06,
      "loss": 0.172,
      "step": 1643
    },
    {
      "epoch": 3.033240997229917,
      "grad_norm": 1.9520834684371948,
      "learning_rate": 4.923e-06,
      "loss": 0.2029,
      "step": 1644
    },
    {
      "epoch": 3.0350877192982457,
      "grad_norm": 1.4923053979873657,
      "learning_rate": 4.926000000000001e-06,
      "loss": 0.1399,
      "step": 1645
    },
    {
      "epoch": 3.0369344413665744,
      "grad_norm": 1.0933345556259155,
      "learning_rate": 4.929000000000001e-06,
      "loss": 0.1687,
      "step": 1646
    },
    {
      "epoch": 3.038781163434903,
      "grad_norm": 2.112572193145752,
      "learning_rate": 4.9319999999999995e-06,
      "loss": 0.2484,
      "step": 1647
    },
    {
      "epoch": 3.040627885503232,
      "grad_norm": 1.7236456871032715,
      "learning_rate": 4.935e-06,
      "loss": 0.1377,
      "step": 1648
    },
    {
      "epoch": 3.0424746075715605,
      "grad_norm": 1.726274847984314,
      "learning_rate": 4.938e-06,
      "loss": 0.1647,
      "step": 1649
    },
    {
      "epoch": 3.044321329639889,
      "grad_norm": 1.6629983186721802,
      "learning_rate": 4.941000000000001e-06,
      "loss": 0.1538,
      "step": 1650
    },
    {
      "epoch": 3.046168051708218,
      "grad_norm": 1.8645832538604736,
      "learning_rate": 4.9440000000000004e-06,
      "loss": 0.1484,
      "step": 1651
    },
    {
      "epoch": 3.0480147737765466,
      "grad_norm": 1.8150570392608643,
      "learning_rate": 4.947e-06,
      "loss": 0.1349,
      "step": 1652
    },
    {
      "epoch": 3.0498614958448753,
      "grad_norm": 2.544318199157715,
      "learning_rate": 4.95e-06,
      "loss": 0.1593,
      "step": 1653
    },
    {
      "epoch": 3.051708217913204,
      "grad_norm": 2.0313374996185303,
      "learning_rate": 4.953e-06,
      "loss": 0.1239,
      "step": 1654
    },
    {
      "epoch": 3.0535549399815327,
      "grad_norm": 1.766607642173767,
      "learning_rate": 4.9560000000000005e-06,
      "loss": 0.1705,
      "step": 1655
    },
    {
      "epoch": 3.0554016620498614,
      "grad_norm": 1.6488720178604126,
      "learning_rate": 4.959e-06,
      "loss": 0.1477,
      "step": 1656
    },
    {
      "epoch": 3.05724838411819,
      "grad_norm": 1.9187687635421753,
      "learning_rate": 4.962e-06,
      "loss": 0.1626,
      "step": 1657
    },
    {
      "epoch": 3.059095106186519,
      "grad_norm": 1.9352442026138306,
      "learning_rate": 4.965000000000001e-06,
      "loss": 0.1291,
      "step": 1658
    },
    {
      "epoch": 3.0609418282548475,
      "grad_norm": 2.122387647628784,
      "learning_rate": 4.968e-06,
      "loss": 0.2403,
      "step": 1659
    },
    {
      "epoch": 3.062788550323176,
      "grad_norm": 1.7077113389968872,
      "learning_rate": 4.9709999999999995e-06,
      "loss": 0.1439,
      "step": 1660
    },
    {
      "epoch": 3.064635272391505,
      "grad_norm": 2.4755544662475586,
      "learning_rate": 4.974e-06,
      "loss": 0.1117,
      "step": 1661
    },
    {
      "epoch": 3.0664819944598336,
      "grad_norm": 1.930274248123169,
      "learning_rate": 4.977e-06,
      "loss": 0.1821,
      "step": 1662
    },
    {
      "epoch": 3.0683287165281623,
      "grad_norm": 1.8199743032455444,
      "learning_rate": 4.980000000000001e-06,
      "loss": 0.1167,
      "step": 1663
    },
    {
      "epoch": 3.0701754385964914,
      "grad_norm": 1.5116584300994873,
      "learning_rate": 4.983e-06,
      "loss": 0.1242,
      "step": 1664
    },
    {
      "epoch": 3.07202216066482,
      "grad_norm": 1.9836688041687012,
      "learning_rate": 4.985999999999999e-06,
      "loss": 0.186,
      "step": 1665
    },
    {
      "epoch": 3.073868882733149,
      "grad_norm": 2.199866533279419,
      "learning_rate": 4.989e-06,
      "loss": 0.1463,
      "step": 1666
    },
    {
      "epoch": 3.0757156048014775,
      "grad_norm": 1.57767653465271,
      "learning_rate": 4.992e-06,
      "loss": 0.1495,
      "step": 1667
    },
    {
      "epoch": 3.0775623268698062,
      "grad_norm": 9.19756031036377,
      "learning_rate": 4.9950000000000005e-06,
      "loss": 0.1426,
      "step": 1668
    },
    {
      "epoch": 3.079409048938135,
      "grad_norm": 2.502916097640991,
      "learning_rate": 4.998e-06,
      "loss": 0.1973,
      "step": 1669
    },
    {
      "epoch": 3.0812557710064636,
      "grad_norm": 2.128638982772827,
      "learning_rate": 5.001e-06,
      "loss": 0.1319,
      "step": 1670
    },
    {
      "epoch": 3.0831024930747923,
      "grad_norm": 1.7738232612609863,
      "learning_rate": 5.004e-06,
      "loss": 0.155,
      "step": 1671
    },
    {
      "epoch": 3.084949215143121,
      "grad_norm": 2.350332260131836,
      "learning_rate": 5.007e-06,
      "loss": 0.1543,
      "step": 1672
    },
    {
      "epoch": 3.0867959372114497,
      "grad_norm": 1.9206435680389404,
      "learning_rate": 5.01e-06,
      "loss": 0.1919,
      "step": 1673
    },
    {
      "epoch": 3.0886426592797784,
      "grad_norm": 2.0797274112701416,
      "learning_rate": 5.013e-06,
      "loss": 0.1699,
      "step": 1674
    },
    {
      "epoch": 3.090489381348107,
      "grad_norm": 1.9949361085891724,
      "learning_rate": 5.016e-06,
      "loss": 0.178,
      "step": 1675
    },
    {
      "epoch": 3.092336103416436,
      "grad_norm": 2.23211669921875,
      "learning_rate": 5.0190000000000006e-06,
      "loss": 0.1716,
      "step": 1676
    },
    {
      "epoch": 3.0941828254847645,
      "grad_norm": 4.083524703979492,
      "learning_rate": 5.0219999999999995e-06,
      "loss": 0.6273,
      "step": 1677
    },
    {
      "epoch": 3.0960295475530932,
      "grad_norm": 1.6188116073608398,
      "learning_rate": 5.025e-06,
      "loss": 0.5839,
      "step": 1678
    },
    {
      "epoch": 3.097876269621422,
      "grad_norm": 2.0627925395965576,
      "learning_rate": 5.028e-06,
      "loss": 0.5534,
      "step": 1679
    },
    {
      "epoch": 3.0997229916897506,
      "grad_norm": 2.4019577503204346,
      "learning_rate": 5.031e-06,
      "loss": 0.5056,
      "step": 1680
    },
    {
      "epoch": 3.1015697137580793,
      "grad_norm": 2.0087342262268066,
      "learning_rate": 5.034e-06,
      "loss": 0.4375,
      "step": 1681
    },
    {
      "epoch": 3.103416435826408,
      "grad_norm": 1.7077587842941284,
      "learning_rate": 5.037e-06,
      "loss": 0.4366,
      "step": 1682
    },
    {
      "epoch": 3.1052631578947367,
      "grad_norm": 1.9419629573822021,
      "learning_rate": 5.04e-06,
      "loss": 0.3506,
      "step": 1683
    },
    {
      "epoch": 3.1071098799630654,
      "grad_norm": 1.676218867301941,
      "learning_rate": 5.043e-06,
      "loss": 0.3371,
      "step": 1684
    },
    {
      "epoch": 3.108956602031394,
      "grad_norm": 1.9952783584594727,
      "learning_rate": 5.046e-06,
      "loss": 0.3364,
      "step": 1685
    },
    {
      "epoch": 3.110803324099723,
      "grad_norm": 2.8741376399993896,
      "learning_rate": 5.049e-06,
      "loss": 0.3141,
      "step": 1686
    },
    {
      "epoch": 3.1126500461680515,
      "grad_norm": 1.4315117597579956,
      "learning_rate": 5.052e-06,
      "loss": 0.2986,
      "step": 1687
    },
    {
      "epoch": 3.11449676823638,
      "grad_norm": 1.3568508625030518,
      "learning_rate": 5.055000000000001e-06,
      "loss": 0.2689,
      "step": 1688
    },
    {
      "epoch": 3.1163434903047094,
      "grad_norm": 2.1754391193389893,
      "learning_rate": 5.0580000000000005e-06,
      "loss": 0.2782,
      "step": 1689
    },
    {
      "epoch": 3.118190212373038,
      "grad_norm": 2.0119006633758545,
      "learning_rate": 5.0609999999999995e-06,
      "loss": 0.2859,
      "step": 1690
    },
    {
      "epoch": 3.1200369344413668,
      "grad_norm": 1.4242953062057495,
      "learning_rate": 5.064e-06,
      "loss": 0.2626,
      "step": 1691
    },
    {
      "epoch": 3.1218836565096955,
      "grad_norm": 1.5842236280441284,
      "learning_rate": 5.067e-06,
      "loss": 0.2223,
      "step": 1692
    },
    {
      "epoch": 3.123730378578024,
      "grad_norm": 4.055980205535889,
      "learning_rate": 5.070000000000001e-06,
      "loss": 0.1274,
      "step": 1693
    },
    {
      "epoch": 3.125577100646353,
      "grad_norm": 2.017376184463501,
      "learning_rate": 5.073e-06,
      "loss": 0.2235,
      "step": 1694
    },
    {
      "epoch": 3.1274238227146816,
      "grad_norm": 1.4071455001831055,
      "learning_rate": 5.076e-06,
      "loss": 0.1488,
      "step": 1695
    },
    {
      "epoch": 3.1292705447830103,
      "grad_norm": 1.4524205923080444,
      "learning_rate": 5.079e-06,
      "loss": 0.1208,
      "step": 1696
    },
    {
      "epoch": 3.131117266851339,
      "grad_norm": 2.6339752674102783,
      "learning_rate": 5.082e-06,
      "loss": 0.1519,
      "step": 1697
    },
    {
      "epoch": 3.1329639889196677,
      "grad_norm": 1.3864768743515015,
      "learning_rate": 5.0850000000000004e-06,
      "loss": 0.1285,
      "step": 1698
    },
    {
      "epoch": 3.1348107109879964,
      "grad_norm": 1.4206315279006958,
      "learning_rate": 5.088e-06,
      "loss": 0.1302,
      "step": 1699
    },
    {
      "epoch": 3.136657433056325,
      "grad_norm": 1.1164093017578125,
      "learning_rate": 5.091e-06,
      "loss": 0.1233,
      "step": 1700
    },
    {
      "epoch": 3.1385041551246537,
      "grad_norm": 2.3856983184814453,
      "learning_rate": 5.094000000000001e-06,
      "loss": 0.1948,
      "step": 1701
    },
    {
      "epoch": 3.1403508771929824,
      "grad_norm": 1.6560757160186768,
      "learning_rate": 5.097e-06,
      "loss": 0.1125,
      "step": 1702
    },
    {
      "epoch": 3.142197599261311,
      "grad_norm": 1.2046996355056763,
      "learning_rate": 5.1e-06,
      "loss": 0.1099,
      "step": 1703
    },
    {
      "epoch": 3.14404432132964,
      "grad_norm": 2.4310736656188965,
      "learning_rate": 5.103e-06,
      "loss": 0.1507,
      "step": 1704
    },
    {
      "epoch": 3.1458910433979685,
      "grad_norm": 1.898454189300537,
      "learning_rate": 5.106e-06,
      "loss": 0.1306,
      "step": 1705
    },
    {
      "epoch": 3.1477377654662972,
      "grad_norm": 2.1356546878814697,
      "learning_rate": 5.1090000000000006e-06,
      "loss": 0.1367,
      "step": 1706
    },
    {
      "epoch": 3.149584487534626,
      "grad_norm": 2.0182154178619385,
      "learning_rate": 5.112e-06,
      "loss": 0.1339,
      "step": 1707
    },
    {
      "epoch": 3.1514312096029546,
      "grad_norm": 1.420379638671875,
      "learning_rate": 5.115e-06,
      "loss": 0.1415,
      "step": 1708
    },
    {
      "epoch": 3.1532779316712833,
      "grad_norm": 9.14201831817627,
      "learning_rate": 5.118e-06,
      "loss": 0.144,
      "step": 1709
    },
    {
      "epoch": 3.155124653739612,
      "grad_norm": 2.5935096740722656,
      "learning_rate": 5.121e-06,
      "loss": 0.1447,
      "step": 1710
    },
    {
      "epoch": 3.1569713758079407,
      "grad_norm": 2.102509021759033,
      "learning_rate": 5.124e-06,
      "loss": 0.1698,
      "step": 1711
    },
    {
      "epoch": 3.1588180978762694,
      "grad_norm": 1.7125483751296997,
      "learning_rate": 5.127e-06,
      "loss": 0.1426,
      "step": 1712
    },
    {
      "epoch": 3.160664819944598,
      "grad_norm": 2.138498306274414,
      "learning_rate": 5.130000000000001e-06,
      "loss": 0.1363,
      "step": 1713
    },
    {
      "epoch": 3.1625115420129273,
      "grad_norm": 2.0215048789978027,
      "learning_rate": 5.133e-06,
      "loss": 0.1309,
      "step": 1714
    },
    {
      "epoch": 3.164358264081256,
      "grad_norm": 3.2613959312438965,
      "learning_rate": 5.136e-06,
      "loss": 0.1733,
      "step": 1715
    },
    {
      "epoch": 3.1662049861495847,
      "grad_norm": 2.8282909393310547,
      "learning_rate": 5.139e-06,
      "loss": 0.1563,
      "step": 1716
    },
    {
      "epoch": 3.1680517082179134,
      "grad_norm": 1.946689486503601,
      "learning_rate": 5.142e-06,
      "loss": 0.1375,
      "step": 1717
    },
    {
      "epoch": 3.169898430286242,
      "grad_norm": 1.8160659074783325,
      "learning_rate": 5.145000000000001e-06,
      "loss": 0.1319,
      "step": 1718
    },
    {
      "epoch": 3.1717451523545708,
      "grad_norm": 1.7862443923950195,
      "learning_rate": 5.1480000000000005e-06,
      "loss": 0.1782,
      "step": 1719
    },
    {
      "epoch": 3.1735918744228995,
      "grad_norm": 2.174208879470825,
      "learning_rate": 5.1509999999999995e-06,
      "loss": 0.1747,
      "step": 1720
    },
    {
      "epoch": 3.175438596491228,
      "grad_norm": 2.127056121826172,
      "learning_rate": 5.154e-06,
      "loss": 0.1822,
      "step": 1721
    },
    {
      "epoch": 3.177285318559557,
      "grad_norm": 1.744895577430725,
      "learning_rate": 5.157e-06,
      "loss": 0.1204,
      "step": 1722
    },
    {
      "epoch": 3.1791320406278856,
      "grad_norm": 1.6898212432861328,
      "learning_rate": 5.16e-06,
      "loss": 0.1292,
      "step": 1723
    },
    {
      "epoch": 3.1809787626962143,
      "grad_norm": 1.7619431018829346,
      "learning_rate": 5.163e-06,
      "loss": 0.1442,
      "step": 1724
    },
    {
      "epoch": 3.182825484764543,
      "grad_norm": 1.6283921003341675,
      "learning_rate": 5.166e-06,
      "loss": 0.1423,
      "step": 1725
    },
    {
      "epoch": 3.1846722068328717,
      "grad_norm": 3.300206184387207,
      "learning_rate": 5.169e-06,
      "loss": 0.1778,
      "step": 1726
    },
    {
      "epoch": 3.1865189289012004,
      "grad_norm": 1.6233912706375122,
      "learning_rate": 5.172e-06,
      "loss": 0.534,
      "step": 1727
    },
    {
      "epoch": 3.188365650969529,
      "grad_norm": 2.2645161151885986,
      "learning_rate": 5.175e-06,
      "loss": 0.589,
      "step": 1728
    },
    {
      "epoch": 3.1902123730378578,
      "grad_norm": 4.905712127685547,
      "learning_rate": 5.178e-06,
      "loss": 0.4501,
      "step": 1729
    },
    {
      "epoch": 3.1920590951061865,
      "grad_norm": 2.5921883583068848,
      "learning_rate": 5.181e-06,
      "loss": 0.4044,
      "step": 1730
    },
    {
      "epoch": 3.193905817174515,
      "grad_norm": 1.6506649255752563,
      "learning_rate": 5.184000000000001e-06,
      "loss": 0.347,
      "step": 1731
    },
    {
      "epoch": 3.195752539242844,
      "grad_norm": 1.9817479848861694,
      "learning_rate": 5.1870000000000005e-06,
      "loss": 0.4327,
      "step": 1732
    },
    {
      "epoch": 3.1975992613111726,
      "grad_norm": 2.370718240737915,
      "learning_rate": 5.1899999999999994e-06,
      "loss": 0.3587,
      "step": 1733
    },
    {
      "epoch": 3.1994459833795013,
      "grad_norm": 1.4260053634643555,
      "learning_rate": 5.193e-06,
      "loss": 0.2848,
      "step": 1734
    },
    {
      "epoch": 3.20129270544783,
      "grad_norm": 1.3881109952926636,
      "learning_rate": 5.196e-06,
      "loss": 0.2891,
      "step": 1735
    },
    {
      "epoch": 3.2031394275161587,
      "grad_norm": 1.8376153707504272,
      "learning_rate": 5.1990000000000005e-06,
      "loss": 0.3453,
      "step": 1736
    },
    {
      "epoch": 3.2049861495844874,
      "grad_norm": 2.22748064994812,
      "learning_rate": 5.202e-06,
      "loss": 0.4393,
      "step": 1737
    },
    {
      "epoch": 3.206832871652816,
      "grad_norm": 1.5622321367263794,
      "learning_rate": 5.205e-06,
      "loss": 0.2238,
      "step": 1738
    },
    {
      "epoch": 3.208679593721145,
      "grad_norm": 2.117229461669922,
      "learning_rate": 5.208e-06,
      "loss": 0.2132,
      "step": 1739
    },
    {
      "epoch": 3.2105263157894735,
      "grad_norm": 5.6894073486328125,
      "learning_rate": 5.211e-06,
      "loss": 0.1495,
      "step": 1740
    },
    {
      "epoch": 3.2123730378578026,
      "grad_norm": 1.2661709785461426,
      "learning_rate": 5.214e-06,
      "loss": 0.1761,
      "step": 1741
    },
    {
      "epoch": 3.2142197599261313,
      "grad_norm": 1.3423945903778076,
      "learning_rate": 5.217e-06,
      "loss": 0.1238,
      "step": 1742
    },
    {
      "epoch": 3.21606648199446,
      "grad_norm": 1.421120047569275,
      "learning_rate": 5.22e-06,
      "loss": 0.1806,
      "step": 1743
    },
    {
      "epoch": 3.2179132040627887,
      "grad_norm": 1.6990739107131958,
      "learning_rate": 5.223000000000001e-06,
      "loss": 0.1248,
      "step": 1744
    },
    {
      "epoch": 3.2197599261311174,
      "grad_norm": 2.5295536518096924,
      "learning_rate": 5.226e-06,
      "loss": 0.1372,
      "step": 1745
    },
    {
      "epoch": 3.221606648199446,
      "grad_norm": 1.3577442169189453,
      "learning_rate": 5.229e-06,
      "loss": 0.145,
      "step": 1746
    },
    {
      "epoch": 3.223453370267775,
      "grad_norm": 1.1989928483963013,
      "learning_rate": 5.232e-06,
      "loss": 0.0898,
      "step": 1747
    },
    {
      "epoch": 3.2253000923361035,
      "grad_norm": 2.8232219219207764,
      "learning_rate": 5.235e-06,
      "loss": 0.164,
      "step": 1748
    },
    {
      "epoch": 3.227146814404432,
      "grad_norm": 2.043588876724243,
      "learning_rate": 5.2380000000000005e-06,
      "loss": 0.1451,
      "step": 1749
    },
    {
      "epoch": 3.228993536472761,
      "grad_norm": 2.160836696624756,
      "learning_rate": 5.241e-06,
      "loss": 0.1178,
      "step": 1750
    },
    {
      "epoch": 3.2308402585410896,
      "grad_norm": 1.4974416494369507,
      "learning_rate": 5.244e-06,
      "loss": 0.1553,
      "step": 1751
    },
    {
      "epoch": 3.2326869806094183,
      "grad_norm": 2.6034817695617676,
      "learning_rate": 5.247e-06,
      "loss": 0.172,
      "step": 1752
    },
    {
      "epoch": 3.234533702677747,
      "grad_norm": 1.3825095891952515,
      "learning_rate": 5.25e-06,
      "loss": 0.138,
      "step": 1753
    },
    {
      "epoch": 3.2363804247460757,
      "grad_norm": 1.5406382083892822,
      "learning_rate": 5.253e-06,
      "loss": 0.1471,
      "step": 1754
    },
    {
      "epoch": 3.2382271468144044,
      "grad_norm": 2.92680287361145,
      "learning_rate": 5.256e-06,
      "loss": 0.1445,
      "step": 1755
    },
    {
      "epoch": 3.240073868882733,
      "grad_norm": 4.907720565795898,
      "learning_rate": 5.259000000000001e-06,
      "loss": 0.1616,
      "step": 1756
    },
    {
      "epoch": 3.2419205909510618,
      "grad_norm": 1.7645773887634277,
      "learning_rate": 5.262e-06,
      "loss": 0.1392,
      "step": 1757
    },
    {
      "epoch": 3.2437673130193905,
      "grad_norm": 2.366370677947998,
      "learning_rate": 5.2649999999999996e-06,
      "loss": 0.1177,
      "step": 1758
    },
    {
      "epoch": 3.245614035087719,
      "grad_norm": 3.015122890472412,
      "learning_rate": 5.268e-06,
      "loss": 0.1436,
      "step": 1759
    },
    {
      "epoch": 3.247460757156048,
      "grad_norm": 1.498371958732605,
      "learning_rate": 5.271e-06,
      "loss": 0.1338,
      "step": 1760
    },
    {
      "epoch": 3.2493074792243766,
      "grad_norm": 1.8410630226135254,
      "learning_rate": 5.274000000000001e-06,
      "loss": 0.1333,
      "step": 1761
    },
    {
      "epoch": 3.2511542012927053,
      "grad_norm": 1.737515926361084,
      "learning_rate": 5.2770000000000005e-06,
      "loss": 0.1199,
      "step": 1762
    },
    {
      "epoch": 3.253000923361034,
      "grad_norm": 1.9851926565170288,
      "learning_rate": 5.279999999999999e-06,
      "loss": 0.1505,
      "step": 1763
    },
    {
      "epoch": 3.254847645429363,
      "grad_norm": 5.256008148193359,
      "learning_rate": 5.283e-06,
      "loss": 0.1445,
      "step": 1764
    },
    {
      "epoch": 3.2566943674976914,
      "grad_norm": 1.6933232545852661,
      "learning_rate": 5.286e-06,
      "loss": 0.1317,
      "step": 1765
    },
    {
      "epoch": 3.2585410895660205,
      "grad_norm": 2.7634260654449463,
      "learning_rate": 5.2890000000000005e-06,
      "loss": 0.1561,
      "step": 1766
    },
    {
      "epoch": 3.260387811634349,
      "grad_norm": 2.7004668712615967,
      "learning_rate": 5.292e-06,
      "loss": 0.1626,
      "step": 1767
    },
    {
      "epoch": 3.262234533702678,
      "grad_norm": 1.7570018768310547,
      "learning_rate": 5.295e-06,
      "loss": 0.1528,
      "step": 1768
    },
    {
      "epoch": 3.2640812557710066,
      "grad_norm": 2.9642531871795654,
      "learning_rate": 5.298e-06,
      "loss": 0.11,
      "step": 1769
    },
    {
      "epoch": 3.2659279778393353,
      "grad_norm": 2.9844934940338135,
      "learning_rate": 5.301e-06,
      "loss": 0.1467,
      "step": 1770
    },
    {
      "epoch": 3.267774699907664,
      "grad_norm": 1.9152250289916992,
      "learning_rate": 5.304e-06,
      "loss": 0.1505,
      "step": 1771
    },
    {
      "epoch": 3.2696214219759927,
      "grad_norm": 2.3223955631256104,
      "learning_rate": 5.307e-06,
      "loss": 0.1601,
      "step": 1772
    },
    {
      "epoch": 3.2714681440443214,
      "grad_norm": 1.631397008895874,
      "learning_rate": 5.31e-06,
      "loss": 0.1601,
      "step": 1773
    },
    {
      "epoch": 3.27331486611265,
      "grad_norm": 2.114636182785034,
      "learning_rate": 5.313000000000001e-06,
      "loss": 0.1995,
      "step": 1774
    },
    {
      "epoch": 3.275161588180979,
      "grad_norm": 1.5473500490188599,
      "learning_rate": 5.3160000000000004e-06,
      "loss": 0.145,
      "step": 1775
    },
    {
      "epoch": 3.2770083102493075,
      "grad_norm": 2.362990379333496,
      "learning_rate": 5.319e-06,
      "loss": 0.2343,
      "step": 1776
    },
    {
      "epoch": 3.278855032317636,
      "grad_norm": 3.1484501361846924,
      "learning_rate": 5.322e-06,
      "loss": 0.5929,
      "step": 1777
    },
    {
      "epoch": 3.280701754385965,
      "grad_norm": 5.884219646453857,
      "learning_rate": 5.325e-06,
      "loss": 0.666,
      "step": 1778
    },
    {
      "epoch": 3.2825484764542936,
      "grad_norm": 1.5649977922439575,
      "learning_rate": 5.3280000000000005e-06,
      "loss": 0.4397,
      "step": 1779
    },
    {
      "epoch": 3.2843951985226223,
      "grad_norm": 2.324662446975708,
      "learning_rate": 5.331e-06,
      "loss": 0.4573,
      "step": 1780
    },
    {
      "epoch": 3.286241920590951,
      "grad_norm": 1.8567326068878174,
      "learning_rate": 5.334000000000001e-06,
      "loss": 0.3962,
      "step": 1781
    },
    {
      "epoch": 3.2880886426592797,
      "grad_norm": 2.1249706745147705,
      "learning_rate": 5.337e-06,
      "loss": 0.3903,
      "step": 1782
    },
    {
      "epoch": 3.2899353647276084,
      "grad_norm": 2.6666476726531982,
      "learning_rate": 5.34e-06,
      "loss": 0.3403,
      "step": 1783
    },
    {
      "epoch": 3.291782086795937,
      "grad_norm": 1.413912296295166,
      "learning_rate": 5.343e-06,
      "loss": 0.3315,
      "step": 1784
    },
    {
      "epoch": 3.293628808864266,
      "grad_norm": 2.263784885406494,
      "learning_rate": 5.346e-06,
      "loss": 0.3878,
      "step": 1785
    },
    {
      "epoch": 3.2954755309325945,
      "grad_norm": 1.5043153762817383,
      "learning_rate": 5.349e-06,
      "loss": 0.3375,
      "step": 1786
    },
    {
      "epoch": 3.297322253000923,
      "grad_norm": 1.6676349639892578,
      "learning_rate": 5.352000000000001e-06,
      "loss": 0.2909,
      "step": 1787
    },
    {
      "epoch": 3.299168975069252,
      "grad_norm": 2.5972864627838135,
      "learning_rate": 5.3549999999999996e-06,
      "loss": 0.3257,
      "step": 1788
    },
    {
      "epoch": 3.301015697137581,
      "grad_norm": 1.6890889406204224,
      "learning_rate": 5.358e-06,
      "loss": 0.2643,
      "step": 1789
    },
    {
      "epoch": 3.3028624192059093,
      "grad_norm": 1.1364909410476685,
      "learning_rate": 5.361e-06,
      "loss": 0.2016,
      "step": 1790
    },
    {
      "epoch": 3.3047091412742384,
      "grad_norm": 1.5025843381881714,
      "learning_rate": 5.364e-06,
      "loss": 0.2205,
      "step": 1791
    },
    {
      "epoch": 3.306555863342567,
      "grad_norm": 1.2549930810928345,
      "learning_rate": 5.3670000000000005e-06,
      "loss": 0.2018,
      "step": 1792
    },
    {
      "epoch": 3.308402585410896,
      "grad_norm": 2.2258710861206055,
      "learning_rate": 5.37e-06,
      "loss": 0.1963,
      "step": 1793
    },
    {
      "epoch": 3.3102493074792245,
      "grad_norm": 1.2186377048492432,
      "learning_rate": 5.373e-06,
      "loss": 0.1456,
      "step": 1794
    },
    {
      "epoch": 3.312096029547553,
      "grad_norm": 1.2291967868804932,
      "learning_rate": 5.376e-06,
      "loss": 0.1224,
      "step": 1795
    },
    {
      "epoch": 3.313942751615882,
      "grad_norm": 0.9632380604743958,
      "learning_rate": 5.379e-06,
      "loss": 0.0991,
      "step": 1796
    },
    {
      "epoch": 3.3157894736842106,
      "grad_norm": 1.194434404373169,
      "learning_rate": 5.382e-06,
      "loss": 0.095,
      "step": 1797
    },
    {
      "epoch": 3.3176361957525393,
      "grad_norm": 1.77876877784729,
      "learning_rate": 5.385e-06,
      "loss": 0.1738,
      "step": 1798
    },
    {
      "epoch": 3.319482917820868,
      "grad_norm": 2.20609974861145,
      "learning_rate": 5.388000000000001e-06,
      "loss": 0.105,
      "step": 1799
    },
    {
      "epoch": 3.3213296398891967,
      "grad_norm": 1.5217195749282837,
      "learning_rate": 5.391e-06,
      "loss": 0.1241,
      "step": 1800
    },
    {
      "epoch": 3.3231763619575254,
      "grad_norm": 1.194684386253357,
      "learning_rate": 5.3939999999999995e-06,
      "loss": 0.1365,
      "step": 1801
    },
    {
      "epoch": 3.325023084025854,
      "grad_norm": 1.244110345840454,
      "learning_rate": 5.397e-06,
      "loss": 0.1279,
      "step": 1802
    },
    {
      "epoch": 3.326869806094183,
      "grad_norm": 2.9739558696746826,
      "learning_rate": 5.4e-06,
      "loss": 0.135,
      "step": 1803
    },
    {
      "epoch": 3.3287165281625115,
      "grad_norm": 1.8225957155227661,
      "learning_rate": 5.403000000000001e-06,
      "loss": 0.1005,
      "step": 1804
    },
    {
      "epoch": 3.33056325023084,
      "grad_norm": 1.5515562295913696,
      "learning_rate": 5.406e-06,
      "loss": 0.1708,
      "step": 1805
    },
    {
      "epoch": 3.332409972299169,
      "grad_norm": 1.735154151916504,
      "learning_rate": 5.408999999999999e-06,
      "loss": 0.1091,
      "step": 1806
    },
    {
      "epoch": 3.3342566943674976,
      "grad_norm": 1.587384819984436,
      "learning_rate": 5.412e-06,
      "loss": 0.1558,
      "step": 1807
    },
    {
      "epoch": 3.3361034164358263,
      "grad_norm": 1.638647198677063,
      "learning_rate": 5.415e-06,
      "loss": 0.1064,
      "step": 1808
    },
    {
      "epoch": 3.337950138504155,
      "grad_norm": 1.2404173612594604,
      "learning_rate": 5.4180000000000005e-06,
      "loss": 0.1065,
      "step": 1809
    },
    {
      "epoch": 3.3397968605724837,
      "grad_norm": 3.201526165008545,
      "learning_rate": 5.421e-06,
      "loss": 0.1334,
      "step": 1810
    },
    {
      "epoch": 3.3416435826408124,
      "grad_norm": 2.358724355697632,
      "learning_rate": 5.424e-06,
      "loss": 0.134,
      "step": 1811
    },
    {
      "epoch": 3.343490304709141,
      "grad_norm": 1.3843193054199219,
      "learning_rate": 5.427e-06,
      "loss": 0.0946,
      "step": 1812
    },
    {
      "epoch": 3.34533702677747,
      "grad_norm": 2.5044610500335693,
      "learning_rate": 5.43e-06,
      "loss": 0.1552,
      "step": 1813
    },
    {
      "epoch": 3.347183748845799,
      "grad_norm": 1.6049011945724487,
      "learning_rate": 5.433e-06,
      "loss": 0.154,
      "step": 1814
    },
    {
      "epoch": 3.349030470914127,
      "grad_norm": 1.7652519941329956,
      "learning_rate": 5.436e-06,
      "loss": 0.1191,
      "step": 1815
    },
    {
      "epoch": 3.3508771929824563,
      "grad_norm": 1.646363615989685,
      "learning_rate": 5.439e-06,
      "loss": 0.1366,
      "step": 1816
    },
    {
      "epoch": 3.352723915050785,
      "grad_norm": 2.438486337661743,
      "learning_rate": 5.442000000000001e-06,
      "loss": 0.1325,
      "step": 1817
    },
    {
      "epoch": 3.3545706371191137,
      "grad_norm": 1.6104596853256226,
      "learning_rate": 5.445e-06,
      "loss": 0.1316,
      "step": 1818
    },
    {
      "epoch": 3.3564173591874424,
      "grad_norm": 1.6542346477508545,
      "learning_rate": 5.448e-06,
      "loss": 0.1435,
      "step": 1819
    },
    {
      "epoch": 3.358264081255771,
      "grad_norm": 1.6384179592132568,
      "learning_rate": 5.451e-06,
      "loss": 0.1408,
      "step": 1820
    },
    {
      "epoch": 3.3601108033241,
      "grad_norm": 2.035884380340576,
      "learning_rate": 5.454e-06,
      "loss": 0.1251,
      "step": 1821
    },
    {
      "epoch": 3.3619575253924285,
      "grad_norm": 2.1199872493743896,
      "learning_rate": 5.4570000000000004e-06,
      "loss": 0.1318,
      "step": 1822
    },
    {
      "epoch": 3.3638042474607572,
      "grad_norm": 2.0500221252441406,
      "learning_rate": 5.46e-06,
      "loss": 0.1293,
      "step": 1823
    },
    {
      "epoch": 3.365650969529086,
      "grad_norm": 1.677475094795227,
      "learning_rate": 5.463000000000001e-06,
      "loss": 0.2024,
      "step": 1824
    },
    {
      "epoch": 3.3674976915974146,
      "grad_norm": 1.9693593978881836,
      "learning_rate": 5.466e-06,
      "loss": 0.1779,
      "step": 1825
    },
    {
      "epoch": 3.3693444136657433,
      "grad_norm": 1.5180773735046387,
      "learning_rate": 5.469e-06,
      "loss": 0.1339,
      "step": 1826
    },
    {
      "epoch": 3.371191135734072,
      "grad_norm": 2.3797709941864014,
      "learning_rate": 5.472e-06,
      "loss": 0.589,
      "step": 1827
    },
    {
      "epoch": 3.3730378578024007,
      "grad_norm": 2.1648378372192383,
      "learning_rate": 5.475e-06,
      "loss": 0.5639,
      "step": 1828
    },
    {
      "epoch": 3.3748845798707294,
      "grad_norm": 1.9987053871154785,
      "learning_rate": 5.478000000000001e-06,
      "loss": 0.4686,
      "step": 1829
    },
    {
      "epoch": 3.376731301939058,
      "grad_norm": 2.010835886001587,
      "learning_rate": 5.4810000000000005e-06,
      "loss": 0.4747,
      "step": 1830
    },
    {
      "epoch": 3.378578024007387,
      "grad_norm": 1.4874591827392578,
      "learning_rate": 5.4839999999999995e-06,
      "loss": 0.3482,
      "step": 1831
    },
    {
      "epoch": 3.3804247460757155,
      "grad_norm": 1.4924468994140625,
      "learning_rate": 5.487e-06,
      "loss": 0.3022,
      "step": 1832
    },
    {
      "epoch": 3.3822714681440442,
      "grad_norm": 1.5955424308776855,
      "learning_rate": 5.49e-06,
      "loss": 0.3013,
      "step": 1833
    },
    {
      "epoch": 3.384118190212373,
      "grad_norm": 1.8055914640426636,
      "learning_rate": 5.493000000000001e-06,
      "loss": 0.4244,
      "step": 1834
    },
    {
      "epoch": 3.3859649122807016,
      "grad_norm": 1.6258183717727661,
      "learning_rate": 5.496e-06,
      "loss": 0.3016,
      "step": 1835
    },
    {
      "epoch": 3.3878116343490303,
      "grad_norm": 1.7717417478561401,
      "learning_rate": 5.499e-06,
      "loss": 0.3108,
      "step": 1836
    },
    {
      "epoch": 3.389658356417359,
      "grad_norm": 2.136167049407959,
      "learning_rate": 5.502e-06,
      "loss": 0.2754,
      "step": 1837
    },
    {
      "epoch": 3.3915050784856877,
      "grad_norm": 1.1560344696044922,
      "learning_rate": 5.505e-06,
      "loss": 0.193,
      "step": 1838
    },
    {
      "epoch": 3.393351800554017,
      "grad_norm": 1.2215458154678345,
      "learning_rate": 5.5080000000000005e-06,
      "loss": 0.1924,
      "step": 1839
    },
    {
      "epoch": 3.395198522622345,
      "grad_norm": 1.3697618246078491,
      "learning_rate": 5.511e-06,
      "loss": 0.1704,
      "step": 1840
    },
    {
      "epoch": 3.3970452446906743,
      "grad_norm": 1.5075842142105103,
      "learning_rate": 5.514e-06,
      "loss": 0.1752,
      "step": 1841
    },
    {
      "epoch": 3.398891966759003,
      "grad_norm": 1.186893343925476,
      "learning_rate": 5.517000000000001e-06,
      "loss": 0.1683,
      "step": 1842
    },
    {
      "epoch": 3.4007386888273317,
      "grad_norm": 1.2870360612869263,
      "learning_rate": 5.52e-06,
      "loss": 0.1104,
      "step": 1843
    },
    {
      "epoch": 3.4025854108956604,
      "grad_norm": 1.3535423278808594,
      "learning_rate": 5.523e-06,
      "loss": 0.1625,
      "step": 1844
    },
    {
      "epoch": 3.404432132963989,
      "grad_norm": 0.9633287191390991,
      "learning_rate": 5.526e-06,
      "loss": 0.1042,
      "step": 1845
    },
    {
      "epoch": 3.4062788550323178,
      "grad_norm": 1.3369865417480469,
      "learning_rate": 5.529e-06,
      "loss": 0.1994,
      "step": 1846
    },
    {
      "epoch": 3.4081255771006465,
      "grad_norm": 1.342253565788269,
      "learning_rate": 5.5320000000000006e-06,
      "loss": 0.1389,
      "step": 1847
    },
    {
      "epoch": 3.409972299168975,
      "grad_norm": 2.2483065128326416,
      "learning_rate": 5.535e-06,
      "loss": 0.1297,
      "step": 1848
    },
    {
      "epoch": 3.411819021237304,
      "grad_norm": 1.504280686378479,
      "learning_rate": 5.537999999999999e-06,
      "loss": 0.1155,
      "step": 1849
    },
    {
      "epoch": 3.4136657433056325,
      "grad_norm": 1.8840082883834839,
      "learning_rate": 5.541e-06,
      "loss": 0.1765,
      "step": 1850
    },
    {
      "epoch": 3.4155124653739612,
      "grad_norm": 1.5378448963165283,
      "learning_rate": 5.544e-06,
      "loss": 0.1434,
      "step": 1851
    },
    {
      "epoch": 3.41735918744229,
      "grad_norm": 1.1429051160812378,
      "learning_rate": 5.547e-06,
      "loss": 0.1191,
      "step": 1852
    },
    {
      "epoch": 3.4192059095106186,
      "grad_norm": 1.4453113079071045,
      "learning_rate": 5.55e-06,
      "loss": 0.089,
      "step": 1853
    },
    {
      "epoch": 3.4210526315789473,
      "grad_norm": 1.3558111190795898,
      "learning_rate": 5.553e-06,
      "loss": 0.1171,
      "step": 1854
    },
    {
      "epoch": 3.422899353647276,
      "grad_norm": 3.135983943939209,
      "learning_rate": 5.556e-06,
      "loss": 0.1268,
      "step": 1855
    },
    {
      "epoch": 3.4247460757156047,
      "grad_norm": 1.6183189153671265,
      "learning_rate": 5.559e-06,
      "loss": 0.119,
      "step": 1856
    },
    {
      "epoch": 3.4265927977839334,
      "grad_norm": 1.4396922588348389,
      "learning_rate": 5.562e-06,
      "loss": 0.1557,
      "step": 1857
    },
    {
      "epoch": 3.428439519852262,
      "grad_norm": 1.64681077003479,
      "learning_rate": 5.565e-06,
      "loss": 0.0944,
      "step": 1858
    },
    {
      "epoch": 3.430286241920591,
      "grad_norm": 1.7854353189468384,
      "learning_rate": 5.568e-06,
      "loss": 0.1203,
      "step": 1859
    },
    {
      "epoch": 3.4321329639889195,
      "grad_norm": 1.3215970993041992,
      "learning_rate": 5.5710000000000005e-06,
      "loss": 0.0962,
      "step": 1860
    },
    {
      "epoch": 3.4339796860572482,
      "grad_norm": 1.485531210899353,
      "learning_rate": 5.574e-06,
      "loss": 0.1415,
      "step": 1861
    },
    {
      "epoch": 3.435826408125577,
      "grad_norm": 1.915353775024414,
      "learning_rate": 5.577e-06,
      "loss": 0.1219,
      "step": 1862
    },
    {
      "epoch": 3.4376731301939056,
      "grad_norm": 5.033260345458984,
      "learning_rate": 5.58e-06,
      "loss": 0.1146,
      "step": 1863
    },
    {
      "epoch": 3.439519852262235,
      "grad_norm": 1.7047919034957886,
      "learning_rate": 5.583e-06,
      "loss": 0.1072,
      "step": 1864
    },
    {
      "epoch": 3.441366574330563,
      "grad_norm": 1.481224536895752,
      "learning_rate": 5.586e-06,
      "loss": 0.1166,
      "step": 1865
    },
    {
      "epoch": 3.443213296398892,
      "grad_norm": 1.5594127178192139,
      "learning_rate": 5.589e-06,
      "loss": 0.1386,
      "step": 1866
    },
    {
      "epoch": 3.445060018467221,
      "grad_norm": 1.7530744075775146,
      "learning_rate": 5.592000000000001e-06,
      "loss": 0.1145,
      "step": 1867
    },
    {
      "epoch": 3.4469067405355496,
      "grad_norm": 1.3681082725524902,
      "learning_rate": 5.595e-06,
      "loss": 0.1197,
      "step": 1868
    },
    {
      "epoch": 3.4487534626038783,
      "grad_norm": 1.802425503730774,
      "learning_rate": 5.598e-06,
      "loss": 0.1206,
      "step": 1869
    },
    {
      "epoch": 3.450600184672207,
      "grad_norm": 1.789586067199707,
      "learning_rate": 5.601e-06,
      "loss": 0.1961,
      "step": 1870
    },
    {
      "epoch": 3.4524469067405357,
      "grad_norm": 1.5348275899887085,
      "learning_rate": 5.604e-06,
      "loss": 0.1543,
      "step": 1871
    },
    {
      "epoch": 3.4542936288088644,
      "grad_norm": 1.5526305437088013,
      "learning_rate": 5.607000000000001e-06,
      "loss": 0.1006,
      "step": 1872
    },
    {
      "epoch": 3.456140350877193,
      "grad_norm": 4.2244415283203125,
      "learning_rate": 5.6100000000000005e-06,
      "loss": 0.1398,
      "step": 1873
    },
    {
      "epoch": 3.4579870729455218,
      "grad_norm": 2.305570602416992,
      "learning_rate": 5.6129999999999995e-06,
      "loss": 0.1257,
      "step": 1874
    },
    {
      "epoch": 3.4598337950138505,
      "grad_norm": 1.7376497983932495,
      "learning_rate": 5.616e-06,
      "loss": 0.1504,
      "step": 1875
    },
    {
      "epoch": 3.461680517082179,
      "grad_norm": 3.7213504314422607,
      "learning_rate": 5.619e-06,
      "loss": 0.2539,
      "step": 1876
    },
    {
      "epoch": 3.463527239150508,
      "grad_norm": 3.0193634033203125,
      "learning_rate": 5.6220000000000006e-06,
      "loss": 0.6193,
      "step": 1877
    },
    {
      "epoch": 3.4653739612188366,
      "grad_norm": 2.032485008239746,
      "learning_rate": 5.625e-06,
      "loss": 0.3857,
      "step": 1878
    },
    {
      "epoch": 3.4672206832871653,
      "grad_norm": 1.407131314277649,
      "learning_rate": 5.628e-06,
      "loss": 0.455,
      "step": 1879
    },
    {
      "epoch": 3.469067405355494,
      "grad_norm": 1.7298333644866943,
      "learning_rate": 5.631e-06,
      "loss": 0.4029,
      "step": 1880
    },
    {
      "epoch": 3.4709141274238227,
      "grad_norm": 1.9133267402648926,
      "learning_rate": 5.634e-06,
      "loss": 0.438,
      "step": 1881
    },
    {
      "epoch": 3.4727608494921514,
      "grad_norm": 1.6703765392303467,
      "learning_rate": 5.637e-06,
      "loss": 0.4298,
      "step": 1882
    },
    {
      "epoch": 3.47460757156048,
      "grad_norm": 1.2871016263961792,
      "learning_rate": 5.64e-06,
      "loss": 0.3063,
      "step": 1883
    },
    {
      "epoch": 3.4764542936288088,
      "grad_norm": 2.3849496841430664,
      "learning_rate": 5.643e-06,
      "loss": 0.4411,
      "step": 1884
    },
    {
      "epoch": 3.4783010156971375,
      "grad_norm": 1.4804643392562866,
      "learning_rate": 5.646000000000001e-06,
      "loss": 0.27,
      "step": 1885
    },
    {
      "epoch": 3.480147737765466,
      "grad_norm": 2.092325210571289,
      "learning_rate": 5.649e-06,
      "loss": 0.2392,
      "step": 1886
    },
    {
      "epoch": 3.481994459833795,
      "grad_norm": 1.2303647994995117,
      "learning_rate": 5.652e-06,
      "loss": 0.1759,
      "step": 1887
    },
    {
      "epoch": 3.4838411819021236,
      "grad_norm": 1.8435465097427368,
      "learning_rate": 5.655e-06,
      "loss": 0.2126,
      "step": 1888
    },
    {
      "epoch": 3.4856879039704527,
      "grad_norm": 1.3646825551986694,
      "learning_rate": 5.658e-06,
      "loss": 0.213,
      "step": 1889
    },
    {
      "epoch": 3.487534626038781,
      "grad_norm": 1.2749691009521484,
      "learning_rate": 5.6610000000000005e-06,
      "loss": 0.1466,
      "step": 1890
    },
    {
      "epoch": 3.48938134810711,
      "grad_norm": 1.3183910846710205,
      "learning_rate": 5.664e-06,
      "loss": 0.1894,
      "step": 1891
    },
    {
      "epoch": 3.4912280701754383,
      "grad_norm": 1.5395432710647583,
      "learning_rate": 5.667e-06,
      "loss": 0.1734,
      "step": 1892
    },
    {
      "epoch": 3.4930747922437675,
      "grad_norm": 2.1787004470825195,
      "learning_rate": 5.67e-06,
      "loss": 0.1379,
      "step": 1893
    },
    {
      "epoch": 3.494921514312096,
      "grad_norm": 1.3665242195129395,
      "learning_rate": 5.673e-06,
      "loss": 0.1335,
      "step": 1894
    },
    {
      "epoch": 3.496768236380425,
      "grad_norm": 5.0550103187561035,
      "learning_rate": 5.676e-06,
      "loss": 0.1146,
      "step": 1895
    },
    {
      "epoch": 3.4986149584487536,
      "grad_norm": 2.3658273220062256,
      "learning_rate": 5.679e-06,
      "loss": 0.1421,
      "step": 1896
    },
    {
      "epoch": 3.5004616805170823,
      "grad_norm": 1.0035806894302368,
      "learning_rate": 5.682000000000001e-06,
      "loss": 0.1141,
      "step": 1897
    },
    {
      "epoch": 3.502308402585411,
      "grad_norm": 6.410629749298096,
      "learning_rate": 5.685e-06,
      "loss": 0.1171,
      "step": 1898
    },
    {
      "epoch": 3.5041551246537397,
      "grad_norm": 1.1340585947036743,
      "learning_rate": 5.688e-06,
      "loss": 0.1112,
      "step": 1899
    },
    {
      "epoch": 3.5060018467220684,
      "grad_norm": 1.2341041564941406,
      "learning_rate": 5.691e-06,
      "loss": 0.129,
      "step": 1900
    },
    {
      "epoch": 3.507848568790397,
      "grad_norm": 2.3341636657714844,
      "learning_rate": 5.694e-06,
      "loss": 0.116,
      "step": 1901
    },
    {
      "epoch": 3.509695290858726,
      "grad_norm": 1.6923784017562866,
      "learning_rate": 5.697000000000001e-06,
      "loss": 0.1143,
      "step": 1902
    },
    {
      "epoch": 3.5115420129270545,
      "grad_norm": 2.5729382038116455,
      "learning_rate": 5.7000000000000005e-06,
      "loss": 0.1099,
      "step": 1903
    },
    {
      "epoch": 3.513388734995383,
      "grad_norm": 1.2444424629211426,
      "learning_rate": 5.703e-06,
      "loss": 0.1303,
      "step": 1904
    },
    {
      "epoch": 3.515235457063712,
      "grad_norm": 1.2153400182724,
      "learning_rate": 5.706e-06,
      "loss": 0.0941,
      "step": 1905
    },
    {
      "epoch": 3.5170821791320406,
      "grad_norm": 1.57122004032135,
      "learning_rate": 5.709e-06,
      "loss": 0.1996,
      "step": 1906
    },
    {
      "epoch": 3.5189289012003693,
      "grad_norm": 1.2429941892623901,
      "learning_rate": 5.7120000000000005e-06,
      "loss": 0.1322,
      "step": 1907
    },
    {
      "epoch": 3.520775623268698,
      "grad_norm": 1.3662152290344238,
      "learning_rate": 5.715e-06,
      "loss": 0.0835,
      "step": 1908
    },
    {
      "epoch": 3.5226223453370267,
      "grad_norm": 4.294366359710693,
      "learning_rate": 5.718e-06,
      "loss": 0.1686,
      "step": 1909
    },
    {
      "epoch": 3.5244690674053554,
      "grad_norm": 2.29258394241333,
      "learning_rate": 5.721000000000001e-06,
      "loss": 0.1044,
      "step": 1910
    },
    {
      "epoch": 3.526315789473684,
      "grad_norm": 1.248388648033142,
      "learning_rate": 5.724e-06,
      "loss": 0.0925,
      "step": 1911
    },
    {
      "epoch": 3.5281625115420128,
      "grad_norm": 1.540623426437378,
      "learning_rate": 5.7269999999999995e-06,
      "loss": 0.1123,
      "step": 1912
    },
    {
      "epoch": 3.5300092336103415,
      "grad_norm": 2.4571754932403564,
      "learning_rate": 5.73e-06,
      "loss": 0.1247,
      "step": 1913
    },
    {
      "epoch": 3.5318559556786706,
      "grad_norm": 1.2694852352142334,
      "learning_rate": 5.733e-06,
      "loss": 0.1746,
      "step": 1914
    },
    {
      "epoch": 3.533702677746999,
      "grad_norm": 1.9375823736190796,
      "learning_rate": 5.736000000000001e-06,
      "loss": 0.1105,
      "step": 1915
    },
    {
      "epoch": 3.535549399815328,
      "grad_norm": 1.210279107093811,
      "learning_rate": 5.7390000000000004e-06,
      "loss": 0.1333,
      "step": 1916
    },
    {
      "epoch": 3.5373961218836563,
      "grad_norm": 1.4333395957946777,
      "learning_rate": 5.741999999999999e-06,
      "loss": 0.1251,
      "step": 1917
    },
    {
      "epoch": 3.5392428439519854,
      "grad_norm": 1.5113213062286377,
      "learning_rate": 5.745e-06,
      "loss": 0.105,
      "step": 1918
    },
    {
      "epoch": 3.541089566020314,
      "grad_norm": 1.5336154699325562,
      "learning_rate": 5.748e-06,
      "loss": 0.1238,
      "step": 1919
    },
    {
      "epoch": 3.542936288088643,
      "grad_norm": 1.4925528764724731,
      "learning_rate": 5.7510000000000005e-06,
      "loss": 0.1156,
      "step": 1920
    },
    {
      "epoch": 3.5447830101569715,
      "grad_norm": 1.5090351104736328,
      "learning_rate": 5.754e-06,
      "loss": 0.1149,
      "step": 1921
    },
    {
      "epoch": 3.5466297322253,
      "grad_norm": 1.419166922569275,
      "learning_rate": 5.757e-06,
      "loss": 0.1477,
      "step": 1922
    },
    {
      "epoch": 3.548476454293629,
      "grad_norm": 1.3856487274169922,
      "learning_rate": 5.76e-06,
      "loss": 0.1166,
      "step": 1923
    },
    {
      "epoch": 3.5503231763619576,
      "grad_norm": 1.7472511529922485,
      "learning_rate": 5.763e-06,
      "loss": 0.146,
      "step": 1924
    },
    {
      "epoch": 3.5521698984302863,
      "grad_norm": 1.4180855751037598,
      "learning_rate": 5.766e-06,
      "loss": 0.1507,
      "step": 1925
    },
    {
      "epoch": 3.554016620498615,
      "grad_norm": 2.02449107170105,
      "learning_rate": 5.769e-06,
      "loss": 0.1547,
      "step": 1926
    },
    {
      "epoch": 3.5558633425669437,
      "grad_norm": 4.551610469818115,
      "learning_rate": 5.772e-06,
      "loss": 0.5175,
      "step": 1927
    },
    {
      "epoch": 3.5577100646352724,
      "grad_norm": 1.604357123374939,
      "learning_rate": 5.775000000000001e-06,
      "loss": 0.5595,
      "step": 1928
    },
    {
      "epoch": 3.559556786703601,
      "grad_norm": 1.855041265487671,
      "learning_rate": 5.7779999999999996e-06,
      "loss": 0.4088,
      "step": 1929
    },
    {
      "epoch": 3.56140350877193,
      "grad_norm": 1.8530044555664062,
      "learning_rate": 5.781e-06,
      "loss": 0.3441,
      "step": 1930
    },
    {
      "epoch": 3.5632502308402585,
      "grad_norm": 2.204385280609131,
      "learning_rate": 5.784e-06,
      "loss": 0.5021,
      "step": 1931
    },
    {
      "epoch": 3.565096952908587,
      "grad_norm": 2.2241737842559814,
      "learning_rate": 5.787e-06,
      "loss": 0.3674,
      "step": 1932
    },
    {
      "epoch": 3.566943674976916,
      "grad_norm": 1.4868899583816528,
      "learning_rate": 5.7900000000000005e-06,
      "loss": 0.297,
      "step": 1933
    },
    {
      "epoch": 3.5687903970452446,
      "grad_norm": 2.145841121673584,
      "learning_rate": 5.793e-06,
      "loss": 0.292,
      "step": 1934
    },
    {
      "epoch": 3.5706371191135733,
      "grad_norm": 1.8769104480743408,
      "learning_rate": 5.796e-06,
      "loss": 0.263,
      "step": 1935
    },
    {
      "epoch": 3.572483841181902,
      "grad_norm": 4.806507110595703,
      "learning_rate": 5.799e-06,
      "loss": 0.3297,
      "step": 1936
    },
    {
      "epoch": 3.5743305632502307,
      "grad_norm": 1.6606978178024292,
      "learning_rate": 5.802e-06,
      "loss": 0.2308,
      "step": 1937
    },
    {
      "epoch": 3.5761772853185594,
      "grad_norm": 1.4710232019424438,
      "learning_rate": 5.805e-06,
      "loss": 0.2831,
      "step": 1938
    },
    {
      "epoch": 3.5780240073868885,
      "grad_norm": 2.6567530632019043,
      "learning_rate": 5.808e-06,
      "loss": 0.2106,
      "step": 1939
    },
    {
      "epoch": 3.579870729455217,
      "grad_norm": 1.5181427001953125,
      "learning_rate": 5.811000000000001e-06,
      "loss": 0.2107,
      "step": 1940
    },
    {
      "epoch": 3.581717451523546,
      "grad_norm": 0.9983479976654053,
      "learning_rate": 5.814e-06,
      "loss": 0.1584,
      "step": 1941
    },
    {
      "epoch": 3.583564173591874,
      "grad_norm": 1.196229338645935,
      "learning_rate": 5.8169999999999995e-06,
      "loss": 0.1357,
      "step": 1942
    },
    {
      "epoch": 3.5854108956602033,
      "grad_norm": 1.343436598777771,
      "learning_rate": 5.82e-06,
      "loss": 0.1194,
      "step": 1943
    },
    {
      "epoch": 3.587257617728532,
      "grad_norm": 1.8224588632583618,
      "learning_rate": 5.823e-06,
      "loss": 0.1649,
      "step": 1944
    },
    {
      "epoch": 3.5891043397968607,
      "grad_norm": 1.064284324645996,
      "learning_rate": 5.826000000000001e-06,
      "loss": 0.106,
      "step": 1945
    },
    {
      "epoch": 3.5909510618651894,
      "grad_norm": 1.1194217205047607,
      "learning_rate": 5.8290000000000004e-06,
      "loss": 0.1177,
      "step": 1946
    },
    {
      "epoch": 3.592797783933518,
      "grad_norm": 1.6041698455810547,
      "learning_rate": 5.832e-06,
      "loss": 0.133,
      "step": 1947
    },
    {
      "epoch": 3.594644506001847,
      "grad_norm": 1.0496641397476196,
      "learning_rate": 5.835e-06,
      "loss": 0.0764,
      "step": 1948
    },
    {
      "epoch": 3.5964912280701755,
      "grad_norm": 1.0636441707611084,
      "learning_rate": 5.838e-06,
      "loss": 0.1288,
      "step": 1949
    },
    {
      "epoch": 3.598337950138504,
      "grad_norm": 1.3821836709976196,
      "learning_rate": 5.8410000000000005e-06,
      "loss": 0.1299,
      "step": 1950
    },
    {
      "epoch": 3.600184672206833,
      "grad_norm": 1.1364189386367798,
      "learning_rate": 5.844e-06,
      "loss": 0.1114,
      "step": 1951
    },
    {
      "epoch": 3.6020313942751616,
      "grad_norm": 1.2574251890182495,
      "learning_rate": 5.847e-06,
      "loss": 0.1286,
      "step": 1952
    },
    {
      "epoch": 3.6038781163434903,
      "grad_norm": 1.209403157234192,
      "learning_rate": 5.850000000000001e-06,
      "loss": 0.1498,
      "step": 1953
    },
    {
      "epoch": 3.605724838411819,
      "grad_norm": 1.3319051265716553,
      "learning_rate": 5.853e-06,
      "loss": 0.1073,
      "step": 1954
    },
    {
      "epoch": 3.6075715604801477,
      "grad_norm": 1.2135307788848877,
      "learning_rate": 5.856e-06,
      "loss": 0.106,
      "step": 1955
    },
    {
      "epoch": 3.6094182825484764,
      "grad_norm": 1.4304325580596924,
      "learning_rate": 5.859e-06,
      "loss": 0.1141,
      "step": 1956
    },
    {
      "epoch": 3.611265004616805,
      "grad_norm": 5.753938674926758,
      "learning_rate": 5.862e-06,
      "loss": 0.1105,
      "step": 1957
    },
    {
      "epoch": 3.613111726685134,
      "grad_norm": 2.1282248497009277,
      "learning_rate": 5.865000000000001e-06,
      "loss": 0.1175,
      "step": 1958
    },
    {
      "epoch": 3.6149584487534625,
      "grad_norm": 1.8825041055679321,
      "learning_rate": 5.868e-06,
      "loss": 0.1184,
      "step": 1959
    },
    {
      "epoch": 3.616805170821791,
      "grad_norm": 1.0043890476226807,
      "learning_rate": 5.871e-06,
      "loss": 0.0976,
      "step": 1960
    },
    {
      "epoch": 3.61865189289012,
      "grad_norm": 2.1982343196868896,
      "learning_rate": 5.874e-06,
      "loss": 0.0953,
      "step": 1961
    },
    {
      "epoch": 3.6204986149584486,
      "grad_norm": 1.6441147327423096,
      "learning_rate": 5.877e-06,
      "loss": 0.1208,
      "step": 1962
    },
    {
      "epoch": 3.6223453370267773,
      "grad_norm": 1.4551324844360352,
      "learning_rate": 5.8800000000000005e-06,
      "loss": 0.1104,
      "step": 1963
    },
    {
      "epoch": 3.6241920590951064,
      "grad_norm": 1.2573660612106323,
      "learning_rate": 5.883e-06,
      "loss": 0.1104,
      "step": 1964
    },
    {
      "epoch": 3.6260387811634347,
      "grad_norm": 1.6478545665740967,
      "learning_rate": 5.886000000000001e-06,
      "loss": 0.1119,
      "step": 1965
    },
    {
      "epoch": 3.627885503231764,
      "grad_norm": 1.328403115272522,
      "learning_rate": 5.889e-06,
      "loss": 0.1182,
      "step": 1966
    },
    {
      "epoch": 3.629732225300092,
      "grad_norm": 2.832441806793213,
      "learning_rate": 5.892e-06,
      "loss": 0.1433,
      "step": 1967
    },
    {
      "epoch": 3.6315789473684212,
      "grad_norm": 1.5062068700790405,
      "learning_rate": 5.895e-06,
      "loss": 0.0838,
      "step": 1968
    },
    {
      "epoch": 3.6334256694367495,
      "grad_norm": 1.5999783277511597,
      "learning_rate": 5.898e-06,
      "loss": 0.0968,
      "step": 1969
    },
    {
      "epoch": 3.6352723915050786,
      "grad_norm": 1.426970362663269,
      "learning_rate": 5.901000000000001e-06,
      "loss": 0.1154,
      "step": 1970
    },
    {
      "epoch": 3.6371191135734073,
      "grad_norm": 3.131009340286255,
      "learning_rate": 5.9040000000000006e-06,
      "loss": 0.1042,
      "step": 1971
    },
    {
      "epoch": 3.638965835641736,
      "grad_norm": 1.6253173351287842,
      "learning_rate": 5.9069999999999995e-06,
      "loss": 0.1047,
      "step": 1972
    },
    {
      "epoch": 3.6408125577100647,
      "grad_norm": 1.677268385887146,
      "learning_rate": 5.91e-06,
      "loss": 0.115,
      "step": 1973
    },
    {
      "epoch": 3.6426592797783934,
      "grad_norm": 2.1825778484344482,
      "learning_rate": 5.913e-06,
      "loss": 0.1139,
      "step": 1974
    },
    {
      "epoch": 3.644506001846722,
      "grad_norm": 2.2541301250457764,
      "learning_rate": 5.916e-06,
      "loss": 0.1551,
      "step": 1975
    },
    {
      "epoch": 3.646352723915051,
      "grad_norm": 1.7591955661773682,
      "learning_rate": 5.919e-06,
      "loss": 0.2356,
      "step": 1976
    },
    {
      "epoch": 3.6481994459833795,
      "grad_norm": 2.3737945556640625,
      "learning_rate": 5.922e-06,
      "loss": 0.5899,
      "step": 1977
    },
    {
      "epoch": 3.6500461680517082,
      "grad_norm": 1.9221827983856201,
      "learning_rate": 5.925e-06,
      "loss": 0.4634,
      "step": 1978
    },
    {
      "epoch": 3.651892890120037,
      "grad_norm": 2.4015867710113525,
      "learning_rate": 5.928e-06,
      "loss": 0.4463,
      "step": 1979
    },
    {
      "epoch": 3.6537396121883656,
      "grad_norm": 4.889469146728516,
      "learning_rate": 5.931e-06,
      "loss": 0.426,
      "step": 1980
    },
    {
      "epoch": 3.6555863342566943,
      "grad_norm": 2.2621586322784424,
      "learning_rate": 5.934e-06,
      "loss": 0.4606,
      "step": 1981
    },
    {
      "epoch": 3.657433056325023,
      "grad_norm": 1.894276738166809,
      "learning_rate": 5.937e-06,
      "loss": 0.3655,
      "step": 1982
    },
    {
      "epoch": 3.6592797783933517,
      "grad_norm": 3.954047203063965,
      "learning_rate": 5.940000000000001e-06,
      "loss": 0.3543,
      "step": 1983
    },
    {
      "epoch": 3.6611265004616804,
      "grad_norm": 1.4562526941299438,
      "learning_rate": 5.943e-06,
      "loss": 0.3325,
      "step": 1984
    },
    {
      "epoch": 3.662973222530009,
      "grad_norm": 2.2910425662994385,
      "learning_rate": 5.9459999999999995e-06,
      "loss": 0.3744,
      "step": 1985
    },
    {
      "epoch": 3.664819944598338,
      "grad_norm": 1.5386607646942139,
      "learning_rate": 5.949e-06,
      "loss": 0.2899,
      "step": 1986
    },
    {
      "epoch": 3.6666666666666665,
      "grad_norm": 1.333303451538086,
      "learning_rate": 5.952e-06,
      "loss": 0.2228,
      "step": 1987
    },
    {
      "epoch": 3.668513388734995,
      "grad_norm": 3.049942970275879,
      "learning_rate": 5.955000000000001e-06,
      "loss": 0.2197,
      "step": 1988
    },
    {
      "epoch": 3.6703601108033244,
      "grad_norm": 2.74458384513855,
      "learning_rate": 5.958e-06,
      "loss": 0.2569,
      "step": 1989
    },
    {
      "epoch": 3.6722068328716526,
      "grad_norm": 1.1316044330596924,
      "learning_rate": 5.961e-06,
      "loss": 0.1435,
      "step": 1990
    },
    {
      "epoch": 3.6740535549399818,
      "grad_norm": 1.435760498046875,
      "learning_rate": 5.964e-06,
      "loss": 0.165,
      "step": 1991
    },
    {
      "epoch": 3.67590027700831,
      "grad_norm": 1.0853261947631836,
      "learning_rate": 5.967e-06,
      "loss": 0.0853,
      "step": 1992
    },
    {
      "epoch": 3.677746999076639,
      "grad_norm": 1.67013418674469,
      "learning_rate": 5.9700000000000004e-06,
      "loss": 0.2136,
      "step": 1993
    },
    {
      "epoch": 3.6795937211449674,
      "grad_norm": 1.1271677017211914,
      "learning_rate": 5.973e-06,
      "loss": 0.0932,
      "step": 1994
    },
    {
      "epoch": 3.6814404432132966,
      "grad_norm": 1.1513522863388062,
      "learning_rate": 5.976e-06,
      "loss": 0.1335,
      "step": 1995
    },
    {
      "epoch": 3.6832871652816253,
      "grad_norm": 1.074549674987793,
      "learning_rate": 5.979000000000001e-06,
      "loss": 0.0735,
      "step": 1996
    },
    {
      "epoch": 3.685133887349954,
      "grad_norm": 1.177593469619751,
      "learning_rate": 5.982e-06,
      "loss": 0.2005,
      "step": 1997
    },
    {
      "epoch": 3.6869806094182827,
      "grad_norm": 1.4211703538894653,
      "learning_rate": 5.985e-06,
      "loss": 0.143,
      "step": 1998
    },
    {
      "epoch": 3.6888273314866113,
      "grad_norm": 1.6892846822738647,
      "learning_rate": 5.988e-06,
      "loss": 0.1164,
      "step": 1999
    },
    {
      "epoch": 3.69067405355494,
      "grad_norm": 1.2731127738952637,
      "learning_rate": 5.991e-06,
      "loss": 0.092,
      "step": 2000
    },
    {
      "epoch": 3.69067405355494,
      "eval_cer": 0.15273349405085931,
      "eval_loss": 0.5228983163833618,
      "eval_runtime": 15.933,
      "eval_samples_per_second": 19.08,
      "eval_steps_per_second": 0.628,
      "eval_wer": 0.5397160399079048,
      "step": 2000
    },
    {
      "epoch": 3.6925207756232687,
      "grad_norm": 1.0816723108291626,
      "learning_rate": 5.9940000000000005e-06,
      "loss": 0.1102,
      "step": 2001
    },
    {
      "epoch": 3.6943674976915974,
      "grad_norm": 1.175575852394104,
      "learning_rate": 5.997e-06,
      "loss": 0.1103,
      "step": 2002
    },
    {
      "epoch": 3.696214219759926,
      "grad_norm": 1.5876294374465942,
      "learning_rate": 6e-06,
      "loss": 0.1206,
      "step": 2003
    },
    {
      "epoch": 3.698060941828255,
      "grad_norm": 1.339050054550171,
      "learning_rate": 6.003e-06,
      "loss": 0.0987,
      "step": 2004
    },
    {
      "epoch": 3.6999076638965835,
      "grad_norm": 2.4920217990875244,
      "learning_rate": 6.006e-06,
      "loss": 0.0947,
      "step": 2005
    },
    {
      "epoch": 3.7017543859649122,
      "grad_norm": 1.912536859512329,
      "learning_rate": 6.009e-06,
      "loss": 0.1419,
      "step": 2006
    },
    {
      "epoch": 3.703601108033241,
      "grad_norm": 1.2025752067565918,
      "learning_rate": 6.012e-06,
      "loss": 0.1129,
      "step": 2007
    },
    {
      "epoch": 3.7054478301015696,
      "grad_norm": 3.4578535556793213,
      "learning_rate": 6.015000000000001e-06,
      "loss": 0.1331,
      "step": 2008
    },
    {
      "epoch": 3.7072945521698983,
      "grad_norm": 1.6533622741699219,
      "learning_rate": 6.018e-06,
      "loss": 0.1541,
      "step": 2009
    },
    {
      "epoch": 3.709141274238227,
      "grad_norm": 1.2792508602142334,
      "learning_rate": 6.021e-06,
      "loss": 0.0943,
      "step": 2010
    },
    {
      "epoch": 3.7109879963065557,
      "grad_norm": 1.6220040321350098,
      "learning_rate": 6.024e-06,
      "loss": 0.1202,
      "step": 2011
    },
    {
      "epoch": 3.7128347183748844,
      "grad_norm": 1.363448143005371,
      "learning_rate": 6.027e-06,
      "loss": 0.1145,
      "step": 2012
    },
    {
      "epoch": 3.714681440443213,
      "grad_norm": 1.4746122360229492,
      "learning_rate": 6.030000000000001e-06,
      "loss": 0.083,
      "step": 2013
    },
    {
      "epoch": 3.7165281625115423,
      "grad_norm": 1.3242028951644897,
      "learning_rate": 6.0330000000000005e-06,
      "loss": 0.0927,
      "step": 2014
    },
    {
      "epoch": 3.7183748845798705,
      "grad_norm": 1.3828983306884766,
      "learning_rate": 6.0359999999999995e-06,
      "loss": 0.112,
      "step": 2015
    },
    {
      "epoch": 3.7202216066481997,
      "grad_norm": 1.7886730432510376,
      "learning_rate": 6.039e-06,
      "loss": 0.1225,
      "step": 2016
    },
    {
      "epoch": 3.722068328716528,
      "grad_norm": 4.497995853424072,
      "learning_rate": 6.042e-06,
      "loss": 0.0945,
      "step": 2017
    },
    {
      "epoch": 3.723915050784857,
      "grad_norm": 1.2448220252990723,
      "learning_rate": 6.0450000000000006e-06,
      "loss": 0.0997,
      "step": 2018
    },
    {
      "epoch": 3.7257617728531853,
      "grad_norm": 1.299957513809204,
      "learning_rate": 6.048e-06,
      "loss": 0.1073,
      "step": 2019
    },
    {
      "epoch": 3.7276084949215145,
      "grad_norm": 1.221881628036499,
      "learning_rate": 6.051e-06,
      "loss": 0.1206,
      "step": 2020
    },
    {
      "epoch": 3.729455216989843,
      "grad_norm": 1.9921565055847168,
      "learning_rate": 6.054e-06,
      "loss": 0.1125,
      "step": 2021
    },
    {
      "epoch": 3.731301939058172,
      "grad_norm": 1.4520527124404907,
      "learning_rate": 6.057e-06,
      "loss": 0.0999,
      "step": 2022
    },
    {
      "epoch": 3.7331486611265006,
      "grad_norm": 1.3707276582717896,
      "learning_rate": 6.0600000000000004e-06,
      "loss": 0.1016,
      "step": 2023
    },
    {
      "epoch": 3.7349953831948293,
      "grad_norm": 1.359123945236206,
      "learning_rate": 6.063e-06,
      "loss": 0.1095,
      "step": 2024
    },
    {
      "epoch": 3.736842105263158,
      "grad_norm": 3.0324485301971436,
      "learning_rate": 6.066e-06,
      "loss": 0.143,
      "step": 2025
    },
    {
      "epoch": 3.7386888273314867,
      "grad_norm": 2.0628585815429688,
      "learning_rate": 6.069000000000001e-06,
      "loss": 0.1456,
      "step": 2026
    },
    {
      "epoch": 3.7405355493998154,
      "grad_norm": 5.545530319213867,
      "learning_rate": 6.072e-06,
      "loss": 0.5264,
      "step": 2027
    },
    {
      "epoch": 3.742382271468144,
      "grad_norm": 2.2468793392181396,
      "learning_rate": 6.075e-06,
      "loss": 0.542,
      "step": 2028
    },
    {
      "epoch": 3.7442289935364728,
      "grad_norm": 1.6987916231155396,
      "learning_rate": 6.078e-06,
      "loss": 0.3783,
      "step": 2029
    },
    {
      "epoch": 3.7460757156048015,
      "grad_norm": 1.487799882888794,
      "learning_rate": 6.081e-06,
      "loss": 0.426,
      "step": 2030
    },
    {
      "epoch": 3.74792243767313,
      "grad_norm": 1.8354527950286865,
      "learning_rate": 6.0840000000000005e-06,
      "loss": 0.3765,
      "step": 2031
    },
    {
      "epoch": 3.749769159741459,
      "grad_norm": 1.3133721351623535,
      "learning_rate": 6.087e-06,
      "loss": 0.2697,
      "step": 2032
    },
    {
      "epoch": 3.7516158818097876,
      "grad_norm": 1.655531406402588,
      "learning_rate": 6.090000000000001e-06,
      "loss": 0.4274,
      "step": 2033
    },
    {
      "epoch": 3.7534626038781163,
      "grad_norm": 1.0694929361343384,
      "learning_rate": 6.093e-06,
      "loss": 0.3198,
      "step": 2034
    },
    {
      "epoch": 3.755309325946445,
      "grad_norm": 1.8194829225540161,
      "learning_rate": 6.096e-06,
      "loss": 0.3143,
      "step": 2035
    },
    {
      "epoch": 3.7571560480147737,
      "grad_norm": 1.8747495412826538,
      "learning_rate": 6.099e-06,
      "loss": 0.2571,
      "step": 2036
    },
    {
      "epoch": 3.7590027700831024,
      "grad_norm": 1.4677729606628418,
      "learning_rate": 6.102e-06,
      "loss": 0.2237,
      "step": 2037
    },
    {
      "epoch": 3.760849492151431,
      "grad_norm": 1.2673132419586182,
      "learning_rate": 6.105e-06,
      "loss": 0.2766,
      "step": 2038
    },
    {
      "epoch": 3.76269621421976,
      "grad_norm": 2.505880117416382,
      "learning_rate": 6.108000000000001e-06,
      "loss": 0.3313,
      "step": 2039
    },
    {
      "epoch": 3.7645429362880884,
      "grad_norm": 2.244453191757202,
      "learning_rate": 6.111e-06,
      "loss": 0.1818,
      "step": 2040
    },
    {
      "epoch": 3.7663896583564176,
      "grad_norm": 1.4449100494384766,
      "learning_rate": 6.114e-06,
      "loss": 0.1676,
      "step": 2041
    },
    {
      "epoch": 3.768236380424746,
      "grad_norm": 1.4164530038833618,
      "learning_rate": 6.117e-06,
      "loss": 0.128,
      "step": 2042
    },
    {
      "epoch": 3.770083102493075,
      "grad_norm": 1.2722538709640503,
      "learning_rate": 6.12e-06,
      "loss": 0.1526,
      "step": 2043
    },
    {
      "epoch": 3.7719298245614032,
      "grad_norm": 1.0496044158935547,
      "learning_rate": 6.1230000000000005e-06,
      "loss": 0.1081,
      "step": 2044
    },
    {
      "epoch": 3.7737765466297324,
      "grad_norm": 0.9792291522026062,
      "learning_rate": 6.126e-06,
      "loss": 0.1203,
      "step": 2045
    },
    {
      "epoch": 3.775623268698061,
      "grad_norm": 0.8785455822944641,
      "learning_rate": 6.129e-06,
      "loss": 0.0942,
      "step": 2046
    },
    {
      "epoch": 3.77746999076639,
      "grad_norm": 2.594404935836792,
      "learning_rate": 6.132e-06,
      "loss": 0.1138,
      "step": 2047
    },
    {
      "epoch": 3.7793167128347185,
      "grad_norm": 1.3891233205795288,
      "learning_rate": 6.135e-06,
      "loss": 0.1122,
      "step": 2048
    },
    {
      "epoch": 3.781163434903047,
      "grad_norm": 1.6807653903961182,
      "learning_rate": 6.138e-06,
      "loss": 0.1289,
      "step": 2049
    },
    {
      "epoch": 3.783010156971376,
      "grad_norm": 1.5894526243209839,
      "learning_rate": 6.141e-06,
      "loss": 0.1294,
      "step": 2050
    },
    {
      "epoch": 3.7848568790397046,
      "grad_norm": 1.5388537645339966,
      "learning_rate": 6.144000000000001e-06,
      "loss": 0.1347,
      "step": 2051
    },
    {
      "epoch": 3.7867036011080333,
      "grad_norm": 1.0728822946548462,
      "learning_rate": 6.147e-06,
      "loss": 0.0983,
      "step": 2052
    },
    {
      "epoch": 3.788550323176362,
      "grad_norm": 1.2530133724212646,
      "learning_rate": 6.1499999999999996e-06,
      "loss": 0.1288,
      "step": 2053
    },
    {
      "epoch": 3.7903970452446907,
      "grad_norm": 1.2521809339523315,
      "learning_rate": 6.153e-06,
      "loss": 0.101,
      "step": 2054
    },
    {
      "epoch": 3.7922437673130194,
      "grad_norm": 1.032920002937317,
      "learning_rate": 6.156e-06,
      "loss": 0.0842,
      "step": 2055
    },
    {
      "epoch": 3.794090489381348,
      "grad_norm": 1.201108694076538,
      "learning_rate": 6.159000000000001e-06,
      "loss": 0.0904,
      "step": 2056
    },
    {
      "epoch": 3.7959372114496768,
      "grad_norm": 1.3609286546707153,
      "learning_rate": 6.1620000000000005e-06,
      "loss": 0.0745,
      "step": 2057
    },
    {
      "epoch": 3.7977839335180055,
      "grad_norm": 1.2280689477920532,
      "learning_rate": 6.164999999999999e-06,
      "loss": 0.1081,
      "step": 2058
    },
    {
      "epoch": 3.799630655586334,
      "grad_norm": 1.2103352546691895,
      "learning_rate": 6.168e-06,
      "loss": 0.1182,
      "step": 2059
    },
    {
      "epoch": 3.801477377654663,
      "grad_norm": 1.2055014371871948,
      "learning_rate": 6.171e-06,
      "loss": 0.1004,
      "step": 2060
    },
    {
      "epoch": 3.8033240997229916,
      "grad_norm": 1.1704323291778564,
      "learning_rate": 6.1740000000000005e-06,
      "loss": 0.1044,
      "step": 2061
    },
    {
      "epoch": 3.8051708217913203,
      "grad_norm": 1.461685061454773,
      "learning_rate": 6.177e-06,
      "loss": 0.0929,
      "step": 2062
    },
    {
      "epoch": 3.807017543859649,
      "grad_norm": 1.297804594039917,
      "learning_rate": 6.18e-06,
      "loss": 0.1194,
      "step": 2063
    },
    {
      "epoch": 3.808864265927978,
      "grad_norm": 2.321072816848755,
      "learning_rate": 6.183e-06,
      "loss": 0.1522,
      "step": 2064
    },
    {
      "epoch": 3.8107109879963064,
      "grad_norm": 1.7555663585662842,
      "learning_rate": 6.186e-06,
      "loss": 0.0904,
      "step": 2065
    },
    {
      "epoch": 3.8125577100646355,
      "grad_norm": 1.1725404262542725,
      "learning_rate": 6.189e-06,
      "loss": 0.1416,
      "step": 2066
    },
    {
      "epoch": 3.8144044321329638,
      "grad_norm": 1.6720110177993774,
      "learning_rate": 6.192e-06,
      "loss": 0.1405,
      "step": 2067
    },
    {
      "epoch": 3.816251154201293,
      "grad_norm": 1.382629632949829,
      "learning_rate": 6.195e-06,
      "loss": 0.0963,
      "step": 2068
    },
    {
      "epoch": 3.818097876269621,
      "grad_norm": 1.3463071584701538,
      "learning_rate": 6.198000000000001e-06,
      "loss": 0.0951,
      "step": 2069
    },
    {
      "epoch": 3.8199445983379503,
      "grad_norm": 1.4168822765350342,
      "learning_rate": 6.201e-06,
      "loss": 0.0903,
      "step": 2070
    },
    {
      "epoch": 3.821791320406279,
      "grad_norm": 1.5806570053100586,
      "learning_rate": 6.204e-06,
      "loss": 0.0922,
      "step": 2071
    },
    {
      "epoch": 3.8236380424746077,
      "grad_norm": 1.9736270904541016,
      "learning_rate": 6.207e-06,
      "loss": 0.1037,
      "step": 2072
    },
    {
      "epoch": 3.8254847645429364,
      "grad_norm": 1.1318566799163818,
      "learning_rate": 6.21e-06,
      "loss": 0.0986,
      "step": 2073
    },
    {
      "epoch": 3.827331486611265,
      "grad_norm": 1.464217185974121,
      "learning_rate": 6.2130000000000005e-06,
      "loss": 0.1275,
      "step": 2074
    },
    {
      "epoch": 3.829178208679594,
      "grad_norm": 2.048211097717285,
      "learning_rate": 6.216e-06,
      "loss": 0.132,
      "step": 2075
    },
    {
      "epoch": 3.8310249307479225,
      "grad_norm": 2.4093711376190186,
      "learning_rate": 6.219000000000001e-06,
      "loss": 0.1212,
      "step": 2076
    },
    {
      "epoch": 3.832871652816251,
      "grad_norm": 1.6214300394058228,
      "learning_rate": 6.222e-06,
      "loss": 0.4919,
      "step": 2077
    },
    {
      "epoch": 3.83471837488458,
      "grad_norm": 1.7599939107894897,
      "learning_rate": 6.225e-06,
      "loss": 0.3767,
      "step": 2078
    },
    {
      "epoch": 3.8365650969529086,
      "grad_norm": 3.717496871948242,
      "learning_rate": 6.228e-06,
      "loss": 0.4679,
      "step": 2079
    },
    {
      "epoch": 3.8384118190212373,
      "grad_norm": 1.5492913722991943,
      "learning_rate": 6.231e-06,
      "loss": 0.4594,
      "step": 2080
    },
    {
      "epoch": 3.840258541089566,
      "grad_norm": 1.411643147468567,
      "learning_rate": 6.234000000000001e-06,
      "loss": 0.3923,
      "step": 2081
    },
    {
      "epoch": 3.8421052631578947,
      "grad_norm": 1.8489406108856201,
      "learning_rate": 6.237000000000001e-06,
      "loss": 0.2718,
      "step": 2082
    },
    {
      "epoch": 3.8439519852262234,
      "grad_norm": 1.3735231161117554,
      "learning_rate": 6.2399999999999995e-06,
      "loss": 0.3215,
      "step": 2083
    },
    {
      "epoch": 3.845798707294552,
      "grad_norm": 1.6795915365219116,
      "learning_rate": 6.243e-06,
      "loss": 0.2696,
      "step": 2084
    },
    {
      "epoch": 3.847645429362881,
      "grad_norm": 1.3095409870147705,
      "learning_rate": 6.246e-06,
      "loss": 0.3077,
      "step": 2085
    },
    {
      "epoch": 3.8494921514312095,
      "grad_norm": 1.792952537536621,
      "learning_rate": 6.249000000000001e-06,
      "loss": 0.2491,
      "step": 2086
    },
    {
      "epoch": 3.851338873499538,
      "grad_norm": 1.0679808855056763,
      "learning_rate": 6.2520000000000004e-06,
      "loss": 0.2769,
      "step": 2087
    },
    {
      "epoch": 3.853185595567867,
      "grad_norm": 1.1066974401474,
      "learning_rate": 6.255e-06,
      "loss": 0.2073,
      "step": 2088
    },
    {
      "epoch": 3.855032317636196,
      "grad_norm": 2.68408465385437,
      "learning_rate": 6.258e-06,
      "loss": 0.3588,
      "step": 2089
    },
    {
      "epoch": 3.8568790397045243,
      "grad_norm": 5.881558895111084,
      "learning_rate": 6.261e-06,
      "loss": 0.2474,
      "step": 2090
    },
    {
      "epoch": 3.8587257617728534,
      "grad_norm": 1.1949398517608643,
      "learning_rate": 6.2640000000000005e-06,
      "loss": 0.1647,
      "step": 2091
    },
    {
      "epoch": 3.8605724838411817,
      "grad_norm": 2.1825714111328125,
      "learning_rate": 6.267e-06,
      "loss": 0.1624,
      "step": 2092
    },
    {
      "epoch": 3.862419205909511,
      "grad_norm": 4.6003546714782715,
      "learning_rate": 6.27e-06,
      "loss": 0.1243,
      "step": 2093
    },
    {
      "epoch": 3.864265927977839,
      "grad_norm": 1.0134408473968506,
      "learning_rate": 6.273000000000001e-06,
      "loss": 0.1175,
      "step": 2094
    },
    {
      "epoch": 3.866112650046168,
      "grad_norm": 1.549644112586975,
      "learning_rate": 6.276e-06,
      "loss": 0.0978,
      "step": 2095
    },
    {
      "epoch": 3.867959372114497,
      "grad_norm": 1.3444812297821045,
      "learning_rate": 6.279e-06,
      "loss": 0.1493,
      "step": 2096
    },
    {
      "epoch": 3.8698060941828256,
      "grad_norm": 1.0612382888793945,
      "learning_rate": 6.282e-06,
      "loss": 0.1069,
      "step": 2097
    },
    {
      "epoch": 3.8716528162511543,
      "grad_norm": 1.1147220134735107,
      "learning_rate": 6.285e-06,
      "loss": 0.1083,
      "step": 2098
    },
    {
      "epoch": 3.873499538319483,
      "grad_norm": 1.3620564937591553,
      "learning_rate": 6.288000000000001e-06,
      "loss": 0.0893,
      "step": 2099
    },
    {
      "epoch": 3.8753462603878117,
      "grad_norm": 3.2398104667663574,
      "learning_rate": 6.291e-06,
      "loss": 0.1036,
      "step": 2100
    },
    {
      "epoch": 3.8771929824561404,
      "grad_norm": 1.14120614528656,
      "learning_rate": 6.293999999999999e-06,
      "loss": 0.1057,
      "step": 2101
    },
    {
      "epoch": 3.879039704524469,
      "grad_norm": 1.2740801572799683,
      "learning_rate": 6.297e-06,
      "loss": 0.1245,
      "step": 2102
    },
    {
      "epoch": 3.880886426592798,
      "grad_norm": 1.1721112728118896,
      "learning_rate": 6.3e-06,
      "loss": 0.1275,
      "step": 2103
    },
    {
      "epoch": 3.8827331486611265,
      "grad_norm": 1.4239356517791748,
      "learning_rate": 6.3030000000000005e-06,
      "loss": 0.2154,
      "step": 2104
    },
    {
      "epoch": 3.884579870729455,
      "grad_norm": 1.0123012065887451,
      "learning_rate": 6.306e-06,
      "loss": 0.0966,
      "step": 2105
    },
    {
      "epoch": 3.886426592797784,
      "grad_norm": 1.8538200855255127,
      "learning_rate": 6.309e-06,
      "loss": 0.1693,
      "step": 2106
    },
    {
      "epoch": 3.8882733148661126,
      "grad_norm": 0.9299476742744446,
      "learning_rate": 6.312e-06,
      "loss": 0.0891,
      "step": 2107
    },
    {
      "epoch": 3.8901200369344413,
      "grad_norm": 1.3803143501281738,
      "learning_rate": 6.315e-06,
      "loss": 0.1548,
      "step": 2108
    },
    {
      "epoch": 3.89196675900277,
      "grad_norm": 1.0537395477294922,
      "learning_rate": 6.318e-06,
      "loss": 0.072,
      "step": 2109
    },
    {
      "epoch": 3.8938134810710987,
      "grad_norm": 1.2850028276443481,
      "learning_rate": 6.321e-06,
      "loss": 0.0819,
      "step": 2110
    },
    {
      "epoch": 3.8956602031394274,
      "grad_norm": 1.0578045845031738,
      "learning_rate": 6.324e-06,
      "loss": 0.1024,
      "step": 2111
    },
    {
      "epoch": 3.897506925207756,
      "grad_norm": 1.1398320198059082,
      "learning_rate": 6.327000000000001e-06,
      "loss": 0.0915,
      "step": 2112
    },
    {
      "epoch": 3.899353647276085,
      "grad_norm": 1.2906196117401123,
      "learning_rate": 6.3299999999999995e-06,
      "loss": 0.14,
      "step": 2113
    },
    {
      "epoch": 3.901200369344414,
      "grad_norm": 1.2287468910217285,
      "learning_rate": 6.333e-06,
      "loss": 0.0972,
      "step": 2114
    },
    {
      "epoch": 3.903047091412742,
      "grad_norm": 1.1388696432113647,
      "learning_rate": 6.336e-06,
      "loss": 0.0845,
      "step": 2115
    },
    {
      "epoch": 3.9048938134810713,
      "grad_norm": 1.6295071840286255,
      "learning_rate": 6.339e-06,
      "loss": 0.0994,
      "step": 2116
    },
    {
      "epoch": 3.9067405355493996,
      "grad_norm": 1.361403226852417,
      "learning_rate": 6.3420000000000004e-06,
      "loss": 0.094,
      "step": 2117
    },
    {
      "epoch": 3.9085872576177287,
      "grad_norm": 1.7775616645812988,
      "learning_rate": 6.345e-06,
      "loss": 0.1288,
      "step": 2118
    },
    {
      "epoch": 3.910433979686057,
      "grad_norm": 1.6608407497406006,
      "learning_rate": 6.348000000000001e-06,
      "loss": 0.1267,
      "step": 2119
    },
    {
      "epoch": 3.912280701754386,
      "grad_norm": 2.0776705741882324,
      "learning_rate": 6.351e-06,
      "loss": 0.1085,
      "step": 2120
    },
    {
      "epoch": 3.914127423822715,
      "grad_norm": 1.467939853668213,
      "learning_rate": 6.354e-06,
      "loss": 0.114,
      "step": 2121
    },
    {
      "epoch": 3.9159741458910435,
      "grad_norm": 1.66560697555542,
      "learning_rate": 6.357e-06,
      "loss": 0.1166,
      "step": 2122
    },
    {
      "epoch": 3.9178208679593722,
      "grad_norm": 2.018527030944824,
      "learning_rate": 6.36e-06,
      "loss": 0.2089,
      "step": 2123
    },
    {
      "epoch": 3.919667590027701,
      "grad_norm": 1.6574705839157104,
      "learning_rate": 6.363000000000001e-06,
      "loss": 0.0896,
      "step": 2124
    },
    {
      "epoch": 3.9215143120960296,
      "grad_norm": 1.3335787057876587,
      "learning_rate": 6.3660000000000005e-06,
      "loss": 0.1261,
      "step": 2125
    },
    {
      "epoch": 3.9233610341643583,
      "grad_norm": 1.821389079093933,
      "learning_rate": 6.3689999999999995e-06,
      "loss": 0.1744,
      "step": 2126
    },
    {
      "epoch": 3.925207756232687,
      "grad_norm": 3.1028544902801514,
      "learning_rate": 6.372e-06,
      "loss": 0.5456,
      "step": 2127
    },
    {
      "epoch": 3.9270544783010157,
      "grad_norm": 1.4182378053665161,
      "learning_rate": 6.375e-06,
      "loss": 0.4202,
      "step": 2128
    },
    {
      "epoch": 3.9289012003693444,
      "grad_norm": 1.26549232006073,
      "learning_rate": 6.378000000000001e-06,
      "loss": 0.3366,
      "step": 2129
    },
    {
      "epoch": 3.930747922437673,
      "grad_norm": 1.560637354850769,
      "learning_rate": 6.381e-06,
      "loss": 0.3467,
      "step": 2130
    },
    {
      "epoch": 3.932594644506002,
      "grad_norm": 2.5093820095062256,
      "learning_rate": 6.384e-06,
      "loss": 0.4014,
      "step": 2131
    },
    {
      "epoch": 3.9344413665743305,
      "grad_norm": 1.0541409254074097,
      "learning_rate": 6.387e-06,
      "loss": 0.2948,
      "step": 2132
    },
    {
      "epoch": 3.936288088642659,
      "grad_norm": 1.5808500051498413,
      "learning_rate": 6.39e-06,
      "loss": 0.2711,
      "step": 2133
    },
    {
      "epoch": 3.938134810710988,
      "grad_norm": 1.3126081228256226,
      "learning_rate": 6.3930000000000005e-06,
      "loss": 0.2422,
      "step": 2134
    },
    {
      "epoch": 3.9399815327793166,
      "grad_norm": 1.0315762758255005,
      "learning_rate": 6.396e-06,
      "loss": 0.1848,
      "step": 2135
    },
    {
      "epoch": 3.9418282548476453,
      "grad_norm": 1.1984046697616577,
      "learning_rate": 6.399e-06,
      "loss": 0.1908,
      "step": 2136
    },
    {
      "epoch": 3.943674976915974,
      "grad_norm": 0.9420778155326843,
      "learning_rate": 6.402000000000001e-06,
      "loss": 0.1425,
      "step": 2137
    },
    {
      "epoch": 3.9455216989843027,
      "grad_norm": 1.1212126016616821,
      "learning_rate": 6.405e-06,
      "loss": 0.1209,
      "step": 2138
    },
    {
      "epoch": 3.9473684210526314,
      "grad_norm": 0.931004524230957,
      "learning_rate": 6.408e-06,
      "loss": 0.1815,
      "step": 2139
    },
    {
      "epoch": 3.94921514312096,
      "grad_norm": 0.990565836429596,
      "learning_rate": 6.411e-06,
      "loss": 0.0673,
      "step": 2140
    },
    {
      "epoch": 3.9510618651892893,
      "grad_norm": 1.6753158569335938,
      "learning_rate": 6.414e-06,
      "loss": 0.1226,
      "step": 2141
    },
    {
      "epoch": 3.9529085872576175,
      "grad_norm": 1.456681251525879,
      "learning_rate": 6.4170000000000006e-06,
      "loss": 0.1283,
      "step": 2142
    },
    {
      "epoch": 3.9547553093259467,
      "grad_norm": 1.3051668405532837,
      "learning_rate": 6.42e-06,
      "loss": 0.1209,
      "step": 2143
    },
    {
      "epoch": 3.956602031394275,
      "grad_norm": 1.251253604888916,
      "learning_rate": 6.423e-06,
      "loss": 0.0885,
      "step": 2144
    },
    {
      "epoch": 3.958448753462604,
      "grad_norm": 1.232637882232666,
      "learning_rate": 6.426e-06,
      "loss": 0.1096,
      "step": 2145
    },
    {
      "epoch": 3.9602954755309328,
      "grad_norm": 1.351613163948059,
      "learning_rate": 6.429e-06,
      "loss": 0.0987,
      "step": 2146
    },
    {
      "epoch": 3.9621421975992615,
      "grad_norm": 1.1840687990188599,
      "learning_rate": 6.432e-06,
      "loss": 0.0946,
      "step": 2147
    },
    {
      "epoch": 3.96398891966759,
      "grad_norm": 1.0928349494934082,
      "learning_rate": 6.435e-06,
      "loss": 0.1247,
      "step": 2148
    },
    {
      "epoch": 3.965835641735919,
      "grad_norm": 1.0318615436553955,
      "learning_rate": 6.438000000000001e-06,
      "loss": 0.1022,
      "step": 2149
    },
    {
      "epoch": 3.9676823638042475,
      "grad_norm": 0.8724531531333923,
      "learning_rate": 6.441e-06,
      "loss": 0.0947,
      "step": 2150
    },
    {
      "epoch": 3.9695290858725762,
      "grad_norm": 1.2489190101623535,
      "learning_rate": 6.444e-06,
      "loss": 0.0858,
      "step": 2151
    },
    {
      "epoch": 3.971375807940905,
      "grad_norm": 1.392446517944336,
      "learning_rate": 6.447e-06,
      "loss": 0.0902,
      "step": 2152
    },
    {
      "epoch": 3.9732225300092336,
      "grad_norm": 1.2182376384735107,
      "learning_rate": 6.45e-06,
      "loss": 0.0871,
      "step": 2153
    },
    {
      "epoch": 3.9750692520775623,
      "grad_norm": 1.9293391704559326,
      "learning_rate": 6.453000000000001e-06,
      "loss": 0.1234,
      "step": 2154
    },
    {
      "epoch": 3.976915974145891,
      "grad_norm": 1.5359190702438354,
      "learning_rate": 6.4560000000000005e-06,
      "loss": 0.1224,
      "step": 2155
    },
    {
      "epoch": 3.9787626962142197,
      "grad_norm": 1.3025881052017212,
      "learning_rate": 6.4589999999999995e-06,
      "loss": 0.1664,
      "step": 2156
    },
    {
      "epoch": 3.9806094182825484,
      "grad_norm": 1.6642571687698364,
      "learning_rate": 6.462e-06,
      "loss": 0.0904,
      "step": 2157
    },
    {
      "epoch": 3.982456140350877,
      "grad_norm": 1.8174163103103638,
      "learning_rate": 6.465e-06,
      "loss": 0.1102,
      "step": 2158
    },
    {
      "epoch": 3.984302862419206,
      "grad_norm": 2.3735907077789307,
      "learning_rate": 6.468000000000001e-06,
      "loss": 0.1019,
      "step": 2159
    },
    {
      "epoch": 3.9861495844875345,
      "grad_norm": 1.7283834218978882,
      "learning_rate": 6.471e-06,
      "loss": 0.0737,
      "step": 2160
    },
    {
      "epoch": 3.9879963065558632,
      "grad_norm": 5.035957336425781,
      "learning_rate": 6.474e-06,
      "loss": 0.105,
      "step": 2161
    },
    {
      "epoch": 3.989843028624192,
      "grad_norm": 1.3808785676956177,
      "learning_rate": 6.477000000000001e-06,
      "loss": 0.0946,
      "step": 2162
    },
    {
      "epoch": 3.9916897506925206,
      "grad_norm": 1.2071834802627563,
      "learning_rate": 6.48e-06,
      "loss": 0.1027,
      "step": 2163
    },
    {
      "epoch": 3.9935364727608493,
      "grad_norm": 1.04526948928833,
      "learning_rate": 6.483e-06,
      "loss": 0.0779,
      "step": 2164
    },
    {
      "epoch": 3.995383194829178,
      "grad_norm": 1.663916826248169,
      "learning_rate": 6.486e-06,
      "loss": 0.1289,
      "step": 2165
    },
    {
      "epoch": 3.997229916897507,
      "grad_norm": 1.2232156991958618,
      "learning_rate": 6.489e-06,
      "loss": 0.0969,
      "step": 2166
    },
    {
      "epoch": 3.9990766389658354,
      "grad_norm": 1.5424480438232422,
      "learning_rate": 6.492000000000001e-06,
      "loss": 0.1221,
      "step": 2167
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.185368299484253,
      "learning_rate": 6.4950000000000005e-06,
      "loss": 0.0528,
      "step": 2168
    },
    {
      "epoch": 4.001846722068329,
      "grad_norm": 1.911673665046692,
      "learning_rate": 6.4979999999999994e-06,
      "loss": 0.5474,
      "step": 2169
    },
    {
      "epoch": 4.003693444136657,
      "grad_norm": 1.5852998495101929,
      "learning_rate": 6.501e-06,
      "loss": 0.4573,
      "step": 2170
    },
    {
      "epoch": 4.0055401662049865,
      "grad_norm": 2.0277857780456543,
      "learning_rate": 6.504e-06,
      "loss": 0.3894,
      "step": 2171
    },
    {
      "epoch": 4.007386888273315,
      "grad_norm": 2.6222214698791504,
      "learning_rate": 6.5070000000000005e-06,
      "loss": 0.454,
      "step": 2172
    },
    {
      "epoch": 4.009233610341644,
      "grad_norm": 1.520296573638916,
      "learning_rate": 6.51e-06,
      "loss": 0.3785,
      "step": 2173
    },
    {
      "epoch": 4.011080332409972,
      "grad_norm": 1.2729988098144531,
      "learning_rate": 6.513e-06,
      "loss": 0.3233,
      "step": 2174
    },
    {
      "epoch": 4.012927054478301,
      "grad_norm": 1.6835660934448242,
      "learning_rate": 6.516e-06,
      "loss": 0.2551,
      "step": 2175
    },
    {
      "epoch": 4.01477377654663,
      "grad_norm": 1.376844048500061,
      "learning_rate": 6.519e-06,
      "loss": 0.2445,
      "step": 2176
    },
    {
      "epoch": 4.016620498614959,
      "grad_norm": 1.2151410579681396,
      "learning_rate": 6.522e-06,
      "loss": 0.336,
      "step": 2177
    },
    {
      "epoch": 4.018467220683287,
      "grad_norm": 1.3106422424316406,
      "learning_rate": 6.525e-06,
      "loss": 0.3283,
      "step": 2178
    },
    {
      "epoch": 4.020313942751616,
      "grad_norm": 1.0718411207199097,
      "learning_rate": 6.528e-06,
      "loss": 0.2019,
      "step": 2179
    },
    {
      "epoch": 4.022160664819944,
      "grad_norm": 1.2590523958206177,
      "learning_rate": 6.531000000000001e-06,
      "loss": 0.1917,
      "step": 2180
    },
    {
      "epoch": 4.0240073868882735,
      "grad_norm": 1.6002975702285767,
      "learning_rate": 6.534e-06,
      "loss": 0.2292,
      "step": 2181
    },
    {
      "epoch": 4.025854108956602,
      "grad_norm": 0.8854488730430603,
      "learning_rate": 6.537e-06,
      "loss": 0.1327,
      "step": 2182
    },
    {
      "epoch": 4.027700831024931,
      "grad_norm": 1.2897201776504517,
      "learning_rate": 6.54e-06,
      "loss": 0.1739,
      "step": 2183
    },
    {
      "epoch": 4.029547553093259,
      "grad_norm": 1.1138757467269897,
      "learning_rate": 6.543e-06,
      "loss": 0.1351,
      "step": 2184
    },
    {
      "epoch": 4.031394275161588,
      "grad_norm": 1.1339983940124512,
      "learning_rate": 6.5460000000000005e-06,
      "loss": 0.1039,
      "step": 2185
    },
    {
      "epoch": 4.033240997229917,
      "grad_norm": 0.9088255167007446,
      "learning_rate": 6.549e-06,
      "loss": 0.1187,
      "step": 2186
    },
    {
      "epoch": 4.035087719298246,
      "grad_norm": 1.0308983325958252,
      "learning_rate": 6.552e-06,
      "loss": 0.08,
      "step": 2187
    },
    {
      "epoch": 4.036934441366574,
      "grad_norm": 2.115433692932129,
      "learning_rate": 6.555e-06,
      "loss": 0.106,
      "step": 2188
    },
    {
      "epoch": 4.038781163434903,
      "grad_norm": 1.531274437904358,
      "learning_rate": 6.558e-06,
      "loss": 0.1008,
      "step": 2189
    },
    {
      "epoch": 4.040627885503231,
      "grad_norm": 1.2350906133651733,
      "learning_rate": 6.561e-06,
      "loss": 0.0928,
      "step": 2190
    },
    {
      "epoch": 4.0424746075715605,
      "grad_norm": 0.9239581227302551,
      "learning_rate": 6.564e-06,
      "loss": 0.0777,
      "step": 2191
    },
    {
      "epoch": 4.044321329639889,
      "grad_norm": 0.9720399975776672,
      "learning_rate": 6.567000000000001e-06,
      "loss": 0.0563,
      "step": 2192
    },
    {
      "epoch": 4.046168051708218,
      "grad_norm": 0.8621312379837036,
      "learning_rate": 6.57e-06,
      "loss": 0.0581,
      "step": 2193
    },
    {
      "epoch": 4.048014773776547,
      "grad_norm": 3.7505435943603516,
      "learning_rate": 6.573e-06,
      "loss": 0.0927,
      "step": 2194
    },
    {
      "epoch": 4.049861495844875,
      "grad_norm": 1.2696458101272583,
      "learning_rate": 6.576e-06,
      "loss": 0.0902,
      "step": 2195
    },
    {
      "epoch": 4.0517082179132045,
      "grad_norm": 2.1792521476745605,
      "learning_rate": 6.579e-06,
      "loss": 0.1256,
      "step": 2196
    },
    {
      "epoch": 4.053554939981533,
      "grad_norm": 1.2086941003799438,
      "learning_rate": 6.582000000000001e-06,
      "loss": 0.1057,
      "step": 2197
    },
    {
      "epoch": 4.055401662049862,
      "grad_norm": 1.3100030422210693,
      "learning_rate": 6.5850000000000005e-06,
      "loss": 0.0875,
      "step": 2198
    },
    {
      "epoch": 4.05724838411819,
      "grad_norm": 1.3629885911941528,
      "learning_rate": 6.5879999999999994e-06,
      "loss": 0.1275,
      "step": 2199
    },
    {
      "epoch": 4.059095106186519,
      "grad_norm": 4.132709980010986,
      "learning_rate": 6.591e-06,
      "loss": 0.0776,
      "step": 2200
    },
    {
      "epoch": 4.0609418282548475,
      "grad_norm": 1.144539475440979,
      "learning_rate": 6.594e-06,
      "loss": 0.0939,
      "step": 2201
    },
    {
      "epoch": 4.062788550323177,
      "grad_norm": 1.3136900663375854,
      "learning_rate": 6.5970000000000005e-06,
      "loss": 0.0804,
      "step": 2202
    },
    {
      "epoch": 4.064635272391505,
      "grad_norm": 1.6249648332595825,
      "learning_rate": 6.6e-06,
      "loss": 0.0888,
      "step": 2203
    },
    {
      "epoch": 4.066481994459834,
      "grad_norm": 1.2383456230163574,
      "learning_rate": 6.603e-06,
      "loss": 0.0901,
      "step": 2204
    },
    {
      "epoch": 4.068328716528162,
      "grad_norm": 1.428479552268982,
      "learning_rate": 6.606000000000001e-06,
      "loss": 0.1368,
      "step": 2205
    },
    {
      "epoch": 4.0701754385964914,
      "grad_norm": 1.5173728466033936,
      "learning_rate": 6.609e-06,
      "loss": 0.0862,
      "step": 2206
    },
    {
      "epoch": 4.07202216066482,
      "grad_norm": 1.3671185970306396,
      "learning_rate": 6.612e-06,
      "loss": 0.0681,
      "step": 2207
    },
    {
      "epoch": 4.073868882733149,
      "grad_norm": 1.0532758235931396,
      "learning_rate": 6.615e-06,
      "loss": 0.0832,
      "step": 2208
    },
    {
      "epoch": 4.075715604801477,
      "grad_norm": 1.199684977531433,
      "learning_rate": 6.618e-06,
      "loss": 0.0798,
      "step": 2209
    },
    {
      "epoch": 4.077562326869806,
      "grad_norm": 2.115050792694092,
      "learning_rate": 6.621000000000001e-06,
      "loss": 0.1533,
      "step": 2210
    },
    {
      "epoch": 4.0794090489381345,
      "grad_norm": 1.1212481260299683,
      "learning_rate": 6.6240000000000004e-06,
      "loss": 0.0706,
      "step": 2211
    },
    {
      "epoch": 4.081255771006464,
      "grad_norm": 1.539804458618164,
      "learning_rate": 6.627e-06,
      "loss": 0.0979,
      "step": 2212
    },
    {
      "epoch": 4.083102493074792,
      "grad_norm": 1.9622479677200317,
      "learning_rate": 6.63e-06,
      "loss": 0.1041,
      "step": 2213
    },
    {
      "epoch": 4.084949215143121,
      "grad_norm": 1.1550004482269287,
      "learning_rate": 6.633e-06,
      "loss": 0.0762,
      "step": 2214
    },
    {
      "epoch": 4.086795937211449,
      "grad_norm": 1.2409030199050903,
      "learning_rate": 6.6360000000000005e-06,
      "loss": 0.1235,
      "step": 2215
    },
    {
      "epoch": 4.088642659279778,
      "grad_norm": 1.394726276397705,
      "learning_rate": 6.639e-06,
      "loss": 0.0847,
      "step": 2216
    },
    {
      "epoch": 4.090489381348107,
      "grad_norm": 1.7710202932357788,
      "learning_rate": 6.642000000000001e-06,
      "loss": 0.1489,
      "step": 2217
    },
    {
      "epoch": 4.092336103416436,
      "grad_norm": 1.8116905689239502,
      "learning_rate": 6.645e-06,
      "loss": 0.1583,
      "step": 2218
    },
    {
      "epoch": 4.094182825484765,
      "grad_norm": 2.0445168018341064,
      "learning_rate": 6.648e-06,
      "loss": 0.4672,
      "step": 2219
    },
    {
      "epoch": 4.096029547553093,
      "grad_norm": 1.6109222173690796,
      "learning_rate": 6.651e-06,
      "loss": 0.3575,
      "step": 2220
    },
    {
      "epoch": 4.097876269621422,
      "grad_norm": 1.006513237953186,
      "learning_rate": 6.654e-06,
      "loss": 0.3371,
      "step": 2221
    },
    {
      "epoch": 4.099722991689751,
      "grad_norm": 1.294262409210205,
      "learning_rate": 6.657e-06,
      "loss": 0.4369,
      "step": 2222
    },
    {
      "epoch": 4.10156971375808,
      "grad_norm": 2.5763139724731445,
      "learning_rate": 6.660000000000001e-06,
      "loss": 0.3756,
      "step": 2223
    },
    {
      "epoch": 4.103416435826408,
      "grad_norm": 1.9553343057632446,
      "learning_rate": 6.6629999999999996e-06,
      "loss": 0.3584,
      "step": 2224
    },
    {
      "epoch": 4.105263157894737,
      "grad_norm": 1.4016964435577393,
      "learning_rate": 6.666e-06,
      "loss": 0.2867,
      "step": 2225
    },
    {
      "epoch": 4.107109879963065,
      "grad_norm": 0.8154698610305786,
      "learning_rate": 6.669e-06,
      "loss": 0.2086,
      "step": 2226
    },
    {
      "epoch": 4.108956602031395,
      "grad_norm": 1.5167957544326782,
      "learning_rate": 6.672e-06,
      "loss": 0.2631,
      "step": 2227
    },
    {
      "epoch": 4.110803324099723,
      "grad_norm": 1.1658596992492676,
      "learning_rate": 6.6750000000000005e-06,
      "loss": 0.3025,
      "step": 2228
    },
    {
      "epoch": 4.112650046168052,
      "grad_norm": 1.6568107604980469,
      "learning_rate": 6.678e-06,
      "loss": 0.2013,
      "step": 2229
    },
    {
      "epoch": 4.11449676823638,
      "grad_norm": 1.6511210203170776,
      "learning_rate": 6.681e-06,
      "loss": 0.2295,
      "step": 2230
    },
    {
      "epoch": 4.116343490304709,
      "grad_norm": 0.9506266713142395,
      "learning_rate": 6.684e-06,
      "loss": 0.1298,
      "step": 2231
    },
    {
      "epoch": 4.118190212373038,
      "grad_norm": 1.0544261932373047,
      "learning_rate": 6.687e-06,
      "loss": 0.1572,
      "step": 2232
    },
    {
      "epoch": 4.120036934441367,
      "grad_norm": 1.4018826484680176,
      "learning_rate": 6.69e-06,
      "loss": 0.1093,
      "step": 2233
    },
    {
      "epoch": 4.121883656509695,
      "grad_norm": 0.9542692303657532,
      "learning_rate": 6.693e-06,
      "loss": 0.112,
      "step": 2234
    },
    {
      "epoch": 4.123730378578024,
      "grad_norm": 1.1946085691452026,
      "learning_rate": 6.696000000000001e-06,
      "loss": 0.0899,
      "step": 2235
    },
    {
      "epoch": 4.125577100646352,
      "grad_norm": 0.9538592100143433,
      "learning_rate": 6.699e-06,
      "loss": 0.0872,
      "step": 2236
    },
    {
      "epoch": 4.127423822714682,
      "grad_norm": 1.4511198997497559,
      "learning_rate": 6.7019999999999995e-06,
      "loss": 0.0991,
      "step": 2237
    },
    {
      "epoch": 4.12927054478301,
      "grad_norm": 1.070308804512024,
      "learning_rate": 6.705e-06,
      "loss": 0.1096,
      "step": 2238
    },
    {
      "epoch": 4.131117266851339,
      "grad_norm": 0.942848265171051,
      "learning_rate": 6.708e-06,
      "loss": 0.0773,
      "step": 2239
    },
    {
      "epoch": 4.132963988919667,
      "grad_norm": 1.3079289197921753,
      "learning_rate": 6.711000000000001e-06,
      "loss": 0.1019,
      "step": 2240
    },
    {
      "epoch": 4.134810710987996,
      "grad_norm": 1.2125372886657715,
      "learning_rate": 6.7140000000000004e-06,
      "loss": 0.1144,
      "step": 2241
    },
    {
      "epoch": 4.136657433056325,
      "grad_norm": 1.0549064874649048,
      "learning_rate": 6.716999999999999e-06,
      "loss": 0.088,
      "step": 2242
    },
    {
      "epoch": 4.138504155124654,
      "grad_norm": 1.2122917175292969,
      "learning_rate": 6.72e-06,
      "loss": 0.1125,
      "step": 2243
    },
    {
      "epoch": 4.140350877192983,
      "grad_norm": 0.9945119619369507,
      "learning_rate": 6.723e-06,
      "loss": 0.0933,
      "step": 2244
    },
    {
      "epoch": 4.142197599261311,
      "grad_norm": 1.1362215280532837,
      "learning_rate": 6.7260000000000005e-06,
      "loss": 0.1095,
      "step": 2245
    },
    {
      "epoch": 4.14404432132964,
      "grad_norm": 1.3205327987670898,
      "learning_rate": 6.729e-06,
      "loss": 0.1808,
      "step": 2246
    },
    {
      "epoch": 4.1458910433979685,
      "grad_norm": 1.3926762342453003,
      "learning_rate": 6.732e-06,
      "loss": 0.1172,
      "step": 2247
    },
    {
      "epoch": 4.147737765466298,
      "grad_norm": 1.0345947742462158,
      "learning_rate": 6.735000000000001e-06,
      "loss": 0.0861,
      "step": 2248
    },
    {
      "epoch": 4.149584487534626,
      "grad_norm": 1.600387692451477,
      "learning_rate": 6.738e-06,
      "loss": 0.127,
      "step": 2249
    },
    {
      "epoch": 4.151431209602955,
      "grad_norm": 1.2428592443466187,
      "learning_rate": 6.741e-06,
      "loss": 0.1019,
      "step": 2250
    },
    {
      "epoch": 4.153277931671283,
      "grad_norm": 1.1123014688491821,
      "learning_rate": 6.744e-06,
      "loss": 0.0856,
      "step": 2251
    },
    {
      "epoch": 4.1551246537396125,
      "grad_norm": 1.348739504814148,
      "learning_rate": 6.747e-06,
      "loss": 0.1091,
      "step": 2252
    },
    {
      "epoch": 4.156971375807941,
      "grad_norm": 1.0651084184646606,
      "learning_rate": 6.750000000000001e-06,
      "loss": 0.0778,
      "step": 2253
    },
    {
      "epoch": 4.15881809787627,
      "grad_norm": 1.460421085357666,
      "learning_rate": 6.753e-06,
      "loss": 0.078,
      "step": 2254
    },
    {
      "epoch": 4.160664819944598,
      "grad_norm": 0.9551540017127991,
      "learning_rate": 6.756e-06,
      "loss": 0.0792,
      "step": 2255
    },
    {
      "epoch": 4.162511542012927,
      "grad_norm": 1.3960237503051758,
      "learning_rate": 6.759e-06,
      "loss": 0.068,
      "step": 2256
    },
    {
      "epoch": 4.1643582640812555,
      "grad_norm": 1.785151481628418,
      "learning_rate": 6.762e-06,
      "loss": 0.0848,
      "step": 2257
    },
    {
      "epoch": 4.166204986149585,
      "grad_norm": 1.0868968963623047,
      "learning_rate": 6.7650000000000005e-06,
      "loss": 0.0919,
      "step": 2258
    },
    {
      "epoch": 4.168051708217913,
      "grad_norm": 1.7152469158172607,
      "learning_rate": 6.768e-06,
      "loss": 0.1152,
      "step": 2259
    },
    {
      "epoch": 4.169898430286242,
      "grad_norm": 1.454818606376648,
      "learning_rate": 6.771000000000001e-06,
      "loss": 0.1094,
      "step": 2260
    },
    {
      "epoch": 4.17174515235457,
      "grad_norm": 1.5478395223617554,
      "learning_rate": 6.774e-06,
      "loss": 0.1006,
      "step": 2261
    },
    {
      "epoch": 4.1735918744228995,
      "grad_norm": 1.4219355583190918,
      "learning_rate": 6.777e-06,
      "loss": 0.1003,
      "step": 2262
    },
    {
      "epoch": 4.175438596491228,
      "grad_norm": 1.2621865272521973,
      "learning_rate": 6.78e-06,
      "loss": 0.0945,
      "step": 2263
    },
    {
      "epoch": 4.177285318559557,
      "grad_norm": 1.3872356414794922,
      "learning_rate": 6.783e-06,
      "loss": 0.07,
      "step": 2264
    },
    {
      "epoch": 4.179132040627885,
      "grad_norm": 1.5787618160247803,
      "learning_rate": 6.786000000000001e-06,
      "loss": 0.0851,
      "step": 2265
    },
    {
      "epoch": 4.180978762696214,
      "grad_norm": 1.548042893409729,
      "learning_rate": 6.7890000000000006e-06,
      "loss": 0.11,
      "step": 2266
    },
    {
      "epoch": 4.1828254847645425,
      "grad_norm": 1.557713270187378,
      "learning_rate": 6.7919999999999995e-06,
      "loss": 0.097,
      "step": 2267
    },
    {
      "epoch": 4.184672206832872,
      "grad_norm": 1.7162854671478271,
      "learning_rate": 6.795e-06,
      "loss": 0.1412,
      "step": 2268
    },
    {
      "epoch": 4.186518928901201,
      "grad_norm": 4.086080551147461,
      "learning_rate": 6.798e-06,
      "loss": 0.4405,
      "step": 2269
    },
    {
      "epoch": 4.188365650969529,
      "grad_norm": 3.8445074558258057,
      "learning_rate": 6.801000000000001e-06,
      "loss": 0.438,
      "step": 2270
    },
    {
      "epoch": 4.190212373037858,
      "grad_norm": 2.2936015129089355,
      "learning_rate": 6.804e-06,
      "loss": 0.415,
      "step": 2271
    },
    {
      "epoch": 4.1920590951061865,
      "grad_norm": 1.6141554117202759,
      "learning_rate": 6.807e-06,
      "loss": 0.3014,
      "step": 2272
    },
    {
      "epoch": 4.193905817174516,
      "grad_norm": 2.340399742126465,
      "learning_rate": 6.81e-06,
      "loss": 0.347,
      "step": 2273
    },
    {
      "epoch": 4.195752539242844,
      "grad_norm": 2.091618299484253,
      "learning_rate": 6.813e-06,
      "loss": 0.3336,
      "step": 2274
    },
    {
      "epoch": 4.197599261311173,
      "grad_norm": 1.398072361946106,
      "learning_rate": 6.8160000000000005e-06,
      "loss": 0.2575,
      "step": 2275
    },
    {
      "epoch": 4.199445983379501,
      "grad_norm": 1.7019211053848267,
      "learning_rate": 6.819e-06,
      "loss": 0.358,
      "step": 2276
    },
    {
      "epoch": 4.20129270544783,
      "grad_norm": 2.1351137161254883,
      "learning_rate": 6.822e-06,
      "loss": 0.3326,
      "step": 2277
    },
    {
      "epoch": 4.203139427516159,
      "grad_norm": 0.9898922443389893,
      "learning_rate": 6.825000000000001e-06,
      "loss": 0.2067,
      "step": 2278
    },
    {
      "epoch": 4.204986149584488,
      "grad_norm": 1.372165560722351,
      "learning_rate": 6.828e-06,
      "loss": 0.213,
      "step": 2279
    },
    {
      "epoch": 4.206832871652816,
      "grad_norm": 1.582594633102417,
      "learning_rate": 6.831e-06,
      "loss": 0.2519,
      "step": 2280
    },
    {
      "epoch": 4.208679593721145,
      "grad_norm": 1.4090359210968018,
      "learning_rate": 6.834e-06,
      "loss": 0.2676,
      "step": 2281
    },
    {
      "epoch": 4.2105263157894735,
      "grad_norm": 1.1971672773361206,
      "learning_rate": 6.837e-06,
      "loss": 0.1274,
      "step": 2282
    },
    {
      "epoch": 4.212373037857803,
      "grad_norm": 1.5907059907913208,
      "learning_rate": 6.840000000000001e-06,
      "loss": 0.1727,
      "step": 2283
    },
    {
      "epoch": 4.214219759926131,
      "grad_norm": 1.0141587257385254,
      "learning_rate": 6.843e-06,
      "loss": 0.1094,
      "step": 2284
    },
    {
      "epoch": 4.21606648199446,
      "grad_norm": 1.7501100301742554,
      "learning_rate": 6.845999999999999e-06,
      "loss": 0.1339,
      "step": 2285
    },
    {
      "epoch": 4.217913204062788,
      "grad_norm": 1.2493884563446045,
      "learning_rate": 6.849e-06,
      "loss": 0.1022,
      "step": 2286
    },
    {
      "epoch": 4.219759926131117,
      "grad_norm": 1.592990517616272,
      "learning_rate": 6.852e-06,
      "loss": 0.0944,
      "step": 2287
    },
    {
      "epoch": 4.221606648199446,
      "grad_norm": 1.0477226972579956,
      "learning_rate": 6.8550000000000004e-06,
      "loss": 0.0782,
      "step": 2288
    },
    {
      "epoch": 4.223453370267775,
      "grad_norm": 1.2984161376953125,
      "learning_rate": 6.858e-06,
      "loss": 0.1491,
      "step": 2289
    },
    {
      "epoch": 4.225300092336103,
      "grad_norm": 1.1166598796844482,
      "learning_rate": 6.861e-06,
      "loss": 0.1148,
      "step": 2290
    },
    {
      "epoch": 4.227146814404432,
      "grad_norm": 1.163062572479248,
      "learning_rate": 6.864000000000001e-06,
      "loss": 0.0927,
      "step": 2291
    },
    {
      "epoch": 4.22899353647276,
      "grad_norm": 0.9511471390724182,
      "learning_rate": 6.867e-06,
      "loss": 0.0874,
      "step": 2292
    },
    {
      "epoch": 4.23084025854109,
      "grad_norm": 1.0879565477371216,
      "learning_rate": 6.87e-06,
      "loss": 0.0665,
      "step": 2293
    },
    {
      "epoch": 4.232686980609419,
      "grad_norm": 1.0754992961883545,
      "learning_rate": 6.873e-06,
      "loss": 0.0981,
      "step": 2294
    },
    {
      "epoch": 4.234533702677747,
      "grad_norm": 1.0654864311218262,
      "learning_rate": 6.876e-06,
      "loss": 0.0671,
      "step": 2295
    },
    {
      "epoch": 4.236380424746076,
      "grad_norm": 1.311405897140503,
      "learning_rate": 6.8790000000000005e-06,
      "loss": 0.084,
      "step": 2296
    },
    {
      "epoch": 4.238227146814404,
      "grad_norm": 1.3197026252746582,
      "learning_rate": 6.882e-06,
      "loss": 0.0855,
      "step": 2297
    },
    {
      "epoch": 4.2400738688827335,
      "grad_norm": 1.2920937538146973,
      "learning_rate": 6.885e-06,
      "loss": 0.1308,
      "step": 2298
    },
    {
      "epoch": 4.241920590951062,
      "grad_norm": 1.233723759651184,
      "learning_rate": 6.888e-06,
      "loss": 0.0839,
      "step": 2299
    },
    {
      "epoch": 4.243767313019391,
      "grad_norm": 1.194996953010559,
      "learning_rate": 6.891e-06,
      "loss": 0.0765,
      "step": 2300
    },
    {
      "epoch": 4.245614035087719,
      "grad_norm": 1.2588831186294556,
      "learning_rate": 6.894e-06,
      "loss": 0.0886,
      "step": 2301
    },
    {
      "epoch": 4.247460757156048,
      "grad_norm": 0.7167526483535767,
      "learning_rate": 6.897e-06,
      "loss": 0.0516,
      "step": 2302
    },
    {
      "epoch": 4.249307479224377,
      "grad_norm": 0.928087055683136,
      "learning_rate": 6.900000000000001e-06,
      "loss": 0.0712,
      "step": 2303
    },
    {
      "epoch": 4.251154201292706,
      "grad_norm": 1.2723053693771362,
      "learning_rate": 6.903e-06,
      "loss": 0.1029,
      "step": 2304
    },
    {
      "epoch": 4.253000923361034,
      "grad_norm": 1.5679049491882324,
      "learning_rate": 6.906e-06,
      "loss": 0.0929,
      "step": 2305
    },
    {
      "epoch": 4.254847645429363,
      "grad_norm": 1.2212111949920654,
      "learning_rate": 6.909e-06,
      "loss": 0.0773,
      "step": 2306
    },
    {
      "epoch": 4.256694367497691,
      "grad_norm": 0.8988540768623352,
      "learning_rate": 6.912e-06,
      "loss": 0.085,
      "step": 2307
    },
    {
      "epoch": 4.2585410895660205,
      "grad_norm": 1.235801100730896,
      "learning_rate": 6.915000000000001e-06,
      "loss": 0.1045,
      "step": 2308
    },
    {
      "epoch": 4.260387811634349,
      "grad_norm": 1.007037878036499,
      "learning_rate": 6.9180000000000005e-06,
      "loss": 0.0806,
      "step": 2309
    },
    {
      "epoch": 4.262234533702678,
      "grad_norm": 1.2111518383026123,
      "learning_rate": 6.9209999999999995e-06,
      "loss": 0.0723,
      "step": 2310
    },
    {
      "epoch": 4.264081255771006,
      "grad_norm": 1.328084945678711,
      "learning_rate": 6.924e-06,
      "loss": 0.1017,
      "step": 2311
    },
    {
      "epoch": 4.265927977839335,
      "grad_norm": 1.5174875259399414,
      "learning_rate": 6.927e-06,
      "loss": 0.0943,
      "step": 2312
    },
    {
      "epoch": 4.267774699907664,
      "grad_norm": 1.4504024982452393,
      "learning_rate": 6.9300000000000006e-06,
      "loss": 0.0932,
      "step": 2313
    },
    {
      "epoch": 4.269621421975993,
      "grad_norm": 1.2458757162094116,
      "learning_rate": 6.933e-06,
      "loss": 0.1036,
      "step": 2314
    },
    {
      "epoch": 4.271468144044321,
      "grad_norm": 1.3173937797546387,
      "learning_rate": 6.936e-06,
      "loss": 0.1401,
      "step": 2315
    },
    {
      "epoch": 4.27331486611265,
      "grad_norm": 1.7283494472503662,
      "learning_rate": 6.939e-06,
      "loss": 0.1783,
      "step": 2316
    },
    {
      "epoch": 4.275161588180978,
      "grad_norm": 2.916658639907837,
      "learning_rate": 6.942e-06,
      "loss": 0.0841,
      "step": 2317
    },
    {
      "epoch": 4.2770083102493075,
      "grad_norm": 1.529240608215332,
      "learning_rate": 6.945e-06,
      "loss": 0.1205,
      "step": 2318
    },
    {
      "epoch": 4.278855032317637,
      "grad_norm": 2.7654285430908203,
      "learning_rate": 6.948e-06,
      "loss": 0.5173,
      "step": 2319
    },
    {
      "epoch": 4.280701754385965,
      "grad_norm": 1.5961171388626099,
      "learning_rate": 6.951e-06,
      "loss": 0.4608,
      "step": 2320
    },
    {
      "epoch": 4.282548476454294,
      "grad_norm": 1.9149647951126099,
      "learning_rate": 6.954000000000001e-06,
      "loss": 0.3426,
      "step": 2321
    },
    {
      "epoch": 4.284395198522622,
      "grad_norm": 1.4873987436294556,
      "learning_rate": 6.957e-06,
      "loss": 0.3897,
      "step": 2322
    },
    {
      "epoch": 4.286241920590951,
      "grad_norm": 1.1254777908325195,
      "learning_rate": 6.96e-06,
      "loss": 0.338,
      "step": 2323
    },
    {
      "epoch": 4.28808864265928,
      "grad_norm": 1.5503439903259277,
      "learning_rate": 6.963e-06,
      "loss": 0.4296,
      "step": 2324
    },
    {
      "epoch": 4.289935364727609,
      "grad_norm": 1.42909574508667,
      "learning_rate": 6.966e-06,
      "loss": 0.2739,
      "step": 2325
    },
    {
      "epoch": 4.291782086795937,
      "grad_norm": 1.3683276176452637,
      "learning_rate": 6.9690000000000005e-06,
      "loss": 0.2861,
      "step": 2326
    },
    {
      "epoch": 4.293628808864266,
      "grad_norm": 1.140487790107727,
      "learning_rate": 6.972e-06,
      "loss": 0.1559,
      "step": 2327
    },
    {
      "epoch": 4.2954755309325945,
      "grad_norm": 1.4923127889633179,
      "learning_rate": 6.975e-06,
      "loss": 0.2232,
      "step": 2328
    },
    {
      "epoch": 4.297322253000924,
      "grad_norm": 1.2142882347106934,
      "learning_rate": 6.978e-06,
      "loss": 0.2329,
      "step": 2329
    },
    {
      "epoch": 4.299168975069252,
      "grad_norm": 1.9966238737106323,
      "learning_rate": 6.981e-06,
      "loss": 0.238,
      "step": 2330
    },
    {
      "epoch": 4.301015697137581,
      "grad_norm": 2.1601319313049316,
      "learning_rate": 6.984e-06,
      "loss": 0.2103,
      "step": 2331
    },
    {
      "epoch": 4.302862419205909,
      "grad_norm": 1.0669894218444824,
      "learning_rate": 6.987e-06,
      "loss": 0.1036,
      "step": 2332
    },
    {
      "epoch": 4.304709141274238,
      "grad_norm": 1.0218197107315063,
      "learning_rate": 6.990000000000001e-06,
      "loss": 0.1374,
      "step": 2333
    },
    {
      "epoch": 4.306555863342567,
      "grad_norm": 0.9428393244743347,
      "learning_rate": 6.993000000000001e-06,
      "loss": 0.1168,
      "step": 2334
    },
    {
      "epoch": 4.308402585410896,
      "grad_norm": 0.9400322437286377,
      "learning_rate": 6.996e-06,
      "loss": 0.0822,
      "step": 2335
    },
    {
      "epoch": 4.310249307479224,
      "grad_norm": 1.2620066404342651,
      "learning_rate": 6.999e-06,
      "loss": 0.1182,
      "step": 2336
    },
    {
      "epoch": 4.312096029547553,
      "grad_norm": 0.9195238351821899,
      "learning_rate": 7.002e-06,
      "loss": 0.0916,
      "step": 2337
    },
    {
      "epoch": 4.3139427516158815,
      "grad_norm": 1.087558388710022,
      "learning_rate": 7.005000000000001e-06,
      "loss": 0.1864,
      "step": 2338
    },
    {
      "epoch": 4.315789473684211,
      "grad_norm": 1.180001974105835,
      "learning_rate": 7.0080000000000005e-06,
      "loss": 0.0721,
      "step": 2339
    },
    {
      "epoch": 4.317636195752539,
      "grad_norm": 1.4501054286956787,
      "learning_rate": 7.011e-06,
      "loss": 0.1138,
      "step": 2340
    },
    {
      "epoch": 4.319482917820868,
      "grad_norm": 0.8645310401916504,
      "learning_rate": 7.014e-06,
      "loss": 0.0771,
      "step": 2341
    },
    {
      "epoch": 4.321329639889196,
      "grad_norm": 0.9253374338150024,
      "learning_rate": 7.017e-06,
      "loss": 0.0759,
      "step": 2342
    },
    {
      "epoch": 4.323176361957525,
      "grad_norm": 0.88505619764328,
      "learning_rate": 7.0200000000000006e-06,
      "loss": 0.0791,
      "step": 2343
    },
    {
      "epoch": 4.325023084025855,
      "grad_norm": 1.6287795305252075,
      "learning_rate": 7.023e-06,
      "loss": 0.1053,
      "step": 2344
    },
    {
      "epoch": 4.326869806094183,
      "grad_norm": 0.9386304616928101,
      "learning_rate": 7.026e-06,
      "loss": 0.0667,
      "step": 2345
    },
    {
      "epoch": 4.328716528162512,
      "grad_norm": 0.9785365462303162,
      "learning_rate": 7.029000000000001e-06,
      "loss": 0.0814,
      "step": 2346
    },
    {
      "epoch": 4.33056325023084,
      "grad_norm": 1.2905381917953491,
      "learning_rate": 7.032e-06,
      "loss": 0.0959,
      "step": 2347
    },
    {
      "epoch": 4.332409972299169,
      "grad_norm": 1.4337656497955322,
      "learning_rate": 7.0349999999999996e-06,
      "loss": 0.0682,
      "step": 2348
    },
    {
      "epoch": 4.334256694367498,
      "grad_norm": 0.8870456218719482,
      "learning_rate": 7.038e-06,
      "loss": 0.069,
      "step": 2349
    },
    {
      "epoch": 4.336103416435827,
      "grad_norm": 1.0684574842453003,
      "learning_rate": 7.041e-06,
      "loss": 0.0877,
      "step": 2350
    },
    {
      "epoch": 4.337950138504155,
      "grad_norm": 1.24235999584198,
      "learning_rate": 7.044000000000001e-06,
      "loss": 0.0994,
      "step": 2351
    },
    {
      "epoch": 4.339796860572484,
      "grad_norm": 1.1478756666183472,
      "learning_rate": 7.0470000000000005e-06,
      "loss": 0.0899,
      "step": 2352
    },
    {
      "epoch": 4.341643582640812,
      "grad_norm": 0.9353871941566467,
      "learning_rate": 7.049999999999999e-06,
      "loss": 0.0829,
      "step": 2353
    },
    {
      "epoch": 4.3434903047091415,
      "grad_norm": 1.3551195859909058,
      "learning_rate": 7.053e-06,
      "loss": 0.1088,
      "step": 2354
    },
    {
      "epoch": 4.34533702677747,
      "grad_norm": 1.18695068359375,
      "learning_rate": 7.056e-06,
      "loss": 0.1109,
      "step": 2355
    },
    {
      "epoch": 4.347183748845799,
      "grad_norm": 1.437421202659607,
      "learning_rate": 7.0590000000000005e-06,
      "loss": 0.0998,
      "step": 2356
    },
    {
      "epoch": 4.349030470914127,
      "grad_norm": 0.967350959777832,
      "learning_rate": 7.062e-06,
      "loss": 0.0864,
      "step": 2357
    },
    {
      "epoch": 4.350877192982456,
      "grad_norm": 1.421446442604065,
      "learning_rate": 7.065e-06,
      "loss": 0.1081,
      "step": 2358
    },
    {
      "epoch": 4.352723915050785,
      "grad_norm": 1.1982686519622803,
      "learning_rate": 7.068e-06,
      "loss": 0.0771,
      "step": 2359
    },
    {
      "epoch": 4.354570637119114,
      "grad_norm": 1.0983085632324219,
      "learning_rate": 7.071e-06,
      "loss": 0.0861,
      "step": 2360
    },
    {
      "epoch": 4.356417359187442,
      "grad_norm": 2.7813031673431396,
      "learning_rate": 7.074e-06,
      "loss": 0.1023,
      "step": 2361
    },
    {
      "epoch": 4.358264081255771,
      "grad_norm": 1.3879581689834595,
      "learning_rate": 7.077e-06,
      "loss": 0.082,
      "step": 2362
    },
    {
      "epoch": 4.360110803324099,
      "grad_norm": 1.6462188959121704,
      "learning_rate": 7.08e-06,
      "loss": 0.0811,
      "step": 2363
    },
    {
      "epoch": 4.3619575253924285,
      "grad_norm": 1.314080834388733,
      "learning_rate": 7.083000000000001e-06,
      "loss": 0.083,
      "step": 2364
    },
    {
      "epoch": 4.363804247460757,
      "grad_norm": 1.1373904943466187,
      "learning_rate": 7.086e-06,
      "loss": 0.0778,
      "step": 2365
    },
    {
      "epoch": 4.365650969529086,
      "grad_norm": 1.0565918684005737,
      "learning_rate": 7.089e-06,
      "loss": 0.0757,
      "step": 2366
    },
    {
      "epoch": 4.367497691597414,
      "grad_norm": 1.1691209077835083,
      "learning_rate": 7.092e-06,
      "loss": 0.0895,
      "step": 2367
    },
    {
      "epoch": 4.369344413665743,
      "grad_norm": 1.9542254209518433,
      "learning_rate": 7.095e-06,
      "loss": 0.1602,
      "step": 2368
    },
    {
      "epoch": 4.3711911357340725,
      "grad_norm": 4.830984115600586,
      "learning_rate": 7.0980000000000005e-06,
      "loss": 0.419,
      "step": 2369
    },
    {
      "epoch": 4.373037857802401,
      "grad_norm": 2.0716261863708496,
      "learning_rate": 7.101e-06,
      "loss": 0.4215,
      "step": 2370
    },
    {
      "epoch": 4.374884579870729,
      "grad_norm": 1.3052515983581543,
      "learning_rate": 7.104e-06,
      "loss": 0.3627,
      "step": 2371
    },
    {
      "epoch": 4.376731301939058,
      "grad_norm": 1.3134573698043823,
      "learning_rate": 7.107e-06,
      "loss": 0.3632,
      "step": 2372
    },
    {
      "epoch": 4.378578024007387,
      "grad_norm": 2.2939159870147705,
      "learning_rate": 7.11e-06,
      "loss": 0.3344,
      "step": 2373
    },
    {
      "epoch": 4.3804247460757155,
      "grad_norm": 1.4574236869812012,
      "learning_rate": 7.113e-06,
      "loss": 0.2472,
      "step": 2374
    },
    {
      "epoch": 4.382271468144045,
      "grad_norm": 1.3681788444519043,
      "learning_rate": 7.116e-06,
      "loss": 0.298,
      "step": 2375
    },
    {
      "epoch": 4.384118190212373,
      "grad_norm": 1.1303521394729614,
      "learning_rate": 7.119000000000001e-06,
      "loss": 0.2606,
      "step": 2376
    },
    {
      "epoch": 4.385964912280702,
      "grad_norm": 1.774362325668335,
      "learning_rate": 7.122000000000001e-06,
      "loss": 0.3622,
      "step": 2377
    },
    {
      "epoch": 4.38781163434903,
      "grad_norm": 1.030994176864624,
      "learning_rate": 7.1249999999999995e-06,
      "loss": 0.2323,
      "step": 2378
    },
    {
      "epoch": 4.3896583564173595,
      "grad_norm": 1.397836446762085,
      "learning_rate": 7.128e-06,
      "loss": 0.2153,
      "step": 2379
    },
    {
      "epoch": 4.391505078485688,
      "grad_norm": 0.9883468151092529,
      "learning_rate": 7.131e-06,
      "loss": 0.1679,
      "step": 2380
    },
    {
      "epoch": 4.393351800554017,
      "grad_norm": 1.6239488124847412,
      "learning_rate": 7.134000000000001e-06,
      "loss": 0.233,
      "step": 2381
    },
    {
      "epoch": 4.395198522622345,
      "grad_norm": 1.3051291704177856,
      "learning_rate": 7.1370000000000004e-06,
      "loss": 0.1932,
      "step": 2382
    },
    {
      "epoch": 4.397045244690674,
      "grad_norm": 1.332719326019287,
      "learning_rate": 7.14e-06,
      "loss": 0.1189,
      "step": 2383
    },
    {
      "epoch": 4.3988919667590025,
      "grad_norm": 1.0142273902893066,
      "learning_rate": 7.143e-06,
      "loss": 0.1665,
      "step": 2384
    },
    {
      "epoch": 4.400738688827332,
      "grad_norm": 0.8044421076774597,
      "learning_rate": 7.146e-06,
      "loss": 0.0814,
      "step": 2385
    },
    {
      "epoch": 4.40258541089566,
      "grad_norm": 1.1707408428192139,
      "learning_rate": 7.1490000000000005e-06,
      "loss": 0.1247,
      "step": 2386
    },
    {
      "epoch": 4.404432132963989,
      "grad_norm": 1.0917186737060547,
      "learning_rate": 7.152e-06,
      "loss": 0.0814,
      "step": 2387
    },
    {
      "epoch": 4.406278855032317,
      "grad_norm": 1.094245433807373,
      "learning_rate": 7.155e-06,
      "loss": 0.0785,
      "step": 2388
    },
    {
      "epoch": 4.4081255771006465,
      "grad_norm": 1.2464826107025146,
      "learning_rate": 7.158000000000001e-06,
      "loss": 0.0984,
      "step": 2389
    },
    {
      "epoch": 4.409972299168975,
      "grad_norm": 1.0168637037277222,
      "learning_rate": 7.161e-06,
      "loss": 0.0717,
      "step": 2390
    },
    {
      "epoch": 4.411819021237304,
      "grad_norm": 1.3358882665634155,
      "learning_rate": 7.164e-06,
      "loss": 0.0797,
      "step": 2391
    },
    {
      "epoch": 4.413665743305632,
      "grad_norm": 1.3539047241210938,
      "learning_rate": 7.167e-06,
      "loss": 0.1148,
      "step": 2392
    },
    {
      "epoch": 4.415512465373961,
      "grad_norm": 1.1813031435012817,
      "learning_rate": 7.17e-06,
      "loss": 0.0852,
      "step": 2393
    },
    {
      "epoch": 4.41735918744229,
      "grad_norm": 1.0633906126022339,
      "learning_rate": 7.173000000000001e-06,
      "loss": 0.1041,
      "step": 2394
    },
    {
      "epoch": 4.419205909510619,
      "grad_norm": 1.2666691541671753,
      "learning_rate": 7.176e-06,
      "loss": 0.0974,
      "step": 2395
    },
    {
      "epoch": 4.421052631578947,
      "grad_norm": 0.8168101906776428,
      "learning_rate": 7.179e-06,
      "loss": 0.0798,
      "step": 2396
    },
    {
      "epoch": 4.422899353647276,
      "grad_norm": 1.5207481384277344,
      "learning_rate": 7.182e-06,
      "loss": 0.0763,
      "step": 2397
    },
    {
      "epoch": 4.424746075715605,
      "grad_norm": 1.416662335395813,
      "learning_rate": 7.185e-06,
      "loss": 0.1024,
      "step": 2398
    },
    {
      "epoch": 4.426592797783933,
      "grad_norm": 1.1562248468399048,
      "learning_rate": 7.1880000000000005e-06,
      "loss": 0.0812,
      "step": 2399
    },
    {
      "epoch": 4.428439519852263,
      "grad_norm": 1.0115388631820679,
      "learning_rate": 7.191e-06,
      "loss": 0.0604,
      "step": 2400
    },
    {
      "epoch": 4.430286241920591,
      "grad_norm": 1.4185620546340942,
      "learning_rate": 7.194000000000001e-06,
      "loss": 0.0885,
      "step": 2401
    },
    {
      "epoch": 4.43213296398892,
      "grad_norm": 1.5242897272109985,
      "learning_rate": 7.197e-06,
      "loss": 0.0708,
      "step": 2402
    },
    {
      "epoch": 4.433979686057248,
      "grad_norm": 1.4109538793563843,
      "learning_rate": 7.2e-06,
      "loss": 0.108,
      "step": 2403
    },
    {
      "epoch": 4.435826408125577,
      "grad_norm": 0.8287237882614136,
      "learning_rate": 7.203e-06,
      "loss": 0.0723,
      "step": 2404
    },
    {
      "epoch": 4.437673130193906,
      "grad_norm": 0.9260662198066711,
      "learning_rate": 7.206e-06,
      "loss": 0.0702,
      "step": 2405
    },
    {
      "epoch": 4.439519852262235,
      "grad_norm": 1.1125154495239258,
      "learning_rate": 7.209000000000001e-06,
      "loss": 0.1022,
      "step": 2406
    },
    {
      "epoch": 4.441366574330563,
      "grad_norm": 1.1801170110702515,
      "learning_rate": 7.2120000000000006e-06,
      "loss": 0.0964,
      "step": 2407
    },
    {
      "epoch": 4.443213296398892,
      "grad_norm": 1.0909982919692993,
      "learning_rate": 7.2149999999999995e-06,
      "loss": 0.0881,
      "step": 2408
    },
    {
      "epoch": 4.44506001846722,
      "grad_norm": 1.1723145246505737,
      "learning_rate": 7.218e-06,
      "loss": 0.0888,
      "step": 2409
    },
    {
      "epoch": 4.44690674053555,
      "grad_norm": 1.0578376054763794,
      "learning_rate": 7.221e-06,
      "loss": 0.0733,
      "step": 2410
    },
    {
      "epoch": 4.448753462603878,
      "grad_norm": 1.6445304155349731,
      "learning_rate": 7.224e-06,
      "loss": 0.083,
      "step": 2411
    },
    {
      "epoch": 4.450600184672207,
      "grad_norm": 1.5911602973937988,
      "learning_rate": 7.2270000000000004e-06,
      "loss": 0.0677,
      "step": 2412
    },
    {
      "epoch": 4.452446906740535,
      "grad_norm": 1.4749010801315308,
      "learning_rate": 7.23e-06,
      "loss": 0.0882,
      "step": 2413
    },
    {
      "epoch": 4.454293628808864,
      "grad_norm": 1.1367706060409546,
      "learning_rate": 7.233e-06,
      "loss": 0.0699,
      "step": 2414
    },
    {
      "epoch": 4.456140350877193,
      "grad_norm": 1.34004545211792,
      "learning_rate": 7.236e-06,
      "loss": 0.0888,
      "step": 2415
    },
    {
      "epoch": 4.457987072945522,
      "grad_norm": 1.772800087928772,
      "learning_rate": 7.239e-06,
      "loss": 0.0837,
      "step": 2416
    },
    {
      "epoch": 4.45983379501385,
      "grad_norm": 2.071884870529175,
      "learning_rate": 7.242e-06,
      "loss": 0.1246,
      "step": 2417
    },
    {
      "epoch": 4.461680517082179,
      "grad_norm": 3.1921818256378174,
      "learning_rate": 7.245e-06,
      "loss": 0.1084,
      "step": 2418
    },
    {
      "epoch": 4.463527239150508,
      "grad_norm": 1.80342435836792,
      "learning_rate": 7.248000000000001e-06,
      "loss": 0.4917,
      "step": 2419
    },
    {
      "epoch": 4.465373961218837,
      "grad_norm": 1.9937363862991333,
      "learning_rate": 7.2510000000000005e-06,
      "loss": 0.3945,
      "step": 2420
    },
    {
      "epoch": 4.467220683287165,
      "grad_norm": 1.4467228651046753,
      "learning_rate": 7.2539999999999995e-06,
      "loss": 0.3882,
      "step": 2421
    },
    {
      "epoch": 4.469067405355494,
      "grad_norm": 1.2707688808441162,
      "learning_rate": 7.257e-06,
      "loss": 0.3186,
      "step": 2422
    },
    {
      "epoch": 4.470914127423823,
      "grad_norm": 1.1893327236175537,
      "learning_rate": 7.26e-06,
      "loss": 0.3115,
      "step": 2423
    },
    {
      "epoch": 4.472760849492151,
      "grad_norm": 1.9630070924758911,
      "learning_rate": 7.263000000000001e-06,
      "loss": 0.2778,
      "step": 2424
    },
    {
      "epoch": 4.4746075715604805,
      "grad_norm": 1.1939640045166016,
      "learning_rate": 7.266e-06,
      "loss": 0.2441,
      "step": 2425
    },
    {
      "epoch": 4.476454293628809,
      "grad_norm": 1.2537411451339722,
      "learning_rate": 7.269e-06,
      "loss": 0.2923,
      "step": 2426
    },
    {
      "epoch": 4.478301015697138,
      "grad_norm": 1.9572612047195435,
      "learning_rate": 7.272e-06,
      "loss": 0.2752,
      "step": 2427
    },
    {
      "epoch": 4.480147737765466,
      "grad_norm": 1.1313151121139526,
      "learning_rate": 7.275e-06,
      "loss": 0.2391,
      "step": 2428
    },
    {
      "epoch": 4.481994459833795,
      "grad_norm": 1.2185724973678589,
      "learning_rate": 7.2780000000000005e-06,
      "loss": 0.2191,
      "step": 2429
    },
    {
      "epoch": 4.4838411819021236,
      "grad_norm": 1.0316160917282104,
      "learning_rate": 7.281e-06,
      "loss": 0.1636,
      "step": 2430
    },
    {
      "epoch": 4.485687903970453,
      "grad_norm": 0.8509050607681274,
      "learning_rate": 7.284e-06,
      "loss": 0.1291,
      "step": 2431
    },
    {
      "epoch": 4.487534626038781,
      "grad_norm": 1.032467246055603,
      "learning_rate": 7.287000000000001e-06,
      "loss": 0.0756,
      "step": 2432
    },
    {
      "epoch": 4.48938134810711,
      "grad_norm": 1.5208518505096436,
      "learning_rate": 7.29e-06,
      "loss": 0.1552,
      "step": 2433
    },
    {
      "epoch": 4.491228070175438,
      "grad_norm": 1.3500417470932007,
      "learning_rate": 7.293e-06,
      "loss": 0.1275,
      "step": 2434
    },
    {
      "epoch": 4.4930747922437675,
      "grad_norm": 0.8810822367668152,
      "learning_rate": 7.296e-06,
      "loss": 0.0998,
      "step": 2435
    },
    {
      "epoch": 4.494921514312096,
      "grad_norm": 0.7776767611503601,
      "learning_rate": 7.299e-06,
      "loss": 0.1161,
      "step": 2436
    },
    {
      "epoch": 4.496768236380425,
      "grad_norm": 1.5567034482955933,
      "learning_rate": 7.3020000000000006e-06,
      "loss": 0.0988,
      "step": 2437
    },
    {
      "epoch": 4.498614958448753,
      "grad_norm": 1.1697642803192139,
      "learning_rate": 7.305e-06,
      "loss": 0.0954,
      "step": 2438
    },
    {
      "epoch": 4.500461680517082,
      "grad_norm": 0.9913340210914612,
      "learning_rate": 7.308e-06,
      "loss": 0.0634,
      "step": 2439
    },
    {
      "epoch": 4.5023084025854105,
      "grad_norm": 1.37468421459198,
      "learning_rate": 7.311e-06,
      "loss": 0.0615,
      "step": 2440
    },
    {
      "epoch": 4.50415512465374,
      "grad_norm": 1.6015678644180298,
      "learning_rate": 7.314e-06,
      "loss": 0.1341,
      "step": 2441
    },
    {
      "epoch": 4.506001846722068,
      "grad_norm": 1.031510591506958,
      "learning_rate": 7.317e-06,
      "loss": 0.0822,
      "step": 2442
    },
    {
      "epoch": 4.507848568790397,
      "grad_norm": 1.2176783084869385,
      "learning_rate": 7.32e-06,
      "loss": 0.0852,
      "step": 2443
    },
    {
      "epoch": 4.509695290858726,
      "grad_norm": 1.1300382614135742,
      "learning_rate": 7.323000000000001e-06,
      "loss": 0.159,
      "step": 2444
    },
    {
      "epoch": 4.5115420129270545,
      "grad_norm": 1.0426396131515503,
      "learning_rate": 7.326e-06,
      "loss": 0.0848,
      "step": 2445
    },
    {
      "epoch": 4.513388734995383,
      "grad_norm": 1.8430628776550293,
      "learning_rate": 7.329e-06,
      "loss": 0.0891,
      "step": 2446
    },
    {
      "epoch": 4.515235457063712,
      "grad_norm": 1.6290236711502075,
      "learning_rate": 7.332e-06,
      "loss": 0.065,
      "step": 2447
    },
    {
      "epoch": 4.517082179132041,
      "grad_norm": 0.8595914840698242,
      "learning_rate": 7.335e-06,
      "loss": 0.0682,
      "step": 2448
    },
    {
      "epoch": 4.518928901200369,
      "grad_norm": 0.9311236143112183,
      "learning_rate": 7.338000000000001e-06,
      "loss": 0.069,
      "step": 2449
    },
    {
      "epoch": 4.520775623268698,
      "grad_norm": 0.777802050113678,
      "learning_rate": 7.3410000000000005e-06,
      "loss": 0.0656,
      "step": 2450
    },
    {
      "epoch": 4.522622345337027,
      "grad_norm": 1.0638920068740845,
      "learning_rate": 7.3439999999999995e-06,
      "loss": 0.0717,
      "step": 2451
    },
    {
      "epoch": 4.524469067405356,
      "grad_norm": 1.1997246742248535,
      "learning_rate": 7.347e-06,
      "loss": 0.0818,
      "step": 2452
    },
    {
      "epoch": 4.526315789473684,
      "grad_norm": 1.017382025718689,
      "learning_rate": 7.35e-06,
      "loss": 0.0821,
      "step": 2453
    },
    {
      "epoch": 4.528162511542013,
      "grad_norm": 1.1677801609039307,
      "learning_rate": 7.353000000000001e-06,
      "loss": 0.0609,
      "step": 2454
    },
    {
      "epoch": 4.5300092336103415,
      "grad_norm": 1.519492506980896,
      "learning_rate": 7.356e-06,
      "loss": 0.0954,
      "step": 2455
    },
    {
      "epoch": 4.531855955678671,
      "grad_norm": 1.160491704940796,
      "learning_rate": 7.359e-06,
      "loss": 0.1009,
      "step": 2456
    },
    {
      "epoch": 4.533702677746999,
      "grad_norm": 0.9831103086471558,
      "learning_rate": 7.362e-06,
      "loss": 0.0919,
      "step": 2457
    },
    {
      "epoch": 4.535549399815328,
      "grad_norm": 1.2899953126907349,
      "learning_rate": 7.365e-06,
      "loss": 0.0827,
      "step": 2458
    },
    {
      "epoch": 4.537396121883656,
      "grad_norm": 1.563739538192749,
      "learning_rate": 7.3680000000000004e-06,
      "loss": 0.0721,
      "step": 2459
    },
    {
      "epoch": 4.539242843951985,
      "grad_norm": 1.4064379930496216,
      "learning_rate": 7.371e-06,
      "loss": 0.0823,
      "step": 2460
    },
    {
      "epoch": 4.541089566020314,
      "grad_norm": 1.5630841255187988,
      "learning_rate": 7.374e-06,
      "loss": 0.0935,
      "step": 2461
    },
    {
      "epoch": 4.542936288088643,
      "grad_norm": 2.0699667930603027,
      "learning_rate": 7.377000000000001e-06,
      "loss": 0.1116,
      "step": 2462
    },
    {
      "epoch": 4.544783010156971,
      "grad_norm": 1.2968478202819824,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 0.0637,
      "step": 2463
    },
    {
      "epoch": 4.5466297322253,
      "grad_norm": 1.3118655681610107,
      "learning_rate": 7.383e-06,
      "loss": 0.0797,
      "step": 2464
    },
    {
      "epoch": 4.5484764542936285,
      "grad_norm": 3.162208318710327,
      "learning_rate": 7.386e-06,
      "loss": 0.102,
      "step": 2465
    },
    {
      "epoch": 4.550323176361958,
      "grad_norm": 2.0861332416534424,
      "learning_rate": 7.389e-06,
      "loss": 0.11,
      "step": 2466
    },
    {
      "epoch": 4.552169898430286,
      "grad_norm": 1.3612501621246338,
      "learning_rate": 7.3920000000000005e-06,
      "loss": 0.1025,
      "step": 2467
    },
    {
      "epoch": 4.554016620498615,
      "grad_norm": 1.6249091625213623,
      "learning_rate": 7.395e-06,
      "loss": 0.1311,
      "step": 2468
    },
    {
      "epoch": 4.555863342566944,
      "grad_norm": 2.665177345275879,
      "learning_rate": 7.398000000000001e-06,
      "loss": 0.3888,
      "step": 2469
    },
    {
      "epoch": 4.557710064635272,
      "grad_norm": 2.436779260635376,
      "learning_rate": 7.401e-06,
      "loss": 0.3954,
      "step": 2470
    },
    {
      "epoch": 4.559556786703601,
      "grad_norm": 1.1689409017562866,
      "learning_rate": 7.404e-06,
      "loss": 0.34,
      "step": 2471
    },
    {
      "epoch": 4.56140350877193,
      "grad_norm": 1.4799636602401733,
      "learning_rate": 7.407e-06,
      "loss": 0.3183,
      "step": 2472
    },
    {
      "epoch": 4.563250230840259,
      "grad_norm": 2.0600669384002686,
      "learning_rate": 7.41e-06,
      "loss": 0.2514,
      "step": 2473
    },
    {
      "epoch": 4.565096952908587,
      "grad_norm": 1.1407296657562256,
      "learning_rate": 7.413e-06,
      "loss": 0.2809,
      "step": 2474
    },
    {
      "epoch": 4.566943674976916,
      "grad_norm": 1.2882170677185059,
      "learning_rate": 7.416000000000001e-06,
      "loss": 0.2511,
      "step": 2475
    },
    {
      "epoch": 4.568790397045245,
      "grad_norm": 1.4791680574417114,
      "learning_rate": 7.419e-06,
      "loss": 0.2621,
      "step": 2476
    },
    {
      "epoch": 4.570637119113574,
      "grad_norm": 1.2269575595855713,
      "learning_rate": 7.422e-06,
      "loss": 0.2375,
      "step": 2477
    },
    {
      "epoch": 4.572483841181902,
      "grad_norm": 1.0275866985321045,
      "learning_rate": 7.425e-06,
      "loss": 0.2019,
      "step": 2478
    },
    {
      "epoch": 4.574330563250231,
      "grad_norm": 2.3359060287475586,
      "learning_rate": 7.428e-06,
      "loss": 0.2382,
      "step": 2479
    },
    {
      "epoch": 4.576177285318559,
      "grad_norm": 0.9369570016860962,
      "learning_rate": 7.4310000000000005e-06,
      "loss": 0.1561,
      "step": 2480
    },
    {
      "epoch": 4.5780240073868885,
      "grad_norm": 1.2175058126449585,
      "learning_rate": 7.434e-06,
      "loss": 0.184,
      "step": 2481
    },
    {
      "epoch": 4.579870729455217,
      "grad_norm": 2.063988447189331,
      "learning_rate": 7.437e-06,
      "loss": 0.2426,
      "step": 2482
    },
    {
      "epoch": 4.581717451523546,
      "grad_norm": 1.105177640914917,
      "learning_rate": 7.44e-06,
      "loss": 0.1359,
      "step": 2483
    },
    {
      "epoch": 4.583564173591874,
      "grad_norm": 1.8581008911132812,
      "learning_rate": 7.443e-06,
      "loss": 0.1228,
      "step": 2484
    },
    {
      "epoch": 4.585410895660203,
      "grad_norm": 0.8880102634429932,
      "learning_rate": 7.446e-06,
      "loss": 0.1086,
      "step": 2485
    },
    {
      "epoch": 4.587257617728532,
      "grad_norm": 0.9845847487449646,
      "learning_rate": 7.449e-06,
      "loss": 0.0594,
      "step": 2486
    },
    {
      "epoch": 4.589104339796861,
      "grad_norm": 0.7028189897537231,
      "learning_rate": 7.452000000000001e-06,
      "loss": 0.0764,
      "step": 2487
    },
    {
      "epoch": 4.590951061865189,
      "grad_norm": 1.0322048664093018,
      "learning_rate": 7.455e-06,
      "loss": 0.0814,
      "step": 2488
    },
    {
      "epoch": 4.592797783933518,
      "grad_norm": 1.1624642610549927,
      "learning_rate": 7.4579999999999996e-06,
      "loss": 0.08,
      "step": 2489
    },
    {
      "epoch": 4.594644506001846,
      "grad_norm": 0.8388569951057434,
      "learning_rate": 7.461e-06,
      "loss": 0.0655,
      "step": 2490
    },
    {
      "epoch": 4.5964912280701755,
      "grad_norm": 0.7668055295944214,
      "learning_rate": 7.464e-06,
      "loss": 0.0857,
      "step": 2491
    },
    {
      "epoch": 4.598337950138504,
      "grad_norm": 1.1497136354446411,
      "learning_rate": 7.467000000000001e-06,
      "loss": 0.07,
      "step": 2492
    },
    {
      "epoch": 4.600184672206833,
      "grad_norm": 2.340991497039795,
      "learning_rate": 7.4700000000000005e-06,
      "loss": 0.0676,
      "step": 2493
    },
    {
      "epoch": 4.602031394275162,
      "grad_norm": 0.8587968349456787,
      "learning_rate": 7.4729999999999994e-06,
      "loss": 0.0703,
      "step": 2494
    },
    {
      "epoch": 4.60387811634349,
      "grad_norm": 2.3311848640441895,
      "learning_rate": 7.476e-06,
      "loss": 0.0863,
      "step": 2495
    },
    {
      "epoch": 4.605724838411819,
      "grad_norm": 0.7522993683815002,
      "learning_rate": 7.479e-06,
      "loss": 0.0512,
      "step": 2496
    },
    {
      "epoch": 4.607571560480148,
      "grad_norm": 0.8542543053627014,
      "learning_rate": 7.4820000000000005e-06,
      "loss": 0.0746,
      "step": 2497
    },
    {
      "epoch": 4.609418282548477,
      "grad_norm": 1.0095270872116089,
      "learning_rate": 7.485e-06,
      "loss": 0.0673,
      "step": 2498
    },
    {
      "epoch": 4.611265004616805,
      "grad_norm": 1.6802910566329956,
      "learning_rate": 7.488e-06,
      "loss": 0.1941,
      "step": 2499
    },
    {
      "epoch": 4.613111726685134,
      "grad_norm": 1.132878065109253,
      "learning_rate": 7.491e-06,
      "loss": 0.0946,
      "step": 2500
    },
    {
      "epoch": 4.6149584487534625,
      "grad_norm": 0.8775937557220459,
      "learning_rate": 7.494e-06,
      "loss": 0.0651,
      "step": 2501
    },
    {
      "epoch": 4.616805170821792,
      "grad_norm": 1.0279693603515625,
      "learning_rate": 7.497e-06,
      "loss": 0.0837,
      "step": 2502
    },
    {
      "epoch": 4.61865189289012,
      "grad_norm": 1.2879340648651123,
      "learning_rate": 7.5e-06,
      "loss": 0.0532,
      "step": 2503
    },
    {
      "epoch": 4.620498614958449,
      "grad_norm": 1.112121820449829,
      "learning_rate": 7.503e-06,
      "loss": 0.0604,
      "step": 2504
    },
    {
      "epoch": 4.622345337026777,
      "grad_norm": 1.0696443319320679,
      "learning_rate": 7.506e-06,
      "loss": 0.067,
      "step": 2505
    },
    {
      "epoch": 4.624192059095106,
      "grad_norm": 0.9257684350013733,
      "learning_rate": 7.5090000000000004e-06,
      "loss": 0.0728,
      "step": 2506
    },
    {
      "epoch": 4.626038781163435,
      "grad_norm": 1.5705357789993286,
      "learning_rate": 7.512e-06,
      "loss": 0.0926,
      "step": 2507
    },
    {
      "epoch": 4.627885503231764,
      "grad_norm": 1.1221293210983276,
      "learning_rate": 7.515e-06,
      "loss": 0.0859,
      "step": 2508
    },
    {
      "epoch": 4.629732225300092,
      "grad_norm": 1.043793797492981,
      "learning_rate": 7.518e-06,
      "loss": 0.0649,
      "step": 2509
    },
    {
      "epoch": 4.631578947368421,
      "grad_norm": 0.9244360327720642,
      "learning_rate": 7.521e-06,
      "loss": 0.0654,
      "step": 2510
    },
    {
      "epoch": 4.6334256694367495,
      "grad_norm": 1.8804733753204346,
      "learning_rate": 7.524000000000001e-06,
      "loss": 0.0901,
      "step": 2511
    },
    {
      "epoch": 4.635272391505079,
      "grad_norm": 2.1051440238952637,
      "learning_rate": 7.527000000000001e-06,
      "loss": 0.0578,
      "step": 2512
    },
    {
      "epoch": 4.637119113573407,
      "grad_norm": 1.860549807548523,
      "learning_rate": 7.53e-06,
      "loss": 0.1183,
      "step": 2513
    },
    {
      "epoch": 4.638965835641736,
      "grad_norm": 1.026725172996521,
      "learning_rate": 7.533e-06,
      "loss": 0.0936,
      "step": 2514
    },
    {
      "epoch": 4.640812557710064,
      "grad_norm": 1.1471177339553833,
      "learning_rate": 7.5359999999999995e-06,
      "loss": 0.0937,
      "step": 2515
    },
    {
      "epoch": 4.642659279778393,
      "grad_norm": 1.392022967338562,
      "learning_rate": 7.539000000000001e-06,
      "loss": 0.1107,
      "step": 2516
    },
    {
      "epoch": 4.644506001846722,
      "grad_norm": 1.7344036102294922,
      "learning_rate": 7.542000000000001e-06,
      "loss": 0.1169,
      "step": 2517
    },
    {
      "epoch": 4.646352723915051,
      "grad_norm": 1.535583734512329,
      "learning_rate": 7.545000000000001e-06,
      "loss": 0.1065,
      "step": 2518
    },
    {
      "epoch": 4.64819944598338,
      "grad_norm": 1.7330703735351562,
      "learning_rate": 7.5479999999999996e-06,
      "loss": 0.4917,
      "step": 2519
    },
    {
      "epoch": 4.650046168051708,
      "grad_norm": 1.3887149095535278,
      "learning_rate": 7.550999999999999e-06,
      "loss": 0.4331,
      "step": 2520
    },
    {
      "epoch": 4.6518928901200365,
      "grad_norm": 1.4955347776412964,
      "learning_rate": 7.554000000000001e-06,
      "loss": 0.3801,
      "step": 2521
    },
    {
      "epoch": 4.653739612188366,
      "grad_norm": 1.291242241859436,
      "learning_rate": 7.557000000000001e-06,
      "loss": 0.2607,
      "step": 2522
    },
    {
      "epoch": 4.655586334256695,
      "grad_norm": 1.6524022817611694,
      "learning_rate": 7.5600000000000005e-06,
      "loss": 0.3722,
      "step": 2523
    },
    {
      "epoch": 4.657433056325023,
      "grad_norm": 1.3549121618270874,
      "learning_rate": 7.563e-06,
      "loss": 0.2796,
      "step": 2524
    },
    {
      "epoch": 4.659279778393352,
      "grad_norm": 1.261399507522583,
      "learning_rate": 7.565999999999999e-06,
      "loss": 0.2414,
      "step": 2525
    },
    {
      "epoch": 4.66112650046168,
      "grad_norm": 1.4139667749404907,
      "learning_rate": 7.569000000000001e-06,
      "loss": 0.263,
      "step": 2526
    },
    {
      "epoch": 4.66297322253001,
      "grad_norm": 1.3193572759628296,
      "learning_rate": 7.5720000000000005e-06,
      "loss": 0.2835,
      "step": 2527
    },
    {
      "epoch": 4.664819944598338,
      "grad_norm": 1.1930153369903564,
      "learning_rate": 7.575e-06,
      "loss": 0.1887,
      "step": 2528
    },
    {
      "epoch": 4.666666666666667,
      "grad_norm": 1.1165175437927246,
      "learning_rate": 7.578e-06,
      "loss": 0.2039,
      "step": 2529
    },
    {
      "epoch": 4.668513388734995,
      "grad_norm": 1.2176852226257324,
      "learning_rate": 7.581e-06,
      "loss": 0.1993,
      "step": 2530
    },
    {
      "epoch": 4.670360110803324,
      "grad_norm": 1.6196171045303345,
      "learning_rate": 7.5840000000000006e-06,
      "loss": 0.3159,
      "step": 2531
    },
    {
      "epoch": 4.672206832871653,
      "grad_norm": 1.2227379083633423,
      "learning_rate": 7.587e-06,
      "loss": 0.1302,
      "step": 2532
    },
    {
      "epoch": 4.674053554939982,
      "grad_norm": 1.3840858936309814,
      "learning_rate": 7.59e-06,
      "loss": 0.1816,
      "step": 2533
    },
    {
      "epoch": 4.67590027700831,
      "grad_norm": 1.2817070484161377,
      "learning_rate": 7.593e-06,
      "loss": 0.1301,
      "step": 2534
    },
    {
      "epoch": 4.677746999076639,
      "grad_norm": 1.2893495559692383,
      "learning_rate": 7.596e-06,
      "loss": 0.1599,
      "step": 2535
    },
    {
      "epoch": 4.679593721144967,
      "grad_norm": 0.8995867967605591,
      "learning_rate": 7.599000000000001e-06,
      "loss": 0.0656,
      "step": 2536
    },
    {
      "epoch": 4.6814404432132966,
      "grad_norm": 1.0696057081222534,
      "learning_rate": 7.602e-06,
      "loss": 0.0714,
      "step": 2537
    },
    {
      "epoch": 4.683287165281625,
      "grad_norm": 0.9868329167366028,
      "learning_rate": 7.605e-06,
      "loss": 0.0726,
      "step": 2538
    },
    {
      "epoch": 4.685133887349954,
      "grad_norm": 1.6628823280334473,
      "learning_rate": 7.608e-06,
      "loss": 0.071,
      "step": 2539
    },
    {
      "epoch": 4.686980609418282,
      "grad_norm": 0.901323676109314,
      "learning_rate": 7.611e-06,
      "loss": 0.0726,
      "step": 2540
    },
    {
      "epoch": 4.688827331486611,
      "grad_norm": 0.9900251626968384,
      "learning_rate": 7.614000000000001e-06,
      "loss": 0.0818,
      "step": 2541
    },
    {
      "epoch": 4.69067405355494,
      "grad_norm": 0.9476147890090942,
      "learning_rate": 7.617000000000001e-06,
      "loss": 0.0863,
      "step": 2542
    },
    {
      "epoch": 4.692520775623269,
      "grad_norm": 1.3561960458755493,
      "learning_rate": 7.62e-06,
      "loss": 0.1161,
      "step": 2543
    },
    {
      "epoch": 4.694367497691598,
      "grad_norm": 0.802712619304657,
      "learning_rate": 7.623e-06,
      "loss": 0.0772,
      "step": 2544
    },
    {
      "epoch": 4.696214219759926,
      "grad_norm": 0.976687490940094,
      "learning_rate": 7.6259999999999995e-06,
      "loss": 0.0709,
      "step": 2545
    },
    {
      "epoch": 4.698060941828254,
      "grad_norm": 1.0002014636993408,
      "learning_rate": 7.629000000000001e-06,
      "loss": 0.0803,
      "step": 2546
    },
    {
      "epoch": 4.6999076638965835,
      "grad_norm": 1.0644595623016357,
      "learning_rate": 7.632e-06,
      "loss": 0.068,
      "step": 2547
    },
    {
      "epoch": 4.701754385964913,
      "grad_norm": 0.7027485370635986,
      "learning_rate": 7.635e-06,
      "loss": 0.0655,
      "step": 2548
    },
    {
      "epoch": 4.703601108033241,
      "grad_norm": 1.1962902545928955,
      "learning_rate": 7.638e-06,
      "loss": 0.0897,
      "step": 2549
    },
    {
      "epoch": 4.70544783010157,
      "grad_norm": 1.5846956968307495,
      "learning_rate": 7.641e-06,
      "loss": 0.1526,
      "step": 2550
    },
    {
      "epoch": 4.707294552169898,
      "grad_norm": 1.1864464282989502,
      "learning_rate": 7.644000000000002e-06,
      "loss": 0.086,
      "step": 2551
    },
    {
      "epoch": 4.7091412742382275,
      "grad_norm": 0.921712338924408,
      "learning_rate": 7.647000000000001e-06,
      "loss": 0.079,
      "step": 2552
    },
    {
      "epoch": 4.710987996306556,
      "grad_norm": 1.6384105682373047,
      "learning_rate": 7.65e-06,
      "loss": 0.0882,
      "step": 2553
    },
    {
      "epoch": 4.712834718374885,
      "grad_norm": 1.165614366531372,
      "learning_rate": 7.653e-06,
      "loss": 0.0982,
      "step": 2554
    },
    {
      "epoch": 4.714681440443213,
      "grad_norm": 0.9998936653137207,
      "learning_rate": 7.656e-06,
      "loss": 0.053,
      "step": 2555
    },
    {
      "epoch": 4.716528162511542,
      "grad_norm": 1.1780154705047607,
      "learning_rate": 7.659e-06,
      "loss": 0.0766,
      "step": 2556
    },
    {
      "epoch": 4.7183748845798705,
      "grad_norm": 0.8436416983604431,
      "learning_rate": 7.662e-06,
      "loss": 0.0739,
      "step": 2557
    },
    {
      "epoch": 4.7202216066482,
      "grad_norm": 0.9713981747627258,
      "learning_rate": 7.665e-06,
      "loss": 0.0897,
      "step": 2558
    },
    {
      "epoch": 4.722068328716528,
      "grad_norm": 0.9224056005477905,
      "learning_rate": 7.668e-06,
      "loss": 0.076,
      "step": 2559
    },
    {
      "epoch": 4.723915050784857,
      "grad_norm": 0.9956642389297485,
      "learning_rate": 7.671e-06,
      "loss": 0.0644,
      "step": 2560
    },
    {
      "epoch": 4.725761772853185,
      "grad_norm": 1.1300500631332397,
      "learning_rate": 7.674000000000001e-06,
      "loss": 0.0849,
      "step": 2561
    },
    {
      "epoch": 4.7276084949215145,
      "grad_norm": 1.3759992122650146,
      "learning_rate": 7.677000000000001e-06,
      "loss": 0.1103,
      "step": 2562
    },
    {
      "epoch": 4.729455216989843,
      "grad_norm": 1.100203514099121,
      "learning_rate": 7.680000000000001e-06,
      "loss": 0.0744,
      "step": 2563
    },
    {
      "epoch": 4.731301939058172,
      "grad_norm": 0.9975654482841492,
      "learning_rate": 7.683e-06,
      "loss": 0.0706,
      "step": 2564
    },
    {
      "epoch": 4.7331486611265,
      "grad_norm": 1.2560099363327026,
      "learning_rate": 7.685999999999999e-06,
      "loss": 0.0995,
      "step": 2565
    },
    {
      "epoch": 4.734995383194829,
      "grad_norm": 1.2191094160079956,
      "learning_rate": 7.688999999999999e-06,
      "loss": 0.0927,
      "step": 2566
    },
    {
      "epoch": 4.7368421052631575,
      "grad_norm": 1.7450473308563232,
      "learning_rate": 7.692e-06,
      "loss": 0.1354,
      "step": 2567
    },
    {
      "epoch": 4.738688827331487,
      "grad_norm": 1.9262582063674927,
      "learning_rate": 7.695e-06,
      "loss": 0.1264,
      "step": 2568
    },
    {
      "epoch": 4.740535549399816,
      "grad_norm": 2.491466999053955,
      "learning_rate": 7.698e-06,
      "loss": 0.4965,
      "step": 2569
    },
    {
      "epoch": 4.742382271468144,
      "grad_norm": 1.3091697692871094,
      "learning_rate": 7.701e-06,
      "loss": 0.3776,
      "step": 2570
    },
    {
      "epoch": 4.744228993536472,
      "grad_norm": 1.6606218814849854,
      "learning_rate": 7.704e-06,
      "loss": 0.3546,
      "step": 2571
    },
    {
      "epoch": 4.7460757156048015,
      "grad_norm": 1.1433162689208984,
      "learning_rate": 7.707000000000001e-06,
      "loss": 0.3365,
      "step": 2572
    },
    {
      "epoch": 4.747922437673131,
      "grad_norm": 1.1602765321731567,
      "learning_rate": 7.71e-06,
      "loss": 0.2869,
      "step": 2573
    },
    {
      "epoch": 4.749769159741459,
      "grad_norm": 1.4442332983016968,
      "learning_rate": 7.713e-06,
      "loss": 0.3196,
      "step": 2574
    },
    {
      "epoch": 4.751615881809788,
      "grad_norm": 1.1443811655044556,
      "learning_rate": 7.716e-06,
      "loss": 0.2504,
      "step": 2575
    },
    {
      "epoch": 4.753462603878116,
      "grad_norm": 1.018218994140625,
      "learning_rate": 7.719e-06,
      "loss": 0.2439,
      "step": 2576
    },
    {
      "epoch": 4.755309325946445,
      "grad_norm": 1.1695661544799805,
      "learning_rate": 7.722e-06,
      "loss": 0.2409,
      "step": 2577
    },
    {
      "epoch": 4.757156048014774,
      "grad_norm": 1.1299222707748413,
      "learning_rate": 7.725e-06,
      "loss": 0.1832,
      "step": 2578
    },
    {
      "epoch": 4.759002770083103,
      "grad_norm": 1.2177181243896484,
      "learning_rate": 7.728e-06,
      "loss": 0.2338,
      "step": 2579
    },
    {
      "epoch": 4.760849492151431,
      "grad_norm": 1.2508996725082397,
      "learning_rate": 7.731e-06,
      "loss": 0.1498,
      "step": 2580
    },
    {
      "epoch": 4.76269621421976,
      "grad_norm": 1.3955219984054565,
      "learning_rate": 7.733999999999999e-06,
      "loss": 0.1899,
      "step": 2581
    },
    {
      "epoch": 4.7645429362880884,
      "grad_norm": 1.0323314666748047,
      "learning_rate": 7.737e-06,
      "loss": 0.1173,
      "step": 2582
    },
    {
      "epoch": 4.766389658356418,
      "grad_norm": 0.8515747785568237,
      "learning_rate": 7.74e-06,
      "loss": 0.1056,
      "step": 2583
    },
    {
      "epoch": 4.768236380424746,
      "grad_norm": 0.7767705321311951,
      "learning_rate": 7.743e-06,
      "loss": 0.0839,
      "step": 2584
    },
    {
      "epoch": 4.770083102493075,
      "grad_norm": 0.8640003800392151,
      "learning_rate": 7.746e-06,
      "loss": 0.0661,
      "step": 2585
    },
    {
      "epoch": 4.771929824561403,
      "grad_norm": 0.7137645483016968,
      "learning_rate": 7.749e-06,
      "loss": 0.0633,
      "step": 2586
    },
    {
      "epoch": 4.773776546629732,
      "grad_norm": 0.9822434186935425,
      "learning_rate": 7.752000000000001e-06,
      "loss": 0.0957,
      "step": 2587
    },
    {
      "epoch": 4.775623268698061,
      "grad_norm": 0.9581803679466248,
      "learning_rate": 7.755000000000001e-06,
      "loss": 0.1328,
      "step": 2588
    },
    {
      "epoch": 4.77746999076639,
      "grad_norm": 1.1790051460266113,
      "learning_rate": 7.758000000000001e-06,
      "loss": 0.0914,
      "step": 2589
    },
    {
      "epoch": 4.779316712834718,
      "grad_norm": 0.988778829574585,
      "learning_rate": 7.760999999999999e-06,
      "loss": 0.0594,
      "step": 2590
    },
    {
      "epoch": 4.781163434903047,
      "grad_norm": 0.9305976629257202,
      "learning_rate": 7.763999999999999e-06,
      "loss": 0.0762,
      "step": 2591
    },
    {
      "epoch": 4.783010156971375,
      "grad_norm": 0.9188739061355591,
      "learning_rate": 7.767e-06,
      "loss": 0.0626,
      "step": 2592
    },
    {
      "epoch": 4.784856879039705,
      "grad_norm": 0.8577688336372375,
      "learning_rate": 7.77e-06,
      "loss": 0.0565,
      "step": 2593
    },
    {
      "epoch": 4.786703601108034,
      "grad_norm": 0.9031362533569336,
      "learning_rate": 7.773e-06,
      "loss": 0.0855,
      "step": 2594
    },
    {
      "epoch": 4.788550323176362,
      "grad_norm": 0.8325183987617493,
      "learning_rate": 7.776e-06,
      "loss": 0.0761,
      "step": 2595
    },
    {
      "epoch": 4.79039704524469,
      "grad_norm": 1.1880937814712524,
      "learning_rate": 7.779e-06,
      "loss": 0.0664,
      "step": 2596
    },
    {
      "epoch": 4.792243767313019,
      "grad_norm": 1.0195281505584717,
      "learning_rate": 7.782000000000001e-06,
      "loss": 0.0855,
      "step": 2597
    },
    {
      "epoch": 4.7940904893813485,
      "grad_norm": 0.9046306014060974,
      "learning_rate": 7.785000000000001e-06,
      "loss": 0.0784,
      "step": 2598
    },
    {
      "epoch": 4.795937211449677,
      "grad_norm": 0.8892627954483032,
      "learning_rate": 7.788e-06,
      "loss": 0.0648,
      "step": 2599
    },
    {
      "epoch": 4.797783933518006,
      "grad_norm": 1.1928547620773315,
      "learning_rate": 7.791e-06,
      "loss": 0.061,
      "step": 2600
    },
    {
      "epoch": 4.799630655586334,
      "grad_norm": 0.8939501047134399,
      "learning_rate": 7.794e-06,
      "loss": 0.0674,
      "step": 2601
    },
    {
      "epoch": 4.801477377654663,
      "grad_norm": 1.0067095756530762,
      "learning_rate": 7.797e-06,
      "loss": 0.0752,
      "step": 2602
    },
    {
      "epoch": 4.803324099722992,
      "grad_norm": 0.9221670627593994,
      "learning_rate": 7.8e-06,
      "loss": 0.0568,
      "step": 2603
    },
    {
      "epoch": 4.805170821791321,
      "grad_norm": 0.8903829455375671,
      "learning_rate": 7.803e-06,
      "loss": 0.0748,
      "step": 2604
    },
    {
      "epoch": 4.807017543859649,
      "grad_norm": 1.2078776359558105,
      "learning_rate": 7.806e-06,
      "loss": 0.0936,
      "step": 2605
    },
    {
      "epoch": 4.808864265927978,
      "grad_norm": 1.2025963068008423,
      "learning_rate": 7.809e-06,
      "loss": 0.0626,
      "step": 2606
    },
    {
      "epoch": 4.810710987996306,
      "grad_norm": 2.6398403644561768,
      "learning_rate": 7.812e-06,
      "loss": 0.0786,
      "step": 2607
    },
    {
      "epoch": 4.8125577100646355,
      "grad_norm": 1.0301458835601807,
      "learning_rate": 7.815e-06,
      "loss": 0.0955,
      "step": 2608
    },
    {
      "epoch": 4.814404432132964,
      "grad_norm": 1.054529070854187,
      "learning_rate": 7.818e-06,
      "loss": 0.0897,
      "step": 2609
    },
    {
      "epoch": 4.816251154201293,
      "grad_norm": 0.6876257658004761,
      "learning_rate": 7.821e-06,
      "loss": 0.0672,
      "step": 2610
    },
    {
      "epoch": 4.818097876269621,
      "grad_norm": 1.0190913677215576,
      "learning_rate": 7.824e-06,
      "loss": 0.0908,
      "step": 2611
    },
    {
      "epoch": 4.81994459833795,
      "grad_norm": 1.0644663572311401,
      "learning_rate": 7.827000000000001e-06,
      "loss": 0.0886,
      "step": 2612
    },
    {
      "epoch": 4.821791320406279,
      "grad_norm": 1.4151818752288818,
      "learning_rate": 7.830000000000001e-06,
      "loss": 0.0797,
      "step": 2613
    },
    {
      "epoch": 4.823638042474608,
      "grad_norm": 1.2000187635421753,
      "learning_rate": 7.833e-06,
      "loss": 0.079,
      "step": 2614
    },
    {
      "epoch": 4.825484764542936,
      "grad_norm": 0.8478805422782898,
      "learning_rate": 7.836e-06,
      "loss": 0.0714,
      "step": 2615
    },
    {
      "epoch": 4.827331486611265,
      "grad_norm": 0.9508643746376038,
      "learning_rate": 7.838999999999999e-06,
      "loss": 0.0558,
      "step": 2616
    },
    {
      "epoch": 4.829178208679593,
      "grad_norm": 1.2372348308563232,
      "learning_rate": 7.842e-06,
      "loss": 0.1049,
      "step": 2617
    },
    {
      "epoch": 4.8310249307479225,
      "grad_norm": 1.6176644563674927,
      "learning_rate": 7.845e-06,
      "loss": 0.1252,
      "step": 2618
    },
    {
      "epoch": 4.832871652816252,
      "grad_norm": 1.7493839263916016,
      "learning_rate": 7.848e-06,
      "loss": 0.5302,
      "step": 2619
    },
    {
      "epoch": 4.83471837488458,
      "grad_norm": 1.7353014945983887,
      "learning_rate": 7.851e-06,
      "loss": 0.4991,
      "step": 2620
    },
    {
      "epoch": 4.836565096952908,
      "grad_norm": 1.6778644323349,
      "learning_rate": 7.854e-06,
      "loss": 0.2939,
      "step": 2621
    },
    {
      "epoch": 4.838411819021237,
      "grad_norm": 1.1693828105926514,
      "learning_rate": 7.857000000000001e-06,
      "loss": 0.3351,
      "step": 2622
    },
    {
      "epoch": 4.840258541089566,
      "grad_norm": 1.0527209043502808,
      "learning_rate": 7.860000000000001e-06,
      "loss": 0.3107,
      "step": 2623
    },
    {
      "epoch": 4.842105263157895,
      "grad_norm": 1.8866925239562988,
      "learning_rate": 7.863e-06,
      "loss": 0.3261,
      "step": 2624
    },
    {
      "epoch": 4.843951985226224,
      "grad_norm": 1.1227540969848633,
      "learning_rate": 7.866e-06,
      "loss": 0.2156,
      "step": 2625
    },
    {
      "epoch": 4.845798707294552,
      "grad_norm": 1.33430016040802,
      "learning_rate": 7.868999999999999e-06,
      "loss": 0.2171,
      "step": 2626
    },
    {
      "epoch": 4.847645429362881,
      "grad_norm": 1.201172113418579,
      "learning_rate": 7.872e-06,
      "loss": 0.2086,
      "step": 2627
    },
    {
      "epoch": 4.8494921514312095,
      "grad_norm": 1.0619585514068604,
      "learning_rate": 7.875e-06,
      "loss": 0.1949,
      "step": 2628
    },
    {
      "epoch": 4.851338873499539,
      "grad_norm": 1.5394426584243774,
      "learning_rate": 7.878e-06,
      "loss": 0.1855,
      "step": 2629
    },
    {
      "epoch": 4.853185595567867,
      "grad_norm": 1.0706628561019897,
      "learning_rate": 7.881e-06,
      "loss": 0.1521,
      "step": 2630
    },
    {
      "epoch": 4.855032317636196,
      "grad_norm": 2.8826377391815186,
      "learning_rate": 7.884e-06,
      "loss": 0.1697,
      "step": 2631
    },
    {
      "epoch": 4.856879039704524,
      "grad_norm": 1.1852144002914429,
      "learning_rate": 7.887000000000001e-06,
      "loss": 0.1184,
      "step": 2632
    },
    {
      "epoch": 4.858725761772853,
      "grad_norm": 1.0162383317947388,
      "learning_rate": 7.89e-06,
      "loss": 0.1214,
      "step": 2633
    },
    {
      "epoch": 4.860572483841182,
      "grad_norm": 1.0937857627868652,
      "learning_rate": 7.893e-06,
      "loss": 0.095,
      "step": 2634
    },
    {
      "epoch": 4.862419205909511,
      "grad_norm": 0.9450917840003967,
      "learning_rate": 7.896e-06,
      "loss": 0.1052,
      "step": 2635
    },
    {
      "epoch": 4.864265927977839,
      "grad_norm": 2.6459155082702637,
      "learning_rate": 7.899e-06,
      "loss": 0.1139,
      "step": 2636
    },
    {
      "epoch": 4.866112650046168,
      "grad_norm": 1.0539472103118896,
      "learning_rate": 7.902000000000002e-06,
      "loss": 0.1326,
      "step": 2637
    },
    {
      "epoch": 4.8679593721144965,
      "grad_norm": 0.7081450819969177,
      "learning_rate": 7.905000000000001e-06,
      "loss": 0.0584,
      "step": 2638
    },
    {
      "epoch": 4.869806094182826,
      "grad_norm": 1.5462771654129028,
      "learning_rate": 7.908e-06,
      "loss": 0.0813,
      "step": 2639
    },
    {
      "epoch": 4.871652816251154,
      "grad_norm": 1.0687777996063232,
      "learning_rate": 7.911e-06,
      "loss": 0.0553,
      "step": 2640
    },
    {
      "epoch": 4.873499538319483,
      "grad_norm": 0.6855216026306152,
      "learning_rate": 7.913999999999999e-06,
      "loss": 0.0657,
      "step": 2641
    },
    {
      "epoch": 4.875346260387811,
      "grad_norm": 0.9286635518074036,
      "learning_rate": 7.917e-06,
      "loss": 0.0749,
      "step": 2642
    },
    {
      "epoch": 4.87719298245614,
      "grad_norm": 1.0680339336395264,
      "learning_rate": 7.92e-06,
      "loss": 0.0824,
      "step": 2643
    },
    {
      "epoch": 4.87903970452447,
      "grad_norm": 1.5928995609283447,
      "learning_rate": 7.923e-06,
      "loss": 0.0837,
      "step": 2644
    },
    {
      "epoch": 4.880886426592798,
      "grad_norm": 0.8490452170372009,
      "learning_rate": 7.926e-06,
      "loss": 0.0656,
      "step": 2645
    },
    {
      "epoch": 4.882733148661126,
      "grad_norm": 1.2324718236923218,
      "learning_rate": 7.929e-06,
      "loss": 0.0564,
      "step": 2646
    },
    {
      "epoch": 4.884579870729455,
      "grad_norm": 1.1055073738098145,
      "learning_rate": 7.932000000000001e-06,
      "loss": 0.0643,
      "step": 2647
    },
    {
      "epoch": 4.886426592797784,
      "grad_norm": 0.7276487350463867,
      "learning_rate": 7.935000000000001e-06,
      "loss": 0.0695,
      "step": 2648
    },
    {
      "epoch": 4.888273314866113,
      "grad_norm": 1.3281259536743164,
      "learning_rate": 7.938000000000001e-06,
      "loss": 0.0939,
      "step": 2649
    },
    {
      "epoch": 4.890120036934442,
      "grad_norm": 0.9045303463935852,
      "learning_rate": 7.941e-06,
      "loss": 0.0646,
      "step": 2650
    },
    {
      "epoch": 4.89196675900277,
      "grad_norm": 0.9664737582206726,
      "learning_rate": 7.943999999999999e-06,
      "loss": 0.0664,
      "step": 2651
    },
    {
      "epoch": 4.893813481071099,
      "grad_norm": 0.933375895023346,
      "learning_rate": 7.947e-06,
      "loss": 0.0645,
      "step": 2652
    },
    {
      "epoch": 4.895660203139427,
      "grad_norm": 1.1591026782989502,
      "learning_rate": 7.95e-06,
      "loss": 0.077,
      "step": 2653
    },
    {
      "epoch": 4.8975069252077565,
      "grad_norm": 1.2264528274536133,
      "learning_rate": 7.953e-06,
      "loss": 0.1174,
      "step": 2654
    },
    {
      "epoch": 4.899353647276085,
      "grad_norm": 1.124033808708191,
      "learning_rate": 7.956e-06,
      "loss": 0.0751,
      "step": 2655
    },
    {
      "epoch": 4.901200369344414,
      "grad_norm": 0.8462250232696533,
      "learning_rate": 7.959e-06,
      "loss": 0.0722,
      "step": 2656
    },
    {
      "epoch": 4.903047091412742,
      "grad_norm": 1.0723295211791992,
      "learning_rate": 7.962000000000001e-06,
      "loss": 0.0808,
      "step": 2657
    },
    {
      "epoch": 4.904893813481071,
      "grad_norm": 1.098881483078003,
      "learning_rate": 7.965e-06,
      "loss": 0.0609,
      "step": 2658
    },
    {
      "epoch": 4.9067405355494,
      "grad_norm": 0.9864880442619324,
      "learning_rate": 7.968e-06,
      "loss": 0.0778,
      "step": 2659
    },
    {
      "epoch": 4.908587257617729,
      "grad_norm": 0.7928457856178284,
      "learning_rate": 7.971e-06,
      "loss": 0.0657,
      "step": 2660
    },
    {
      "epoch": 4.910433979686057,
      "grad_norm": 1.0289839506149292,
      "learning_rate": 7.974e-06,
      "loss": 0.0713,
      "step": 2661
    },
    {
      "epoch": 4.912280701754386,
      "grad_norm": 1.0452907085418701,
      "learning_rate": 7.977000000000002e-06,
      "loss": 0.0615,
      "step": 2662
    },
    {
      "epoch": 4.914127423822714,
      "grad_norm": 0.8897021412849426,
      "learning_rate": 7.98e-06,
      "loss": 0.0674,
      "step": 2663
    },
    {
      "epoch": 4.9159741458910435,
      "grad_norm": 1.2523964643478394,
      "learning_rate": 7.983e-06,
      "loss": 0.0769,
      "step": 2664
    },
    {
      "epoch": 4.917820867959372,
      "grad_norm": 1.1305283308029175,
      "learning_rate": 7.986e-06,
      "loss": 0.0674,
      "step": 2665
    },
    {
      "epoch": 4.919667590027701,
      "grad_norm": 1.435927391052246,
      "learning_rate": 7.989e-06,
      "loss": 0.0695,
      "step": 2666
    },
    {
      "epoch": 4.921514312096029,
      "grad_norm": 1.2424691915512085,
      "learning_rate": 7.992e-06,
      "loss": 0.0838,
      "step": 2667
    },
    {
      "epoch": 4.923361034164358,
      "grad_norm": 1.2204331159591675,
      "learning_rate": 7.995e-06,
      "loss": 0.0693,
      "step": 2668
    },
    {
      "epoch": 4.9252077562326875,
      "grad_norm": 2.060532331466675,
      "learning_rate": 7.998e-06,
      "loss": 0.4751,
      "step": 2669
    },
    {
      "epoch": 4.927054478301016,
      "grad_norm": 1.386932611465454,
      "learning_rate": 8.001e-06,
      "loss": 0.3697,
      "step": 2670
    },
    {
      "epoch": 4.928901200369344,
      "grad_norm": 1.1898244619369507,
      "learning_rate": 8.004e-06,
      "loss": 0.3438,
      "step": 2671
    },
    {
      "epoch": 4.930747922437673,
      "grad_norm": 10.301701545715332,
      "learning_rate": 8.007000000000001e-06,
      "loss": 0.3882,
      "step": 2672
    },
    {
      "epoch": 4.932594644506002,
      "grad_norm": 1.6150550842285156,
      "learning_rate": 8.010000000000001e-06,
      "loss": 0.2899,
      "step": 2673
    },
    {
      "epoch": 4.9344413665743305,
      "grad_norm": 1.1645244359970093,
      "learning_rate": 8.013000000000001e-06,
      "loss": 0.2327,
      "step": 2674
    },
    {
      "epoch": 4.93628808864266,
      "grad_norm": 1.136450171470642,
      "learning_rate": 8.016e-06,
      "loss": 0.2859,
      "step": 2675
    },
    {
      "epoch": 4.938134810710988,
      "grad_norm": 1.1616744995117188,
      "learning_rate": 8.018999999999999e-06,
      "loss": 0.2183,
      "step": 2676
    },
    {
      "epoch": 4.939981532779317,
      "grad_norm": 1.301045536994934,
      "learning_rate": 8.022e-06,
      "loss": 0.2169,
      "step": 2677
    },
    {
      "epoch": 4.941828254847645,
      "grad_norm": 0.9416823983192444,
      "learning_rate": 8.025e-06,
      "loss": 0.1446,
      "step": 2678
    },
    {
      "epoch": 4.9436749769159745,
      "grad_norm": 1.5150599479675293,
      "learning_rate": 8.028e-06,
      "loss": 0.1712,
      "step": 2679
    },
    {
      "epoch": 4.945521698984303,
      "grad_norm": 0.7367508411407471,
      "learning_rate": 8.031e-06,
      "loss": 0.096,
      "step": 2680
    },
    {
      "epoch": 4.947368421052632,
      "grad_norm": 0.8093917965888977,
      "learning_rate": 8.034e-06,
      "loss": 0.0702,
      "step": 2681
    },
    {
      "epoch": 4.94921514312096,
      "grad_norm": 0.9393030405044556,
      "learning_rate": 8.037000000000001e-06,
      "loss": 0.0895,
      "step": 2682
    },
    {
      "epoch": 4.951061865189289,
      "grad_norm": 0.849078893661499,
      "learning_rate": 8.040000000000001e-06,
      "loss": 0.0874,
      "step": 2683
    },
    {
      "epoch": 4.9529085872576175,
      "grad_norm": 1.019606590270996,
      "learning_rate": 8.043e-06,
      "loss": 0.0846,
      "step": 2684
    },
    {
      "epoch": 4.954755309325947,
      "grad_norm": 0.8608704805374146,
      "learning_rate": 8.046e-06,
      "loss": 0.0578,
      "step": 2685
    },
    {
      "epoch": 4.956602031394275,
      "grad_norm": 1.1765228509902954,
      "learning_rate": 8.049e-06,
      "loss": 0.1218,
      "step": 2686
    },
    {
      "epoch": 4.958448753462604,
      "grad_norm": 0.8517112135887146,
      "learning_rate": 8.052000000000002e-06,
      "loss": 0.0685,
      "step": 2687
    },
    {
      "epoch": 4.960295475530932,
      "grad_norm": 0.9209573864936829,
      "learning_rate": 8.055e-06,
      "loss": 0.0558,
      "step": 2688
    },
    {
      "epoch": 4.9621421975992615,
      "grad_norm": 0.9230965971946716,
      "learning_rate": 8.058e-06,
      "loss": 0.0647,
      "step": 2689
    },
    {
      "epoch": 4.96398891966759,
      "grad_norm": 1.2790827751159668,
      "learning_rate": 8.061e-06,
      "loss": 0.1008,
      "step": 2690
    },
    {
      "epoch": 4.965835641735919,
      "grad_norm": 1.0466139316558838,
      "learning_rate": 8.064e-06,
      "loss": 0.0827,
      "step": 2691
    },
    {
      "epoch": 4.967682363804247,
      "grad_norm": 1.233939528465271,
      "learning_rate": 8.067e-06,
      "loss": 0.0685,
      "step": 2692
    },
    {
      "epoch": 4.969529085872576,
      "grad_norm": 1.100265622138977,
      "learning_rate": 8.07e-06,
      "loss": 0.055,
      "step": 2693
    },
    {
      "epoch": 4.971375807940905,
      "grad_norm": 1.711647868156433,
      "learning_rate": 8.073e-06,
      "loss": 0.063,
      "step": 2694
    },
    {
      "epoch": 4.973222530009234,
      "grad_norm": 0.9649916887283325,
      "learning_rate": 8.076e-06,
      "loss": 0.066,
      "step": 2695
    },
    {
      "epoch": 4.975069252077562,
      "grad_norm": 1.0252290964126587,
      "learning_rate": 8.079e-06,
      "loss": 0.064,
      "step": 2696
    },
    {
      "epoch": 4.976915974145891,
      "grad_norm": 0.9084988236427307,
      "learning_rate": 8.082e-06,
      "loss": 0.0637,
      "step": 2697
    },
    {
      "epoch": 4.97876269621422,
      "grad_norm": 0.6036769151687622,
      "learning_rate": 8.085000000000001e-06,
      "loss": 0.0437,
      "step": 2698
    },
    {
      "epoch": 4.980609418282548,
      "grad_norm": 1.1912896633148193,
      "learning_rate": 8.088000000000001e-06,
      "loss": 0.1391,
      "step": 2699
    },
    {
      "epoch": 4.982456140350877,
      "grad_norm": 1.3569239377975464,
      "learning_rate": 8.091e-06,
      "loss": 0.0746,
      "step": 2700
    },
    {
      "epoch": 4.984302862419206,
      "grad_norm": 1.2725087404251099,
      "learning_rate": 8.093999999999999e-06,
      "loss": 0.0636,
      "step": 2701
    },
    {
      "epoch": 4.986149584487535,
      "grad_norm": 2.4054908752441406,
      "learning_rate": 8.096999999999999e-06,
      "loss": 0.0793,
      "step": 2702
    },
    {
      "epoch": 4.987996306555863,
      "grad_norm": 1.1511203050613403,
      "learning_rate": 8.1e-06,
      "loss": 0.0836,
      "step": 2703
    },
    {
      "epoch": 4.989843028624192,
      "grad_norm": 1.0939910411834717,
      "learning_rate": 8.103e-06,
      "loss": 0.0942,
      "step": 2704
    },
    {
      "epoch": 4.991689750692521,
      "grad_norm": 1.0632432699203491,
      "learning_rate": 8.106e-06,
      "loss": 0.085,
      "step": 2705
    },
    {
      "epoch": 4.99353647276085,
      "grad_norm": 1.101324439048767,
      "learning_rate": 8.109e-06,
      "loss": 0.0752,
      "step": 2706
    },
    {
      "epoch": 4.995383194829178,
      "grad_norm": 1.3632900714874268,
      "learning_rate": 8.112e-06,
      "loss": 0.0974,
      "step": 2707
    },
    {
      "epoch": 4.997229916897507,
      "grad_norm": 1.1220369338989258,
      "learning_rate": 8.115000000000001e-06,
      "loss": 0.0876,
      "step": 2708
    },
    {
      "epoch": 4.999076638965835,
      "grad_norm": 1.160844087600708,
      "learning_rate": 8.118000000000001e-06,
      "loss": 0.0957,
      "step": 2709
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.7586629986763,
      "learning_rate": 8.121e-06,
      "loss": 0.0444,
      "step": 2710
    },
    {
      "epoch": 5.001846722068329,
      "grad_norm": 3.188286781311035,
      "learning_rate": 8.124e-06,
      "loss": 0.5124,
      "step": 2711
    },
    {
      "epoch": 5.003693444136657,
      "grad_norm": 1.9085744619369507,
      "learning_rate": 8.126999999999999e-06,
      "loss": 0.3812,
      "step": 2712
    },
    {
      "epoch": 5.0055401662049865,
      "grad_norm": 1.2450673580169678,
      "learning_rate": 8.13e-06,
      "loss": 0.2909,
      "step": 2713
    },
    {
      "epoch": 5.007386888273315,
      "grad_norm": 2.713906764984131,
      "learning_rate": 8.133e-06,
      "loss": 0.3009,
      "step": 2714
    },
    {
      "epoch": 5.009233610341644,
      "grad_norm": 1.3698234558105469,
      "learning_rate": 8.136e-06,
      "loss": 0.2429,
      "step": 2715
    },
    {
      "epoch": 5.011080332409972,
      "grad_norm": 1.40462064743042,
      "learning_rate": 8.139e-06,
      "loss": 0.1949,
      "step": 2716
    },
    {
      "epoch": 5.012927054478301,
      "grad_norm": 1.9252722263336182,
      "learning_rate": 8.142e-06,
      "loss": 0.2235,
      "step": 2717
    },
    {
      "epoch": 5.01477377654663,
      "grad_norm": 1.615817666053772,
      "learning_rate": 8.145e-06,
      "loss": 0.3593,
      "step": 2718
    },
    {
      "epoch": 5.016620498614959,
      "grad_norm": 0.8248971104621887,
      "learning_rate": 8.148e-06,
      "loss": 0.164,
      "step": 2719
    },
    {
      "epoch": 5.018467220683287,
      "grad_norm": 1.431121826171875,
      "learning_rate": 8.151e-06,
      "loss": 0.1736,
      "step": 2720
    },
    {
      "epoch": 5.020313942751616,
      "grad_norm": 2.1033153533935547,
      "learning_rate": 8.154e-06,
      "loss": 0.1873,
      "step": 2721
    },
    {
      "epoch": 5.022160664819944,
      "grad_norm": 0.9570878148078918,
      "learning_rate": 8.157e-06,
      "loss": 0.1279,
      "step": 2722
    },
    {
      "epoch": 5.0240073868882735,
      "grad_norm": 0.952393651008606,
      "learning_rate": 8.160000000000001e-06,
      "loss": 0.146,
      "step": 2723
    },
    {
      "epoch": 5.025854108956602,
      "grad_norm": 1.0486096143722534,
      "learning_rate": 8.163000000000001e-06,
      "loss": 0.1244,
      "step": 2724
    },
    {
      "epoch": 5.027700831024931,
      "grad_norm": 0.9891552329063416,
      "learning_rate": 8.166e-06,
      "loss": 0.1664,
      "step": 2725
    },
    {
      "epoch": 5.029547553093259,
      "grad_norm": 1.0865612030029297,
      "learning_rate": 8.169e-06,
      "loss": 0.1581,
      "step": 2726
    },
    {
      "epoch": 5.031394275161588,
      "grad_norm": 0.7026161551475525,
      "learning_rate": 8.171999999999999e-06,
      "loss": 0.0739,
      "step": 2727
    },
    {
      "epoch": 5.033240997229917,
      "grad_norm": 0.732552170753479,
      "learning_rate": 8.175e-06,
      "loss": 0.0691,
      "step": 2728
    },
    {
      "epoch": 5.035087719298246,
      "grad_norm": 0.8606172204017639,
      "learning_rate": 8.178e-06,
      "loss": 0.0759,
      "step": 2729
    },
    {
      "epoch": 5.036934441366574,
      "grad_norm": 0.9478769898414612,
      "learning_rate": 8.181e-06,
      "loss": 0.0805,
      "step": 2730
    },
    {
      "epoch": 5.038781163434903,
      "grad_norm": 1.0164841413497925,
      "learning_rate": 8.184e-06,
      "loss": 0.0735,
      "step": 2731
    },
    {
      "epoch": 5.040627885503231,
      "grad_norm": 0.8229144811630249,
      "learning_rate": 8.187e-06,
      "loss": 0.0715,
      "step": 2732
    },
    {
      "epoch": 5.0424746075715605,
      "grad_norm": 0.8915043473243713,
      "learning_rate": 8.190000000000001e-06,
      "loss": 0.064,
      "step": 2733
    },
    {
      "epoch": 5.044321329639889,
      "grad_norm": 0.8569628000259399,
      "learning_rate": 8.193000000000001e-06,
      "loss": 0.0529,
      "step": 2734
    },
    {
      "epoch": 5.046168051708218,
      "grad_norm": 0.8679004311561584,
      "learning_rate": 8.196e-06,
      "loss": 0.0498,
      "step": 2735
    },
    {
      "epoch": 5.048014773776547,
      "grad_norm": 0.686722457408905,
      "learning_rate": 8.199e-06,
      "loss": 0.0619,
      "step": 2736
    },
    {
      "epoch": 5.049861495844875,
      "grad_norm": 0.7008607387542725,
      "learning_rate": 8.201999999999999e-06,
      "loss": 0.0639,
      "step": 2737
    },
    {
      "epoch": 5.0517082179132045,
      "grad_norm": 0.7146738767623901,
      "learning_rate": 8.205e-06,
      "loss": 0.0553,
      "step": 2738
    },
    {
      "epoch": 5.053554939981533,
      "grad_norm": 1.0253753662109375,
      "learning_rate": 8.208e-06,
      "loss": 0.0733,
      "step": 2739
    },
    {
      "epoch": 5.055401662049862,
      "grad_norm": 0.7840259671211243,
      "learning_rate": 8.211e-06,
      "loss": 0.0603,
      "step": 2740
    },
    {
      "epoch": 5.05724838411819,
      "grad_norm": 0.9601883888244629,
      "learning_rate": 8.214e-06,
      "loss": 0.0807,
      "step": 2741
    },
    {
      "epoch": 5.059095106186519,
      "grad_norm": 1.1228169202804565,
      "learning_rate": 8.217e-06,
      "loss": 0.0719,
      "step": 2742
    },
    {
      "epoch": 5.0609418282548475,
      "grad_norm": 0.6802356243133545,
      "learning_rate": 8.220000000000001e-06,
      "loss": 0.0376,
      "step": 2743
    },
    {
      "epoch": 5.062788550323177,
      "grad_norm": 0.7134103775024414,
      "learning_rate": 8.223e-06,
      "loss": 0.0496,
      "step": 2744
    },
    {
      "epoch": 5.064635272391505,
      "grad_norm": 1.4297904968261719,
      "learning_rate": 8.226e-06,
      "loss": 0.0489,
      "step": 2745
    },
    {
      "epoch": 5.066481994459834,
      "grad_norm": 1.171958088874817,
      "learning_rate": 8.229e-06,
      "loss": 0.0549,
      "step": 2746
    },
    {
      "epoch": 5.068328716528162,
      "grad_norm": 0.9930132627487183,
      "learning_rate": 8.232e-06,
      "loss": 0.0595,
      "step": 2747
    },
    {
      "epoch": 5.0701754385964914,
      "grad_norm": 0.6678305864334106,
      "learning_rate": 8.235000000000002e-06,
      "loss": 0.0474,
      "step": 2748
    },
    {
      "epoch": 5.07202216066482,
      "grad_norm": 1.126110315322876,
      "learning_rate": 8.238e-06,
      "loss": 0.0738,
      "step": 2749
    },
    {
      "epoch": 5.073868882733149,
      "grad_norm": 1.4353430271148682,
      "learning_rate": 8.241e-06,
      "loss": 0.0755,
      "step": 2750
    },
    {
      "epoch": 5.075715604801477,
      "grad_norm": 1.0827726125717163,
      "learning_rate": 8.244e-06,
      "loss": 0.0937,
      "step": 2751
    },
    {
      "epoch": 5.077562326869806,
      "grad_norm": 1.1548527479171753,
      "learning_rate": 8.246999999999999e-06,
      "loss": 0.0702,
      "step": 2752
    },
    {
      "epoch": 5.0794090489381345,
      "grad_norm": 1.123356819152832,
      "learning_rate": 8.25e-06,
      "loss": 0.0709,
      "step": 2753
    },
    {
      "epoch": 5.081255771006464,
      "grad_norm": 1.2064100503921509,
      "learning_rate": 8.253e-06,
      "loss": 0.0902,
      "step": 2754
    },
    {
      "epoch": 5.083102493074792,
      "grad_norm": 1.0698316097259521,
      "learning_rate": 8.256e-06,
      "loss": 0.068,
      "step": 2755
    },
    {
      "epoch": 5.084949215143121,
      "grad_norm": 0.9638137817382812,
      "learning_rate": 8.259e-06,
      "loss": 0.0562,
      "step": 2756
    },
    {
      "epoch": 5.086795937211449,
      "grad_norm": 0.931310772895813,
      "learning_rate": 8.262e-06,
      "loss": 0.0577,
      "step": 2757
    },
    {
      "epoch": 5.088642659279778,
      "grad_norm": 2.2462685108184814,
      "learning_rate": 8.265000000000001e-06,
      "loss": 0.0998,
      "step": 2758
    },
    {
      "epoch": 5.090489381348107,
      "grad_norm": 1.3807369470596313,
      "learning_rate": 8.268000000000001e-06,
      "loss": 0.085,
      "step": 2759
    },
    {
      "epoch": 5.092336103416436,
      "grad_norm": 1.427977204322815,
      "learning_rate": 8.271000000000001e-06,
      "loss": 0.079,
      "step": 2760
    },
    {
      "epoch": 5.094182825484765,
      "grad_norm": 1.4242230653762817,
      "learning_rate": 8.274e-06,
      "loss": 0.3582,
      "step": 2761
    },
    {
      "epoch": 5.096029547553093,
      "grad_norm": 1.1141996383666992,
      "learning_rate": 8.276999999999999e-06,
      "loss": 0.3369,
      "step": 2762
    },
    {
      "epoch": 5.097876269621422,
      "grad_norm": 0.9965702295303345,
      "learning_rate": 8.28e-06,
      "loss": 0.3146,
      "step": 2763
    },
    {
      "epoch": 5.099722991689751,
      "grad_norm": 1.071621298789978,
      "learning_rate": 8.283e-06,
      "loss": 0.3233,
      "step": 2764
    },
    {
      "epoch": 5.10156971375808,
      "grad_norm": 0.9454147219657898,
      "learning_rate": 8.286e-06,
      "loss": 0.2085,
      "step": 2765
    },
    {
      "epoch": 5.103416435826408,
      "grad_norm": 2.2237796783447266,
      "learning_rate": 8.289e-06,
      "loss": 0.2989,
      "step": 2766
    },
    {
      "epoch": 5.105263157894737,
      "grad_norm": 1.0580756664276123,
      "learning_rate": 8.292e-06,
      "loss": 0.2079,
      "step": 2767
    },
    {
      "epoch": 5.107109879963065,
      "grad_norm": 0.9704204201698303,
      "learning_rate": 8.295000000000001e-06,
      "loss": 0.2233,
      "step": 2768
    },
    {
      "epoch": 5.108956602031395,
      "grad_norm": 0.8892659544944763,
      "learning_rate": 8.298000000000001e-06,
      "loss": 0.1835,
      "step": 2769
    },
    {
      "epoch": 5.110803324099723,
      "grad_norm": 1.0301181077957153,
      "learning_rate": 8.301e-06,
      "loss": 0.2356,
      "step": 2770
    },
    {
      "epoch": 5.112650046168052,
      "grad_norm": 1.033872127532959,
      "learning_rate": 8.304e-06,
      "loss": 0.1523,
      "step": 2771
    },
    {
      "epoch": 5.11449676823638,
      "grad_norm": 0.9132250547409058,
      "learning_rate": 8.307e-06,
      "loss": 0.137,
      "step": 2772
    },
    {
      "epoch": 5.116343490304709,
      "grad_norm": 0.7538077235221863,
      "learning_rate": 8.310000000000002e-06,
      "loss": 0.1323,
      "step": 2773
    },
    {
      "epoch": 5.118190212373038,
      "grad_norm": 0.8518770933151245,
      "learning_rate": 8.313e-06,
      "loss": 0.1159,
      "step": 2774
    },
    {
      "epoch": 5.120036934441367,
      "grad_norm": 0.7349260449409485,
      "learning_rate": 8.316e-06,
      "loss": 0.0928,
      "step": 2775
    },
    {
      "epoch": 5.121883656509695,
      "grad_norm": 0.7747098803520203,
      "learning_rate": 8.319e-06,
      "loss": 0.1134,
      "step": 2776
    },
    {
      "epoch": 5.123730378578024,
      "grad_norm": 0.9408161640167236,
      "learning_rate": 8.322e-06,
      "loss": 0.1117,
      "step": 2777
    },
    {
      "epoch": 5.125577100646352,
      "grad_norm": 0.9748533368110657,
      "learning_rate": 8.325e-06,
      "loss": 0.1154,
      "step": 2778
    },
    {
      "epoch": 5.127423822714682,
      "grad_norm": 0.6912077069282532,
      "learning_rate": 8.328e-06,
      "loss": 0.0687,
      "step": 2779
    },
    {
      "epoch": 5.12927054478301,
      "grad_norm": 0.6826411485671997,
      "learning_rate": 8.331e-06,
      "loss": 0.066,
      "step": 2780
    },
    {
      "epoch": 5.131117266851339,
      "grad_norm": 0.971698522567749,
      "learning_rate": 8.334e-06,
      "loss": 0.1017,
      "step": 2781
    },
    {
      "epoch": 5.132963988919667,
      "grad_norm": 0.8723594546318054,
      "learning_rate": 8.337e-06,
      "loss": 0.0594,
      "step": 2782
    },
    {
      "epoch": 5.134810710987996,
      "grad_norm": 0.8311214447021484,
      "learning_rate": 8.340000000000001e-06,
      "loss": 0.0874,
      "step": 2783
    },
    {
      "epoch": 5.136657433056325,
      "grad_norm": 0.7077476978302002,
      "learning_rate": 8.343000000000001e-06,
      "loss": 0.0512,
      "step": 2784
    },
    {
      "epoch": 5.138504155124654,
      "grad_norm": 0.9080737233161926,
      "learning_rate": 8.346000000000001e-06,
      "loss": 0.0783,
      "step": 2785
    },
    {
      "epoch": 5.140350877192983,
      "grad_norm": 0.6995786428451538,
      "learning_rate": 8.349e-06,
      "loss": 0.0563,
      "step": 2786
    },
    {
      "epoch": 5.142197599261311,
      "grad_norm": 0.9258413314819336,
      "learning_rate": 8.351999999999999e-06,
      "loss": 0.0596,
      "step": 2787
    },
    {
      "epoch": 5.14404432132964,
      "grad_norm": 1.9468938112258911,
      "learning_rate": 8.355e-06,
      "loss": 0.0477,
      "step": 2788
    },
    {
      "epoch": 5.1458910433979685,
      "grad_norm": 0.7890531420707703,
      "learning_rate": 8.358e-06,
      "loss": 0.0528,
      "step": 2789
    },
    {
      "epoch": 5.147737765466298,
      "grad_norm": 0.7697831988334656,
      "learning_rate": 8.361e-06,
      "loss": 0.0517,
      "step": 2790
    },
    {
      "epoch": 5.149584487534626,
      "grad_norm": 0.9314472675323486,
      "learning_rate": 8.364e-06,
      "loss": 0.0489,
      "step": 2791
    },
    {
      "epoch": 5.151431209602955,
      "grad_norm": 0.9839686751365662,
      "learning_rate": 8.367e-06,
      "loss": 0.0615,
      "step": 2792
    },
    {
      "epoch": 5.153277931671283,
      "grad_norm": 0.616888701915741,
      "learning_rate": 8.370000000000001e-06,
      "loss": 0.0448,
      "step": 2793
    },
    {
      "epoch": 5.1551246537396125,
      "grad_norm": 1.0835751295089722,
      "learning_rate": 8.373000000000001e-06,
      "loss": 0.048,
      "step": 2794
    },
    {
      "epoch": 5.156971375807941,
      "grad_norm": 1.1965879201889038,
      "learning_rate": 8.376e-06,
      "loss": 0.0938,
      "step": 2795
    },
    {
      "epoch": 5.15881809787627,
      "grad_norm": 0.7703157663345337,
      "learning_rate": 8.379e-06,
      "loss": 0.0645,
      "step": 2796
    },
    {
      "epoch": 5.160664819944598,
      "grad_norm": 0.8498938679695129,
      "learning_rate": 8.382e-06,
      "loss": 0.0523,
      "step": 2797
    },
    {
      "epoch": 5.162511542012927,
      "grad_norm": 0.8971265554428101,
      "learning_rate": 8.385e-06,
      "loss": 0.0513,
      "step": 2798
    },
    {
      "epoch": 5.1643582640812555,
      "grad_norm": 0.8701927661895752,
      "learning_rate": 8.388e-06,
      "loss": 0.0518,
      "step": 2799
    },
    {
      "epoch": 5.166204986149585,
      "grad_norm": 0.7444379329681396,
      "learning_rate": 8.391e-06,
      "loss": 0.0429,
      "step": 2800
    },
    {
      "epoch": 5.168051708217913,
      "grad_norm": 1.0041154623031616,
      "learning_rate": 8.394e-06,
      "loss": 0.0661,
      "step": 2801
    },
    {
      "epoch": 5.169898430286242,
      "grad_norm": 0.8491671681404114,
      "learning_rate": 8.397e-06,
      "loss": 0.0752,
      "step": 2802
    },
    {
      "epoch": 5.17174515235457,
      "grad_norm": 0.8518207669258118,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.0624,
      "step": 2803
    },
    {
      "epoch": 5.1735918744228995,
      "grad_norm": 0.9227288961410522,
      "learning_rate": 8.403e-06,
      "loss": 0.0538,
      "step": 2804
    },
    {
      "epoch": 5.175438596491228,
      "grad_norm": 0.8876579999923706,
      "learning_rate": 8.406e-06,
      "loss": 0.0474,
      "step": 2805
    },
    {
      "epoch": 5.177285318559557,
      "grad_norm": 2.571634531021118,
      "learning_rate": 8.409e-06,
      "loss": 0.0456,
      "step": 2806
    },
    {
      "epoch": 5.179132040627885,
      "grad_norm": 0.7877750396728516,
      "learning_rate": 8.412e-06,
      "loss": 0.0517,
      "step": 2807
    },
    {
      "epoch": 5.180978762696214,
      "grad_norm": 1.27826988697052,
      "learning_rate": 8.415000000000002e-06,
      "loss": 0.109,
      "step": 2808
    },
    {
      "epoch": 5.1828254847645425,
      "grad_norm": 1.2227567434310913,
      "learning_rate": 8.418000000000001e-06,
      "loss": 0.0814,
      "step": 2809
    },
    {
      "epoch": 5.184672206832872,
      "grad_norm": 1.4295302629470825,
      "learning_rate": 8.421000000000001e-06,
      "loss": 0.0789,
      "step": 2810
    },
    {
      "epoch": 5.186518928901201,
      "grad_norm": 2.2919962406158447,
      "learning_rate": 8.424e-06,
      "loss": 0.3895,
      "step": 2811
    },
    {
      "epoch": 5.188365650969529,
      "grad_norm": 1.3867071866989136,
      "learning_rate": 8.426999999999999e-06,
      "loss": 0.3742,
      "step": 2812
    },
    {
      "epoch": 5.190212373037858,
      "grad_norm": 1.6625173091888428,
      "learning_rate": 8.43e-06,
      "loss": 0.4138,
      "step": 2813
    },
    {
      "epoch": 5.1920590951061865,
      "grad_norm": 1.237267255783081,
      "learning_rate": 8.433e-06,
      "loss": 0.3137,
      "step": 2814
    },
    {
      "epoch": 5.193905817174516,
      "grad_norm": 1.9351158142089844,
      "learning_rate": 8.436e-06,
      "loss": 0.3284,
      "step": 2815
    },
    {
      "epoch": 5.195752539242844,
      "grad_norm": 0.9686079621315002,
      "learning_rate": 8.439e-06,
      "loss": 0.249,
      "step": 2816
    },
    {
      "epoch": 5.197599261311173,
      "grad_norm": 0.9176819324493408,
      "learning_rate": 8.442e-06,
      "loss": 0.2494,
      "step": 2817
    },
    {
      "epoch": 5.199445983379501,
      "grad_norm": 1.0931744575500488,
      "learning_rate": 8.445e-06,
      "loss": 0.2237,
      "step": 2818
    },
    {
      "epoch": 5.20129270544783,
      "grad_norm": 1.0970206260681152,
      "learning_rate": 8.448000000000001e-06,
      "loss": 0.2043,
      "step": 2819
    },
    {
      "epoch": 5.203139427516159,
      "grad_norm": 1.022572636604309,
      "learning_rate": 8.451000000000001e-06,
      "loss": 0.2139,
      "step": 2820
    },
    {
      "epoch": 5.204986149584488,
      "grad_norm": 1.361502766609192,
      "learning_rate": 8.454e-06,
      "loss": 0.1792,
      "step": 2821
    },
    {
      "epoch": 5.206832871652816,
      "grad_norm": 1.2826979160308838,
      "learning_rate": 8.457e-06,
      "loss": 0.2567,
      "step": 2822
    },
    {
      "epoch": 5.208679593721145,
      "grad_norm": 1.0315555334091187,
      "learning_rate": 8.459999999999999e-06,
      "loss": 0.1461,
      "step": 2823
    },
    {
      "epoch": 5.2105263157894735,
      "grad_norm": 1.2108699083328247,
      "learning_rate": 8.463e-06,
      "loss": 0.1435,
      "step": 2824
    },
    {
      "epoch": 5.212373037857803,
      "grad_norm": 1.0025715827941895,
      "learning_rate": 8.466e-06,
      "loss": 0.1251,
      "step": 2825
    },
    {
      "epoch": 5.214219759926131,
      "grad_norm": 0.7117714881896973,
      "learning_rate": 8.469e-06,
      "loss": 0.1303,
      "step": 2826
    },
    {
      "epoch": 5.21606648199446,
      "grad_norm": 0.5328357219696045,
      "learning_rate": 8.472e-06,
      "loss": 0.0406,
      "step": 2827
    },
    {
      "epoch": 5.217913204062788,
      "grad_norm": 1.2464380264282227,
      "learning_rate": 8.475e-06,
      "loss": 0.1048,
      "step": 2828
    },
    {
      "epoch": 5.219759926131117,
      "grad_norm": 0.9050692915916443,
      "learning_rate": 8.478e-06,
      "loss": 0.1046,
      "step": 2829
    },
    {
      "epoch": 5.221606648199446,
      "grad_norm": 0.7401684522628784,
      "learning_rate": 8.481e-06,
      "loss": 0.0854,
      "step": 2830
    },
    {
      "epoch": 5.223453370267775,
      "grad_norm": 0.5116982460021973,
      "learning_rate": 8.484e-06,
      "loss": 0.0457,
      "step": 2831
    },
    {
      "epoch": 5.225300092336103,
      "grad_norm": 0.7157394886016846,
      "learning_rate": 8.487e-06,
      "loss": 0.0564,
      "step": 2832
    },
    {
      "epoch": 5.227146814404432,
      "grad_norm": 1.1383388042449951,
      "learning_rate": 8.49e-06,
      "loss": 0.0675,
      "step": 2833
    },
    {
      "epoch": 5.22899353647276,
      "grad_norm": 0.8179400563240051,
      "learning_rate": 8.493000000000002e-06,
      "loss": 0.0565,
      "step": 2834
    },
    {
      "epoch": 5.23084025854109,
      "grad_norm": 0.9336811304092407,
      "learning_rate": 8.496e-06,
      "loss": 0.0617,
      "step": 2835
    },
    {
      "epoch": 5.232686980609419,
      "grad_norm": 0.8065788745880127,
      "learning_rate": 8.499e-06,
      "loss": 0.039,
      "step": 2836
    },
    {
      "epoch": 5.234533702677747,
      "grad_norm": 0.8431143760681152,
      "learning_rate": 8.502e-06,
      "loss": 0.0662,
      "step": 2837
    },
    {
      "epoch": 5.236380424746076,
      "grad_norm": 0.9685320854187012,
      "learning_rate": 8.504999999999999e-06,
      "loss": 0.056,
      "step": 2838
    },
    {
      "epoch": 5.238227146814404,
      "grad_norm": 1.681692123413086,
      "learning_rate": 8.508e-06,
      "loss": 0.0725,
      "step": 2839
    },
    {
      "epoch": 5.2400738688827335,
      "grad_norm": 0.757462739944458,
      "learning_rate": 8.511e-06,
      "loss": 0.0623,
      "step": 2840
    },
    {
      "epoch": 5.241920590951062,
      "grad_norm": 1.066636323928833,
      "learning_rate": 8.514e-06,
      "loss": 0.0556,
      "step": 2841
    },
    {
      "epoch": 5.243767313019391,
      "grad_norm": 0.7465166449546814,
      "learning_rate": 8.517e-06,
      "loss": 0.0405,
      "step": 2842
    },
    {
      "epoch": 5.245614035087719,
      "grad_norm": 0.8556094765663147,
      "learning_rate": 8.52e-06,
      "loss": 0.0518,
      "step": 2843
    },
    {
      "epoch": 5.247460757156048,
      "grad_norm": 0.8630668520927429,
      "learning_rate": 8.523000000000001e-06,
      "loss": 0.0534,
      "step": 2844
    },
    {
      "epoch": 5.249307479224377,
      "grad_norm": 0.9147830605506897,
      "learning_rate": 8.526000000000001e-06,
      "loss": 0.0472,
      "step": 2845
    },
    {
      "epoch": 5.251154201292706,
      "grad_norm": 0.936911404132843,
      "learning_rate": 8.529e-06,
      "loss": 0.0474,
      "step": 2846
    },
    {
      "epoch": 5.253000923361034,
      "grad_norm": 0.8163319230079651,
      "learning_rate": 8.532e-06,
      "loss": 0.0652,
      "step": 2847
    },
    {
      "epoch": 5.254847645429363,
      "grad_norm": 0.9939672946929932,
      "learning_rate": 8.534999999999999e-06,
      "loss": 0.0633,
      "step": 2848
    },
    {
      "epoch": 5.256694367497691,
      "grad_norm": 1.0096458196640015,
      "learning_rate": 8.538e-06,
      "loss": 0.0553,
      "step": 2849
    },
    {
      "epoch": 5.2585410895660205,
      "grad_norm": 1.5377824306488037,
      "learning_rate": 8.541e-06,
      "loss": 0.0983,
      "step": 2850
    },
    {
      "epoch": 5.260387811634349,
      "grad_norm": 0.9376013875007629,
      "learning_rate": 8.544e-06,
      "loss": 0.0647,
      "step": 2851
    },
    {
      "epoch": 5.262234533702678,
      "grad_norm": 1.0527558326721191,
      "learning_rate": 8.547e-06,
      "loss": 0.1096,
      "step": 2852
    },
    {
      "epoch": 5.264081255771006,
      "grad_norm": 0.8554616570472717,
      "learning_rate": 8.55e-06,
      "loss": 0.0608,
      "step": 2853
    },
    {
      "epoch": 5.265927977839335,
      "grad_norm": 0.9472962021827698,
      "learning_rate": 8.553000000000001e-06,
      "loss": 0.0812,
      "step": 2854
    },
    {
      "epoch": 5.267774699907664,
      "grad_norm": 0.8088082075119019,
      "learning_rate": 8.556e-06,
      "loss": 0.0601,
      "step": 2855
    },
    {
      "epoch": 5.269621421975993,
      "grad_norm": 0.9127931594848633,
      "learning_rate": 8.559e-06,
      "loss": 0.0693,
      "step": 2856
    },
    {
      "epoch": 5.271468144044321,
      "grad_norm": 0.7901133894920349,
      "learning_rate": 8.562e-06,
      "loss": 0.0515,
      "step": 2857
    },
    {
      "epoch": 5.27331486611265,
      "grad_norm": 1.4003362655639648,
      "learning_rate": 8.565e-06,
      "loss": 0.0728,
      "step": 2858
    },
    {
      "epoch": 5.275161588180978,
      "grad_norm": 1.0627663135528564,
      "learning_rate": 8.568000000000002e-06,
      "loss": 0.0749,
      "step": 2859
    },
    {
      "epoch": 5.2770083102493075,
      "grad_norm": 2.196826457977295,
      "learning_rate": 8.571e-06,
      "loss": 0.0995,
      "step": 2860
    },
    {
      "epoch": 5.278855032317637,
      "grad_norm": 1.8018981218338013,
      "learning_rate": 8.574e-06,
      "loss": 0.3984,
      "step": 2861
    },
    {
      "epoch": 5.280701754385965,
      "grad_norm": 1.298303246498108,
      "learning_rate": 8.577e-06,
      "loss": 0.4234,
      "step": 2862
    },
    {
      "epoch": 5.282548476454294,
      "grad_norm": 1.3109056949615479,
      "learning_rate": 8.58e-06,
      "loss": 0.3235,
      "step": 2863
    },
    {
      "epoch": 5.284395198522622,
      "grad_norm": 1.5986768007278442,
      "learning_rate": 8.583e-06,
      "loss": 0.282,
      "step": 2864
    },
    {
      "epoch": 5.286241920590951,
      "grad_norm": 1.181259274482727,
      "learning_rate": 8.586e-06,
      "loss": 0.2529,
      "step": 2865
    },
    {
      "epoch": 5.28808864265928,
      "grad_norm": 0.8804054260253906,
      "learning_rate": 8.589e-06,
      "loss": 0.2063,
      "step": 2866
    },
    {
      "epoch": 5.289935364727609,
      "grad_norm": 1.4291982650756836,
      "learning_rate": 8.592e-06,
      "loss": 0.2953,
      "step": 2867
    },
    {
      "epoch": 5.291782086795937,
      "grad_norm": 1.3200314044952393,
      "learning_rate": 8.595e-06,
      "loss": 0.2368,
      "step": 2868
    },
    {
      "epoch": 5.293628808864266,
      "grad_norm": 1.1014959812164307,
      "learning_rate": 8.598000000000001e-06,
      "loss": 0.2103,
      "step": 2869
    },
    {
      "epoch": 5.2954755309325945,
      "grad_norm": 0.9992939233779907,
      "learning_rate": 8.601000000000001e-06,
      "loss": 0.1908,
      "step": 2870
    },
    {
      "epoch": 5.297322253000924,
      "grad_norm": 0.9212896227836609,
      "learning_rate": 8.604000000000001e-06,
      "loss": 0.1221,
      "step": 2871
    },
    {
      "epoch": 5.299168975069252,
      "grad_norm": 0.8088216185569763,
      "learning_rate": 8.606999999999999e-06,
      "loss": 0.1182,
      "step": 2872
    },
    {
      "epoch": 5.301015697137581,
      "grad_norm": 0.789671778678894,
      "learning_rate": 8.609999999999999e-06,
      "loss": 0.1241,
      "step": 2873
    },
    {
      "epoch": 5.302862419205909,
      "grad_norm": 1.0368704795837402,
      "learning_rate": 8.613e-06,
      "loss": 0.0944,
      "step": 2874
    },
    {
      "epoch": 5.304709141274238,
      "grad_norm": 1.2252488136291504,
      "learning_rate": 8.616e-06,
      "loss": 0.1023,
      "step": 2875
    },
    {
      "epoch": 5.306555863342567,
      "grad_norm": 1.0052070617675781,
      "learning_rate": 8.619e-06,
      "loss": 0.0693,
      "step": 2876
    },
    {
      "epoch": 5.308402585410896,
      "grad_norm": 0.6709212064743042,
      "learning_rate": 8.622e-06,
      "loss": 0.0674,
      "step": 2877
    },
    {
      "epoch": 5.310249307479224,
      "grad_norm": 0.5491957664489746,
      "learning_rate": 8.625e-06,
      "loss": 0.0614,
      "step": 2878
    },
    {
      "epoch": 5.312096029547553,
      "grad_norm": 0.8416413068771362,
      "learning_rate": 8.628000000000001e-06,
      "loss": 0.0663,
      "step": 2879
    },
    {
      "epoch": 5.3139427516158815,
      "grad_norm": 0.8481470346450806,
      "learning_rate": 8.631000000000001e-06,
      "loss": 0.0751,
      "step": 2880
    },
    {
      "epoch": 5.315789473684211,
      "grad_norm": 1.0984277725219727,
      "learning_rate": 8.634e-06,
      "loss": 0.0693,
      "step": 2881
    },
    {
      "epoch": 5.317636195752539,
      "grad_norm": 0.9302229285240173,
      "learning_rate": 8.637e-06,
      "loss": 0.0926,
      "step": 2882
    },
    {
      "epoch": 5.319482917820868,
      "grad_norm": 0.49322235584259033,
      "learning_rate": 8.64e-06,
      "loss": 0.0357,
      "step": 2883
    },
    {
      "epoch": 5.321329639889196,
      "grad_norm": 0.730798602104187,
      "learning_rate": 8.643e-06,
      "loss": 0.0418,
      "step": 2884
    },
    {
      "epoch": 5.323176361957525,
      "grad_norm": 0.689988374710083,
      "learning_rate": 8.646e-06,
      "loss": 0.069,
      "step": 2885
    },
    {
      "epoch": 5.325023084025855,
      "grad_norm": 0.9584289789199829,
      "learning_rate": 8.649e-06,
      "loss": 0.0699,
      "step": 2886
    },
    {
      "epoch": 5.326869806094183,
      "grad_norm": 0.8632835745811462,
      "learning_rate": 8.652e-06,
      "loss": 0.0638,
      "step": 2887
    },
    {
      "epoch": 5.328716528162512,
      "grad_norm": 1.3212529420852661,
      "learning_rate": 8.655e-06,
      "loss": 0.0458,
      "step": 2888
    },
    {
      "epoch": 5.33056325023084,
      "grad_norm": 0.7729990482330322,
      "learning_rate": 8.658e-06,
      "loss": 0.0644,
      "step": 2889
    },
    {
      "epoch": 5.332409972299169,
      "grad_norm": 1.917777180671692,
      "learning_rate": 8.661e-06,
      "loss": 0.1455,
      "step": 2890
    },
    {
      "epoch": 5.334256694367498,
      "grad_norm": 1.0701407194137573,
      "learning_rate": 8.664e-06,
      "loss": 0.0824,
      "step": 2891
    },
    {
      "epoch": 5.336103416435827,
      "grad_norm": 0.8621686697006226,
      "learning_rate": 8.667e-06,
      "loss": 0.0592,
      "step": 2892
    },
    {
      "epoch": 5.337950138504155,
      "grad_norm": 0.9155268669128418,
      "learning_rate": 8.67e-06,
      "loss": 0.0749,
      "step": 2893
    },
    {
      "epoch": 5.339796860572484,
      "grad_norm": 0.8312194347381592,
      "learning_rate": 8.673000000000001e-06,
      "loss": 0.0484,
      "step": 2894
    },
    {
      "epoch": 5.341643582640812,
      "grad_norm": 1.4102102518081665,
      "learning_rate": 8.676000000000001e-06,
      "loss": 0.0683,
      "step": 2895
    },
    {
      "epoch": 5.3434903047091415,
      "grad_norm": 1.0904110670089722,
      "learning_rate": 8.679000000000001e-06,
      "loss": 0.0677,
      "step": 2896
    },
    {
      "epoch": 5.34533702677747,
      "grad_norm": 0.8273285627365112,
      "learning_rate": 8.682e-06,
      "loss": 0.0669,
      "step": 2897
    },
    {
      "epoch": 5.347183748845799,
      "grad_norm": 1.3562661409378052,
      "learning_rate": 8.684999999999999e-06,
      "loss": 0.0727,
      "step": 2898
    },
    {
      "epoch": 5.349030470914127,
      "grad_norm": 1.7254983186721802,
      "learning_rate": 8.688e-06,
      "loss": 0.065,
      "step": 2899
    },
    {
      "epoch": 5.350877192982456,
      "grad_norm": 0.9011834859848022,
      "learning_rate": 8.691e-06,
      "loss": 0.0575,
      "step": 2900
    },
    {
      "epoch": 5.352723915050785,
      "grad_norm": 0.9402887225151062,
      "learning_rate": 8.694e-06,
      "loss": 0.0535,
      "step": 2901
    },
    {
      "epoch": 5.354570637119114,
      "grad_norm": 0.9309573769569397,
      "learning_rate": 8.697e-06,
      "loss": 0.0476,
      "step": 2902
    },
    {
      "epoch": 5.356417359187442,
      "grad_norm": 0.9273201823234558,
      "learning_rate": 8.7e-06,
      "loss": 0.0502,
      "step": 2903
    },
    {
      "epoch": 5.358264081255771,
      "grad_norm": 0.978259265422821,
      "learning_rate": 8.703000000000001e-06,
      "loss": 0.0678,
      "step": 2904
    },
    {
      "epoch": 5.360110803324099,
      "grad_norm": 0.7556988000869751,
      "learning_rate": 8.706000000000001e-06,
      "loss": 0.0551,
      "step": 2905
    },
    {
      "epoch": 5.3619575253924285,
      "grad_norm": 0.8013013601303101,
      "learning_rate": 8.709e-06,
      "loss": 0.0589,
      "step": 2906
    },
    {
      "epoch": 5.363804247460757,
      "grad_norm": 0.9943859577178955,
      "learning_rate": 8.712e-06,
      "loss": 0.0761,
      "step": 2907
    },
    {
      "epoch": 5.365650969529086,
      "grad_norm": 1.665634036064148,
      "learning_rate": 8.715e-06,
      "loss": 0.0771,
      "step": 2908
    },
    {
      "epoch": 5.367497691597414,
      "grad_norm": 1.0174927711486816,
      "learning_rate": 8.718e-06,
      "loss": 0.0794,
      "step": 2909
    },
    {
      "epoch": 5.369344413665743,
      "grad_norm": 0.9980137944221497,
      "learning_rate": 8.721e-06,
      "loss": 0.0642,
      "step": 2910
    },
    {
      "epoch": 5.3711911357340725,
      "grad_norm": 1.8771740198135376,
      "learning_rate": 8.724e-06,
      "loss": 0.4153,
      "step": 2911
    },
    {
      "epoch": 5.373037857802401,
      "grad_norm": 2.773858070373535,
      "learning_rate": 8.727e-06,
      "loss": 0.4126,
      "step": 2912
    },
    {
      "epoch": 5.374884579870729,
      "grad_norm": 1.1791237592697144,
      "learning_rate": 8.73e-06,
      "loss": 0.286,
      "step": 2913
    },
    {
      "epoch": 5.376731301939058,
      "grad_norm": 1.0450700521469116,
      "learning_rate": 8.733000000000001e-06,
      "loss": 0.2737,
      "step": 2914
    },
    {
      "epoch": 5.378578024007387,
      "grad_norm": 0.9821802973747253,
      "learning_rate": 8.736e-06,
      "loss": 0.3204,
      "step": 2915
    },
    {
      "epoch": 5.3804247460757155,
      "grad_norm": 2.578831911087036,
      "learning_rate": 8.739e-06,
      "loss": 0.3898,
      "step": 2916
    },
    {
      "epoch": 5.382271468144045,
      "grad_norm": 1.2257970571517944,
      "learning_rate": 8.742e-06,
      "loss": 0.2362,
      "step": 2917
    },
    {
      "epoch": 5.384118190212373,
      "grad_norm": 1.0278574228286743,
      "learning_rate": 8.745e-06,
      "loss": 0.1894,
      "step": 2918
    },
    {
      "epoch": 5.385964912280702,
      "grad_norm": 1.2053768634796143,
      "learning_rate": 8.748000000000002e-06,
      "loss": 0.2346,
      "step": 2919
    },
    {
      "epoch": 5.38781163434903,
      "grad_norm": 1.3017722368240356,
      "learning_rate": 8.751000000000001e-06,
      "loss": 0.2011,
      "step": 2920
    },
    {
      "epoch": 5.3896583564173595,
      "grad_norm": 0.7116005420684814,
      "learning_rate": 8.754e-06,
      "loss": 0.1469,
      "step": 2921
    },
    {
      "epoch": 5.391505078485688,
      "grad_norm": 2.64703106880188,
      "learning_rate": 8.757e-06,
      "loss": 0.1881,
      "step": 2922
    },
    {
      "epoch": 5.393351800554017,
      "grad_norm": 1.7918483018875122,
      "learning_rate": 8.759999999999999e-06,
      "loss": 0.1773,
      "step": 2923
    },
    {
      "epoch": 5.395198522622345,
      "grad_norm": 0.6470117568969727,
      "learning_rate": 8.763e-06,
      "loss": 0.0867,
      "step": 2924
    },
    {
      "epoch": 5.397045244690674,
      "grad_norm": 2.316588878631592,
      "learning_rate": 8.766e-06,
      "loss": 0.193,
      "step": 2925
    },
    {
      "epoch": 5.3988919667590025,
      "grad_norm": 1.2009514570236206,
      "learning_rate": 8.769e-06,
      "loss": 0.102,
      "step": 2926
    },
    {
      "epoch": 5.400738688827332,
      "grad_norm": 1.333717703819275,
      "learning_rate": 8.772e-06,
      "loss": 0.0831,
      "step": 2927
    },
    {
      "epoch": 5.40258541089566,
      "grad_norm": 1.4752955436706543,
      "learning_rate": 8.775e-06,
      "loss": 0.075,
      "step": 2928
    },
    {
      "epoch": 5.404432132963989,
      "grad_norm": 0.9960635900497437,
      "learning_rate": 8.778000000000001e-06,
      "loss": 0.0757,
      "step": 2929
    },
    {
      "epoch": 5.406278855032317,
      "grad_norm": 0.7506101131439209,
      "learning_rate": 8.781000000000001e-06,
      "loss": 0.0787,
      "step": 2930
    },
    {
      "epoch": 5.4081255771006465,
      "grad_norm": 0.7151127457618713,
      "learning_rate": 8.784000000000001e-06,
      "loss": 0.0586,
      "step": 2931
    },
    {
      "epoch": 5.409972299168975,
      "grad_norm": 0.9094399213790894,
      "learning_rate": 8.787e-06,
      "loss": 0.072,
      "step": 2932
    },
    {
      "epoch": 5.411819021237304,
      "grad_norm": 0.6951301097869873,
      "learning_rate": 8.79e-06,
      "loss": 0.0861,
      "step": 2933
    },
    {
      "epoch": 5.413665743305632,
      "grad_norm": 0.6631410717964172,
      "learning_rate": 8.793e-06,
      "loss": 0.0524,
      "step": 2934
    },
    {
      "epoch": 5.415512465373961,
      "grad_norm": 0.6533665060997009,
      "learning_rate": 8.796e-06,
      "loss": 0.0517,
      "step": 2935
    },
    {
      "epoch": 5.41735918744229,
      "grad_norm": 0.8334203362464905,
      "learning_rate": 8.799e-06,
      "loss": 0.0591,
      "step": 2936
    },
    {
      "epoch": 5.419205909510619,
      "grad_norm": 0.786499559879303,
      "learning_rate": 8.802e-06,
      "loss": 0.061,
      "step": 2937
    },
    {
      "epoch": 5.421052631578947,
      "grad_norm": 0.8949714303016663,
      "learning_rate": 8.805e-06,
      "loss": 0.076,
      "step": 2938
    },
    {
      "epoch": 5.422899353647276,
      "grad_norm": 0.8435794711112976,
      "learning_rate": 8.808000000000001e-06,
      "loss": 0.0617,
      "step": 2939
    },
    {
      "epoch": 5.424746075715605,
      "grad_norm": 0.7329889535903931,
      "learning_rate": 8.811000000000001e-06,
      "loss": 0.0463,
      "step": 2940
    },
    {
      "epoch": 5.426592797783933,
      "grad_norm": 0.694934070110321,
      "learning_rate": 8.814e-06,
      "loss": 0.0467,
      "step": 2941
    },
    {
      "epoch": 5.428439519852263,
      "grad_norm": 0.8145605325698853,
      "learning_rate": 8.817e-06,
      "loss": 0.0664,
      "step": 2942
    },
    {
      "epoch": 5.430286241920591,
      "grad_norm": 0.9637972712516785,
      "learning_rate": 8.82e-06,
      "loss": 0.0718,
      "step": 2943
    },
    {
      "epoch": 5.43213296398892,
      "grad_norm": 0.6921423673629761,
      "learning_rate": 8.823e-06,
      "loss": 0.0524,
      "step": 2944
    },
    {
      "epoch": 5.433979686057248,
      "grad_norm": 0.7768904566764832,
      "learning_rate": 8.826000000000002e-06,
      "loss": 0.0479,
      "step": 2945
    },
    {
      "epoch": 5.435826408125577,
      "grad_norm": 0.941554069519043,
      "learning_rate": 8.829e-06,
      "loss": 0.0358,
      "step": 2946
    },
    {
      "epoch": 5.437673130193906,
      "grad_norm": 1.102005124092102,
      "learning_rate": 8.832e-06,
      "loss": 0.0757,
      "step": 2947
    },
    {
      "epoch": 5.439519852262235,
      "grad_norm": 1.0594245195388794,
      "learning_rate": 8.835e-06,
      "loss": 0.0408,
      "step": 2948
    },
    {
      "epoch": 5.441366574330563,
      "grad_norm": 0.686547577381134,
      "learning_rate": 8.837999999999999e-06,
      "loss": 0.0511,
      "step": 2949
    },
    {
      "epoch": 5.443213296398892,
      "grad_norm": 0.8735717535018921,
      "learning_rate": 8.841e-06,
      "loss": 0.0484,
      "step": 2950
    },
    {
      "epoch": 5.44506001846722,
      "grad_norm": 0.9363495707511902,
      "learning_rate": 8.844e-06,
      "loss": 0.0892,
      "step": 2951
    },
    {
      "epoch": 5.44690674053555,
      "grad_norm": 0.9038994908332825,
      "learning_rate": 8.847e-06,
      "loss": 0.0466,
      "step": 2952
    },
    {
      "epoch": 5.448753462603878,
      "grad_norm": 1.1139678955078125,
      "learning_rate": 8.85e-06,
      "loss": 0.0843,
      "step": 2953
    },
    {
      "epoch": 5.450600184672207,
      "grad_norm": 1.3504257202148438,
      "learning_rate": 8.853e-06,
      "loss": 0.0634,
      "step": 2954
    },
    {
      "epoch": 5.452446906740535,
      "grad_norm": 0.7875360250473022,
      "learning_rate": 8.856000000000001e-06,
      "loss": 0.0557,
      "step": 2955
    },
    {
      "epoch": 5.454293628808864,
      "grad_norm": 1.0293179750442505,
      "learning_rate": 8.859000000000001e-06,
      "loss": 0.0516,
      "step": 2956
    },
    {
      "epoch": 5.456140350877193,
      "grad_norm": 1.1746110916137695,
      "learning_rate": 8.862000000000001e-06,
      "loss": 0.0404,
      "step": 2957
    },
    {
      "epoch": 5.457987072945522,
      "grad_norm": 0.681563138961792,
      "learning_rate": 8.864999999999999e-06,
      "loss": 0.0412,
      "step": 2958
    },
    {
      "epoch": 5.45983379501385,
      "grad_norm": 1.1411263942718506,
      "learning_rate": 8.867999999999999e-06,
      "loss": 0.0845,
      "step": 2959
    },
    {
      "epoch": 5.461680517082179,
      "grad_norm": 1.3232712745666504,
      "learning_rate": 8.871e-06,
      "loss": 0.1214,
      "step": 2960
    },
    {
      "epoch": 5.463527239150508,
      "grad_norm": 1.824684739112854,
      "learning_rate": 8.874e-06,
      "loss": 0.4566,
      "step": 2961
    },
    {
      "epoch": 5.465373961218837,
      "grad_norm": 1.305968165397644,
      "learning_rate": 8.877e-06,
      "loss": 0.4121,
      "step": 2962
    },
    {
      "epoch": 5.467220683287165,
      "grad_norm": 1.9621754884719849,
      "learning_rate": 8.88e-06,
      "loss": 0.252,
      "step": 2963
    },
    {
      "epoch": 5.469067405355494,
      "grad_norm": 1.6577662229537964,
      "learning_rate": 8.883e-06,
      "loss": 0.306,
      "step": 2964
    },
    {
      "epoch": 5.470914127423823,
      "grad_norm": 1.061056137084961,
      "learning_rate": 8.886000000000001e-06,
      "loss": 0.2434,
      "step": 2965
    },
    {
      "epoch": 5.472760849492151,
      "grad_norm": 1.2487244606018066,
      "learning_rate": 8.889e-06,
      "loss": 0.2658,
      "step": 2966
    },
    {
      "epoch": 5.4746075715604805,
      "grad_norm": 1.0350834131240845,
      "learning_rate": 8.892e-06,
      "loss": 0.2689,
      "step": 2967
    },
    {
      "epoch": 5.476454293628809,
      "grad_norm": 0.811021089553833,
      "learning_rate": 8.895e-06,
      "loss": 0.1559,
      "step": 2968
    },
    {
      "epoch": 5.478301015697138,
      "grad_norm": 0.9796880483627319,
      "learning_rate": 8.898e-06,
      "loss": 0.3147,
      "step": 2969
    },
    {
      "epoch": 5.480147737765466,
      "grad_norm": 1.0987402200698853,
      "learning_rate": 8.901e-06,
      "loss": 0.1979,
      "step": 2970
    },
    {
      "epoch": 5.481994459833795,
      "grad_norm": 1.0269370079040527,
      "learning_rate": 8.904e-06,
      "loss": 0.1935,
      "step": 2971
    },
    {
      "epoch": 5.4838411819021236,
      "grad_norm": 0.9793119430541992,
      "learning_rate": 8.907e-06,
      "loss": 0.1526,
      "step": 2972
    },
    {
      "epoch": 5.485687903970453,
      "grad_norm": 0.7819326519966125,
      "learning_rate": 8.91e-06,
      "loss": 0.1513,
      "step": 2973
    },
    {
      "epoch": 5.487534626038781,
      "grad_norm": 0.9465290307998657,
      "learning_rate": 8.913e-06,
      "loss": 0.1285,
      "step": 2974
    },
    {
      "epoch": 5.48938134810711,
      "grad_norm": 0.9836941957473755,
      "learning_rate": 8.916e-06,
      "loss": 0.1069,
      "step": 2975
    },
    {
      "epoch": 5.491228070175438,
      "grad_norm": 0.9329361915588379,
      "learning_rate": 8.919e-06,
      "loss": 0.1254,
      "step": 2976
    },
    {
      "epoch": 5.4930747922437675,
      "grad_norm": 1.0424612760543823,
      "learning_rate": 8.922e-06,
      "loss": 0.107,
      "step": 2977
    },
    {
      "epoch": 5.494921514312096,
      "grad_norm": 1.5464701652526855,
      "learning_rate": 8.925e-06,
      "loss": 0.0687,
      "step": 2978
    },
    {
      "epoch": 5.496768236380425,
      "grad_norm": 0.681675136089325,
      "learning_rate": 8.928e-06,
      "loss": 0.0608,
      "step": 2979
    },
    {
      "epoch": 5.498614958448753,
      "grad_norm": 1.3586539030075073,
      "learning_rate": 8.931000000000001e-06,
      "loss": 0.1181,
      "step": 2980
    },
    {
      "epoch": 5.500461680517082,
      "grad_norm": 1.1351392269134521,
      "learning_rate": 8.934000000000001e-06,
      "loss": 0.0602,
      "step": 2981
    },
    {
      "epoch": 5.5023084025854105,
      "grad_norm": 0.9219081401824951,
      "learning_rate": 8.937000000000001e-06,
      "loss": 0.098,
      "step": 2982
    },
    {
      "epoch": 5.50415512465374,
      "grad_norm": 1.2950011491775513,
      "learning_rate": 8.939999999999999e-06,
      "loss": 0.0701,
      "step": 2983
    },
    {
      "epoch": 5.506001846722068,
      "grad_norm": 0.7642313838005066,
      "learning_rate": 8.942999999999999e-06,
      "loss": 0.0474,
      "step": 2984
    },
    {
      "epoch": 5.507848568790397,
      "grad_norm": 1.5005654096603394,
      "learning_rate": 8.946e-06,
      "loss": 0.0863,
      "step": 2985
    },
    {
      "epoch": 5.509695290858726,
      "grad_norm": 0.8310704231262207,
      "learning_rate": 8.949e-06,
      "loss": 0.0736,
      "step": 2986
    },
    {
      "epoch": 5.5115420129270545,
      "grad_norm": 0.8422238826751709,
      "learning_rate": 8.952e-06,
      "loss": 0.0833,
      "step": 2987
    },
    {
      "epoch": 5.513388734995383,
      "grad_norm": 0.650684654712677,
      "learning_rate": 8.955e-06,
      "loss": 0.0531,
      "step": 2988
    },
    {
      "epoch": 5.515235457063712,
      "grad_norm": 0.9870924949645996,
      "learning_rate": 8.958e-06,
      "loss": 0.0594,
      "step": 2989
    },
    {
      "epoch": 5.517082179132041,
      "grad_norm": 1.1698861122131348,
      "learning_rate": 8.961000000000001e-06,
      "loss": 0.0559,
      "step": 2990
    },
    {
      "epoch": 5.518928901200369,
      "grad_norm": 0.7865745425224304,
      "learning_rate": 8.964000000000001e-06,
      "loss": 0.0697,
      "step": 2991
    },
    {
      "epoch": 5.520775623268698,
      "grad_norm": 1.2937891483306885,
      "learning_rate": 8.967e-06,
      "loss": 0.0656,
      "step": 2992
    },
    {
      "epoch": 5.522622345337027,
      "grad_norm": 1.5481828451156616,
      "learning_rate": 8.97e-06,
      "loss": 0.063,
      "step": 2993
    },
    {
      "epoch": 5.524469067405356,
      "grad_norm": 1.1445446014404297,
      "learning_rate": 8.973e-06,
      "loss": 0.0693,
      "step": 2994
    },
    {
      "epoch": 5.526315789473684,
      "grad_norm": 1.5518786907196045,
      "learning_rate": 8.976e-06,
      "loss": 0.0556,
      "step": 2995
    },
    {
      "epoch": 5.528162511542013,
      "grad_norm": 0.5828351974487305,
      "learning_rate": 8.979e-06,
      "loss": 0.0341,
      "step": 2996
    },
    {
      "epoch": 5.5300092336103415,
      "grad_norm": 0.7857339978218079,
      "learning_rate": 8.982e-06,
      "loss": 0.0552,
      "step": 2997
    },
    {
      "epoch": 5.531855955678671,
      "grad_norm": 0.8015343546867371,
      "learning_rate": 8.985e-06,
      "loss": 0.0654,
      "step": 2998
    },
    {
      "epoch": 5.533702677746999,
      "grad_norm": 1.580649495124817,
      "learning_rate": 8.988e-06,
      "loss": 0.0526,
      "step": 2999
    },
    {
      "epoch": 5.535549399815328,
      "grad_norm": 1.2285726070404053,
      "learning_rate": 8.991e-06,
      "loss": 0.0604,
      "step": 3000
    },
    {
      "epoch": 5.535549399815328,
      "eval_cer": 0.13466573346812868,
      "eval_loss": 0.42114710807800293,
      "eval_runtime": 16.2471,
      "eval_samples_per_second": 18.711,
      "eval_steps_per_second": 0.615,
      "eval_wer": 0.4785111281657713,
      "step": 3000
    },
    {
      "epoch": 5.537396121883656,
      "grad_norm": 2.3396427631378174,
      "learning_rate": 8.994e-06,
      "loss": 0.0483,
      "step": 3001
    },
    {
      "epoch": 5.539242843951985,
      "grad_norm": 1.1838312149047852,
      "learning_rate": 8.997e-06,
      "loss": 0.0643,
      "step": 3002
    },
    {
      "epoch": 5.541089566020314,
      "grad_norm": 1.489532470703125,
      "learning_rate": 9e-06,
      "loss": 0.0414,
      "step": 3003
    },
    {
      "epoch": 5.542936288088643,
      "grad_norm": 0.9960378408432007,
      "learning_rate": 9.003e-06,
      "loss": 0.056,
      "step": 3004
    },
    {
      "epoch": 5.544783010156971,
      "grad_norm": 1.023935079574585,
      "learning_rate": 9.006000000000002e-06,
      "loss": 0.0585,
      "step": 3005
    },
    {
      "epoch": 5.5466297322253,
      "grad_norm": 1.3380311727523804,
      "learning_rate": 9.009000000000001e-06,
      "loss": 0.0695,
      "step": 3006
    },
    {
      "epoch": 5.5484764542936285,
      "grad_norm": 1.2113758325576782,
      "learning_rate": 9.012e-06,
      "loss": 0.0856,
      "step": 3007
    },
    {
      "epoch": 5.550323176361958,
      "grad_norm": 1.1438714265823364,
      "learning_rate": 9.015e-06,
      "loss": 0.0752,
      "step": 3008
    },
    {
      "epoch": 5.552169898430286,
      "grad_norm": 1.388740062713623,
      "learning_rate": 9.017999999999999e-06,
      "loss": 0.0974,
      "step": 3009
    },
    {
      "epoch": 5.554016620498615,
      "grad_norm": 1.1810368299484253,
      "learning_rate": 9.021e-06,
      "loss": 0.1307,
      "step": 3010
    },
    {
      "epoch": 5.555863342566944,
      "grad_norm": 3.170240640640259,
      "learning_rate": 9.024e-06,
      "loss": 0.5096,
      "step": 3011
    },
    {
      "epoch": 5.557710064635272,
      "grad_norm": 1.2501941919326782,
      "learning_rate": 9.027e-06,
      "loss": 0.3023,
      "step": 3012
    },
    {
      "epoch": 5.559556786703601,
      "grad_norm": 1.2062792778015137,
      "learning_rate": 9.03e-06,
      "loss": 0.3017,
      "step": 3013
    },
    {
      "epoch": 5.56140350877193,
      "grad_norm": 0.9803247451782227,
      "learning_rate": 9.033e-06,
      "loss": 0.2733,
      "step": 3014
    },
    {
      "epoch": 5.563250230840259,
      "grad_norm": 0.9785440564155579,
      "learning_rate": 9.036000000000001e-06,
      "loss": 0.2386,
      "step": 3015
    },
    {
      "epoch": 5.565096952908587,
      "grad_norm": 2.5354113578796387,
      "learning_rate": 9.039000000000001e-06,
      "loss": 0.2437,
      "step": 3016
    },
    {
      "epoch": 5.566943674976916,
      "grad_norm": 1.2615872621536255,
      "learning_rate": 9.042e-06,
      "loss": 0.2878,
      "step": 3017
    },
    {
      "epoch": 5.568790397045245,
      "grad_norm": 0.8574416637420654,
      "learning_rate": 9.045e-06,
      "loss": 0.1853,
      "step": 3018
    },
    {
      "epoch": 5.570637119113574,
      "grad_norm": 1.2679378986358643,
      "learning_rate": 9.048e-06,
      "loss": 0.2943,
      "step": 3019
    },
    {
      "epoch": 5.572483841181902,
      "grad_norm": 1.3468197584152222,
      "learning_rate": 9.051e-06,
      "loss": 0.2309,
      "step": 3020
    },
    {
      "epoch": 5.574330563250231,
      "grad_norm": 0.8612178564071655,
      "learning_rate": 9.054e-06,
      "loss": 0.1631,
      "step": 3021
    },
    {
      "epoch": 5.576177285318559,
      "grad_norm": 5.9170989990234375,
      "learning_rate": 9.057e-06,
      "loss": 0.2011,
      "step": 3022
    },
    {
      "epoch": 5.5780240073868885,
      "grad_norm": 1.277970552444458,
      "learning_rate": 9.06e-06,
      "loss": 0.2202,
      "step": 3023
    },
    {
      "epoch": 5.579870729455217,
      "grad_norm": 0.7107061147689819,
      "learning_rate": 9.063e-06,
      "loss": 0.069,
      "step": 3024
    },
    {
      "epoch": 5.581717451523546,
      "grad_norm": 0.9380223751068115,
      "learning_rate": 9.066000000000001e-06,
      "loss": 0.0869,
      "step": 3025
    },
    {
      "epoch": 5.583564173591874,
      "grad_norm": 1.3844622373580933,
      "learning_rate": 9.069e-06,
      "loss": 0.0685,
      "step": 3026
    },
    {
      "epoch": 5.585410895660203,
      "grad_norm": 0.7406550645828247,
      "learning_rate": 9.072e-06,
      "loss": 0.056,
      "step": 3027
    },
    {
      "epoch": 5.587257617728532,
      "grad_norm": 1.171863317489624,
      "learning_rate": 9.075e-06,
      "loss": 0.1099,
      "step": 3028
    },
    {
      "epoch": 5.589104339796861,
      "grad_norm": 0.9505486488342285,
      "learning_rate": 9.078e-06,
      "loss": 0.0639,
      "step": 3029
    },
    {
      "epoch": 5.590951061865189,
      "grad_norm": 0.8913169503211975,
      "learning_rate": 9.081000000000002e-06,
      "loss": 0.1375,
      "step": 3030
    },
    {
      "epoch": 5.592797783933518,
      "grad_norm": 1.2923994064331055,
      "learning_rate": 9.084000000000001e-06,
      "loss": 0.0891,
      "step": 3031
    },
    {
      "epoch": 5.594644506001846,
      "grad_norm": 1.3994338512420654,
      "learning_rate": 9.087e-06,
      "loss": 0.0706,
      "step": 3032
    },
    {
      "epoch": 5.5964912280701755,
      "grad_norm": 0.9740058779716492,
      "learning_rate": 9.09e-06,
      "loss": 0.0825,
      "step": 3033
    },
    {
      "epoch": 5.598337950138504,
      "grad_norm": 0.6192094683647156,
      "learning_rate": 9.093e-06,
      "loss": 0.0527,
      "step": 3034
    },
    {
      "epoch": 5.600184672206833,
      "grad_norm": 0.9898630976676941,
      "learning_rate": 9.096e-06,
      "loss": 0.059,
      "step": 3035
    },
    {
      "epoch": 5.602031394275162,
      "grad_norm": 0.9869064688682556,
      "learning_rate": 9.099e-06,
      "loss": 0.0926,
      "step": 3036
    },
    {
      "epoch": 5.60387811634349,
      "grad_norm": 0.775285542011261,
      "learning_rate": 9.102e-06,
      "loss": 0.0603,
      "step": 3037
    },
    {
      "epoch": 5.605724838411819,
      "grad_norm": 1.0720051527023315,
      "learning_rate": 9.105e-06,
      "loss": 0.0603,
      "step": 3038
    },
    {
      "epoch": 5.607571560480148,
      "grad_norm": 2.360222101211548,
      "learning_rate": 9.108e-06,
      "loss": 0.0702,
      "step": 3039
    },
    {
      "epoch": 5.609418282548477,
      "grad_norm": 0.9469456672668457,
      "learning_rate": 9.111000000000001e-06,
      "loss": 0.0443,
      "step": 3040
    },
    {
      "epoch": 5.611265004616805,
      "grad_norm": 1.5419272184371948,
      "learning_rate": 9.114000000000001e-06,
      "loss": 0.1102,
      "step": 3041
    },
    {
      "epoch": 5.613111726685134,
      "grad_norm": 0.6679257154464722,
      "learning_rate": 9.117000000000001e-06,
      "loss": 0.0591,
      "step": 3042
    },
    {
      "epoch": 5.6149584487534625,
      "grad_norm": 0.9657368063926697,
      "learning_rate": 9.12e-06,
      "loss": 0.0735,
      "step": 3043
    },
    {
      "epoch": 5.616805170821792,
      "grad_norm": 0.9724937677383423,
      "learning_rate": 9.122999999999999e-06,
      "loss": 0.0496,
      "step": 3044
    },
    {
      "epoch": 5.61865189289012,
      "grad_norm": 1.4401867389678955,
      "learning_rate": 9.126e-06,
      "loss": 0.0722,
      "step": 3045
    },
    {
      "epoch": 5.620498614958449,
      "grad_norm": 1.0665438175201416,
      "learning_rate": 9.129e-06,
      "loss": 0.0656,
      "step": 3046
    },
    {
      "epoch": 5.622345337026777,
      "grad_norm": 0.6612203121185303,
      "learning_rate": 9.132e-06,
      "loss": 0.0366,
      "step": 3047
    },
    {
      "epoch": 5.624192059095106,
      "grad_norm": 1.3835797309875488,
      "learning_rate": 9.135e-06,
      "loss": 0.0824,
      "step": 3048
    },
    {
      "epoch": 5.626038781163435,
      "grad_norm": 0.8385351896286011,
      "learning_rate": 9.138e-06,
      "loss": 0.0671,
      "step": 3049
    },
    {
      "epoch": 5.627885503231764,
      "grad_norm": 0.7732231020927429,
      "learning_rate": 9.141000000000001e-06,
      "loss": 0.0547,
      "step": 3050
    },
    {
      "epoch": 5.629732225300092,
      "grad_norm": 0.8967553973197937,
      "learning_rate": 9.144000000000001e-06,
      "loss": 0.04,
      "step": 3051
    },
    {
      "epoch": 5.631578947368421,
      "grad_norm": 0.8584074378013611,
      "learning_rate": 9.147e-06,
      "loss": 0.0607,
      "step": 3052
    },
    {
      "epoch": 5.6334256694367495,
      "grad_norm": 0.8602099418640137,
      "learning_rate": 9.15e-06,
      "loss": 0.052,
      "step": 3053
    },
    {
      "epoch": 5.635272391505079,
      "grad_norm": 1.226857304573059,
      "learning_rate": 9.153e-06,
      "loss": 0.0616,
      "step": 3054
    },
    {
      "epoch": 5.637119113573407,
      "grad_norm": 1.038644552230835,
      "learning_rate": 9.156000000000002e-06,
      "loss": 0.0821,
      "step": 3055
    },
    {
      "epoch": 5.638965835641736,
      "grad_norm": 1.0842891931533813,
      "learning_rate": 9.159e-06,
      "loss": 0.0758,
      "step": 3056
    },
    {
      "epoch": 5.640812557710064,
      "grad_norm": 0.8205846548080444,
      "learning_rate": 9.162e-06,
      "loss": 0.0582,
      "step": 3057
    },
    {
      "epoch": 5.642659279778393,
      "grad_norm": 1.210604190826416,
      "learning_rate": 9.165e-06,
      "loss": 0.0614,
      "step": 3058
    },
    {
      "epoch": 5.644506001846722,
      "grad_norm": 0.8624764084815979,
      "learning_rate": 9.168e-06,
      "loss": 0.0662,
      "step": 3059
    },
    {
      "epoch": 5.646352723915051,
      "grad_norm": 1.716399073600769,
      "learning_rate": 9.171e-06,
      "loss": 0.1065,
      "step": 3060
    },
    {
      "epoch": 5.64819944598338,
      "grad_norm": 1.7060513496398926,
      "learning_rate": 9.174e-06,
      "loss": 0.3817,
      "step": 3061
    },
    {
      "epoch": 5.650046168051708,
      "grad_norm": 1.4844878911972046,
      "learning_rate": 9.177e-06,
      "loss": 0.3893,
      "step": 3062
    },
    {
      "epoch": 5.6518928901200365,
      "grad_norm": 1.199528694152832,
      "learning_rate": 9.18e-06,
      "loss": 0.2969,
      "step": 3063
    },
    {
      "epoch": 5.653739612188366,
      "grad_norm": 1.4310998916625977,
      "learning_rate": 9.183e-06,
      "loss": 0.2899,
      "step": 3064
    },
    {
      "epoch": 5.655586334256695,
      "grad_norm": 1.5166258811950684,
      "learning_rate": 9.186000000000001e-06,
      "loss": 0.363,
      "step": 3065
    },
    {
      "epoch": 5.657433056325023,
      "grad_norm": 1.2674273252487183,
      "learning_rate": 9.189000000000001e-06,
      "loss": 0.2383,
      "step": 3066
    },
    {
      "epoch": 5.659279778393352,
      "grad_norm": 1.0732433795928955,
      "learning_rate": 9.192000000000001e-06,
      "loss": 0.177,
      "step": 3067
    },
    {
      "epoch": 5.66112650046168,
      "grad_norm": 1.4909396171569824,
      "learning_rate": 9.195000000000001e-06,
      "loss": 0.254,
      "step": 3068
    },
    {
      "epoch": 5.66297322253001,
      "grad_norm": 0.8169516324996948,
      "learning_rate": 9.197999999999999e-06,
      "loss": 0.1878,
      "step": 3069
    },
    {
      "epoch": 5.664819944598338,
      "grad_norm": 0.9201124310493469,
      "learning_rate": 9.200999999999999e-06,
      "loss": 0.1467,
      "step": 3070
    },
    {
      "epoch": 5.666666666666667,
      "grad_norm": 2.299650192260742,
      "learning_rate": 9.204e-06,
      "loss": 0.1529,
      "step": 3071
    },
    {
      "epoch": 5.668513388734995,
      "grad_norm": 1.8124525547027588,
      "learning_rate": 9.207e-06,
      "loss": 0.1673,
      "step": 3072
    },
    {
      "epoch": 5.670360110803324,
      "grad_norm": 1.476135015487671,
      "learning_rate": 9.21e-06,
      "loss": 0.2229,
      "step": 3073
    },
    {
      "epoch": 5.672206832871653,
      "grad_norm": 0.6514289379119873,
      "learning_rate": 9.213e-06,
      "loss": 0.0952,
      "step": 3074
    },
    {
      "epoch": 5.674053554939982,
      "grad_norm": 0.6251052618026733,
      "learning_rate": 9.216e-06,
      "loss": 0.0813,
      "step": 3075
    },
    {
      "epoch": 5.67590027700831,
      "grad_norm": 0.7795804738998413,
      "learning_rate": 9.219000000000001e-06,
      "loss": 0.0616,
      "step": 3076
    },
    {
      "epoch": 5.677746999076639,
      "grad_norm": 0.8727070093154907,
      "learning_rate": 9.222e-06,
      "loss": 0.0755,
      "step": 3077
    },
    {
      "epoch": 5.679593721144967,
      "grad_norm": 0.6587457656860352,
      "learning_rate": 9.225e-06,
      "loss": 0.0533,
      "step": 3078
    },
    {
      "epoch": 5.6814404432132966,
      "grad_norm": 0.7225388288497925,
      "learning_rate": 9.228e-06,
      "loss": 0.0547,
      "step": 3079
    },
    {
      "epoch": 5.683287165281625,
      "grad_norm": 2.320605516433716,
      "learning_rate": 9.231e-06,
      "loss": 0.0504,
      "step": 3080
    },
    {
      "epoch": 5.685133887349954,
      "grad_norm": 1.1022893190383911,
      "learning_rate": 9.234e-06,
      "loss": 0.0463,
      "step": 3081
    },
    {
      "epoch": 5.686980609418282,
      "grad_norm": 0.8792228102684021,
      "learning_rate": 9.237e-06,
      "loss": 0.0478,
      "step": 3082
    },
    {
      "epoch": 5.688827331486611,
      "grad_norm": 0.5602882504463196,
      "learning_rate": 9.24e-06,
      "loss": 0.0382,
      "step": 3083
    },
    {
      "epoch": 5.69067405355494,
      "grad_norm": 0.7190514206886292,
      "learning_rate": 9.243e-06,
      "loss": 0.0518,
      "step": 3084
    },
    {
      "epoch": 5.692520775623269,
      "grad_norm": 0.8902875185012817,
      "learning_rate": 9.246e-06,
      "loss": 0.0827,
      "step": 3085
    },
    {
      "epoch": 5.694367497691598,
      "grad_norm": 0.8684836626052856,
      "learning_rate": 9.249e-06,
      "loss": 0.06,
      "step": 3086
    },
    {
      "epoch": 5.696214219759926,
      "grad_norm": 1.0752253532409668,
      "learning_rate": 9.252e-06,
      "loss": 0.0586,
      "step": 3087
    },
    {
      "epoch": 5.698060941828254,
      "grad_norm": 1.1792585849761963,
      "learning_rate": 9.255e-06,
      "loss": 0.132,
      "step": 3088
    },
    {
      "epoch": 5.6999076638965835,
      "grad_norm": 1.2840609550476074,
      "learning_rate": 9.258e-06,
      "loss": 0.1063,
      "step": 3089
    },
    {
      "epoch": 5.701754385964913,
      "grad_norm": 0.7193743586540222,
      "learning_rate": 9.261e-06,
      "loss": 0.0474,
      "step": 3090
    },
    {
      "epoch": 5.703601108033241,
      "grad_norm": 0.9113737344741821,
      "learning_rate": 9.264000000000001e-06,
      "loss": 0.0869,
      "step": 3091
    },
    {
      "epoch": 5.70544783010157,
      "grad_norm": 0.9547917246818542,
      "learning_rate": 9.267000000000001e-06,
      "loss": 0.0489,
      "step": 3092
    },
    {
      "epoch": 5.707294552169898,
      "grad_norm": 3.226956605911255,
      "learning_rate": 9.27e-06,
      "loss": 0.0482,
      "step": 3093
    },
    {
      "epoch": 5.7091412742382275,
      "grad_norm": 0.8785438537597656,
      "learning_rate": 9.272999999999999e-06,
      "loss": 0.0475,
      "step": 3094
    },
    {
      "epoch": 5.710987996306556,
      "grad_norm": 0.9990173578262329,
      "learning_rate": 9.275999999999999e-06,
      "loss": 0.057,
      "step": 3095
    },
    {
      "epoch": 5.712834718374885,
      "grad_norm": 0.8454262018203735,
      "learning_rate": 9.279e-06,
      "loss": 0.0549,
      "step": 3096
    },
    {
      "epoch": 5.714681440443213,
      "grad_norm": 1.1370716094970703,
      "learning_rate": 9.282e-06,
      "loss": 0.0599,
      "step": 3097
    },
    {
      "epoch": 5.716528162511542,
      "grad_norm": 1.1761515140533447,
      "learning_rate": 9.285e-06,
      "loss": 0.0587,
      "step": 3098
    },
    {
      "epoch": 5.7183748845798705,
      "grad_norm": 1.1789077520370483,
      "learning_rate": 9.288e-06,
      "loss": 0.0788,
      "step": 3099
    },
    {
      "epoch": 5.7202216066482,
      "grad_norm": 3.341630458831787,
      "learning_rate": 9.291e-06,
      "loss": 0.0766,
      "step": 3100
    },
    {
      "epoch": 5.722068328716528,
      "grad_norm": 1.213159441947937,
      "learning_rate": 9.294000000000001e-06,
      "loss": 0.0727,
      "step": 3101
    },
    {
      "epoch": 5.723915050784857,
      "grad_norm": 0.9439771175384521,
      "learning_rate": 9.297000000000001e-06,
      "loss": 0.0622,
      "step": 3102
    },
    {
      "epoch": 5.725761772853185,
      "grad_norm": 0.8545637726783752,
      "learning_rate": 9.3e-06,
      "loss": 0.0484,
      "step": 3103
    },
    {
      "epoch": 5.7276084949215145,
      "grad_norm": 0.8875400424003601,
      "learning_rate": 9.303e-06,
      "loss": 0.0614,
      "step": 3104
    },
    {
      "epoch": 5.729455216989843,
      "grad_norm": 1.552079439163208,
      "learning_rate": 9.306e-06,
      "loss": 0.0628,
      "step": 3105
    },
    {
      "epoch": 5.731301939058172,
      "grad_norm": 1.0208044052124023,
      "learning_rate": 9.309e-06,
      "loss": 0.0628,
      "step": 3106
    },
    {
      "epoch": 5.7331486611265,
      "grad_norm": 1.533717155456543,
      "learning_rate": 9.312e-06,
      "loss": 0.0511,
      "step": 3107
    },
    {
      "epoch": 5.734995383194829,
      "grad_norm": 1.1809885501861572,
      "learning_rate": 9.315e-06,
      "loss": 0.0588,
      "step": 3108
    },
    {
      "epoch": 5.7368421052631575,
      "grad_norm": 0.9960464239120483,
      "learning_rate": 9.318e-06,
      "loss": 0.0603,
      "step": 3109
    },
    {
      "epoch": 5.738688827331487,
      "grad_norm": 0.9623493552207947,
      "learning_rate": 9.321e-06,
      "loss": 0.063,
      "step": 3110
    },
    {
      "epoch": 5.740535549399816,
      "grad_norm": 4.995608329772949,
      "learning_rate": 9.324000000000001e-06,
      "loss": 0.4545,
      "step": 3111
    },
    {
      "epoch": 5.742382271468144,
      "grad_norm": 1.443178415298462,
      "learning_rate": 9.327e-06,
      "loss": 0.3614,
      "step": 3112
    },
    {
      "epoch": 5.744228993536472,
      "grad_norm": 3.596264362335205,
      "learning_rate": 9.33e-06,
      "loss": 0.3435,
      "step": 3113
    },
    {
      "epoch": 5.7460757156048015,
      "grad_norm": 2.392505168914795,
      "learning_rate": 9.333e-06,
      "loss": 0.4267,
      "step": 3114
    },
    {
      "epoch": 5.747922437673131,
      "grad_norm": 1.9041658639907837,
      "learning_rate": 9.336e-06,
      "loss": 0.3021,
      "step": 3115
    },
    {
      "epoch": 5.749769159741459,
      "grad_norm": 1.9096932411193848,
      "learning_rate": 9.339000000000002e-06,
      "loss": 0.3205,
      "step": 3116
    },
    {
      "epoch": 5.751615881809788,
      "grad_norm": 3.222970485687256,
      "learning_rate": 9.342000000000001e-06,
      "loss": 0.2452,
      "step": 3117
    },
    {
      "epoch": 5.753462603878116,
      "grad_norm": 1.1893322467803955,
      "learning_rate": 9.345e-06,
      "loss": 0.2122,
      "step": 3118
    },
    {
      "epoch": 5.755309325946445,
      "grad_norm": 1.0547327995300293,
      "learning_rate": 9.348e-06,
      "loss": 0.2251,
      "step": 3119
    },
    {
      "epoch": 5.757156048014774,
      "grad_norm": 0.8825430274009705,
      "learning_rate": 9.350999999999999e-06,
      "loss": 0.1535,
      "step": 3120
    },
    {
      "epoch": 5.759002770083103,
      "grad_norm": 0.8509583473205566,
      "learning_rate": 9.354e-06,
      "loss": 0.1839,
      "step": 3121
    },
    {
      "epoch": 5.760849492151431,
      "grad_norm": 0.9790893793106079,
      "learning_rate": 9.357e-06,
      "loss": 0.1476,
      "step": 3122
    },
    {
      "epoch": 5.76269621421976,
      "grad_norm": 0.9421710968017578,
      "learning_rate": 9.36e-06,
      "loss": 0.1617,
      "step": 3123
    },
    {
      "epoch": 5.7645429362880884,
      "grad_norm": 0.9089346528053284,
      "learning_rate": 9.363e-06,
      "loss": 0.1796,
      "step": 3124
    },
    {
      "epoch": 5.766389658356418,
      "grad_norm": 1.144519567489624,
      "learning_rate": 9.366e-06,
      "loss": 0.0917,
      "step": 3125
    },
    {
      "epoch": 5.768236380424746,
      "grad_norm": 0.8784078359603882,
      "learning_rate": 9.369000000000001e-06,
      "loss": 0.0894,
      "step": 3126
    },
    {
      "epoch": 5.770083102493075,
      "grad_norm": 1.1179689168930054,
      "learning_rate": 9.372000000000001e-06,
      "loss": 0.0808,
      "step": 3127
    },
    {
      "epoch": 5.771929824561403,
      "grad_norm": 0.7009519338607788,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.0506,
      "step": 3128
    },
    {
      "epoch": 5.773776546629732,
      "grad_norm": 0.6712589263916016,
      "learning_rate": 9.378e-06,
      "loss": 0.0589,
      "step": 3129
    },
    {
      "epoch": 5.775623268698061,
      "grad_norm": 0.6814177632331848,
      "learning_rate": 9.380999999999999e-06,
      "loss": 0.0581,
      "step": 3130
    },
    {
      "epoch": 5.77746999076639,
      "grad_norm": 0.7124582529067993,
      "learning_rate": 9.384e-06,
      "loss": 0.0508,
      "step": 3131
    },
    {
      "epoch": 5.779316712834718,
      "grad_norm": 0.8895642161369324,
      "learning_rate": 9.387e-06,
      "loss": 0.0597,
      "step": 3132
    },
    {
      "epoch": 5.781163434903047,
      "grad_norm": 0.9447484016418457,
      "learning_rate": 9.39e-06,
      "loss": 0.0638,
      "step": 3133
    },
    {
      "epoch": 5.783010156971375,
      "grad_norm": 0.7462688684463501,
      "learning_rate": 9.393e-06,
      "loss": 0.0559,
      "step": 3134
    },
    {
      "epoch": 5.784856879039705,
      "grad_norm": 0.95284503698349,
      "learning_rate": 9.396e-06,
      "loss": 0.0563,
      "step": 3135
    },
    {
      "epoch": 5.786703601108034,
      "grad_norm": 0.926132321357727,
      "learning_rate": 9.399000000000001e-06,
      "loss": 0.0669,
      "step": 3136
    },
    {
      "epoch": 5.788550323176362,
      "grad_norm": 1.1923774480819702,
      "learning_rate": 9.402e-06,
      "loss": 0.0613,
      "step": 3137
    },
    {
      "epoch": 5.79039704524469,
      "grad_norm": 0.9798671007156372,
      "learning_rate": 9.405e-06,
      "loss": 0.0487,
      "step": 3138
    },
    {
      "epoch": 5.792243767313019,
      "grad_norm": 0.7260217666625977,
      "learning_rate": 9.408e-06,
      "loss": 0.0465,
      "step": 3139
    },
    {
      "epoch": 5.7940904893813485,
      "grad_norm": 0.6956988573074341,
      "learning_rate": 9.411e-06,
      "loss": 0.0556,
      "step": 3140
    },
    {
      "epoch": 5.795937211449677,
      "grad_norm": 0.7585822939872742,
      "learning_rate": 9.414000000000002e-06,
      "loss": 0.05,
      "step": 3141
    },
    {
      "epoch": 5.797783933518006,
      "grad_norm": 0.8594281077384949,
      "learning_rate": 9.417e-06,
      "loss": 0.067,
      "step": 3142
    },
    {
      "epoch": 5.799630655586334,
      "grad_norm": 0.8605832457542419,
      "learning_rate": 9.42e-06,
      "loss": 0.0506,
      "step": 3143
    },
    {
      "epoch": 5.801477377654663,
      "grad_norm": 0.8311311602592468,
      "learning_rate": 9.423e-06,
      "loss": 0.0428,
      "step": 3144
    },
    {
      "epoch": 5.803324099722992,
      "grad_norm": 0.9315934181213379,
      "learning_rate": 9.426e-06,
      "loss": 0.0493,
      "step": 3145
    },
    {
      "epoch": 5.805170821791321,
      "grad_norm": 1.0022386312484741,
      "learning_rate": 9.429e-06,
      "loss": 0.0762,
      "step": 3146
    },
    {
      "epoch": 5.807017543859649,
      "grad_norm": 0.9364701509475708,
      "learning_rate": 9.432e-06,
      "loss": 0.0718,
      "step": 3147
    },
    {
      "epoch": 5.808864265927978,
      "grad_norm": 1.5895075798034668,
      "learning_rate": 9.435e-06,
      "loss": 0.0597,
      "step": 3148
    },
    {
      "epoch": 5.810710987996306,
      "grad_norm": 0.939860999584198,
      "learning_rate": 9.438e-06,
      "loss": 0.0628,
      "step": 3149
    },
    {
      "epoch": 5.8125577100646355,
      "grad_norm": 1.6381874084472656,
      "learning_rate": 9.441e-06,
      "loss": 0.0701,
      "step": 3150
    },
    {
      "epoch": 5.814404432132964,
      "grad_norm": 1.853114128112793,
      "learning_rate": 9.444000000000001e-06,
      "loss": 0.0668,
      "step": 3151
    },
    {
      "epoch": 5.816251154201293,
      "grad_norm": 1.0566539764404297,
      "learning_rate": 9.447000000000001e-06,
      "loss": 0.0642,
      "step": 3152
    },
    {
      "epoch": 5.818097876269621,
      "grad_norm": 0.9682995676994324,
      "learning_rate": 9.450000000000001e-06,
      "loss": 0.0559,
      "step": 3153
    },
    {
      "epoch": 5.81994459833795,
      "grad_norm": 0.8305920362472534,
      "learning_rate": 9.453e-06,
      "loss": 0.0639,
      "step": 3154
    },
    {
      "epoch": 5.821791320406279,
      "grad_norm": 0.718386173248291,
      "learning_rate": 9.455999999999999e-06,
      "loss": 0.0377,
      "step": 3155
    },
    {
      "epoch": 5.823638042474608,
      "grad_norm": 0.9657241702079773,
      "learning_rate": 9.459e-06,
      "loss": 0.0737,
      "step": 3156
    },
    {
      "epoch": 5.825484764542936,
      "grad_norm": 0.8440473079681396,
      "learning_rate": 9.462e-06,
      "loss": 0.0526,
      "step": 3157
    },
    {
      "epoch": 5.827331486611265,
      "grad_norm": 1.2559155225753784,
      "learning_rate": 9.465e-06,
      "loss": 0.0827,
      "step": 3158
    },
    {
      "epoch": 5.829178208679593,
      "grad_norm": 1.968029499053955,
      "learning_rate": 9.468e-06,
      "loss": 0.076,
      "step": 3159
    },
    {
      "epoch": 5.8310249307479225,
      "grad_norm": 1.369443655014038,
      "learning_rate": 9.471e-06,
      "loss": 0.087,
      "step": 3160
    },
    {
      "epoch": 5.832871652816252,
      "grad_norm": 2.196585178375244,
      "learning_rate": 9.474000000000001e-06,
      "loss": 0.4043,
      "step": 3161
    },
    {
      "epoch": 5.83471837488458,
      "grad_norm": 1.450124740600586,
      "learning_rate": 9.477000000000001e-06,
      "loss": 0.3129,
      "step": 3162
    },
    {
      "epoch": 5.836565096952908,
      "grad_norm": 1.131670594215393,
      "learning_rate": 9.48e-06,
      "loss": 0.3256,
      "step": 3163
    },
    {
      "epoch": 5.838411819021237,
      "grad_norm": 1.5835055112838745,
      "learning_rate": 9.483e-06,
      "loss": 0.2832,
      "step": 3164
    },
    {
      "epoch": 5.840258541089566,
      "grad_norm": 1.0356972217559814,
      "learning_rate": 9.486e-06,
      "loss": 0.2298,
      "step": 3165
    },
    {
      "epoch": 5.842105263157895,
      "grad_norm": 1.202141523361206,
      "learning_rate": 9.489000000000002e-06,
      "loss": 0.3123,
      "step": 3166
    },
    {
      "epoch": 5.843951985226224,
      "grad_norm": 1.4132212400436401,
      "learning_rate": 9.492e-06,
      "loss": 0.2334,
      "step": 3167
    },
    {
      "epoch": 5.845798707294552,
      "grad_norm": 1.1922186613082886,
      "learning_rate": 9.495e-06,
      "loss": 0.1989,
      "step": 3168
    },
    {
      "epoch": 5.847645429362881,
      "grad_norm": 1.1139531135559082,
      "learning_rate": 9.498e-06,
      "loss": 0.1859,
      "step": 3169
    },
    {
      "epoch": 5.8494921514312095,
      "grad_norm": 1.1097437143325806,
      "learning_rate": 9.501e-06,
      "loss": 0.1994,
      "step": 3170
    },
    {
      "epoch": 5.851338873499539,
      "grad_norm": 1.480333685874939,
      "learning_rate": 9.504e-06,
      "loss": 0.2172,
      "step": 3171
    },
    {
      "epoch": 5.853185595567867,
      "grad_norm": 1.7595967054367065,
      "learning_rate": 9.507e-06,
      "loss": 0.1161,
      "step": 3172
    },
    {
      "epoch": 5.855032317636196,
      "grad_norm": 0.9520134329795837,
      "learning_rate": 9.51e-06,
      "loss": 0.1698,
      "step": 3173
    },
    {
      "epoch": 5.856879039704524,
      "grad_norm": 0.9602885246276855,
      "learning_rate": 9.513e-06,
      "loss": 0.1021,
      "step": 3174
    },
    {
      "epoch": 5.858725761772853,
      "grad_norm": 1.421912670135498,
      "learning_rate": 9.516e-06,
      "loss": 0.11,
      "step": 3175
    },
    {
      "epoch": 5.860572483841182,
      "grad_norm": 0.5686153769493103,
      "learning_rate": 9.519000000000002e-06,
      "loss": 0.0552,
      "step": 3176
    },
    {
      "epoch": 5.862419205909511,
      "grad_norm": 0.7969242334365845,
      "learning_rate": 9.522000000000001e-06,
      "loss": 0.0772,
      "step": 3177
    },
    {
      "epoch": 5.864265927977839,
      "grad_norm": 0.5964579582214355,
      "learning_rate": 9.525000000000001e-06,
      "loss": 0.055,
      "step": 3178
    },
    {
      "epoch": 5.866112650046168,
      "grad_norm": 1.263533353805542,
      "learning_rate": 9.528e-06,
      "loss": 0.079,
      "step": 3179
    },
    {
      "epoch": 5.8679593721144965,
      "grad_norm": 1.914730191230774,
      "learning_rate": 9.530999999999999e-06,
      "loss": 0.0612,
      "step": 3180
    },
    {
      "epoch": 5.869806094182826,
      "grad_norm": 0.6636876463890076,
      "learning_rate": 9.534e-06,
      "loss": 0.051,
      "step": 3181
    },
    {
      "epoch": 5.871652816251154,
      "grad_norm": 1.0478947162628174,
      "learning_rate": 9.537e-06,
      "loss": 0.0555,
      "step": 3182
    },
    {
      "epoch": 5.873499538319483,
      "grad_norm": 0.65780109167099,
      "learning_rate": 9.54e-06,
      "loss": 0.0431,
      "step": 3183
    },
    {
      "epoch": 5.875346260387811,
      "grad_norm": 0.979624330997467,
      "learning_rate": 9.543e-06,
      "loss": 0.0671,
      "step": 3184
    },
    {
      "epoch": 5.87719298245614,
      "grad_norm": 0.8512119054794312,
      "learning_rate": 9.546e-06,
      "loss": 0.0714,
      "step": 3185
    },
    {
      "epoch": 5.87903970452447,
      "grad_norm": 0.7912101745605469,
      "learning_rate": 9.549000000000001e-06,
      "loss": 0.0641,
      "step": 3186
    },
    {
      "epoch": 5.880886426592798,
      "grad_norm": 0.670253574848175,
      "learning_rate": 9.552000000000001e-06,
      "loss": 0.0504,
      "step": 3187
    },
    {
      "epoch": 5.882733148661126,
      "grad_norm": 0.7148275971412659,
      "learning_rate": 9.555e-06,
      "loss": 0.0643,
      "step": 3188
    },
    {
      "epoch": 5.884579870729455,
      "grad_norm": 0.9922288060188293,
      "learning_rate": 9.558e-06,
      "loss": 0.0554,
      "step": 3189
    },
    {
      "epoch": 5.886426592797784,
      "grad_norm": 0.6634912490844727,
      "learning_rate": 9.561e-06,
      "loss": 0.0611,
      "step": 3190
    },
    {
      "epoch": 5.888273314866113,
      "grad_norm": 1.007731556892395,
      "learning_rate": 9.564e-06,
      "loss": 0.0503,
      "step": 3191
    },
    {
      "epoch": 5.890120036934442,
      "grad_norm": 0.8372678160667419,
      "learning_rate": 9.567e-06,
      "loss": 0.0423,
      "step": 3192
    },
    {
      "epoch": 5.89196675900277,
      "grad_norm": 0.6674429774284363,
      "learning_rate": 9.57e-06,
      "loss": 0.0487,
      "step": 3193
    },
    {
      "epoch": 5.893813481071099,
      "grad_norm": 0.8218016624450684,
      "learning_rate": 9.573e-06,
      "loss": 0.0496,
      "step": 3194
    },
    {
      "epoch": 5.895660203139427,
      "grad_norm": 0.6602591872215271,
      "learning_rate": 9.576e-06,
      "loss": 0.0436,
      "step": 3195
    },
    {
      "epoch": 5.8975069252077565,
      "grad_norm": 0.7583157420158386,
      "learning_rate": 9.579e-06,
      "loss": 0.0731,
      "step": 3196
    },
    {
      "epoch": 5.899353647276085,
      "grad_norm": 0.7678322792053223,
      "learning_rate": 9.582e-06,
      "loss": 0.0645,
      "step": 3197
    },
    {
      "epoch": 5.901200369344414,
      "grad_norm": 1.025427222251892,
      "learning_rate": 9.585e-06,
      "loss": 0.0662,
      "step": 3198
    },
    {
      "epoch": 5.903047091412742,
      "grad_norm": 0.934942901134491,
      "learning_rate": 9.588e-06,
      "loss": 0.0684,
      "step": 3199
    },
    {
      "epoch": 5.904893813481071,
      "grad_norm": 1.957726001739502,
      "learning_rate": 9.591e-06,
      "loss": 0.0433,
      "step": 3200
    },
    {
      "epoch": 5.9067405355494,
      "grad_norm": 0.7961621880531311,
      "learning_rate": 9.594e-06,
      "loss": 0.0569,
      "step": 3201
    },
    {
      "epoch": 5.908587257617729,
      "grad_norm": 1.2947304248809814,
      "learning_rate": 9.597000000000001e-06,
      "loss": 0.0738,
      "step": 3202
    },
    {
      "epoch": 5.910433979686057,
      "grad_norm": 1.006197214126587,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.0679,
      "step": 3203
    },
    {
      "epoch": 5.912280701754386,
      "grad_norm": 0.720492422580719,
      "learning_rate": 9.603e-06,
      "loss": 0.0399,
      "step": 3204
    },
    {
      "epoch": 5.914127423822714,
      "grad_norm": 0.9666870832443237,
      "learning_rate": 9.606e-06,
      "loss": 0.0717,
      "step": 3205
    },
    {
      "epoch": 5.9159741458910435,
      "grad_norm": 1.069618821144104,
      "learning_rate": 9.608999999999999e-06,
      "loss": 0.0657,
      "step": 3206
    },
    {
      "epoch": 5.917820867959372,
      "grad_norm": 0.7088072299957275,
      "learning_rate": 9.612e-06,
      "loss": 0.0436,
      "step": 3207
    },
    {
      "epoch": 5.919667590027701,
      "grad_norm": 0.7209672331809998,
      "learning_rate": 9.615e-06,
      "loss": 0.0622,
      "step": 3208
    },
    {
      "epoch": 5.921514312096029,
      "grad_norm": 1.0887486934661865,
      "learning_rate": 9.618e-06,
      "loss": 0.0417,
      "step": 3209
    },
    {
      "epoch": 5.923361034164358,
      "grad_norm": 1.448907494544983,
      "learning_rate": 9.621e-06,
      "loss": 0.1023,
      "step": 3210
    },
    {
      "epoch": 5.9252077562326875,
      "grad_norm": 2.245453119277954,
      "learning_rate": 9.624e-06,
      "loss": 0.3862,
      "step": 3211
    },
    {
      "epoch": 5.927054478301016,
      "grad_norm": 1.4048337936401367,
      "learning_rate": 9.627000000000001e-06,
      "loss": 0.3284,
      "step": 3212
    },
    {
      "epoch": 5.928901200369344,
      "grad_norm": 1.3502554893493652,
      "learning_rate": 9.630000000000001e-06,
      "loss": 0.2569,
      "step": 3213
    },
    {
      "epoch": 5.930747922437673,
      "grad_norm": 1.6020257472991943,
      "learning_rate": 9.633e-06,
      "loss": 0.3036,
      "step": 3214
    },
    {
      "epoch": 5.932594644506002,
      "grad_norm": 1.1737459897994995,
      "learning_rate": 9.636e-06,
      "loss": 0.2589,
      "step": 3215
    },
    {
      "epoch": 5.9344413665743305,
      "grad_norm": 1.5264931917190552,
      "learning_rate": 9.638999999999999e-06,
      "loss": 0.1982,
      "step": 3216
    },
    {
      "epoch": 5.93628808864266,
      "grad_norm": 1.0091787576675415,
      "learning_rate": 9.642e-06,
      "loss": 0.241,
      "step": 3217
    },
    {
      "epoch": 5.938134810710988,
      "grad_norm": 1.1220636367797852,
      "learning_rate": 9.645e-06,
      "loss": 0.2123,
      "step": 3218
    },
    {
      "epoch": 5.939981532779317,
      "grad_norm": 1.072777271270752,
      "learning_rate": 9.648e-06,
      "loss": 0.155,
      "step": 3219
    },
    {
      "epoch": 5.941828254847645,
      "grad_norm": 1.2503160238265991,
      "learning_rate": 9.651e-06,
      "loss": 0.2424,
      "step": 3220
    },
    {
      "epoch": 5.9436749769159745,
      "grad_norm": 2.789108991622925,
      "learning_rate": 9.654e-06,
      "loss": 0.1637,
      "step": 3221
    },
    {
      "epoch": 5.945521698984303,
      "grad_norm": 0.9609374403953552,
      "learning_rate": 9.657000000000001e-06,
      "loss": 0.1563,
      "step": 3222
    },
    {
      "epoch": 5.947368421052632,
      "grad_norm": 3.0387258529663086,
      "learning_rate": 9.66e-06,
      "loss": 0.0993,
      "step": 3223
    },
    {
      "epoch": 5.94921514312096,
      "grad_norm": 0.9588826298713684,
      "learning_rate": 9.663e-06,
      "loss": 0.0646,
      "step": 3224
    },
    {
      "epoch": 5.951061865189289,
      "grad_norm": 1.6367828845977783,
      "learning_rate": 9.666e-06,
      "loss": 0.073,
      "step": 3225
    },
    {
      "epoch": 5.9529085872576175,
      "grad_norm": 0.6142951250076294,
      "learning_rate": 9.669e-06,
      "loss": 0.0714,
      "step": 3226
    },
    {
      "epoch": 5.954755309325947,
      "grad_norm": 1.2287521362304688,
      "learning_rate": 9.672000000000002e-06,
      "loss": 0.11,
      "step": 3227
    },
    {
      "epoch": 5.956602031394275,
      "grad_norm": 1.094412088394165,
      "learning_rate": 9.675e-06,
      "loss": 0.0484,
      "step": 3228
    },
    {
      "epoch": 5.958448753462604,
      "grad_norm": 0.4351859986782074,
      "learning_rate": 9.678e-06,
      "loss": 0.0596,
      "step": 3229
    },
    {
      "epoch": 5.960295475530932,
      "grad_norm": 0.7879501581192017,
      "learning_rate": 9.681e-06,
      "loss": 0.0579,
      "step": 3230
    },
    {
      "epoch": 5.9621421975992615,
      "grad_norm": 0.7965291142463684,
      "learning_rate": 9.683999999999999e-06,
      "loss": 0.0707,
      "step": 3231
    },
    {
      "epoch": 5.96398891966759,
      "grad_norm": 0.8367858529090881,
      "learning_rate": 9.687e-06,
      "loss": 0.0547,
      "step": 3232
    },
    {
      "epoch": 5.965835641735919,
      "grad_norm": 0.7109565138816833,
      "learning_rate": 9.69e-06,
      "loss": 0.0538,
      "step": 3233
    },
    {
      "epoch": 5.967682363804247,
      "grad_norm": 0.7826809287071228,
      "learning_rate": 9.693e-06,
      "loss": 0.0403,
      "step": 3234
    },
    {
      "epoch": 5.969529085872576,
      "grad_norm": 0.9229626655578613,
      "learning_rate": 9.696e-06,
      "loss": 0.1065,
      "step": 3235
    },
    {
      "epoch": 5.971375807940905,
      "grad_norm": 0.7484771013259888,
      "learning_rate": 9.699e-06,
      "loss": 0.0458,
      "step": 3236
    },
    {
      "epoch": 5.973222530009234,
      "grad_norm": 0.6975681781768799,
      "learning_rate": 9.702000000000001e-06,
      "loss": 0.0373,
      "step": 3237
    },
    {
      "epoch": 5.975069252077562,
      "grad_norm": 0.6109060645103455,
      "learning_rate": 9.705000000000001e-06,
      "loss": 0.0515,
      "step": 3238
    },
    {
      "epoch": 5.976915974145891,
      "grad_norm": 1.058773159980774,
      "learning_rate": 9.708000000000001e-06,
      "loss": 0.0931,
      "step": 3239
    },
    {
      "epoch": 5.97876269621422,
      "grad_norm": 1.0566517114639282,
      "learning_rate": 9.711e-06,
      "loss": 0.0656,
      "step": 3240
    },
    {
      "epoch": 5.980609418282548,
      "grad_norm": 1.0419882535934448,
      "learning_rate": 9.713999999999999e-06,
      "loss": 0.1314,
      "step": 3241
    },
    {
      "epoch": 5.982456140350877,
      "grad_norm": 1.0443799495697021,
      "learning_rate": 9.717e-06,
      "loss": 0.0758,
      "step": 3242
    },
    {
      "epoch": 5.984302862419206,
      "grad_norm": 0.9809887409210205,
      "learning_rate": 9.72e-06,
      "loss": 0.061,
      "step": 3243
    },
    {
      "epoch": 5.986149584487535,
      "grad_norm": 0.9536335468292236,
      "learning_rate": 9.723e-06,
      "loss": 0.0615,
      "step": 3244
    },
    {
      "epoch": 5.987996306555863,
      "grad_norm": 1.8318169116973877,
      "learning_rate": 9.726e-06,
      "loss": 0.0542,
      "step": 3245
    },
    {
      "epoch": 5.989843028624192,
      "grad_norm": 1.001519799232483,
      "learning_rate": 9.729e-06,
      "loss": 0.0761,
      "step": 3246
    },
    {
      "epoch": 5.991689750692521,
      "grad_norm": 0.8341793417930603,
      "learning_rate": 9.732000000000001e-06,
      "loss": 0.0454,
      "step": 3247
    },
    {
      "epoch": 5.99353647276085,
      "grad_norm": 1.3690731525421143,
      "learning_rate": 9.735e-06,
      "loss": 0.0554,
      "step": 3248
    },
    {
      "epoch": 5.995383194829178,
      "grad_norm": 1.326370120048523,
      "learning_rate": 9.738e-06,
      "loss": 0.1149,
      "step": 3249
    },
    {
      "epoch": 5.997229916897507,
      "grad_norm": 1.7953455448150635,
      "learning_rate": 9.741e-06,
      "loss": 0.1143,
      "step": 3250
    },
    {
      "epoch": 5.999076638965835,
      "grad_norm": 1.6124578714370728,
      "learning_rate": 9.744e-06,
      "loss": 0.0628,
      "step": 3251
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.8392724394798279,
      "learning_rate": 9.747000000000002e-06,
      "loss": 0.0372,
      "step": 3252
    },
    {
      "epoch": 6.001846722068329,
      "grad_norm": 4.04020881652832,
      "learning_rate": 9.75e-06,
      "loss": 0.4651,
      "step": 3253
    },
    {
      "epoch": 6.003693444136657,
      "grad_norm": 1.0397669076919556,
      "learning_rate": 9.753e-06,
      "loss": 0.309,
      "step": 3254
    },
    {
      "epoch": 6.0055401662049865,
      "grad_norm": 0.9423261284828186,
      "learning_rate": 9.756e-06,
      "loss": 0.2448,
      "step": 3255
    },
    {
      "epoch": 6.007386888273315,
      "grad_norm": 1.156479001045227,
      "learning_rate": 9.759e-06,
      "loss": 0.2676,
      "step": 3256
    },
    {
      "epoch": 6.009233610341644,
      "grad_norm": 1.2824240922927856,
      "learning_rate": 9.762e-06,
      "loss": 0.2602,
      "step": 3257
    },
    {
      "epoch": 6.011080332409972,
      "grad_norm": 1.766750693321228,
      "learning_rate": 9.765e-06,
      "loss": 0.2751,
      "step": 3258
    },
    {
      "epoch": 6.012927054478301,
      "grad_norm": 1.0178364515304565,
      "learning_rate": 9.768e-06,
      "loss": 0.2036,
      "step": 3259
    },
    {
      "epoch": 6.01477377654663,
      "grad_norm": 1.1608028411865234,
      "learning_rate": 9.771e-06,
      "loss": 0.2316,
      "step": 3260
    },
    {
      "epoch": 6.016620498614959,
      "grad_norm": 0.907696545124054,
      "learning_rate": 9.774e-06,
      "loss": 0.1741,
      "step": 3261
    },
    {
      "epoch": 6.018467220683287,
      "grad_norm": 0.8990158438682556,
      "learning_rate": 9.777000000000001e-06,
      "loss": 0.1648,
      "step": 3262
    },
    {
      "epoch": 6.020313942751616,
      "grad_norm": 3.445403575897217,
      "learning_rate": 9.780000000000001e-06,
      "loss": 0.1243,
      "step": 3263
    },
    {
      "epoch": 6.022160664819944,
      "grad_norm": 1.2290663719177246,
      "learning_rate": 9.783000000000001e-06,
      "loss": 0.1855,
      "step": 3264
    },
    {
      "epoch": 6.0240073868882735,
      "grad_norm": 1.3456010818481445,
      "learning_rate": 9.785999999999999e-06,
      "loss": 0.1791,
      "step": 3265
    },
    {
      "epoch": 6.025854108956602,
      "grad_norm": 0.5856161713600159,
      "learning_rate": 9.788999999999999e-06,
      "loss": 0.07,
      "step": 3266
    },
    {
      "epoch": 6.027700831024931,
      "grad_norm": 1.00753915309906,
      "learning_rate": 9.792e-06,
      "loss": 0.0652,
      "step": 3267
    },
    {
      "epoch": 6.029547553093259,
      "grad_norm": 2.3274526596069336,
      "learning_rate": 9.795e-06,
      "loss": 0.1259,
      "step": 3268
    },
    {
      "epoch": 6.031394275161588,
      "grad_norm": 0.6562830209732056,
      "learning_rate": 9.798e-06,
      "loss": 0.0732,
      "step": 3269
    },
    {
      "epoch": 6.033240997229917,
      "grad_norm": 0.6230347156524658,
      "learning_rate": 9.801e-06,
      "loss": 0.0442,
      "step": 3270
    },
    {
      "epoch": 6.035087719298246,
      "grad_norm": 1.4482897520065308,
      "learning_rate": 9.804e-06,
      "loss": 0.0849,
      "step": 3271
    },
    {
      "epoch": 6.036934441366574,
      "grad_norm": 0.9868390560150146,
      "learning_rate": 9.807000000000001e-06,
      "loss": 0.1102,
      "step": 3272
    },
    {
      "epoch": 6.038781163434903,
      "grad_norm": 0.7521336078643799,
      "learning_rate": 9.810000000000001e-06,
      "loss": 0.0647,
      "step": 3273
    },
    {
      "epoch": 6.040627885503231,
      "grad_norm": 0.8148708343505859,
      "learning_rate": 9.813e-06,
      "loss": 0.0647,
      "step": 3274
    },
    {
      "epoch": 6.0424746075715605,
      "grad_norm": 0.6797977089881897,
      "learning_rate": 9.816e-06,
      "loss": 0.0503,
      "step": 3275
    },
    {
      "epoch": 6.044321329639889,
      "grad_norm": 0.947433352470398,
      "learning_rate": 9.819e-06,
      "loss": 0.0674,
      "step": 3276
    },
    {
      "epoch": 6.046168051708218,
      "grad_norm": 0.7319151163101196,
      "learning_rate": 9.822e-06,
      "loss": 0.0542,
      "step": 3277
    },
    {
      "epoch": 6.048014773776547,
      "grad_norm": 0.9400285482406616,
      "learning_rate": 9.825e-06,
      "loss": 0.0562,
      "step": 3278
    },
    {
      "epoch": 6.049861495844875,
      "grad_norm": 0.8096341490745544,
      "learning_rate": 9.828e-06,
      "loss": 0.0659,
      "step": 3279
    },
    {
      "epoch": 6.0517082179132045,
      "grad_norm": 0.7864410877227783,
      "learning_rate": 9.831e-06,
      "loss": 0.0398,
      "step": 3280
    },
    {
      "epoch": 6.053554939981533,
      "grad_norm": 0.5510549545288086,
      "learning_rate": 9.834e-06,
      "loss": 0.032,
      "step": 3281
    },
    {
      "epoch": 6.055401662049862,
      "grad_norm": 0.7736632823944092,
      "learning_rate": 9.837000000000001e-06,
      "loss": 0.0455,
      "step": 3282
    },
    {
      "epoch": 6.05724838411819,
      "grad_norm": 0.6081624031066895,
      "learning_rate": 9.84e-06,
      "loss": 0.0418,
      "step": 3283
    },
    {
      "epoch": 6.059095106186519,
      "grad_norm": 0.4802582263946533,
      "learning_rate": 9.843e-06,
      "loss": 0.0372,
      "step": 3284
    },
    {
      "epoch": 6.0609418282548475,
      "grad_norm": 0.7252798080444336,
      "learning_rate": 9.846e-06,
      "loss": 0.0533,
      "step": 3285
    },
    {
      "epoch": 6.062788550323177,
      "grad_norm": 1.2600115537643433,
      "learning_rate": 9.849e-06,
      "loss": 0.0556,
      "step": 3286
    },
    {
      "epoch": 6.064635272391505,
      "grad_norm": 0.8759087324142456,
      "learning_rate": 9.852000000000002e-06,
      "loss": 0.0525,
      "step": 3287
    },
    {
      "epoch": 6.066481994459834,
      "grad_norm": 0.804162323474884,
      "learning_rate": 9.855000000000001e-06,
      "loss": 0.0419,
      "step": 3288
    },
    {
      "epoch": 6.068328716528162,
      "grad_norm": 1.3809750080108643,
      "learning_rate": 9.858000000000001e-06,
      "loss": 0.0605,
      "step": 3289
    },
    {
      "epoch": 6.0701754385964914,
      "grad_norm": 0.7179027199745178,
      "learning_rate": 9.861e-06,
      "loss": 0.0287,
      "step": 3290
    },
    {
      "epoch": 6.07202216066482,
      "grad_norm": 0.9235922694206238,
      "learning_rate": 9.863999999999999e-06,
      "loss": 0.0409,
      "step": 3291
    },
    {
      "epoch": 6.073868882733149,
      "grad_norm": 0.9280220866203308,
      "learning_rate": 9.867e-06,
      "loss": 0.0469,
      "step": 3292
    },
    {
      "epoch": 6.075715604801477,
      "grad_norm": 0.7803265452384949,
      "learning_rate": 9.87e-06,
      "loss": 0.0495,
      "step": 3293
    },
    {
      "epoch": 6.077562326869806,
      "grad_norm": 1.1641448736190796,
      "learning_rate": 9.873e-06,
      "loss": 0.0719,
      "step": 3294
    },
    {
      "epoch": 6.0794090489381345,
      "grad_norm": 1.104889988899231,
      "learning_rate": 9.876e-06,
      "loss": 0.0526,
      "step": 3295
    },
    {
      "epoch": 6.081255771006464,
      "grad_norm": 0.7316168546676636,
      "learning_rate": 9.879e-06,
      "loss": 0.0464,
      "step": 3296
    },
    {
      "epoch": 6.083102493074792,
      "grad_norm": 0.9376199841499329,
      "learning_rate": 9.882000000000001e-06,
      "loss": 0.0445,
      "step": 3297
    },
    {
      "epoch": 6.084949215143121,
      "grad_norm": 1.3941532373428345,
      "learning_rate": 9.885000000000001e-06,
      "loss": 0.0469,
      "step": 3298
    },
    {
      "epoch": 6.086795937211449,
      "grad_norm": 1.170428991317749,
      "learning_rate": 9.888000000000001e-06,
      "loss": 0.0497,
      "step": 3299
    },
    {
      "epoch": 6.088642659279778,
      "grad_norm": 1.5209548473358154,
      "learning_rate": 9.891e-06,
      "loss": 0.0568,
      "step": 3300
    },
    {
      "epoch": 6.090489381348107,
      "grad_norm": 1.155193567276001,
      "learning_rate": 9.894e-06,
      "loss": 0.0565,
      "step": 3301
    },
    {
      "epoch": 6.092336103416436,
      "grad_norm": 2.590714454650879,
      "learning_rate": 9.897e-06,
      "loss": 0.0667,
      "step": 3302
    },
    {
      "epoch": 6.094182825484765,
      "grad_norm": 3.278620719909668,
      "learning_rate": 9.9e-06,
      "loss": 0.3806,
      "step": 3303
    },
    {
      "epoch": 6.096029547553093,
      "grad_norm": 1.313486933708191,
      "learning_rate": 9.903e-06,
      "loss": 0.2242,
      "step": 3304
    },
    {
      "epoch": 6.097876269621422,
      "grad_norm": 1.0118507146835327,
      "learning_rate": 9.906e-06,
      "loss": 0.2646,
      "step": 3305
    },
    {
      "epoch": 6.099722991689751,
      "grad_norm": 1.1909503936767578,
      "learning_rate": 9.909e-06,
      "loss": 0.2251,
      "step": 3306
    },
    {
      "epoch": 6.10156971375808,
      "grad_norm": 0.9937542080879211,
      "learning_rate": 9.912000000000001e-06,
      "loss": 0.3316,
      "step": 3307
    },
    {
      "epoch": 6.103416435826408,
      "grad_norm": 1.0816807746887207,
      "learning_rate": 9.915e-06,
      "loss": 0.2589,
      "step": 3308
    },
    {
      "epoch": 6.105263157894737,
      "grad_norm": 1.0621750354766846,
      "learning_rate": 9.918e-06,
      "loss": 0.1592,
      "step": 3309
    },
    {
      "epoch": 6.107109879963065,
      "grad_norm": 1.1883070468902588,
      "learning_rate": 9.921e-06,
      "loss": 0.1717,
      "step": 3310
    },
    {
      "epoch": 6.108956602031395,
      "grad_norm": 0.8285554051399231,
      "learning_rate": 9.924e-06,
      "loss": 0.1371,
      "step": 3311
    },
    {
      "epoch": 6.110803324099723,
      "grad_norm": 0.8814343810081482,
      "learning_rate": 9.927000000000002e-06,
      "loss": 0.1727,
      "step": 3312
    },
    {
      "epoch": 6.112650046168052,
      "grad_norm": 1.0420233011245728,
      "learning_rate": 9.930000000000001e-06,
      "loss": 0.1265,
      "step": 3313
    },
    {
      "epoch": 6.11449676823638,
      "grad_norm": 2.53212833404541,
      "learning_rate": 9.933e-06,
      "loss": 0.1131,
      "step": 3314
    },
    {
      "epoch": 6.116343490304709,
      "grad_norm": 1.1509301662445068,
      "learning_rate": 9.936e-06,
      "loss": 0.116,
      "step": 3315
    },
    {
      "epoch": 6.118190212373038,
      "grad_norm": 1.1868430376052856,
      "learning_rate": 9.939e-06,
      "loss": 0.1217,
      "step": 3316
    },
    {
      "epoch": 6.120036934441367,
      "grad_norm": 0.652368426322937,
      "learning_rate": 9.941999999999999e-06,
      "loss": 0.0611,
      "step": 3317
    },
    {
      "epoch": 6.121883656509695,
      "grad_norm": 0.6359832882881165,
      "learning_rate": 9.945e-06,
      "loss": 0.0494,
      "step": 3318
    },
    {
      "epoch": 6.123730378578024,
      "grad_norm": 0.8224269151687622,
      "learning_rate": 9.948e-06,
      "loss": 0.0672,
      "step": 3319
    },
    {
      "epoch": 6.125577100646352,
      "grad_norm": 0.6548646092414856,
      "learning_rate": 9.951e-06,
      "loss": 0.0513,
      "step": 3320
    },
    {
      "epoch": 6.127423822714682,
      "grad_norm": 0.6966165900230408,
      "learning_rate": 9.954e-06,
      "loss": 0.0395,
      "step": 3321
    },
    {
      "epoch": 6.12927054478301,
      "grad_norm": 0.4571305513381958,
      "learning_rate": 9.957e-06,
      "loss": 0.0374,
      "step": 3322
    },
    {
      "epoch": 6.131117266851339,
      "grad_norm": 0.6777710318565369,
      "learning_rate": 9.960000000000001e-06,
      "loss": 0.0372,
      "step": 3323
    },
    {
      "epoch": 6.132963988919667,
      "grad_norm": 0.8258435726165771,
      "learning_rate": 9.963000000000001e-06,
      "loss": 0.0537,
      "step": 3324
    },
    {
      "epoch": 6.134810710987996,
      "grad_norm": 0.6122437715530396,
      "learning_rate": 9.966e-06,
      "loss": 0.0385,
      "step": 3325
    },
    {
      "epoch": 6.136657433056325,
      "grad_norm": 0.7694428563117981,
      "learning_rate": 9.969e-06,
      "loss": 0.0367,
      "step": 3326
    },
    {
      "epoch": 6.138504155124654,
      "grad_norm": 0.7247572541236877,
      "learning_rate": 9.971999999999999e-06,
      "loss": 0.0426,
      "step": 3327
    },
    {
      "epoch": 6.140350877192983,
      "grad_norm": 0.7571830153465271,
      "learning_rate": 9.975e-06,
      "loss": 0.0513,
      "step": 3328
    },
    {
      "epoch": 6.142197599261311,
      "grad_norm": 0.9954696893692017,
      "learning_rate": 9.978e-06,
      "loss": 0.0392,
      "step": 3329
    },
    {
      "epoch": 6.14404432132964,
      "grad_norm": 1.1376701593399048,
      "learning_rate": 9.981e-06,
      "loss": 0.0635,
      "step": 3330
    },
    {
      "epoch": 6.1458910433979685,
      "grad_norm": 0.7513870000839233,
      "learning_rate": 9.984e-06,
      "loss": 0.0433,
      "step": 3331
    },
    {
      "epoch": 6.147737765466298,
      "grad_norm": 1.4481124877929688,
      "learning_rate": 9.987e-06,
      "loss": 0.0551,
      "step": 3332
    },
    {
      "epoch": 6.149584487534626,
      "grad_norm": 0.7296556830406189,
      "learning_rate": 9.990000000000001e-06,
      "loss": 0.0488,
      "step": 3333
    },
    {
      "epoch": 6.151431209602955,
      "grad_norm": 0.6062530279159546,
      "learning_rate": 9.993e-06,
      "loss": 0.0348,
      "step": 3334
    },
    {
      "epoch": 6.153277931671283,
      "grad_norm": 0.5732892751693726,
      "learning_rate": 9.996e-06,
      "loss": 0.0416,
      "step": 3335
    },
    {
      "epoch": 6.1551246537396125,
      "grad_norm": 0.7242116332054138,
      "learning_rate": 9.999e-06,
      "loss": 0.0382,
      "step": 3336
    },
    {
      "epoch": 6.156971375807941,
      "grad_norm": 0.647824764251709,
      "learning_rate": 1.0002e-05,
      "loss": 0.0386,
      "step": 3337
    },
    {
      "epoch": 6.15881809787627,
      "grad_norm": 0.9212784767150879,
      "learning_rate": 1.0005000000000002e-05,
      "loss": 0.0536,
      "step": 3338
    },
    {
      "epoch": 6.160664819944598,
      "grad_norm": 0.6482663750648499,
      "learning_rate": 1.0008e-05,
      "loss": 0.0366,
      "step": 3339
    },
    {
      "epoch": 6.162511542012927,
      "grad_norm": 1.459087610244751,
      "learning_rate": 1.0011e-05,
      "loss": 0.0496,
      "step": 3340
    },
    {
      "epoch": 6.1643582640812555,
      "grad_norm": 0.783748984336853,
      "learning_rate": 1.0014e-05,
      "loss": 0.0367,
      "step": 3341
    },
    {
      "epoch": 6.166204986149585,
      "grad_norm": 0.9266418814659119,
      "learning_rate": 1.0016999999999999e-05,
      "loss": 0.0532,
      "step": 3342
    },
    {
      "epoch": 6.168051708217913,
      "grad_norm": 0.7052007913589478,
      "learning_rate": 1.002e-05,
      "loss": 0.0372,
      "step": 3343
    },
    {
      "epoch": 6.169898430286242,
      "grad_norm": 0.6519185900688171,
      "learning_rate": 1.0023e-05,
      "loss": 0.0545,
      "step": 3344
    },
    {
      "epoch": 6.17174515235457,
      "grad_norm": 1.20285964012146,
      "learning_rate": 1.0026e-05,
      "loss": 0.0547,
      "step": 3345
    },
    {
      "epoch": 6.1735918744228995,
      "grad_norm": 0.860680341720581,
      "learning_rate": 1.0029e-05,
      "loss": 0.0724,
      "step": 3346
    },
    {
      "epoch": 6.175438596491228,
      "grad_norm": 1.4246987104415894,
      "learning_rate": 1.0032e-05,
      "loss": 0.0394,
      "step": 3347
    },
    {
      "epoch": 6.177285318559557,
      "grad_norm": 0.8983826041221619,
      "learning_rate": 1.0035000000000001e-05,
      "loss": 0.0629,
      "step": 3348
    },
    {
      "epoch": 6.179132040627885,
      "grad_norm": 1.397816777229309,
      "learning_rate": 1.0038000000000001e-05,
      "loss": 0.1254,
      "step": 3349
    },
    {
      "epoch": 6.180978762696214,
      "grad_norm": 1.3739649057388306,
      "learning_rate": 1.0041000000000001e-05,
      "loss": 0.0647,
      "step": 3350
    },
    {
      "epoch": 6.1828254847645425,
      "grad_norm": 1.1440520286560059,
      "learning_rate": 1.0043999999999999e-05,
      "loss": 0.0846,
      "step": 3351
    },
    {
      "epoch": 6.184672206832872,
      "grad_norm": 0.9402255415916443,
      "learning_rate": 1.0046999999999999e-05,
      "loss": 0.0599,
      "step": 3352
    },
    {
      "epoch": 6.186518928901201,
      "grad_norm": 1.4966752529144287,
      "learning_rate": 1.005e-05,
      "loss": 0.3815,
      "step": 3353
    },
    {
      "epoch": 6.188365650969529,
      "grad_norm": 2.027479410171509,
      "learning_rate": 1.0053e-05,
      "loss": 0.4318,
      "step": 3354
    },
    {
      "epoch": 6.190212373037858,
      "grad_norm": 1.0403293371200562,
      "learning_rate": 1.0056e-05,
      "loss": 0.2883,
      "step": 3355
    },
    {
      "epoch": 6.1920590951061865,
      "grad_norm": 1.2776825428009033,
      "learning_rate": 1.0059e-05,
      "loss": 0.2647,
      "step": 3356
    },
    {
      "epoch": 6.193905817174516,
      "grad_norm": 1.1058181524276733,
      "learning_rate": 1.0062e-05,
      "loss": 0.3148,
      "step": 3357
    },
    {
      "epoch": 6.195752539242844,
      "grad_norm": 1.2859538793563843,
      "learning_rate": 1.0065000000000001e-05,
      "loss": 0.2302,
      "step": 3358
    },
    {
      "epoch": 6.197599261311173,
      "grad_norm": 0.7337931990623474,
      "learning_rate": 1.0068e-05,
      "loss": 0.1434,
      "step": 3359
    },
    {
      "epoch": 6.199445983379501,
      "grad_norm": 0.8204769492149353,
      "learning_rate": 1.0071e-05,
      "loss": 0.171,
      "step": 3360
    },
    {
      "epoch": 6.20129270544783,
      "grad_norm": 1.006606936454773,
      "learning_rate": 1.0074e-05,
      "loss": 0.1954,
      "step": 3361
    },
    {
      "epoch": 6.203139427516159,
      "grad_norm": 1.0071405172348022,
      "learning_rate": 1.0077e-05,
      "loss": 0.1727,
      "step": 3362
    },
    {
      "epoch": 6.204986149584488,
      "grad_norm": 1.1923887729644775,
      "learning_rate": 1.008e-05,
      "loss": 0.1355,
      "step": 3363
    },
    {
      "epoch": 6.206832871652816,
      "grad_norm": 0.8156916499137878,
      "learning_rate": 1.0083e-05,
      "loss": 0.1401,
      "step": 3364
    },
    {
      "epoch": 6.208679593721145,
      "grad_norm": 0.9403882622718811,
      "learning_rate": 1.0086e-05,
      "loss": 0.1544,
      "step": 3365
    },
    {
      "epoch": 6.2105263157894735,
      "grad_norm": 1.117945909500122,
      "learning_rate": 1.0089e-05,
      "loss": 0.1383,
      "step": 3366
    },
    {
      "epoch": 6.212373037857803,
      "grad_norm": 0.7493522763252258,
      "learning_rate": 1.0092e-05,
      "loss": 0.0692,
      "step": 3367
    },
    {
      "epoch": 6.214219759926131,
      "grad_norm": 0.5226987600326538,
      "learning_rate": 1.0095e-05,
      "loss": 0.0472,
      "step": 3368
    },
    {
      "epoch": 6.21606648199446,
      "grad_norm": 0.7401248812675476,
      "learning_rate": 1.0098e-05,
      "loss": 0.1062,
      "step": 3369
    },
    {
      "epoch": 6.217913204062788,
      "grad_norm": 1.0519952774047852,
      "learning_rate": 1.0101e-05,
      "loss": 0.0539,
      "step": 3370
    },
    {
      "epoch": 6.219759926131117,
      "grad_norm": 0.8793032765388489,
      "learning_rate": 1.0104e-05,
      "loss": 0.0369,
      "step": 3371
    },
    {
      "epoch": 6.221606648199446,
      "grad_norm": 0.48777708411216736,
      "learning_rate": 1.0107e-05,
      "loss": 0.0343,
      "step": 3372
    },
    {
      "epoch": 6.223453370267775,
      "grad_norm": 0.5694409608840942,
      "learning_rate": 1.0110000000000001e-05,
      "loss": 0.0572,
      "step": 3373
    },
    {
      "epoch": 6.225300092336103,
      "grad_norm": 0.7707429528236389,
      "learning_rate": 1.0113000000000001e-05,
      "loss": 0.0681,
      "step": 3374
    },
    {
      "epoch": 6.227146814404432,
      "grad_norm": 0.722528338432312,
      "learning_rate": 1.0116000000000001e-05,
      "loss": 0.0453,
      "step": 3375
    },
    {
      "epoch": 6.22899353647276,
      "grad_norm": 0.834453821182251,
      "learning_rate": 1.0119e-05,
      "loss": 0.0526,
      "step": 3376
    },
    {
      "epoch": 6.23084025854109,
      "grad_norm": 0.9907567501068115,
      "learning_rate": 1.0121999999999999e-05,
      "loss": 0.0596,
      "step": 3377
    },
    {
      "epoch": 6.232686980609419,
      "grad_norm": 0.581995964050293,
      "learning_rate": 1.0125e-05,
      "loss": 0.029,
      "step": 3378
    },
    {
      "epoch": 6.234533702677747,
      "grad_norm": 0.6852633357048035,
      "learning_rate": 1.0128e-05,
      "loss": 0.0362,
      "step": 3379
    },
    {
      "epoch": 6.236380424746076,
      "grad_norm": 0.7753766179084778,
      "learning_rate": 1.0131e-05,
      "loss": 0.0437,
      "step": 3380
    },
    {
      "epoch": 6.238227146814404,
      "grad_norm": 0.6445023417472839,
      "learning_rate": 1.0134e-05,
      "loss": 0.0413,
      "step": 3381
    },
    {
      "epoch": 6.2400738688827335,
      "grad_norm": 0.8584773540496826,
      "learning_rate": 1.0137e-05,
      "loss": 0.0544,
      "step": 3382
    },
    {
      "epoch": 6.241920590951062,
      "grad_norm": 0.7375549674034119,
      "learning_rate": 1.0140000000000001e-05,
      "loss": 0.0642,
      "step": 3383
    },
    {
      "epoch": 6.243767313019391,
      "grad_norm": 2.494497299194336,
      "learning_rate": 1.0143000000000001e-05,
      "loss": 0.0714,
      "step": 3384
    },
    {
      "epoch": 6.245614035087719,
      "grad_norm": 0.6620352864265442,
      "learning_rate": 1.0146e-05,
      "loss": 0.0398,
      "step": 3385
    },
    {
      "epoch": 6.247460757156048,
      "grad_norm": 0.6307715773582458,
      "learning_rate": 1.0149e-05,
      "loss": 0.0546,
      "step": 3386
    },
    {
      "epoch": 6.249307479224377,
      "grad_norm": 0.8055571913719177,
      "learning_rate": 1.0152e-05,
      "loss": 0.0494,
      "step": 3387
    },
    {
      "epoch": 6.251154201292706,
      "grad_norm": 0.9565600156784058,
      "learning_rate": 1.0155e-05,
      "loss": 0.0883,
      "step": 3388
    },
    {
      "epoch": 6.253000923361034,
      "grad_norm": 0.9024067521095276,
      "learning_rate": 1.0158e-05,
      "loss": 0.0558,
      "step": 3389
    },
    {
      "epoch": 6.254847645429363,
      "grad_norm": 1.121073603630066,
      "learning_rate": 1.0161e-05,
      "loss": 0.0572,
      "step": 3390
    },
    {
      "epoch": 6.256694367497691,
      "grad_norm": 0.8038824796676636,
      "learning_rate": 1.0164e-05,
      "loss": 0.0578,
      "step": 3391
    },
    {
      "epoch": 6.2585410895660205,
      "grad_norm": 0.7897089719772339,
      "learning_rate": 1.0167e-05,
      "loss": 0.051,
      "step": 3392
    },
    {
      "epoch": 6.260387811634349,
      "grad_norm": 0.7924173474311829,
      "learning_rate": 1.0170000000000001e-05,
      "loss": 0.0607,
      "step": 3393
    },
    {
      "epoch": 6.262234533702678,
      "grad_norm": 0.8015258312225342,
      "learning_rate": 1.0173e-05,
      "loss": 0.0534,
      "step": 3394
    },
    {
      "epoch": 6.264081255771006,
      "grad_norm": 0.79560387134552,
      "learning_rate": 1.0176e-05,
      "loss": 0.0355,
      "step": 3395
    },
    {
      "epoch": 6.265927977839335,
      "grad_norm": 0.7222800254821777,
      "learning_rate": 1.0179e-05,
      "loss": 0.0546,
      "step": 3396
    },
    {
      "epoch": 6.267774699907664,
      "grad_norm": 0.8553308248519897,
      "learning_rate": 1.0182e-05,
      "loss": 0.061,
      "step": 3397
    },
    {
      "epoch": 6.269621421975993,
      "grad_norm": 1.2343560457229614,
      "learning_rate": 1.0185000000000002e-05,
      "loss": 0.0505,
      "step": 3398
    },
    {
      "epoch": 6.271468144044321,
      "grad_norm": 0.755364179611206,
      "learning_rate": 1.0188000000000001e-05,
      "loss": 0.0632,
      "step": 3399
    },
    {
      "epoch": 6.27331486611265,
      "grad_norm": 1.284324288368225,
      "learning_rate": 1.0191e-05,
      "loss": 0.0499,
      "step": 3400
    },
    {
      "epoch": 6.275161588180978,
      "grad_norm": 0.6206008195877075,
      "learning_rate": 1.0194e-05,
      "loss": 0.0418,
      "step": 3401
    },
    {
      "epoch": 6.2770083102493075,
      "grad_norm": 1.1747897863388062,
      "learning_rate": 1.0196999999999999e-05,
      "loss": 0.0765,
      "step": 3402
    },
    {
      "epoch": 6.278855032317637,
      "grad_norm": 1.0085206031799316,
      "learning_rate": 1.02e-05,
      "loss": 0.3388,
      "step": 3403
    },
    {
      "epoch": 6.280701754385965,
      "grad_norm": 1.1853443384170532,
      "learning_rate": 1.0203e-05,
      "loss": 0.3431,
      "step": 3404
    },
    {
      "epoch": 6.282548476454294,
      "grad_norm": 0.9081771969795227,
      "learning_rate": 1.0206e-05,
      "loss": 0.2456,
      "step": 3405
    },
    {
      "epoch": 6.284395198522622,
      "grad_norm": 1.0242339372634888,
      "learning_rate": 1.0209e-05,
      "loss": 0.2397,
      "step": 3406
    },
    {
      "epoch": 6.286241920590951,
      "grad_norm": 0.9105067849159241,
      "learning_rate": 1.0212e-05,
      "loss": 0.2331,
      "step": 3407
    },
    {
      "epoch": 6.28808864265928,
      "grad_norm": 0.6736548542976379,
      "learning_rate": 1.0215000000000001e-05,
      "loss": 0.1877,
      "step": 3408
    },
    {
      "epoch": 6.289935364727609,
      "grad_norm": 0.7968747019767761,
      "learning_rate": 1.0218000000000001e-05,
      "loss": 0.1593,
      "step": 3409
    },
    {
      "epoch": 6.291782086795937,
      "grad_norm": 0.8158178925514221,
      "learning_rate": 1.0221000000000001e-05,
      "loss": 0.1985,
      "step": 3410
    },
    {
      "epoch": 6.293628808864266,
      "grad_norm": 1.3161625862121582,
      "learning_rate": 1.0224e-05,
      "loss": 0.2134,
      "step": 3411
    },
    {
      "epoch": 6.2954755309325945,
      "grad_norm": 0.9767283797264099,
      "learning_rate": 1.0227e-05,
      "loss": 0.2046,
      "step": 3412
    },
    {
      "epoch": 6.297322253000924,
      "grad_norm": 1.8502895832061768,
      "learning_rate": 1.023e-05,
      "loss": 0.1817,
      "step": 3413
    },
    {
      "epoch": 6.299168975069252,
      "grad_norm": 0.7292230129241943,
      "learning_rate": 1.0233e-05,
      "loss": 0.0889,
      "step": 3414
    },
    {
      "epoch": 6.301015697137581,
      "grad_norm": 0.9305237531661987,
      "learning_rate": 1.0236e-05,
      "loss": 0.1109,
      "step": 3415
    },
    {
      "epoch": 6.302862419205909,
      "grad_norm": 0.6266627311706543,
      "learning_rate": 1.0239e-05,
      "loss": 0.0741,
      "step": 3416
    },
    {
      "epoch": 6.304709141274238,
      "grad_norm": 0.7265087962150574,
      "learning_rate": 1.0242e-05,
      "loss": 0.1007,
      "step": 3417
    },
    {
      "epoch": 6.306555863342567,
      "grad_norm": 0.4341509938240051,
      "learning_rate": 1.0245000000000001e-05,
      "loss": 0.0363,
      "step": 3418
    },
    {
      "epoch": 6.308402585410896,
      "grad_norm": 0.3478453755378723,
      "learning_rate": 1.0248e-05,
      "loss": 0.0328,
      "step": 3419
    },
    {
      "epoch": 6.310249307479224,
      "grad_norm": 0.7002639174461365,
      "learning_rate": 1.0251e-05,
      "loss": 0.0809,
      "step": 3420
    },
    {
      "epoch": 6.312096029547553,
      "grad_norm": 0.5919689536094666,
      "learning_rate": 1.0254e-05,
      "loss": 0.0449,
      "step": 3421
    },
    {
      "epoch": 6.3139427516158815,
      "grad_norm": 0.6110935211181641,
      "learning_rate": 1.0257e-05,
      "loss": 0.0483,
      "step": 3422
    },
    {
      "epoch": 6.315789473684211,
      "grad_norm": 0.699510395526886,
      "learning_rate": 1.0260000000000002e-05,
      "loss": 0.0377,
      "step": 3423
    },
    {
      "epoch": 6.317636195752539,
      "grad_norm": 0.6981428861618042,
      "learning_rate": 1.0263000000000002e-05,
      "loss": 0.048,
      "step": 3424
    },
    {
      "epoch": 6.319482917820868,
      "grad_norm": 1.3745615482330322,
      "learning_rate": 1.0266e-05,
      "loss": 0.0736,
      "step": 3425
    },
    {
      "epoch": 6.321329639889196,
      "grad_norm": 0.9349688291549683,
      "learning_rate": 1.0269e-05,
      "loss": 0.0494,
      "step": 3426
    },
    {
      "epoch": 6.323176361957525,
      "grad_norm": 1.4682650566101074,
      "learning_rate": 1.0272e-05,
      "loss": 0.078,
      "step": 3427
    },
    {
      "epoch": 6.325023084025855,
      "grad_norm": 0.8241517543792725,
      "learning_rate": 1.0275e-05,
      "loss": 0.0775,
      "step": 3428
    },
    {
      "epoch": 6.326869806094183,
      "grad_norm": 1.6067695617675781,
      "learning_rate": 1.0278e-05,
      "loss": 0.1381,
      "step": 3429
    },
    {
      "epoch": 6.328716528162512,
      "grad_norm": 0.7926170229911804,
      "learning_rate": 1.0281e-05,
      "loss": 0.0527,
      "step": 3430
    },
    {
      "epoch": 6.33056325023084,
      "grad_norm": 1.9036262035369873,
      "learning_rate": 1.0284e-05,
      "loss": 0.0394,
      "step": 3431
    },
    {
      "epoch": 6.332409972299169,
      "grad_norm": 0.6403128504753113,
      "learning_rate": 1.0287e-05,
      "loss": 0.0793,
      "step": 3432
    },
    {
      "epoch": 6.334256694367498,
      "grad_norm": 0.6075448989868164,
      "learning_rate": 1.0290000000000001e-05,
      "loss": 0.049,
      "step": 3433
    },
    {
      "epoch": 6.336103416435827,
      "grad_norm": 0.9043646454811096,
      "learning_rate": 1.0293000000000001e-05,
      "loss": 0.0559,
      "step": 3434
    },
    {
      "epoch": 6.337950138504155,
      "grad_norm": 0.9564310908317566,
      "learning_rate": 1.0296000000000001e-05,
      "loss": 0.0372,
      "step": 3435
    },
    {
      "epoch": 6.339796860572484,
      "grad_norm": 0.5282099843025208,
      "learning_rate": 1.0299e-05,
      "loss": 0.0328,
      "step": 3436
    },
    {
      "epoch": 6.341643582640812,
      "grad_norm": 1.1436553001403809,
      "learning_rate": 1.0301999999999999e-05,
      "loss": 0.0548,
      "step": 3437
    },
    {
      "epoch": 6.3434903047091415,
      "grad_norm": 0.8616344928741455,
      "learning_rate": 1.0305e-05,
      "loss": 0.0488,
      "step": 3438
    },
    {
      "epoch": 6.34533702677747,
      "grad_norm": 1.3523774147033691,
      "learning_rate": 1.0308e-05,
      "loss": 0.0562,
      "step": 3439
    },
    {
      "epoch": 6.347183748845799,
      "grad_norm": 0.6882216334342957,
      "learning_rate": 1.0311e-05,
      "loss": 0.0426,
      "step": 3440
    },
    {
      "epoch": 6.349030470914127,
      "grad_norm": 1.0950908660888672,
      "learning_rate": 1.0314e-05,
      "loss": 0.0582,
      "step": 3441
    },
    {
      "epoch": 6.350877192982456,
      "grad_norm": 1.109289526939392,
      "learning_rate": 1.0317e-05,
      "loss": 0.0602,
      "step": 3442
    },
    {
      "epoch": 6.352723915050785,
      "grad_norm": 0.8034445643424988,
      "learning_rate": 1.032e-05,
      "loss": 0.0466,
      "step": 3443
    },
    {
      "epoch": 6.354570637119114,
      "grad_norm": 0.7607554197311401,
      "learning_rate": 1.0323000000000001e-05,
      "loss": 0.0351,
      "step": 3444
    },
    {
      "epoch": 6.356417359187442,
      "grad_norm": 0.9530297517776489,
      "learning_rate": 1.0326e-05,
      "loss": 0.0422,
      "step": 3445
    },
    {
      "epoch": 6.358264081255771,
      "grad_norm": 1.066715955734253,
      "learning_rate": 1.0329e-05,
      "loss": 0.0728,
      "step": 3446
    },
    {
      "epoch": 6.360110803324099,
      "grad_norm": 0.8962160348892212,
      "learning_rate": 1.0332e-05,
      "loss": 0.0558,
      "step": 3447
    },
    {
      "epoch": 6.3619575253924285,
      "grad_norm": 0.6709640026092529,
      "learning_rate": 1.0335e-05,
      "loss": 0.0362,
      "step": 3448
    },
    {
      "epoch": 6.363804247460757,
      "grad_norm": 1.028660774230957,
      "learning_rate": 1.0338e-05,
      "loss": 0.0479,
      "step": 3449
    },
    {
      "epoch": 6.365650969529086,
      "grad_norm": 0.7419813275337219,
      "learning_rate": 1.0341e-05,
      "loss": 0.0515,
      "step": 3450
    },
    {
      "epoch": 6.367497691597414,
      "grad_norm": 1.3043344020843506,
      "learning_rate": 1.0344e-05,
      "loss": 0.0608,
      "step": 3451
    },
    {
      "epoch": 6.369344413665743,
      "grad_norm": 0.9880107641220093,
      "learning_rate": 1.0347e-05,
      "loss": 0.0471,
      "step": 3452
    },
    {
      "epoch": 6.3711911357340725,
      "grad_norm": 1.2809970378875732,
      "learning_rate": 1.035e-05,
      "loss": 0.3272,
      "step": 3453
    },
    {
      "epoch": 6.373037857802401,
      "grad_norm": 1.0759568214416504,
      "learning_rate": 1.0353e-05,
      "loss": 0.3555,
      "step": 3454
    },
    {
      "epoch": 6.374884579870729,
      "grad_norm": 1.1627869606018066,
      "learning_rate": 1.0356e-05,
      "loss": 0.3155,
      "step": 3455
    },
    {
      "epoch": 6.376731301939058,
      "grad_norm": 1.6699895858764648,
      "learning_rate": 1.0359e-05,
      "loss": 0.2746,
      "step": 3456
    },
    {
      "epoch": 6.378578024007387,
      "grad_norm": 0.8494102358818054,
      "learning_rate": 1.0362e-05,
      "loss": 0.2381,
      "step": 3457
    },
    {
      "epoch": 6.3804247460757155,
      "grad_norm": 0.9873664975166321,
      "learning_rate": 1.0365e-05,
      "loss": 0.2455,
      "step": 3458
    },
    {
      "epoch": 6.382271468144045,
      "grad_norm": 1.3834065198898315,
      "learning_rate": 1.0368000000000001e-05,
      "loss": 0.2281,
      "step": 3459
    },
    {
      "epoch": 6.384118190212373,
      "grad_norm": 0.9461009502410889,
      "learning_rate": 1.0371000000000001e-05,
      "loss": 0.1776,
      "step": 3460
    },
    {
      "epoch": 6.385964912280702,
      "grad_norm": 1.184063196182251,
      "learning_rate": 1.0374000000000001e-05,
      "loss": 0.2644,
      "step": 3461
    },
    {
      "epoch": 6.38781163434903,
      "grad_norm": 1.0690431594848633,
      "learning_rate": 1.0376999999999999e-05,
      "loss": 0.2075,
      "step": 3462
    },
    {
      "epoch": 6.3896583564173595,
      "grad_norm": 0.8396578431129456,
      "learning_rate": 1.0379999999999999e-05,
      "loss": 0.1208,
      "step": 3463
    },
    {
      "epoch": 6.391505078485688,
      "grad_norm": 1.136763095855713,
      "learning_rate": 1.0383e-05,
      "loss": 0.188,
      "step": 3464
    },
    {
      "epoch": 6.393351800554017,
      "grad_norm": 0.6293352246284485,
      "learning_rate": 1.0386e-05,
      "loss": 0.0897,
      "step": 3465
    },
    {
      "epoch": 6.395198522622345,
      "grad_norm": 0.8428406715393066,
      "learning_rate": 1.0389e-05,
      "loss": 0.0854,
      "step": 3466
    },
    {
      "epoch": 6.397045244690674,
      "grad_norm": 1.0550479888916016,
      "learning_rate": 1.0392e-05,
      "loss": 0.1188,
      "step": 3467
    },
    {
      "epoch": 6.3988919667590025,
      "grad_norm": 0.7489149570465088,
      "learning_rate": 1.0395e-05,
      "loss": 0.0845,
      "step": 3468
    },
    {
      "epoch": 6.400738688827332,
      "grad_norm": 0.5197301506996155,
      "learning_rate": 1.0398000000000001e-05,
      "loss": 0.0488,
      "step": 3469
    },
    {
      "epoch": 6.40258541089566,
      "grad_norm": 1.0635749101638794,
      "learning_rate": 1.0401000000000001e-05,
      "loss": 0.0613,
      "step": 3470
    },
    {
      "epoch": 6.404432132963989,
      "grad_norm": 0.8245949149131775,
      "learning_rate": 1.0404e-05,
      "loss": 0.0741,
      "step": 3471
    },
    {
      "epoch": 6.406278855032317,
      "grad_norm": 1.0676509141921997,
      "learning_rate": 1.0407e-05,
      "loss": 0.1082,
      "step": 3472
    },
    {
      "epoch": 6.4081255771006465,
      "grad_norm": 1.1991374492645264,
      "learning_rate": 1.041e-05,
      "loss": 0.1308,
      "step": 3473
    },
    {
      "epoch": 6.409972299168975,
      "grad_norm": 0.5810908675193787,
      "learning_rate": 1.0413e-05,
      "loss": 0.0451,
      "step": 3474
    },
    {
      "epoch": 6.411819021237304,
      "grad_norm": 0.5625260472297668,
      "learning_rate": 1.0416e-05,
      "loss": 0.0317,
      "step": 3475
    },
    {
      "epoch": 6.413665743305632,
      "grad_norm": 0.7243143916130066,
      "learning_rate": 1.0419e-05,
      "loss": 0.0504,
      "step": 3476
    },
    {
      "epoch": 6.415512465373961,
      "grad_norm": 0.5748608708381653,
      "learning_rate": 1.0422e-05,
      "loss": 0.0411,
      "step": 3477
    },
    {
      "epoch": 6.41735918744229,
      "grad_norm": 0.8418190479278564,
      "learning_rate": 1.0425e-05,
      "loss": 0.0504,
      "step": 3478
    },
    {
      "epoch": 6.419205909510619,
      "grad_norm": 0.5589009523391724,
      "learning_rate": 1.0428e-05,
      "loss": 0.0473,
      "step": 3479
    },
    {
      "epoch": 6.421052631578947,
      "grad_norm": 0.3747293949127197,
      "learning_rate": 1.0431e-05,
      "loss": 0.0211,
      "step": 3480
    },
    {
      "epoch": 6.422899353647276,
      "grad_norm": 0.9748400449752808,
      "learning_rate": 1.0434e-05,
      "loss": 0.0567,
      "step": 3481
    },
    {
      "epoch": 6.424746075715605,
      "grad_norm": 0.5059589743614197,
      "learning_rate": 1.0437e-05,
      "loss": 0.0307,
      "step": 3482
    },
    {
      "epoch": 6.426592797783933,
      "grad_norm": 0.6502093076705933,
      "learning_rate": 1.044e-05,
      "loss": 0.0454,
      "step": 3483
    },
    {
      "epoch": 6.428439519852263,
      "grad_norm": 0.994144082069397,
      "learning_rate": 1.0443000000000001e-05,
      "loss": 0.0495,
      "step": 3484
    },
    {
      "epoch": 6.430286241920591,
      "grad_norm": 0.9719415903091431,
      "learning_rate": 1.0446000000000001e-05,
      "loss": 0.0471,
      "step": 3485
    },
    {
      "epoch": 6.43213296398892,
      "grad_norm": 0.5761688351631165,
      "learning_rate": 1.0449e-05,
      "loss": 0.0358,
      "step": 3486
    },
    {
      "epoch": 6.433979686057248,
      "grad_norm": 0.8566411137580872,
      "learning_rate": 1.0452e-05,
      "loss": 0.0368,
      "step": 3487
    },
    {
      "epoch": 6.435826408125577,
      "grad_norm": 1.0438909530639648,
      "learning_rate": 1.0454999999999999e-05,
      "loss": 0.0525,
      "step": 3488
    },
    {
      "epoch": 6.437673130193906,
      "grad_norm": 0.691596508026123,
      "learning_rate": 1.0458e-05,
      "loss": 0.0307,
      "step": 3489
    },
    {
      "epoch": 6.439519852262235,
      "grad_norm": 1.0400809049606323,
      "learning_rate": 1.0461e-05,
      "loss": 0.0433,
      "step": 3490
    },
    {
      "epoch": 6.441366574330563,
      "grad_norm": 1.1551169157028198,
      "learning_rate": 1.0464e-05,
      "loss": 0.0412,
      "step": 3491
    },
    {
      "epoch": 6.443213296398892,
      "grad_norm": 1.3905787467956543,
      "learning_rate": 1.0467e-05,
      "loss": 0.0476,
      "step": 3492
    },
    {
      "epoch": 6.44506001846722,
      "grad_norm": 0.8146647810935974,
      "learning_rate": 1.047e-05,
      "loss": 0.0689,
      "step": 3493
    },
    {
      "epoch": 6.44690674053555,
      "grad_norm": 0.9274573922157288,
      "learning_rate": 1.0473000000000001e-05,
      "loss": 0.0461,
      "step": 3494
    },
    {
      "epoch": 6.448753462603878,
      "grad_norm": 0.9964638352394104,
      "learning_rate": 1.0476000000000001e-05,
      "loss": 0.0625,
      "step": 3495
    },
    {
      "epoch": 6.450600184672207,
      "grad_norm": 0.6893898248672485,
      "learning_rate": 1.0479e-05,
      "loss": 0.0425,
      "step": 3496
    },
    {
      "epoch": 6.452446906740535,
      "grad_norm": 0.627974808216095,
      "learning_rate": 1.0482e-05,
      "loss": 0.0476,
      "step": 3497
    },
    {
      "epoch": 6.454293628808864,
      "grad_norm": 0.8051822185516357,
      "learning_rate": 1.0485e-05,
      "loss": 0.0471,
      "step": 3498
    },
    {
      "epoch": 6.456140350877193,
      "grad_norm": 0.8309305906295776,
      "learning_rate": 1.0488e-05,
      "loss": 0.0481,
      "step": 3499
    },
    {
      "epoch": 6.457987072945522,
      "grad_norm": 0.6573142409324646,
      "learning_rate": 1.0491e-05,
      "loss": 0.0397,
      "step": 3500
    },
    {
      "epoch": 6.45983379501385,
      "grad_norm": 2.428025960922241,
      "learning_rate": 1.0494e-05,
      "loss": 0.061,
      "step": 3501
    },
    {
      "epoch": 6.461680517082179,
      "grad_norm": 1.2131538391113281,
      "learning_rate": 1.0497e-05,
      "loss": 0.0639,
      "step": 3502
    },
    {
      "epoch": 6.463527239150508,
      "grad_norm": 1.5500845909118652,
      "learning_rate": 1.05e-05,
      "loss": 0.3862,
      "step": 3503
    },
    {
      "epoch": 6.465373961218837,
      "grad_norm": 1.091170072555542,
      "learning_rate": 1.0503000000000001e-05,
      "loss": 0.3148,
      "step": 3504
    },
    {
      "epoch": 6.467220683287165,
      "grad_norm": 1.0645906925201416,
      "learning_rate": 1.0506e-05,
      "loss": 0.3232,
      "step": 3505
    },
    {
      "epoch": 6.469067405355494,
      "grad_norm": 1.3559659719467163,
      "learning_rate": 1.0509e-05,
      "loss": 0.2834,
      "step": 3506
    },
    {
      "epoch": 6.470914127423823,
      "grad_norm": 1.1737184524536133,
      "learning_rate": 1.0512e-05,
      "loss": 0.2639,
      "step": 3507
    },
    {
      "epoch": 6.472760849492151,
      "grad_norm": 1.5199416875839233,
      "learning_rate": 1.0515e-05,
      "loss": 0.2529,
      "step": 3508
    },
    {
      "epoch": 6.4746075715604805,
      "grad_norm": 0.9331461191177368,
      "learning_rate": 1.0518000000000002e-05,
      "loss": 0.2125,
      "step": 3509
    },
    {
      "epoch": 6.476454293628809,
      "grad_norm": 0.8014020919799805,
      "learning_rate": 1.0521000000000001e-05,
      "loss": 0.2295,
      "step": 3510
    },
    {
      "epoch": 6.478301015697138,
      "grad_norm": 1.2764960527420044,
      "learning_rate": 1.0524e-05,
      "loss": 0.2104,
      "step": 3511
    },
    {
      "epoch": 6.480147737765466,
      "grad_norm": 0.8344443440437317,
      "learning_rate": 1.0527e-05,
      "loss": 0.1395,
      "step": 3512
    },
    {
      "epoch": 6.481994459833795,
      "grad_norm": 1.5831360816955566,
      "learning_rate": 1.0529999999999999e-05,
      "loss": 0.1398,
      "step": 3513
    },
    {
      "epoch": 6.4838411819021236,
      "grad_norm": 0.9806162118911743,
      "learning_rate": 1.0533e-05,
      "loss": 0.1731,
      "step": 3514
    },
    {
      "epoch": 6.485687903970453,
      "grad_norm": 0.7452002763748169,
      "learning_rate": 1.0536e-05,
      "loss": 0.1019,
      "step": 3515
    },
    {
      "epoch": 6.487534626038781,
      "grad_norm": 0.7927934527397156,
      "learning_rate": 1.0539e-05,
      "loss": 0.0808,
      "step": 3516
    },
    {
      "epoch": 6.48938134810711,
      "grad_norm": 1.0615880489349365,
      "learning_rate": 1.0542e-05,
      "loss": 0.1983,
      "step": 3517
    },
    {
      "epoch": 6.491228070175438,
      "grad_norm": 0.9302128553390503,
      "learning_rate": 1.0545e-05,
      "loss": 0.1074,
      "step": 3518
    },
    {
      "epoch": 6.4930747922437675,
      "grad_norm": 0.8489500284194946,
      "learning_rate": 1.0548000000000001e-05,
      "loss": 0.0644,
      "step": 3519
    },
    {
      "epoch": 6.494921514312096,
      "grad_norm": 0.49855339527130127,
      "learning_rate": 1.0551000000000001e-05,
      "loss": 0.0435,
      "step": 3520
    },
    {
      "epoch": 6.496768236380425,
      "grad_norm": 0.7827534675598145,
      "learning_rate": 1.0554000000000001e-05,
      "loss": 0.0548,
      "step": 3521
    },
    {
      "epoch": 6.498614958448753,
      "grad_norm": 2.3017821311950684,
      "learning_rate": 1.0557e-05,
      "loss": 0.0593,
      "step": 3522
    },
    {
      "epoch": 6.500461680517082,
      "grad_norm": 0.5665956139564514,
      "learning_rate": 1.0559999999999999e-05,
      "loss": 0.0539,
      "step": 3523
    },
    {
      "epoch": 6.5023084025854105,
      "grad_norm": 0.7230385541915894,
      "learning_rate": 1.0563e-05,
      "loss": 0.0515,
      "step": 3524
    },
    {
      "epoch": 6.50415512465374,
      "grad_norm": 0.799625039100647,
      "learning_rate": 1.0566e-05,
      "loss": 0.0512,
      "step": 3525
    },
    {
      "epoch": 6.506001846722068,
      "grad_norm": 0.7463293075561523,
      "learning_rate": 1.0569e-05,
      "loss": 0.0479,
      "step": 3526
    },
    {
      "epoch": 6.507848568790397,
      "grad_norm": 0.9893417358398438,
      "learning_rate": 1.0572e-05,
      "loss": 0.0445,
      "step": 3527
    },
    {
      "epoch": 6.509695290858726,
      "grad_norm": 0.7986618876457214,
      "learning_rate": 1.0575e-05,
      "loss": 0.0599,
      "step": 3528
    },
    {
      "epoch": 6.5115420129270545,
      "grad_norm": 0.6171774864196777,
      "learning_rate": 1.0578000000000001e-05,
      "loss": 0.0416,
      "step": 3529
    },
    {
      "epoch": 6.513388734995383,
      "grad_norm": 0.6155923008918762,
      "learning_rate": 1.0581e-05,
      "loss": 0.0321,
      "step": 3530
    },
    {
      "epoch": 6.515235457063712,
      "grad_norm": 0.8731139302253723,
      "learning_rate": 1.0584e-05,
      "loss": 0.048,
      "step": 3531
    },
    {
      "epoch": 6.517082179132041,
      "grad_norm": 1.004489541053772,
      "learning_rate": 1.0587e-05,
      "loss": 0.0373,
      "step": 3532
    },
    {
      "epoch": 6.518928901200369,
      "grad_norm": 0.6643251180648804,
      "learning_rate": 1.059e-05,
      "loss": 0.0398,
      "step": 3533
    },
    {
      "epoch": 6.520775623268698,
      "grad_norm": 1.0458920001983643,
      "learning_rate": 1.0593000000000002e-05,
      "loss": 0.0634,
      "step": 3534
    },
    {
      "epoch": 6.522622345337027,
      "grad_norm": 0.7619039416313171,
      "learning_rate": 1.0596e-05,
      "loss": 0.0452,
      "step": 3535
    },
    {
      "epoch": 6.524469067405356,
      "grad_norm": 0.49429938197135925,
      "learning_rate": 1.0599e-05,
      "loss": 0.0252,
      "step": 3536
    },
    {
      "epoch": 6.526315789473684,
      "grad_norm": 0.9544243216514587,
      "learning_rate": 1.0602e-05,
      "loss": 0.0532,
      "step": 3537
    },
    {
      "epoch": 6.528162511542013,
      "grad_norm": 0.6411821246147156,
      "learning_rate": 1.0605e-05,
      "loss": 0.0463,
      "step": 3538
    },
    {
      "epoch": 6.5300092336103415,
      "grad_norm": 0.9624695181846619,
      "learning_rate": 1.0608e-05,
      "loss": 0.0628,
      "step": 3539
    },
    {
      "epoch": 6.531855955678671,
      "grad_norm": 0.7749090194702148,
      "learning_rate": 1.0611e-05,
      "loss": 0.0786,
      "step": 3540
    },
    {
      "epoch": 6.533702677746999,
      "grad_norm": 0.8554633259773254,
      "learning_rate": 1.0614e-05,
      "loss": 0.0492,
      "step": 3541
    },
    {
      "epoch": 6.535549399815328,
      "grad_norm": 0.9086441397666931,
      "learning_rate": 1.0617e-05,
      "loss": 0.0641,
      "step": 3542
    },
    {
      "epoch": 6.537396121883656,
      "grad_norm": 1.2479761838912964,
      "learning_rate": 1.062e-05,
      "loss": 0.0503,
      "step": 3543
    },
    {
      "epoch": 6.539242843951985,
      "grad_norm": 0.9384217858314514,
      "learning_rate": 1.0623000000000001e-05,
      "loss": 0.0434,
      "step": 3544
    },
    {
      "epoch": 6.541089566020314,
      "grad_norm": 0.9494989514350891,
      "learning_rate": 1.0626000000000001e-05,
      "loss": 0.0533,
      "step": 3545
    },
    {
      "epoch": 6.542936288088643,
      "grad_norm": 0.9067333340644836,
      "learning_rate": 1.0629000000000001e-05,
      "loss": 0.0382,
      "step": 3546
    },
    {
      "epoch": 6.544783010156971,
      "grad_norm": 0.49929141998291016,
      "learning_rate": 1.0632000000000001e-05,
      "loss": 0.0316,
      "step": 3547
    },
    {
      "epoch": 6.5466297322253,
      "grad_norm": 0.9056301712989807,
      "learning_rate": 1.0634999999999999e-05,
      "loss": 0.0556,
      "step": 3548
    },
    {
      "epoch": 6.5484764542936285,
      "grad_norm": 0.7503829598426819,
      "learning_rate": 1.0638e-05,
      "loss": 0.0376,
      "step": 3549
    },
    {
      "epoch": 6.550323176361958,
      "grad_norm": 0.8950915336608887,
      "learning_rate": 1.0641e-05,
      "loss": 0.0437,
      "step": 3550
    },
    {
      "epoch": 6.552169898430286,
      "grad_norm": 0.932851254940033,
      "learning_rate": 1.0644e-05,
      "loss": 0.0763,
      "step": 3551
    },
    {
      "epoch": 6.554016620498615,
      "grad_norm": 0.9872285723686218,
      "learning_rate": 1.0647e-05,
      "loss": 0.0564,
      "step": 3552
    },
    {
      "epoch": 6.555863342566944,
      "grad_norm": 1.4405521154403687,
      "learning_rate": 1.065e-05,
      "loss": 0.3025,
      "step": 3553
    },
    {
      "epoch": 6.557710064635272,
      "grad_norm": 1.3220841884613037,
      "learning_rate": 1.0653000000000001e-05,
      "loss": 0.2998,
      "step": 3554
    },
    {
      "epoch": 6.559556786703601,
      "grad_norm": 1.004377007484436,
      "learning_rate": 1.0656000000000001e-05,
      "loss": 0.275,
      "step": 3555
    },
    {
      "epoch": 6.56140350877193,
      "grad_norm": 1.3007129430770874,
      "learning_rate": 1.0659e-05,
      "loss": 0.2442,
      "step": 3556
    },
    {
      "epoch": 6.563250230840259,
      "grad_norm": 0.8836926221847534,
      "learning_rate": 1.0662e-05,
      "loss": 0.2277,
      "step": 3557
    },
    {
      "epoch": 6.565096952908587,
      "grad_norm": 1.2006216049194336,
      "learning_rate": 1.0665e-05,
      "loss": 0.246,
      "step": 3558
    },
    {
      "epoch": 6.566943674976916,
      "grad_norm": 0.9763594269752502,
      "learning_rate": 1.0668000000000002e-05,
      "loss": 0.1581,
      "step": 3559
    },
    {
      "epoch": 6.568790397045245,
      "grad_norm": 0.7451435327529907,
      "learning_rate": 1.0671e-05,
      "loss": 0.178,
      "step": 3560
    },
    {
      "epoch": 6.570637119113574,
      "grad_norm": 0.720901370048523,
      "learning_rate": 1.0674e-05,
      "loss": 0.1629,
      "step": 3561
    },
    {
      "epoch": 6.572483841181902,
      "grad_norm": 0.8330392241477966,
      "learning_rate": 1.0677e-05,
      "loss": 0.1589,
      "step": 3562
    },
    {
      "epoch": 6.574330563250231,
      "grad_norm": 0.9879810810089111,
      "learning_rate": 1.068e-05,
      "loss": 0.1345,
      "step": 3563
    },
    {
      "epoch": 6.576177285318559,
      "grad_norm": 0.8167811632156372,
      "learning_rate": 1.0683000000000001e-05,
      "loss": 0.1357,
      "step": 3564
    },
    {
      "epoch": 6.5780240073868885,
      "grad_norm": 1.456222414970398,
      "learning_rate": 1.0686e-05,
      "loss": 0.1959,
      "step": 3565
    },
    {
      "epoch": 6.579870729455217,
      "grad_norm": 0.9722453951835632,
      "learning_rate": 1.0689e-05,
      "loss": 0.104,
      "step": 3566
    },
    {
      "epoch": 6.581717451523546,
      "grad_norm": 0.7950766086578369,
      "learning_rate": 1.0692e-05,
      "loss": 0.1026,
      "step": 3567
    },
    {
      "epoch": 6.583564173591874,
      "grad_norm": 0.7319908142089844,
      "learning_rate": 1.0695e-05,
      "loss": 0.0835,
      "step": 3568
    },
    {
      "epoch": 6.585410895660203,
      "grad_norm": 0.4791616201400757,
      "learning_rate": 1.0698e-05,
      "loss": 0.0517,
      "step": 3569
    },
    {
      "epoch": 6.587257617728532,
      "grad_norm": 0.8066070675849915,
      "learning_rate": 1.0701000000000001e-05,
      "loss": 0.0591,
      "step": 3570
    },
    {
      "epoch": 6.589104339796861,
      "grad_norm": 0.5366147756576538,
      "learning_rate": 1.0704000000000001e-05,
      "loss": 0.0609,
      "step": 3571
    },
    {
      "epoch": 6.590951061865189,
      "grad_norm": 0.5729362368583679,
      "learning_rate": 1.0707e-05,
      "loss": 0.067,
      "step": 3572
    },
    {
      "epoch": 6.592797783933518,
      "grad_norm": 0.8931047320365906,
      "learning_rate": 1.0709999999999999e-05,
      "loss": 0.0406,
      "step": 3573
    },
    {
      "epoch": 6.594644506001846,
      "grad_norm": 0.8464675545692444,
      "learning_rate": 1.0712999999999999e-05,
      "loss": 0.06,
      "step": 3574
    },
    {
      "epoch": 6.5964912280701755,
      "grad_norm": 0.8957421779632568,
      "learning_rate": 1.0716e-05,
      "loss": 0.0611,
      "step": 3575
    },
    {
      "epoch": 6.598337950138504,
      "grad_norm": 0.6566211581230164,
      "learning_rate": 1.0719e-05,
      "loss": 0.0469,
      "step": 3576
    },
    {
      "epoch": 6.600184672206833,
      "grad_norm": 0.6337366104125977,
      "learning_rate": 1.0722e-05,
      "loss": 0.0475,
      "step": 3577
    },
    {
      "epoch": 6.602031394275162,
      "grad_norm": 0.6649699807167053,
      "learning_rate": 1.0725e-05,
      "loss": 0.0504,
      "step": 3578
    },
    {
      "epoch": 6.60387811634349,
      "grad_norm": 0.8092751502990723,
      "learning_rate": 1.0728e-05,
      "loss": 0.0717,
      "step": 3579
    },
    {
      "epoch": 6.605724838411819,
      "grad_norm": 0.6652283668518066,
      "learning_rate": 1.0731000000000001e-05,
      "loss": 0.0618,
      "step": 3580
    },
    {
      "epoch": 6.607571560480148,
      "grad_norm": 1.0060979127883911,
      "learning_rate": 1.0734000000000001e-05,
      "loss": 0.0732,
      "step": 3581
    },
    {
      "epoch": 6.609418282548477,
      "grad_norm": 0.5705205798149109,
      "learning_rate": 1.0737e-05,
      "loss": 0.0522,
      "step": 3582
    },
    {
      "epoch": 6.611265004616805,
      "grad_norm": 0.7541183233261108,
      "learning_rate": 1.074e-05,
      "loss": 0.0398,
      "step": 3583
    },
    {
      "epoch": 6.613111726685134,
      "grad_norm": 0.9432827830314636,
      "learning_rate": 1.0743e-05,
      "loss": 0.0441,
      "step": 3584
    },
    {
      "epoch": 6.6149584487534625,
      "grad_norm": 0.8581227660179138,
      "learning_rate": 1.0746e-05,
      "loss": 0.0589,
      "step": 3585
    },
    {
      "epoch": 6.616805170821792,
      "grad_norm": 0.9152125716209412,
      "learning_rate": 1.0749e-05,
      "loss": 0.0627,
      "step": 3586
    },
    {
      "epoch": 6.61865189289012,
      "grad_norm": 0.9982067346572876,
      "learning_rate": 1.0752e-05,
      "loss": 0.0527,
      "step": 3587
    },
    {
      "epoch": 6.620498614958449,
      "grad_norm": 0.617465615272522,
      "learning_rate": 1.0755e-05,
      "loss": 0.0346,
      "step": 3588
    },
    {
      "epoch": 6.622345337026777,
      "grad_norm": 0.9789791703224182,
      "learning_rate": 1.0758e-05,
      "loss": 0.0632,
      "step": 3589
    },
    {
      "epoch": 6.624192059095106,
      "grad_norm": 0.8940677046775818,
      "learning_rate": 1.0761e-05,
      "loss": 0.0495,
      "step": 3590
    },
    {
      "epoch": 6.626038781163435,
      "grad_norm": 1.0663244724273682,
      "learning_rate": 1.0764e-05,
      "loss": 0.0678,
      "step": 3591
    },
    {
      "epoch": 6.627885503231764,
      "grad_norm": 0.8719698190689087,
      "learning_rate": 1.0767e-05,
      "loss": 0.0659,
      "step": 3592
    },
    {
      "epoch": 6.629732225300092,
      "grad_norm": 0.6242677569389343,
      "learning_rate": 1.077e-05,
      "loss": 0.0514,
      "step": 3593
    },
    {
      "epoch": 6.631578947368421,
      "grad_norm": 0.9718525409698486,
      "learning_rate": 1.0773e-05,
      "loss": 0.0539,
      "step": 3594
    },
    {
      "epoch": 6.6334256694367495,
      "grad_norm": 1.2668012380599976,
      "learning_rate": 1.0776000000000002e-05,
      "loss": 0.0596,
      "step": 3595
    },
    {
      "epoch": 6.635272391505079,
      "grad_norm": 0.9224687814712524,
      "learning_rate": 1.0779000000000001e-05,
      "loss": 0.0681,
      "step": 3596
    },
    {
      "epoch": 6.637119113573407,
      "grad_norm": 0.7575953602790833,
      "learning_rate": 1.0782e-05,
      "loss": 0.0416,
      "step": 3597
    },
    {
      "epoch": 6.638965835641736,
      "grad_norm": 0.6865046620368958,
      "learning_rate": 1.0785e-05,
      "loss": 0.0355,
      "step": 3598
    },
    {
      "epoch": 6.640812557710064,
      "grad_norm": 1.1411089897155762,
      "learning_rate": 1.0787999999999999e-05,
      "loss": 0.0513,
      "step": 3599
    },
    {
      "epoch": 6.642659279778393,
      "grad_norm": 0.8646680116653442,
      "learning_rate": 1.0791e-05,
      "loss": 0.0561,
      "step": 3600
    },
    {
      "epoch": 6.644506001846722,
      "grad_norm": 2.013439893722534,
      "learning_rate": 1.0794e-05,
      "loss": 0.0872,
      "step": 3601
    },
    {
      "epoch": 6.646352723915051,
      "grad_norm": 1.046962857246399,
      "learning_rate": 1.0797e-05,
      "loss": 0.0669,
      "step": 3602
    },
    {
      "epoch": 6.64819944598338,
      "grad_norm": 2.58697772026062,
      "learning_rate": 1.08e-05,
      "loss": 0.4191,
      "step": 3603
    },
    {
      "epoch": 6.650046168051708,
      "grad_norm": 1.5602034330368042,
      "learning_rate": 1.0803e-05,
      "loss": 0.3947,
      "step": 3604
    },
    {
      "epoch": 6.6518928901200365,
      "grad_norm": 1.0020840167999268,
      "learning_rate": 1.0806000000000001e-05,
      "loss": 0.32,
      "step": 3605
    },
    {
      "epoch": 6.653739612188366,
      "grad_norm": 1.0240564346313477,
      "learning_rate": 1.0809000000000001e-05,
      "loss": 0.2205,
      "step": 3606
    },
    {
      "epoch": 6.655586334256695,
      "grad_norm": 1.3797069787979126,
      "learning_rate": 1.0812e-05,
      "loss": 0.2478,
      "step": 3607
    },
    {
      "epoch": 6.657433056325023,
      "grad_norm": 1.4267139434814453,
      "learning_rate": 1.0815e-05,
      "loss": 0.2508,
      "step": 3608
    },
    {
      "epoch": 6.659279778393352,
      "grad_norm": 1.2222124338150024,
      "learning_rate": 1.0817999999999999e-05,
      "loss": 0.2054,
      "step": 3609
    },
    {
      "epoch": 6.66112650046168,
      "grad_norm": 1.127008318901062,
      "learning_rate": 1.0821e-05,
      "loss": 0.1975,
      "step": 3610
    },
    {
      "epoch": 6.66297322253001,
      "grad_norm": 0.9890972375869751,
      "learning_rate": 1.0824e-05,
      "loss": 0.1881,
      "step": 3611
    },
    {
      "epoch": 6.664819944598338,
      "grad_norm": 1.3429679870605469,
      "learning_rate": 1.0827e-05,
      "loss": 0.2332,
      "step": 3612
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 1.3030575513839722,
      "learning_rate": 1.083e-05,
      "loss": 0.1254,
      "step": 3613
    },
    {
      "epoch": 6.668513388734995,
      "grad_norm": 2.01155948638916,
      "learning_rate": 1.0833e-05,
      "loss": 0.1191,
      "step": 3614
    },
    {
      "epoch": 6.670360110803324,
      "grad_norm": 0.8010526299476624,
      "learning_rate": 1.0836000000000001e-05,
      "loss": 0.122,
      "step": 3615
    },
    {
      "epoch": 6.672206832871653,
      "grad_norm": 1.210434913635254,
      "learning_rate": 1.0839e-05,
      "loss": 0.0684,
      "step": 3616
    },
    {
      "epoch": 6.674053554939982,
      "grad_norm": 0.9653677344322205,
      "learning_rate": 1.0842e-05,
      "loss": 0.0914,
      "step": 3617
    },
    {
      "epoch": 6.67590027700831,
      "grad_norm": 2.1854913234710693,
      "learning_rate": 1.0845e-05,
      "loss": 0.1486,
      "step": 3618
    },
    {
      "epoch": 6.677746999076639,
      "grad_norm": 0.8655308485031128,
      "learning_rate": 1.0848e-05,
      "loss": 0.0763,
      "step": 3619
    },
    {
      "epoch": 6.679593721144967,
      "grad_norm": 0.42856207489967346,
      "learning_rate": 1.0851000000000002e-05,
      "loss": 0.0371,
      "step": 3620
    },
    {
      "epoch": 6.6814404432132966,
      "grad_norm": 0.8060868978500366,
      "learning_rate": 1.0854e-05,
      "loss": 0.0453,
      "step": 3621
    },
    {
      "epoch": 6.683287165281625,
      "grad_norm": 1.4295423030853271,
      "learning_rate": 1.0857e-05,
      "loss": 0.0821,
      "step": 3622
    },
    {
      "epoch": 6.685133887349954,
      "grad_norm": 0.5448370575904846,
      "learning_rate": 1.086e-05,
      "loss": 0.0353,
      "step": 3623
    },
    {
      "epoch": 6.686980609418282,
      "grad_norm": 0.7300196886062622,
      "learning_rate": 1.0863e-05,
      "loss": 0.0528,
      "step": 3624
    },
    {
      "epoch": 6.688827331486611,
      "grad_norm": 0.6349700093269348,
      "learning_rate": 1.0866e-05,
      "loss": 0.0526,
      "step": 3625
    },
    {
      "epoch": 6.69067405355494,
      "grad_norm": 0.401645690202713,
      "learning_rate": 1.0869e-05,
      "loss": 0.023,
      "step": 3626
    },
    {
      "epoch": 6.692520775623269,
      "grad_norm": 0.5231764912605286,
      "learning_rate": 1.0872e-05,
      "loss": 0.0398,
      "step": 3627
    },
    {
      "epoch": 6.694367497691598,
      "grad_norm": 0.9089896082878113,
      "learning_rate": 1.0875e-05,
      "loss": 0.0471,
      "step": 3628
    },
    {
      "epoch": 6.696214219759926,
      "grad_norm": 0.9645804166793823,
      "learning_rate": 1.0878e-05,
      "loss": 0.0676,
      "step": 3629
    },
    {
      "epoch": 6.698060941828254,
      "grad_norm": 0.7476101517677307,
      "learning_rate": 1.0881000000000001e-05,
      "loss": 0.0526,
      "step": 3630
    },
    {
      "epoch": 6.6999076638965835,
      "grad_norm": 0.7427716255187988,
      "learning_rate": 1.0884000000000001e-05,
      "loss": 0.0327,
      "step": 3631
    },
    {
      "epoch": 6.701754385964913,
      "grad_norm": 0.6101641058921814,
      "learning_rate": 1.0887000000000001e-05,
      "loss": 0.0413,
      "step": 3632
    },
    {
      "epoch": 6.703601108033241,
      "grad_norm": 0.7341259717941284,
      "learning_rate": 1.089e-05,
      "loss": 0.0663,
      "step": 3633
    },
    {
      "epoch": 6.70544783010157,
      "grad_norm": 0.7948614954948425,
      "learning_rate": 1.0892999999999999e-05,
      "loss": 0.0883,
      "step": 3634
    },
    {
      "epoch": 6.707294552169898,
      "grad_norm": 0.7815687656402588,
      "learning_rate": 1.0896e-05,
      "loss": 0.0323,
      "step": 3635
    },
    {
      "epoch": 6.7091412742382275,
      "grad_norm": 0.639951765537262,
      "learning_rate": 1.0899e-05,
      "loss": 0.037,
      "step": 3636
    },
    {
      "epoch": 6.710987996306556,
      "grad_norm": 0.6436964273452759,
      "learning_rate": 1.0902e-05,
      "loss": 0.0292,
      "step": 3637
    },
    {
      "epoch": 6.712834718374885,
      "grad_norm": 0.8461018800735474,
      "learning_rate": 1.0905e-05,
      "loss": 0.0442,
      "step": 3638
    },
    {
      "epoch": 6.714681440443213,
      "grad_norm": 0.724323034286499,
      "learning_rate": 1.0908e-05,
      "loss": 0.0471,
      "step": 3639
    },
    {
      "epoch": 6.716528162511542,
      "grad_norm": 1.6644361019134521,
      "learning_rate": 1.0911000000000001e-05,
      "loss": 0.0465,
      "step": 3640
    },
    {
      "epoch": 6.7183748845798705,
      "grad_norm": 1.0163922309875488,
      "learning_rate": 1.0914000000000001e-05,
      "loss": 0.0478,
      "step": 3641
    },
    {
      "epoch": 6.7202216066482,
      "grad_norm": 1.1931281089782715,
      "learning_rate": 1.0917e-05,
      "loss": 0.0705,
      "step": 3642
    },
    {
      "epoch": 6.722068328716528,
      "grad_norm": 0.6734774112701416,
      "learning_rate": 1.092e-05,
      "loss": 0.0346,
      "step": 3643
    },
    {
      "epoch": 6.723915050784857,
      "grad_norm": 0.6857860684394836,
      "learning_rate": 1.0923e-05,
      "loss": 0.0406,
      "step": 3644
    },
    {
      "epoch": 6.725761772853185,
      "grad_norm": 0.9690484404563904,
      "learning_rate": 1.0926000000000002e-05,
      "loss": 0.0689,
      "step": 3645
    },
    {
      "epoch": 6.7276084949215145,
      "grad_norm": 1.1824640035629272,
      "learning_rate": 1.0929e-05,
      "loss": 0.0527,
      "step": 3646
    },
    {
      "epoch": 6.729455216989843,
      "grad_norm": 0.6219495534896851,
      "learning_rate": 1.0932e-05,
      "loss": 0.0349,
      "step": 3647
    },
    {
      "epoch": 6.731301939058172,
      "grad_norm": 1.6175211668014526,
      "learning_rate": 1.0935e-05,
      "loss": 0.0481,
      "step": 3648
    },
    {
      "epoch": 6.7331486611265,
      "grad_norm": 0.951744794845581,
      "learning_rate": 1.0938e-05,
      "loss": 0.0421,
      "step": 3649
    },
    {
      "epoch": 6.734995383194829,
      "grad_norm": 1.0563994646072388,
      "learning_rate": 1.0941e-05,
      "loss": 0.086,
      "step": 3650
    },
    {
      "epoch": 6.7368421052631575,
      "grad_norm": 1.0864402055740356,
      "learning_rate": 1.0944e-05,
      "loss": 0.0374,
      "step": 3651
    },
    {
      "epoch": 6.738688827331487,
      "grad_norm": 1.0111253261566162,
      "learning_rate": 1.0947e-05,
      "loss": 0.0508,
      "step": 3652
    },
    {
      "epoch": 6.740535549399816,
      "grad_norm": 2.369296073913574,
      "learning_rate": 1.095e-05,
      "loss": 0.4443,
      "step": 3653
    },
    {
      "epoch": 6.742382271468144,
      "grad_norm": 2.0507547855377197,
      "learning_rate": 1.0953e-05,
      "loss": 0.3894,
      "step": 3654
    },
    {
      "epoch": 6.744228993536472,
      "grad_norm": 1.1392920017242432,
      "learning_rate": 1.0956000000000001e-05,
      "loss": 0.3178,
      "step": 3655
    },
    {
      "epoch": 6.7460757156048015,
      "grad_norm": 0.9975927472114563,
      "learning_rate": 1.0959000000000001e-05,
      "loss": 0.2191,
      "step": 3656
    },
    {
      "epoch": 6.747922437673131,
      "grad_norm": 1.681461215019226,
      "learning_rate": 1.0962000000000001e-05,
      "loss": 0.2412,
      "step": 3657
    },
    {
      "epoch": 6.749769159741459,
      "grad_norm": 1.2123366594314575,
      "learning_rate": 1.0965e-05,
      "loss": 0.1996,
      "step": 3658
    },
    {
      "epoch": 6.751615881809788,
      "grad_norm": 0.9915981292724609,
      "learning_rate": 1.0967999999999999e-05,
      "loss": 0.2356,
      "step": 3659
    },
    {
      "epoch": 6.753462603878116,
      "grad_norm": 1.0750139951705933,
      "learning_rate": 1.0971e-05,
      "loss": 0.1982,
      "step": 3660
    },
    {
      "epoch": 6.755309325946445,
      "grad_norm": 1.2357497215270996,
      "learning_rate": 1.0974e-05,
      "loss": 0.2488,
      "step": 3661
    },
    {
      "epoch": 6.757156048014774,
      "grad_norm": 1.0140907764434814,
      "learning_rate": 1.0977e-05,
      "loss": 0.1695,
      "step": 3662
    },
    {
      "epoch": 6.759002770083103,
      "grad_norm": 0.7370830774307251,
      "learning_rate": 1.098e-05,
      "loss": 0.1406,
      "step": 3663
    },
    {
      "epoch": 6.760849492151431,
      "grad_norm": 1.0511841773986816,
      "learning_rate": 1.0983e-05,
      "loss": 0.1273,
      "step": 3664
    },
    {
      "epoch": 6.76269621421976,
      "grad_norm": 0.872833251953125,
      "learning_rate": 1.0986000000000001e-05,
      "loss": 0.1098,
      "step": 3665
    },
    {
      "epoch": 6.7645429362880884,
      "grad_norm": 0.6190897226333618,
      "learning_rate": 1.0989000000000001e-05,
      "loss": 0.0684,
      "step": 3666
    },
    {
      "epoch": 6.766389658356418,
      "grad_norm": 0.555406928062439,
      "learning_rate": 1.0992e-05,
      "loss": 0.0599,
      "step": 3667
    },
    {
      "epoch": 6.768236380424746,
      "grad_norm": 0.817185640335083,
      "learning_rate": 1.0995e-05,
      "loss": 0.1271,
      "step": 3668
    },
    {
      "epoch": 6.770083102493075,
      "grad_norm": 0.9192994832992554,
      "learning_rate": 1.0998e-05,
      "loss": 0.0455,
      "step": 3669
    },
    {
      "epoch": 6.771929824561403,
      "grad_norm": 0.5491228103637695,
      "learning_rate": 1.1001e-05,
      "loss": 0.0603,
      "step": 3670
    },
    {
      "epoch": 6.773776546629732,
      "grad_norm": 0.8587676882743835,
      "learning_rate": 1.1004e-05,
      "loss": 0.0706,
      "step": 3671
    },
    {
      "epoch": 6.775623268698061,
      "grad_norm": 0.6606636643409729,
      "learning_rate": 1.1007e-05,
      "loss": 0.0618,
      "step": 3672
    },
    {
      "epoch": 6.77746999076639,
      "grad_norm": 0.4917014539241791,
      "learning_rate": 1.101e-05,
      "loss": 0.0335,
      "step": 3673
    },
    {
      "epoch": 6.779316712834718,
      "grad_norm": 0.627388060092926,
      "learning_rate": 1.1013e-05,
      "loss": 0.0368,
      "step": 3674
    },
    {
      "epoch": 6.781163434903047,
      "grad_norm": 0.827018141746521,
      "learning_rate": 1.1016000000000001e-05,
      "loss": 0.0857,
      "step": 3675
    },
    {
      "epoch": 6.783010156971375,
      "grad_norm": 0.538213312625885,
      "learning_rate": 1.1019e-05,
      "loss": 0.0322,
      "step": 3676
    },
    {
      "epoch": 6.784856879039705,
      "grad_norm": 0.504508376121521,
      "learning_rate": 1.1022e-05,
      "loss": 0.042,
      "step": 3677
    },
    {
      "epoch": 6.786703601108034,
      "grad_norm": 0.72398841381073,
      "learning_rate": 1.1025e-05,
      "loss": 0.0388,
      "step": 3678
    },
    {
      "epoch": 6.788550323176362,
      "grad_norm": 0.9289087653160095,
      "learning_rate": 1.1028e-05,
      "loss": 0.0386,
      "step": 3679
    },
    {
      "epoch": 6.79039704524469,
      "grad_norm": 0.6446580290794373,
      "learning_rate": 1.1031000000000002e-05,
      "loss": 0.0418,
      "step": 3680
    },
    {
      "epoch": 6.792243767313019,
      "grad_norm": 0.7011333107948303,
      "learning_rate": 1.1034000000000001e-05,
      "loss": 0.0596,
      "step": 3681
    },
    {
      "epoch": 6.7940904893813485,
      "grad_norm": 0.6451525092124939,
      "learning_rate": 1.1037000000000001e-05,
      "loss": 0.0462,
      "step": 3682
    },
    {
      "epoch": 6.795937211449677,
      "grad_norm": 0.9175089001655579,
      "learning_rate": 1.104e-05,
      "loss": 0.0433,
      "step": 3683
    },
    {
      "epoch": 6.797783933518006,
      "grad_norm": 0.9872082471847534,
      "learning_rate": 1.1042999999999999e-05,
      "loss": 0.048,
      "step": 3684
    },
    {
      "epoch": 6.799630655586334,
      "grad_norm": 0.7646401524543762,
      "learning_rate": 1.1046e-05,
      "loss": 0.0568,
      "step": 3685
    },
    {
      "epoch": 6.801477377654663,
      "grad_norm": 0.5979199409484863,
      "learning_rate": 1.1049e-05,
      "loss": 0.0415,
      "step": 3686
    },
    {
      "epoch": 6.803324099722992,
      "grad_norm": 0.816354513168335,
      "learning_rate": 1.1052e-05,
      "loss": 0.0618,
      "step": 3687
    },
    {
      "epoch": 6.805170821791321,
      "grad_norm": 1.424713373184204,
      "learning_rate": 1.1055e-05,
      "loss": 0.038,
      "step": 3688
    },
    {
      "epoch": 6.807017543859649,
      "grad_norm": 0.6271553635597229,
      "learning_rate": 1.1058e-05,
      "loss": 0.0324,
      "step": 3689
    },
    {
      "epoch": 6.808864265927978,
      "grad_norm": 0.9549874663352966,
      "learning_rate": 1.1061000000000001e-05,
      "loss": 0.1012,
      "step": 3690
    },
    {
      "epoch": 6.810710987996306,
      "grad_norm": 0.8129993677139282,
      "learning_rate": 1.1064000000000001e-05,
      "loss": 0.0662,
      "step": 3691
    },
    {
      "epoch": 6.8125577100646355,
      "grad_norm": 2.0147483348846436,
      "learning_rate": 1.1067000000000001e-05,
      "loss": 0.0565,
      "step": 3692
    },
    {
      "epoch": 6.814404432132964,
      "grad_norm": 1.7509359121322632,
      "learning_rate": 1.107e-05,
      "loss": 0.0402,
      "step": 3693
    },
    {
      "epoch": 6.816251154201293,
      "grad_norm": 0.9780064225196838,
      "learning_rate": 1.1073e-05,
      "loss": 0.0488,
      "step": 3694
    },
    {
      "epoch": 6.818097876269621,
      "grad_norm": 0.8157630562782288,
      "learning_rate": 1.1075999999999999e-05,
      "loss": 0.0635,
      "step": 3695
    },
    {
      "epoch": 6.81994459833795,
      "grad_norm": 0.9507924318313599,
      "learning_rate": 1.1079e-05,
      "loss": 0.0589,
      "step": 3696
    },
    {
      "epoch": 6.821791320406279,
      "grad_norm": 0.7969843149185181,
      "learning_rate": 1.1082e-05,
      "loss": 0.0434,
      "step": 3697
    },
    {
      "epoch": 6.823638042474608,
      "grad_norm": 10.65235424041748,
      "learning_rate": 1.1085e-05,
      "loss": 0.0229,
      "step": 3698
    },
    {
      "epoch": 6.825484764542936,
      "grad_norm": 0.8875689506530762,
      "learning_rate": 1.1088e-05,
      "loss": 0.0812,
      "step": 3699
    },
    {
      "epoch": 6.827331486611265,
      "grad_norm": 0.5887988805770874,
      "learning_rate": 1.1091e-05,
      "loss": 0.0502,
      "step": 3700
    },
    {
      "epoch": 6.829178208679593,
      "grad_norm": 1.0622601509094238,
      "learning_rate": 1.1094e-05,
      "loss": 0.0771,
      "step": 3701
    },
    {
      "epoch": 6.8310249307479225,
      "grad_norm": 1.051266074180603,
      "learning_rate": 1.1097e-05,
      "loss": 0.0844,
      "step": 3702
    },
    {
      "epoch": 6.832871652816252,
      "grad_norm": 2.6369080543518066,
      "learning_rate": 1.11e-05,
      "loss": 0.4615,
      "step": 3703
    },
    {
      "epoch": 6.83471837488458,
      "grad_norm": 1.35551118850708,
      "learning_rate": 1.1103e-05,
      "loss": 0.2638,
      "step": 3704
    },
    {
      "epoch": 6.836565096952908,
      "grad_norm": 1.556361198425293,
      "learning_rate": 1.1106e-05,
      "loss": 0.3182,
      "step": 3705
    },
    {
      "epoch": 6.838411819021237,
      "grad_norm": 1.2559797763824463,
      "learning_rate": 1.1109000000000002e-05,
      "loss": 0.3061,
      "step": 3706
    },
    {
      "epoch": 6.840258541089566,
      "grad_norm": 1.3399794101715088,
      "learning_rate": 1.1112e-05,
      "loss": 0.297,
      "step": 3707
    },
    {
      "epoch": 6.842105263157895,
      "grad_norm": 1.2586995363235474,
      "learning_rate": 1.1115e-05,
      "loss": 0.2254,
      "step": 3708
    },
    {
      "epoch": 6.843951985226224,
      "grad_norm": 1.0613036155700684,
      "learning_rate": 1.1118e-05,
      "loss": 0.2483,
      "step": 3709
    },
    {
      "epoch": 6.845798707294552,
      "grad_norm": 1.3740078210830688,
      "learning_rate": 1.1120999999999999e-05,
      "loss": 0.2313,
      "step": 3710
    },
    {
      "epoch": 6.847645429362881,
      "grad_norm": 0.8254731297492981,
      "learning_rate": 1.1124e-05,
      "loss": 0.1593,
      "step": 3711
    },
    {
      "epoch": 6.8494921514312095,
      "grad_norm": 0.7780969142913818,
      "learning_rate": 1.1127e-05,
      "loss": 0.1293,
      "step": 3712
    },
    {
      "epoch": 6.851338873499539,
      "grad_norm": 0.97072833776474,
      "learning_rate": 1.113e-05,
      "loss": 0.1721,
      "step": 3713
    },
    {
      "epoch": 6.853185595567867,
      "grad_norm": 0.9659312963485718,
      "learning_rate": 1.1133e-05,
      "loss": 0.1715,
      "step": 3714
    },
    {
      "epoch": 6.855032317636196,
      "grad_norm": 1.041581630706787,
      "learning_rate": 1.1136e-05,
      "loss": 0.2042,
      "step": 3715
    },
    {
      "epoch": 6.856879039704524,
      "grad_norm": 0.5041184425354004,
      "learning_rate": 1.1139000000000001e-05,
      "loss": 0.0725,
      "step": 3716
    },
    {
      "epoch": 6.858725761772853,
      "grad_norm": 0.4641791880130768,
      "learning_rate": 1.1142000000000001e-05,
      "loss": 0.0548,
      "step": 3717
    },
    {
      "epoch": 6.860572483841182,
      "grad_norm": 0.6150814294815063,
      "learning_rate": 1.1145000000000001e-05,
      "loss": 0.0657,
      "step": 3718
    },
    {
      "epoch": 6.862419205909511,
      "grad_norm": 0.9226855039596558,
      "learning_rate": 1.1148e-05,
      "loss": 0.0709,
      "step": 3719
    },
    {
      "epoch": 6.864265927977839,
      "grad_norm": 0.5464257001876831,
      "learning_rate": 1.1150999999999999e-05,
      "loss": 0.0448,
      "step": 3720
    },
    {
      "epoch": 6.866112650046168,
      "grad_norm": 0.9268885850906372,
      "learning_rate": 1.1154e-05,
      "loss": 0.0544,
      "step": 3721
    },
    {
      "epoch": 6.8679593721144965,
      "grad_norm": 0.6879826784133911,
      "learning_rate": 1.1157e-05,
      "loss": 0.0437,
      "step": 3722
    },
    {
      "epoch": 6.869806094182826,
      "grad_norm": 0.6709389686584473,
      "learning_rate": 1.116e-05,
      "loss": 0.0527,
      "step": 3723
    },
    {
      "epoch": 6.871652816251154,
      "grad_norm": 0.3872710168361664,
      "learning_rate": 1.1163e-05,
      "loss": 0.0256,
      "step": 3724
    },
    {
      "epoch": 6.873499538319483,
      "grad_norm": 0.6115924715995789,
      "learning_rate": 1.1166e-05,
      "loss": 0.0489,
      "step": 3725
    },
    {
      "epoch": 6.875346260387811,
      "grad_norm": 0.6553891897201538,
      "learning_rate": 1.1169000000000001e-05,
      "loss": 0.0577,
      "step": 3726
    },
    {
      "epoch": 6.87719298245614,
      "grad_norm": 1.1549395322799683,
      "learning_rate": 1.1172e-05,
      "loss": 0.0787,
      "step": 3727
    },
    {
      "epoch": 6.87903970452447,
      "grad_norm": 0.7184242606163025,
      "learning_rate": 1.1175e-05,
      "loss": 0.0511,
      "step": 3728
    },
    {
      "epoch": 6.880886426592798,
      "grad_norm": 0.7228502035140991,
      "learning_rate": 1.1178e-05,
      "loss": 0.0506,
      "step": 3729
    },
    {
      "epoch": 6.882733148661126,
      "grad_norm": 1.02662193775177,
      "learning_rate": 1.1181e-05,
      "loss": 0.0562,
      "step": 3730
    },
    {
      "epoch": 6.884579870729455,
      "grad_norm": 0.9584940075874329,
      "learning_rate": 1.1184000000000002e-05,
      "loss": 0.0522,
      "step": 3731
    },
    {
      "epoch": 6.886426592797784,
      "grad_norm": 0.7661015391349792,
      "learning_rate": 1.1187e-05,
      "loss": 0.075,
      "step": 3732
    },
    {
      "epoch": 6.888273314866113,
      "grad_norm": 0.6086714267730713,
      "learning_rate": 1.119e-05,
      "loss": 0.0482,
      "step": 3733
    },
    {
      "epoch": 6.890120036934442,
      "grad_norm": 0.8402490019798279,
      "learning_rate": 1.1193e-05,
      "loss": 0.0461,
      "step": 3734
    },
    {
      "epoch": 6.89196675900277,
      "grad_norm": 0.9646185040473938,
      "learning_rate": 1.1196e-05,
      "loss": 0.0348,
      "step": 3735
    },
    {
      "epoch": 6.893813481071099,
      "grad_norm": 0.8491271734237671,
      "learning_rate": 1.1199e-05,
      "loss": 0.0701,
      "step": 3736
    },
    {
      "epoch": 6.895660203139427,
      "grad_norm": 1.3548365831375122,
      "learning_rate": 1.1202e-05,
      "loss": 0.0643,
      "step": 3737
    },
    {
      "epoch": 6.8975069252077565,
      "grad_norm": 1.1035486459732056,
      "learning_rate": 1.1205e-05,
      "loss": 0.0643,
      "step": 3738
    },
    {
      "epoch": 6.899353647276085,
      "grad_norm": 0.5701906085014343,
      "learning_rate": 1.1208e-05,
      "loss": 0.0401,
      "step": 3739
    },
    {
      "epoch": 6.901200369344414,
      "grad_norm": 0.7111980319023132,
      "learning_rate": 1.1211e-05,
      "loss": 0.0554,
      "step": 3740
    },
    {
      "epoch": 6.903047091412742,
      "grad_norm": 0.6047128438949585,
      "learning_rate": 1.1214000000000001e-05,
      "loss": 0.0409,
      "step": 3741
    },
    {
      "epoch": 6.904893813481071,
      "grad_norm": 1.317233681678772,
      "learning_rate": 1.1217000000000001e-05,
      "loss": 0.0716,
      "step": 3742
    },
    {
      "epoch": 6.9067405355494,
      "grad_norm": 0.8000737428665161,
      "learning_rate": 1.1220000000000001e-05,
      "loss": 0.0484,
      "step": 3743
    },
    {
      "epoch": 6.908587257617729,
      "grad_norm": 0.9985203146934509,
      "learning_rate": 1.1222999999999999e-05,
      "loss": 0.0612,
      "step": 3744
    },
    {
      "epoch": 6.910433979686057,
      "grad_norm": 0.8063696026802063,
      "learning_rate": 1.1225999999999999e-05,
      "loss": 0.0494,
      "step": 3745
    },
    {
      "epoch": 6.912280701754386,
      "grad_norm": 0.9671680331230164,
      "learning_rate": 1.1229e-05,
      "loss": 0.0465,
      "step": 3746
    },
    {
      "epoch": 6.914127423822714,
      "grad_norm": 0.8081934452056885,
      "learning_rate": 1.1232e-05,
      "loss": 0.0603,
      "step": 3747
    },
    {
      "epoch": 6.9159741458910435,
      "grad_norm": 0.6315751671791077,
      "learning_rate": 1.1235e-05,
      "loss": 0.0494,
      "step": 3748
    },
    {
      "epoch": 6.917820867959372,
      "grad_norm": 1.0424327850341797,
      "learning_rate": 1.1238e-05,
      "loss": 0.0743,
      "step": 3749
    },
    {
      "epoch": 6.919667590027701,
      "grad_norm": 0.9841300249099731,
      "learning_rate": 1.1241e-05,
      "loss": 0.086,
      "step": 3750
    },
    {
      "epoch": 6.921514312096029,
      "grad_norm": 0.7991226315498352,
      "learning_rate": 1.1244000000000001e-05,
      "loss": 0.045,
      "step": 3751
    },
    {
      "epoch": 6.923361034164358,
      "grad_norm": 1.1342779397964478,
      "learning_rate": 1.1247000000000001e-05,
      "loss": 0.0686,
      "step": 3752
    },
    {
      "epoch": 6.9252077562326875,
      "grad_norm": 1.4662410020828247,
      "learning_rate": 1.125e-05,
      "loss": 0.2982,
      "step": 3753
    },
    {
      "epoch": 6.927054478301016,
      "grad_norm": 1.1601957082748413,
      "learning_rate": 1.1253e-05,
      "loss": 0.3563,
      "step": 3754
    },
    {
      "epoch": 6.928901200369344,
      "grad_norm": 0.7693567872047424,
      "learning_rate": 1.1256e-05,
      "loss": 0.2017,
      "step": 3755
    },
    {
      "epoch": 6.930747922437673,
      "grad_norm": 1.183059573173523,
      "learning_rate": 1.1259e-05,
      "loss": 0.2333,
      "step": 3756
    },
    {
      "epoch": 6.932594644506002,
      "grad_norm": 0.8615180850028992,
      "learning_rate": 1.1262e-05,
      "loss": 0.2091,
      "step": 3757
    },
    {
      "epoch": 6.9344413665743305,
      "grad_norm": 0.9820761680603027,
      "learning_rate": 1.1265e-05,
      "loss": 0.177,
      "step": 3758
    },
    {
      "epoch": 6.93628808864266,
      "grad_norm": 1.0718605518341064,
      "learning_rate": 1.1268e-05,
      "loss": 0.2127,
      "step": 3759
    },
    {
      "epoch": 6.938134810710988,
      "grad_norm": 0.8176330327987671,
      "learning_rate": 1.1271e-05,
      "loss": 0.1607,
      "step": 3760
    },
    {
      "epoch": 6.939981532779317,
      "grad_norm": 0.8427945375442505,
      "learning_rate": 1.1274e-05,
      "loss": 0.1712,
      "step": 3761
    },
    {
      "epoch": 6.941828254847645,
      "grad_norm": 0.6818265318870544,
      "learning_rate": 1.1277e-05,
      "loss": 0.1341,
      "step": 3762
    },
    {
      "epoch": 6.9436749769159745,
      "grad_norm": 0.7700383067131042,
      "learning_rate": 1.128e-05,
      "loss": 0.1158,
      "step": 3763
    },
    {
      "epoch": 6.945521698984303,
      "grad_norm": 0.5608851313591003,
      "learning_rate": 1.1283e-05,
      "loss": 0.0953,
      "step": 3764
    },
    {
      "epoch": 6.947368421052632,
      "grad_norm": 0.6973187923431396,
      "learning_rate": 1.1286e-05,
      "loss": 0.0873,
      "step": 3765
    },
    {
      "epoch": 6.94921514312096,
      "grad_norm": 0.7664932608604431,
      "learning_rate": 1.1289000000000002e-05,
      "loss": 0.1105,
      "step": 3766
    },
    {
      "epoch": 6.951061865189289,
      "grad_norm": 0.8267348408699036,
      "learning_rate": 1.1292000000000001e-05,
      "loss": 0.1101,
      "step": 3767
    },
    {
      "epoch": 6.9529085872576175,
      "grad_norm": 0.5690252184867859,
      "learning_rate": 1.1295000000000001e-05,
      "loss": 0.0683,
      "step": 3768
    },
    {
      "epoch": 6.954755309325947,
      "grad_norm": 0.5724848508834839,
      "learning_rate": 1.1298e-05,
      "loss": 0.0459,
      "step": 3769
    },
    {
      "epoch": 6.956602031394275,
      "grad_norm": 0.646037220954895,
      "learning_rate": 1.1300999999999999e-05,
      "loss": 0.0703,
      "step": 3770
    },
    {
      "epoch": 6.958448753462604,
      "grad_norm": 0.6106911301612854,
      "learning_rate": 1.1304e-05,
      "loss": 0.0535,
      "step": 3771
    },
    {
      "epoch": 6.960295475530932,
      "grad_norm": 0.5185670852661133,
      "learning_rate": 1.1307e-05,
      "loss": 0.0474,
      "step": 3772
    },
    {
      "epoch": 6.9621421975992615,
      "grad_norm": 0.9679078459739685,
      "learning_rate": 1.131e-05,
      "loss": 0.0386,
      "step": 3773
    },
    {
      "epoch": 6.96398891966759,
      "grad_norm": 0.6940216422080994,
      "learning_rate": 1.1313e-05,
      "loss": 0.0564,
      "step": 3774
    },
    {
      "epoch": 6.965835641735919,
      "grad_norm": 0.6831811666488647,
      "learning_rate": 1.1316e-05,
      "loss": 0.0392,
      "step": 3775
    },
    {
      "epoch": 6.967682363804247,
      "grad_norm": 0.7709175944328308,
      "learning_rate": 1.1319000000000001e-05,
      "loss": 0.0409,
      "step": 3776
    },
    {
      "epoch": 6.969529085872576,
      "grad_norm": 0.7472397685050964,
      "learning_rate": 1.1322000000000001e-05,
      "loss": 0.0447,
      "step": 3777
    },
    {
      "epoch": 6.971375807940905,
      "grad_norm": 0.7726459503173828,
      "learning_rate": 1.1325e-05,
      "loss": 0.0454,
      "step": 3778
    },
    {
      "epoch": 6.973222530009234,
      "grad_norm": 0.6965910792350769,
      "learning_rate": 1.1328e-05,
      "loss": 0.0392,
      "step": 3779
    },
    {
      "epoch": 6.975069252077562,
      "grad_norm": 0.45253920555114746,
      "learning_rate": 1.1331e-05,
      "loss": 0.0251,
      "step": 3780
    },
    {
      "epoch": 6.976915974145891,
      "grad_norm": 0.9065711498260498,
      "learning_rate": 1.1334e-05,
      "loss": 0.0474,
      "step": 3781
    },
    {
      "epoch": 6.97876269621422,
      "grad_norm": 0.5685831308364868,
      "learning_rate": 1.1337e-05,
      "loss": 0.0252,
      "step": 3782
    },
    {
      "epoch": 6.980609418282548,
      "grad_norm": 0.6188499927520752,
      "learning_rate": 1.134e-05,
      "loss": 0.0251,
      "step": 3783
    },
    {
      "epoch": 6.982456140350877,
      "grad_norm": 0.655066967010498,
      "learning_rate": 1.1343e-05,
      "loss": 0.0445,
      "step": 3784
    },
    {
      "epoch": 6.984302862419206,
      "grad_norm": 0.9167073369026184,
      "learning_rate": 1.1346e-05,
      "loss": 0.0377,
      "step": 3785
    },
    {
      "epoch": 6.986149584487535,
      "grad_norm": 0.6690863966941833,
      "learning_rate": 1.1349000000000001e-05,
      "loss": 0.0368,
      "step": 3786
    },
    {
      "epoch": 6.987996306555863,
      "grad_norm": 0.4566117525100708,
      "learning_rate": 1.1352e-05,
      "loss": 0.0262,
      "step": 3787
    },
    {
      "epoch": 6.989843028624192,
      "grad_norm": 0.8612862825393677,
      "learning_rate": 1.1355e-05,
      "loss": 0.0529,
      "step": 3788
    },
    {
      "epoch": 6.991689750692521,
      "grad_norm": 0.8186120986938477,
      "learning_rate": 1.1358e-05,
      "loss": 0.055,
      "step": 3789
    },
    {
      "epoch": 6.99353647276085,
      "grad_norm": 1.036474347114563,
      "learning_rate": 1.1361e-05,
      "loss": 0.059,
      "step": 3790
    },
    {
      "epoch": 6.995383194829178,
      "grad_norm": 0.9849673509597778,
      "learning_rate": 1.1364000000000002e-05,
      "loss": 0.0572,
      "step": 3791
    },
    {
      "epoch": 6.997229916897507,
      "grad_norm": 0.7555782198905945,
      "learning_rate": 1.1367000000000001e-05,
      "loss": 0.0437,
      "step": 3792
    },
    {
      "epoch": 6.999076638965835,
      "grad_norm": 1.5732660293579102,
      "learning_rate": 1.137e-05,
      "loss": 0.084,
      "step": 3793
    },
    {
      "epoch": 7.0,
      "grad_norm": 1.2220367193222046,
      "learning_rate": 1.1373e-05,
      "loss": 0.0706,
      "step": 3794
    },
    {
      "epoch": 7.001846722068329,
      "grad_norm": 1.5011942386627197,
      "learning_rate": 1.1376e-05,
      "loss": 0.3325,
      "step": 3795
    },
    {
      "epoch": 7.003693444136657,
      "grad_norm": 1.134968876838684,
      "learning_rate": 1.1379e-05,
      "loss": 0.3102,
      "step": 3796
    },
    {
      "epoch": 7.0055401662049865,
      "grad_norm": 1.1414200067520142,
      "learning_rate": 1.1382e-05,
      "loss": 0.289,
      "step": 3797
    },
    {
      "epoch": 7.007386888273315,
      "grad_norm": 1.4115161895751953,
      "learning_rate": 1.1385e-05,
      "loss": 0.311,
      "step": 3798
    },
    {
      "epoch": 7.009233610341644,
      "grad_norm": 1.1737440824508667,
      "learning_rate": 1.1388e-05,
      "loss": 0.2275,
      "step": 3799
    },
    {
      "epoch": 7.011080332409972,
      "grad_norm": 0.9778410196304321,
      "learning_rate": 1.1391e-05,
      "loss": 0.2252,
      "step": 3800
    },
    {
      "epoch": 7.012927054478301,
      "grad_norm": 1.2007774114608765,
      "learning_rate": 1.1394000000000001e-05,
      "loss": 0.2059,
      "step": 3801
    },
    {
      "epoch": 7.01477377654663,
      "grad_norm": 0.9967468976974487,
      "learning_rate": 1.1397000000000001e-05,
      "loss": 0.1763,
      "step": 3802
    },
    {
      "epoch": 7.016620498614959,
      "grad_norm": 0.8212352395057678,
      "learning_rate": 1.1400000000000001e-05,
      "loss": 0.1688,
      "step": 3803
    },
    {
      "epoch": 7.018467220683287,
      "grad_norm": 0.7178338170051575,
      "learning_rate": 1.1403e-05,
      "loss": 0.1367,
      "step": 3804
    },
    {
      "epoch": 7.020313942751616,
      "grad_norm": 0.8649212121963501,
      "learning_rate": 1.1406e-05,
      "loss": 0.1519,
      "step": 3805
    },
    {
      "epoch": 7.022160664819944,
      "grad_norm": 0.7651941776275635,
      "learning_rate": 1.1409e-05,
      "loss": 0.1076,
      "step": 3806
    },
    {
      "epoch": 7.0240073868882735,
      "grad_norm": 0.6185657978057861,
      "learning_rate": 1.1412e-05,
      "loss": 0.0653,
      "step": 3807
    },
    {
      "epoch": 7.025854108956602,
      "grad_norm": 0.6425574421882629,
      "learning_rate": 1.1415e-05,
      "loss": 0.0571,
      "step": 3808
    },
    {
      "epoch": 7.027700831024931,
      "grad_norm": 0.7995126247406006,
      "learning_rate": 1.1418e-05,
      "loss": 0.0715,
      "step": 3809
    },
    {
      "epoch": 7.029547553093259,
      "grad_norm": 0.525067925453186,
      "learning_rate": 1.1421e-05,
      "loss": 0.0417,
      "step": 3810
    },
    {
      "epoch": 7.031394275161588,
      "grad_norm": 0.8681696057319641,
      "learning_rate": 1.1424000000000001e-05,
      "loss": 0.054,
      "step": 3811
    },
    {
      "epoch": 7.033240997229917,
      "grad_norm": 0.5476704239845276,
      "learning_rate": 1.1427000000000001e-05,
      "loss": 0.0404,
      "step": 3812
    },
    {
      "epoch": 7.035087719298246,
      "grad_norm": 0.5106635689735413,
      "learning_rate": 1.143e-05,
      "loss": 0.0505,
      "step": 3813
    },
    {
      "epoch": 7.036934441366574,
      "grad_norm": 0.837536096572876,
      "learning_rate": 1.1433e-05,
      "loss": 0.0446,
      "step": 3814
    },
    {
      "epoch": 7.038781163434903,
      "grad_norm": 0.47881874442100525,
      "learning_rate": 1.1436e-05,
      "loss": 0.0343,
      "step": 3815
    },
    {
      "epoch": 7.040627885503231,
      "grad_norm": 1.3114343881607056,
      "learning_rate": 1.1439e-05,
      "loss": 0.0292,
      "step": 3816
    },
    {
      "epoch": 7.0424746075715605,
      "grad_norm": 0.8697632551193237,
      "learning_rate": 1.1442000000000002e-05,
      "loss": 0.051,
      "step": 3817
    },
    {
      "epoch": 7.044321329639889,
      "grad_norm": 0.7076679468154907,
      "learning_rate": 1.1445e-05,
      "loss": 0.0379,
      "step": 3818
    },
    {
      "epoch": 7.046168051708218,
      "grad_norm": 0.933702826499939,
      "learning_rate": 1.1448e-05,
      "loss": 0.049,
      "step": 3819
    },
    {
      "epoch": 7.048014773776547,
      "grad_norm": 0.5204328298568726,
      "learning_rate": 1.1451e-05,
      "loss": 0.0374,
      "step": 3820
    },
    {
      "epoch": 7.049861495844875,
      "grad_norm": 0.8326895236968994,
      "learning_rate": 1.1453999999999999e-05,
      "loss": 0.0464,
      "step": 3821
    },
    {
      "epoch": 7.0517082179132045,
      "grad_norm": 1.0531866550445557,
      "learning_rate": 1.1457e-05,
      "loss": 0.0297,
      "step": 3822
    },
    {
      "epoch": 7.053554939981533,
      "grad_norm": 0.4825021028518677,
      "learning_rate": 1.146e-05,
      "loss": 0.0404,
      "step": 3823
    },
    {
      "epoch": 7.055401662049862,
      "grad_norm": 0.8723925352096558,
      "learning_rate": 1.1463e-05,
      "loss": 0.0398,
      "step": 3824
    },
    {
      "epoch": 7.05724838411819,
      "grad_norm": 0.787763237953186,
      "learning_rate": 1.1466e-05,
      "loss": 0.0409,
      "step": 3825
    },
    {
      "epoch": 7.059095106186519,
      "grad_norm": 1.250771164894104,
      "learning_rate": 1.1469e-05,
      "loss": 0.063,
      "step": 3826
    },
    {
      "epoch": 7.0609418282548475,
      "grad_norm": 1.2153242826461792,
      "learning_rate": 1.1472000000000001e-05,
      "loss": 0.0315,
      "step": 3827
    },
    {
      "epoch": 7.062788550323177,
      "grad_norm": 0.47222232818603516,
      "learning_rate": 1.1475000000000001e-05,
      "loss": 0.0348,
      "step": 3828
    },
    {
      "epoch": 7.064635272391505,
      "grad_norm": 1.2347041368484497,
      "learning_rate": 1.1478000000000001e-05,
      "loss": 0.0351,
      "step": 3829
    },
    {
      "epoch": 7.066481994459834,
      "grad_norm": 0.6874457597732544,
      "learning_rate": 1.1480999999999999e-05,
      "loss": 0.0392,
      "step": 3830
    },
    {
      "epoch": 7.068328716528162,
      "grad_norm": 0.6906430125236511,
      "learning_rate": 1.1483999999999999e-05,
      "loss": 0.048,
      "step": 3831
    },
    {
      "epoch": 7.0701754385964914,
      "grad_norm": 0.5939635038375854,
      "learning_rate": 1.1487e-05,
      "loss": 0.0296,
      "step": 3832
    },
    {
      "epoch": 7.07202216066482,
      "grad_norm": 0.6054182052612305,
      "learning_rate": 1.149e-05,
      "loss": 0.0344,
      "step": 3833
    },
    {
      "epoch": 7.073868882733149,
      "grad_norm": 1.7528551816940308,
      "learning_rate": 1.1493e-05,
      "loss": 0.0575,
      "step": 3834
    },
    {
      "epoch": 7.075715604801477,
      "grad_norm": 1.990154504776001,
      "learning_rate": 1.1496e-05,
      "loss": 0.0416,
      "step": 3835
    },
    {
      "epoch": 7.077562326869806,
      "grad_norm": 0.8029806613922119,
      "learning_rate": 1.1499e-05,
      "loss": 0.0413,
      "step": 3836
    },
    {
      "epoch": 7.0794090489381345,
      "grad_norm": 0.7306238412857056,
      "learning_rate": 1.1502000000000001e-05,
      "loss": 0.0336,
      "step": 3837
    },
    {
      "epoch": 7.081255771006464,
      "grad_norm": 0.5533416867256165,
      "learning_rate": 1.1505e-05,
      "loss": 0.0253,
      "step": 3838
    },
    {
      "epoch": 7.083102493074792,
      "grad_norm": 1.076889991760254,
      "learning_rate": 1.1508e-05,
      "loss": 0.0271,
      "step": 3839
    },
    {
      "epoch": 7.084949215143121,
      "grad_norm": 0.8983293175697327,
      "learning_rate": 1.1511e-05,
      "loss": 0.0578,
      "step": 3840
    },
    {
      "epoch": 7.086795937211449,
      "grad_norm": 0.756494402885437,
      "learning_rate": 1.1514e-05,
      "loss": 0.0386,
      "step": 3841
    },
    {
      "epoch": 7.088642659279778,
      "grad_norm": 0.9268624186515808,
      "learning_rate": 1.1517e-05,
      "loss": 0.0378,
      "step": 3842
    },
    {
      "epoch": 7.090489381348107,
      "grad_norm": 0.760355532169342,
      "learning_rate": 1.152e-05,
      "loss": 0.0555,
      "step": 3843
    },
    {
      "epoch": 7.092336103416436,
      "grad_norm": 0.9901613593101501,
      "learning_rate": 1.1523e-05,
      "loss": 0.0419,
      "step": 3844
    },
    {
      "epoch": 7.094182825484765,
      "grad_norm": 0.8592517971992493,
      "learning_rate": 1.1526e-05,
      "loss": 0.2719,
      "step": 3845
    },
    {
      "epoch": 7.096029547553093,
      "grad_norm": 1.2683329582214355,
      "learning_rate": 1.1529e-05,
      "loss": 0.3161,
      "step": 3846
    },
    {
      "epoch": 7.097876269621422,
      "grad_norm": 1.0169157981872559,
      "learning_rate": 1.1532e-05,
      "loss": 0.2672,
      "step": 3847
    },
    {
      "epoch": 7.099722991689751,
      "grad_norm": 0.9672853350639343,
      "learning_rate": 1.1535e-05,
      "loss": 0.2085,
      "step": 3848
    },
    {
      "epoch": 7.10156971375808,
      "grad_norm": 1.0237081050872803,
      "learning_rate": 1.1538e-05,
      "loss": 0.221,
      "step": 3849
    },
    {
      "epoch": 7.103416435826408,
      "grad_norm": 0.6840682625770569,
      "learning_rate": 1.1541e-05,
      "loss": 0.1727,
      "step": 3850
    },
    {
      "epoch": 7.105263157894737,
      "grad_norm": 0.8392327427864075,
      "learning_rate": 1.1544e-05,
      "loss": 0.1953,
      "step": 3851
    },
    {
      "epoch": 7.107109879963065,
      "grad_norm": 1.0461645126342773,
      "learning_rate": 1.1547000000000001e-05,
      "loss": 0.1848,
      "step": 3852
    },
    {
      "epoch": 7.108956602031395,
      "grad_norm": 0.9008057117462158,
      "learning_rate": 1.1550000000000001e-05,
      "loss": 0.1761,
      "step": 3853
    },
    {
      "epoch": 7.110803324099723,
      "grad_norm": 0.653768002986908,
      "learning_rate": 1.1553000000000001e-05,
      "loss": 0.1228,
      "step": 3854
    },
    {
      "epoch": 7.112650046168052,
      "grad_norm": 0.8528527021408081,
      "learning_rate": 1.1555999999999999e-05,
      "loss": 0.119,
      "step": 3855
    },
    {
      "epoch": 7.11449676823638,
      "grad_norm": 0.7448746562004089,
      "learning_rate": 1.1558999999999999e-05,
      "loss": 0.1272,
      "step": 3856
    },
    {
      "epoch": 7.116343490304709,
      "grad_norm": 0.7983085513114929,
      "learning_rate": 1.1562e-05,
      "loss": 0.0873,
      "step": 3857
    },
    {
      "epoch": 7.118190212373038,
      "grad_norm": 0.6713128089904785,
      "learning_rate": 1.1565e-05,
      "loss": 0.0652,
      "step": 3858
    },
    {
      "epoch": 7.120036934441367,
      "grad_norm": 1.0396053791046143,
      "learning_rate": 1.1568e-05,
      "loss": 0.1061,
      "step": 3859
    },
    {
      "epoch": 7.121883656509695,
      "grad_norm": 1.0687631368637085,
      "learning_rate": 1.1571e-05,
      "loss": 0.0647,
      "step": 3860
    },
    {
      "epoch": 7.123730378578024,
      "grad_norm": 0.4817259907722473,
      "learning_rate": 1.1574e-05,
      "loss": 0.0363,
      "step": 3861
    },
    {
      "epoch": 7.125577100646352,
      "grad_norm": 0.6856535077095032,
      "learning_rate": 1.1577000000000001e-05,
      "loss": 0.0456,
      "step": 3862
    },
    {
      "epoch": 7.127423822714682,
      "grad_norm": 0.5848652720451355,
      "learning_rate": 1.1580000000000001e-05,
      "loss": 0.0571,
      "step": 3863
    },
    {
      "epoch": 7.12927054478301,
      "grad_norm": 0.8321937322616577,
      "learning_rate": 1.1583e-05,
      "loss": 0.095,
      "step": 3864
    },
    {
      "epoch": 7.131117266851339,
      "grad_norm": 0.6058899164199829,
      "learning_rate": 1.1586e-05,
      "loss": 0.051,
      "step": 3865
    },
    {
      "epoch": 7.132963988919667,
      "grad_norm": 0.4666897654533386,
      "learning_rate": 1.1589e-05,
      "loss": 0.0349,
      "step": 3866
    },
    {
      "epoch": 7.134810710987996,
      "grad_norm": 1.0036622285842896,
      "learning_rate": 1.1592e-05,
      "loss": 0.045,
      "step": 3867
    },
    {
      "epoch": 7.136657433056325,
      "grad_norm": 0.48680198192596436,
      "learning_rate": 1.1595e-05,
      "loss": 0.033,
      "step": 3868
    },
    {
      "epoch": 7.138504155124654,
      "grad_norm": 1.0338425636291504,
      "learning_rate": 1.1598e-05,
      "loss": 0.0455,
      "step": 3869
    },
    {
      "epoch": 7.140350877192983,
      "grad_norm": 0.6906830668449402,
      "learning_rate": 1.1601e-05,
      "loss": 0.066,
      "step": 3870
    },
    {
      "epoch": 7.142197599261311,
      "grad_norm": 0.5115965008735657,
      "learning_rate": 1.1604e-05,
      "loss": 0.0264,
      "step": 3871
    },
    {
      "epoch": 7.14404432132964,
      "grad_norm": 0.62877357006073,
      "learning_rate": 1.1607000000000001e-05,
      "loss": 0.0313,
      "step": 3872
    },
    {
      "epoch": 7.1458910433979685,
      "grad_norm": 0.562495231628418,
      "learning_rate": 1.161e-05,
      "loss": 0.0365,
      "step": 3873
    },
    {
      "epoch": 7.147737765466298,
      "grad_norm": 1.0443249940872192,
      "learning_rate": 1.1613e-05,
      "loss": 0.044,
      "step": 3874
    },
    {
      "epoch": 7.149584487534626,
      "grad_norm": 0.8314567804336548,
      "learning_rate": 1.1616e-05,
      "loss": 0.0383,
      "step": 3875
    },
    {
      "epoch": 7.151431209602955,
      "grad_norm": 0.8624547719955444,
      "learning_rate": 1.1619e-05,
      "loss": 0.0465,
      "step": 3876
    },
    {
      "epoch": 7.153277931671283,
      "grad_norm": 0.43689149618148804,
      "learning_rate": 1.1622000000000002e-05,
      "loss": 0.0258,
      "step": 3877
    },
    {
      "epoch": 7.1551246537396125,
      "grad_norm": 0.5368509888648987,
      "learning_rate": 1.1625000000000001e-05,
      "loss": 0.0221,
      "step": 3878
    },
    {
      "epoch": 7.156971375807941,
      "grad_norm": 0.41164904832839966,
      "learning_rate": 1.1628e-05,
      "loss": 0.0264,
      "step": 3879
    },
    {
      "epoch": 7.15881809787627,
      "grad_norm": 0.7307270169258118,
      "learning_rate": 1.1631e-05,
      "loss": 0.0417,
      "step": 3880
    },
    {
      "epoch": 7.160664819944598,
      "grad_norm": 0.9499558806419373,
      "learning_rate": 1.1633999999999999e-05,
      "loss": 0.036,
      "step": 3881
    },
    {
      "epoch": 7.162511542012927,
      "grad_norm": 0.8061280250549316,
      "learning_rate": 1.1637e-05,
      "loss": 0.031,
      "step": 3882
    },
    {
      "epoch": 7.1643582640812555,
      "grad_norm": 0.5703011155128479,
      "learning_rate": 1.164e-05,
      "loss": 0.0456,
      "step": 3883
    },
    {
      "epoch": 7.166204986149585,
      "grad_norm": 0.7621678113937378,
      "learning_rate": 1.1643e-05,
      "loss": 0.0279,
      "step": 3884
    },
    {
      "epoch": 7.168051708217913,
      "grad_norm": 0.5694484710693359,
      "learning_rate": 1.1646e-05,
      "loss": 0.0405,
      "step": 3885
    },
    {
      "epoch": 7.169898430286242,
      "grad_norm": 0.5033777952194214,
      "learning_rate": 1.1649e-05,
      "loss": 0.0246,
      "step": 3886
    },
    {
      "epoch": 7.17174515235457,
      "grad_norm": 0.6190872192382812,
      "learning_rate": 1.1652000000000001e-05,
      "loss": 0.0375,
      "step": 3887
    },
    {
      "epoch": 7.1735918744228995,
      "grad_norm": 0.5975775718688965,
      "learning_rate": 1.1655000000000001e-05,
      "loss": 0.0288,
      "step": 3888
    },
    {
      "epoch": 7.175438596491228,
      "grad_norm": 0.7166990041732788,
      "learning_rate": 1.1658000000000001e-05,
      "loss": 0.0409,
      "step": 3889
    },
    {
      "epoch": 7.177285318559557,
      "grad_norm": 0.9195456504821777,
      "learning_rate": 1.1661e-05,
      "loss": 0.0482,
      "step": 3890
    },
    {
      "epoch": 7.179132040627885,
      "grad_norm": 1.0429009199142456,
      "learning_rate": 1.1664e-05,
      "loss": 0.0508,
      "step": 3891
    },
    {
      "epoch": 7.180978762696214,
      "grad_norm": 0.8649551868438721,
      "learning_rate": 1.1667e-05,
      "loss": 0.0636,
      "step": 3892
    },
    {
      "epoch": 7.1828254847645425,
      "grad_norm": 0.7114192247390747,
      "learning_rate": 1.167e-05,
      "loss": 0.0393,
      "step": 3893
    },
    {
      "epoch": 7.184672206832872,
      "grad_norm": 1.0726664066314697,
      "learning_rate": 1.1673e-05,
      "loss": 0.0507,
      "step": 3894
    },
    {
      "epoch": 7.186518928901201,
      "grad_norm": 2.35176944732666,
      "learning_rate": 1.1676e-05,
      "loss": 0.3612,
      "step": 3895
    },
    {
      "epoch": 7.188365650969529,
      "grad_norm": 1.3197213411331177,
      "learning_rate": 1.1679e-05,
      "loss": 0.3027,
      "step": 3896
    },
    {
      "epoch": 7.190212373037858,
      "grad_norm": 1.1717162132263184,
      "learning_rate": 1.1682000000000001e-05,
      "loss": 0.3242,
      "step": 3897
    },
    {
      "epoch": 7.1920590951061865,
      "grad_norm": 1.3300485610961914,
      "learning_rate": 1.1685e-05,
      "loss": 0.2406,
      "step": 3898
    },
    {
      "epoch": 7.193905817174516,
      "grad_norm": 0.8447170853614807,
      "learning_rate": 1.1688e-05,
      "loss": 0.18,
      "step": 3899
    },
    {
      "epoch": 7.195752539242844,
      "grad_norm": 0.9538955092430115,
      "learning_rate": 1.1691e-05,
      "loss": 0.1968,
      "step": 3900
    },
    {
      "epoch": 7.197599261311173,
      "grad_norm": 0.7060114145278931,
      "learning_rate": 1.1694e-05,
      "loss": 0.1535,
      "step": 3901
    },
    {
      "epoch": 7.199445983379501,
      "grad_norm": 1.2481669187545776,
      "learning_rate": 1.1697000000000002e-05,
      "loss": 0.2951,
      "step": 3902
    },
    {
      "epoch": 7.20129270544783,
      "grad_norm": 0.7738790512084961,
      "learning_rate": 1.1700000000000001e-05,
      "loss": 0.1675,
      "step": 3903
    },
    {
      "epoch": 7.203139427516159,
      "grad_norm": 1.215665578842163,
      "learning_rate": 1.1703e-05,
      "loss": 0.1223,
      "step": 3904
    },
    {
      "epoch": 7.204986149584488,
      "grad_norm": 0.5640180706977844,
      "learning_rate": 1.1706e-05,
      "loss": 0.1018,
      "step": 3905
    },
    {
      "epoch": 7.206832871652816,
      "grad_norm": 0.6813676953315735,
      "learning_rate": 1.1709e-05,
      "loss": 0.0955,
      "step": 3906
    },
    {
      "epoch": 7.208679593721145,
      "grad_norm": 0.8873261213302612,
      "learning_rate": 1.1712e-05,
      "loss": 0.1367,
      "step": 3907
    },
    {
      "epoch": 7.2105263157894735,
      "grad_norm": 0.8507635593414307,
      "learning_rate": 1.1715e-05,
      "loss": 0.1125,
      "step": 3908
    },
    {
      "epoch": 7.212373037857803,
      "grad_norm": 0.5705675482749939,
      "learning_rate": 1.1718e-05,
      "loss": 0.0964,
      "step": 3909
    },
    {
      "epoch": 7.214219759926131,
      "grad_norm": 0.47501277923583984,
      "learning_rate": 1.1721e-05,
      "loss": 0.0511,
      "step": 3910
    },
    {
      "epoch": 7.21606648199446,
      "grad_norm": 0.48651376366615295,
      "learning_rate": 1.1724e-05,
      "loss": 0.0397,
      "step": 3911
    },
    {
      "epoch": 7.217913204062788,
      "grad_norm": 0.4294663667678833,
      "learning_rate": 1.1727000000000001e-05,
      "loss": 0.0319,
      "step": 3912
    },
    {
      "epoch": 7.219759926131117,
      "grad_norm": 0.500373125076294,
      "learning_rate": 1.1730000000000001e-05,
      "loss": 0.0353,
      "step": 3913
    },
    {
      "epoch": 7.221606648199446,
      "grad_norm": 0.4776148498058319,
      "learning_rate": 1.1733000000000001e-05,
      "loss": 0.0328,
      "step": 3914
    },
    {
      "epoch": 7.223453370267775,
      "grad_norm": 0.45514485239982605,
      "learning_rate": 1.1736e-05,
      "loss": 0.0373,
      "step": 3915
    },
    {
      "epoch": 7.225300092336103,
      "grad_norm": 0.39044058322906494,
      "learning_rate": 1.1738999999999999e-05,
      "loss": 0.0241,
      "step": 3916
    },
    {
      "epoch": 7.227146814404432,
      "grad_norm": 1.1295156478881836,
      "learning_rate": 1.1742e-05,
      "loss": 0.042,
      "step": 3917
    },
    {
      "epoch": 7.22899353647276,
      "grad_norm": 0.6388261318206787,
      "learning_rate": 1.1745e-05,
      "loss": 0.0614,
      "step": 3918
    },
    {
      "epoch": 7.23084025854109,
      "grad_norm": 0.4165383279323578,
      "learning_rate": 1.1748e-05,
      "loss": 0.0182,
      "step": 3919
    },
    {
      "epoch": 7.232686980609419,
      "grad_norm": 0.6110149025917053,
      "learning_rate": 1.1751e-05,
      "loss": 0.0342,
      "step": 3920
    },
    {
      "epoch": 7.234533702677747,
      "grad_norm": 1.3807674646377563,
      "learning_rate": 1.1754e-05,
      "loss": 0.0954,
      "step": 3921
    },
    {
      "epoch": 7.236380424746076,
      "grad_norm": 1.6706429719924927,
      "learning_rate": 1.1757000000000001e-05,
      "loss": 0.0347,
      "step": 3922
    },
    {
      "epoch": 7.238227146814404,
      "grad_norm": 0.49312475323677063,
      "learning_rate": 1.1760000000000001e-05,
      "loss": 0.0276,
      "step": 3923
    },
    {
      "epoch": 7.2400738688827335,
      "grad_norm": 0.6351246237754822,
      "learning_rate": 1.1763e-05,
      "loss": 0.0343,
      "step": 3924
    },
    {
      "epoch": 7.241920590951062,
      "grad_norm": 0.7758086323738098,
      "learning_rate": 1.1766e-05,
      "loss": 0.0379,
      "step": 3925
    },
    {
      "epoch": 7.243767313019391,
      "grad_norm": 1.6319568157196045,
      "learning_rate": 1.1769e-05,
      "loss": 0.0458,
      "step": 3926
    },
    {
      "epoch": 7.245614035087719,
      "grad_norm": 0.5458925366401672,
      "learning_rate": 1.1772000000000002e-05,
      "loss": 0.0333,
      "step": 3927
    },
    {
      "epoch": 7.247460757156048,
      "grad_norm": 0.924403190612793,
      "learning_rate": 1.1775000000000002e-05,
      "loss": 0.0626,
      "step": 3928
    },
    {
      "epoch": 7.249307479224377,
      "grad_norm": 1.2806562185287476,
      "learning_rate": 1.1778e-05,
      "loss": 0.0589,
      "step": 3929
    },
    {
      "epoch": 7.251154201292706,
      "grad_norm": 0.6534441113471985,
      "learning_rate": 1.1781e-05,
      "loss": 0.0417,
      "step": 3930
    },
    {
      "epoch": 7.253000923361034,
      "grad_norm": 0.7234696745872498,
      "learning_rate": 1.1784e-05,
      "loss": 0.0352,
      "step": 3931
    },
    {
      "epoch": 7.254847645429363,
      "grad_norm": 0.5123321413993835,
      "learning_rate": 1.1787e-05,
      "loss": 0.0338,
      "step": 3932
    },
    {
      "epoch": 7.256694367497691,
      "grad_norm": 0.5848380327224731,
      "learning_rate": 1.179e-05,
      "loss": 0.0425,
      "step": 3933
    },
    {
      "epoch": 7.2585410895660205,
      "grad_norm": 0.6774309873580933,
      "learning_rate": 1.1793e-05,
      "loss": 0.0392,
      "step": 3934
    },
    {
      "epoch": 7.260387811634349,
      "grad_norm": 0.8812325596809387,
      "learning_rate": 1.1796e-05,
      "loss": 0.0424,
      "step": 3935
    },
    {
      "epoch": 7.262234533702678,
      "grad_norm": 0.8823750019073486,
      "learning_rate": 1.1799e-05,
      "loss": 0.046,
      "step": 3936
    },
    {
      "epoch": 7.264081255771006,
      "grad_norm": 0.8765324950218201,
      "learning_rate": 1.1802000000000002e-05,
      "loss": 0.0358,
      "step": 3937
    },
    {
      "epoch": 7.265927977839335,
      "grad_norm": 0.4929245114326477,
      "learning_rate": 1.1805000000000001e-05,
      "loss": 0.0271,
      "step": 3938
    },
    {
      "epoch": 7.267774699907664,
      "grad_norm": 0.555218517780304,
      "learning_rate": 1.1808000000000001e-05,
      "loss": 0.0404,
      "step": 3939
    },
    {
      "epoch": 7.269621421975993,
      "grad_norm": 0.9540738463401794,
      "learning_rate": 1.1811000000000001e-05,
      "loss": 0.0488,
      "step": 3940
    },
    {
      "epoch": 7.271468144044321,
      "grad_norm": 0.7682968378067017,
      "learning_rate": 1.1813999999999999e-05,
      "loss": 0.0509,
      "step": 3941
    },
    {
      "epoch": 7.27331486611265,
      "grad_norm": 1.5334378480911255,
      "learning_rate": 1.1816999999999999e-05,
      "loss": 0.0606,
      "step": 3942
    },
    {
      "epoch": 7.275161588180978,
      "grad_norm": 0.9216906428337097,
      "learning_rate": 1.182e-05,
      "loss": 0.0483,
      "step": 3943
    },
    {
      "epoch": 7.2770083102493075,
      "grad_norm": 0.915949821472168,
      "learning_rate": 1.1823e-05,
      "loss": 0.0442,
      "step": 3944
    },
    {
      "epoch": 7.278855032317637,
      "grad_norm": 1.0504190921783447,
      "learning_rate": 1.1826e-05,
      "loss": 0.2944,
      "step": 3945
    },
    {
      "epoch": 7.280701754385965,
      "grad_norm": 0.9758450388908386,
      "learning_rate": 1.1829e-05,
      "loss": 0.3035,
      "step": 3946
    },
    {
      "epoch": 7.282548476454294,
      "grad_norm": 0.9999068379402161,
      "learning_rate": 1.1832e-05,
      "loss": 0.2148,
      "step": 3947
    },
    {
      "epoch": 7.284395198522622,
      "grad_norm": 2.185529947280884,
      "learning_rate": 1.1835000000000001e-05,
      "loss": 0.2184,
      "step": 3948
    },
    {
      "epoch": 7.286241920590951,
      "grad_norm": 0.9298887252807617,
      "learning_rate": 1.1838e-05,
      "loss": 0.1772,
      "step": 3949
    },
    {
      "epoch": 7.28808864265928,
      "grad_norm": 1.4344820976257324,
      "learning_rate": 1.1841e-05,
      "loss": 0.2368,
      "step": 3950
    },
    {
      "epoch": 7.289935364727609,
      "grad_norm": 1.246220350265503,
      "learning_rate": 1.1844e-05,
      "loss": 0.1624,
      "step": 3951
    },
    {
      "epoch": 7.291782086795937,
      "grad_norm": 0.7392306923866272,
      "learning_rate": 1.1847e-05,
      "loss": 0.1859,
      "step": 3952
    },
    {
      "epoch": 7.293628808864266,
      "grad_norm": 0.8688386678695679,
      "learning_rate": 1.185e-05,
      "loss": 0.1944,
      "step": 3953
    },
    {
      "epoch": 7.2954755309325945,
      "grad_norm": 0.7431932687759399,
      "learning_rate": 1.1853e-05,
      "loss": 0.1005,
      "step": 3954
    },
    {
      "epoch": 7.297322253000924,
      "grad_norm": 0.9904128909111023,
      "learning_rate": 1.1856e-05,
      "loss": 0.1232,
      "step": 3955
    },
    {
      "epoch": 7.299168975069252,
      "grad_norm": 0.7489523887634277,
      "learning_rate": 1.1859e-05,
      "loss": 0.079,
      "step": 3956
    },
    {
      "epoch": 7.301015697137581,
      "grad_norm": 2.4830214977264404,
      "learning_rate": 1.1862e-05,
      "loss": 0.0734,
      "step": 3957
    },
    {
      "epoch": 7.302862419205909,
      "grad_norm": 0.47750627994537354,
      "learning_rate": 1.1865e-05,
      "loss": 0.0462,
      "step": 3958
    },
    {
      "epoch": 7.304709141274238,
      "grad_norm": 0.5852447748184204,
      "learning_rate": 1.1868e-05,
      "loss": 0.0975,
      "step": 3959
    },
    {
      "epoch": 7.306555863342567,
      "grad_norm": 0.586001455783844,
      "learning_rate": 1.1871e-05,
      "loss": 0.0861,
      "step": 3960
    },
    {
      "epoch": 7.308402585410896,
      "grad_norm": 0.6134559512138367,
      "learning_rate": 1.1874e-05,
      "loss": 0.0558,
      "step": 3961
    },
    {
      "epoch": 7.310249307479224,
      "grad_norm": 1.9210140705108643,
      "learning_rate": 1.1877e-05,
      "loss": 0.0457,
      "step": 3962
    },
    {
      "epoch": 7.312096029547553,
      "grad_norm": 1.0939297676086426,
      "learning_rate": 1.1880000000000001e-05,
      "loss": 0.0987,
      "step": 3963
    },
    {
      "epoch": 7.3139427516158815,
      "grad_norm": 0.9990463256835938,
      "learning_rate": 1.1883000000000001e-05,
      "loss": 0.1067,
      "step": 3964
    },
    {
      "epoch": 7.315789473684211,
      "grad_norm": 0.7564393877983093,
      "learning_rate": 1.1886e-05,
      "loss": 0.0613,
      "step": 3965
    },
    {
      "epoch": 7.317636195752539,
      "grad_norm": 0.4570484459400177,
      "learning_rate": 1.1889e-05,
      "loss": 0.0392,
      "step": 3966
    },
    {
      "epoch": 7.319482917820868,
      "grad_norm": 0.593252420425415,
      "learning_rate": 1.1891999999999999e-05,
      "loss": 0.0469,
      "step": 3967
    },
    {
      "epoch": 7.321329639889196,
      "grad_norm": 0.7046050429344177,
      "learning_rate": 1.1895e-05,
      "loss": 0.0436,
      "step": 3968
    },
    {
      "epoch": 7.323176361957525,
      "grad_norm": 0.647758424282074,
      "learning_rate": 1.1898e-05,
      "loss": 0.041,
      "step": 3969
    },
    {
      "epoch": 7.325023084025855,
      "grad_norm": 0.7471902370452881,
      "learning_rate": 1.1901e-05,
      "loss": 0.0623,
      "step": 3970
    },
    {
      "epoch": 7.326869806094183,
      "grad_norm": 0.6938056945800781,
      "learning_rate": 1.1904e-05,
      "loss": 0.0369,
      "step": 3971
    },
    {
      "epoch": 7.328716528162512,
      "grad_norm": 0.6024402976036072,
      "learning_rate": 1.1907e-05,
      "loss": 0.0309,
      "step": 3972
    },
    {
      "epoch": 7.33056325023084,
      "grad_norm": 0.9108567833900452,
      "learning_rate": 1.1910000000000001e-05,
      "loss": 0.0313,
      "step": 3973
    },
    {
      "epoch": 7.332409972299169,
      "grad_norm": 0.8201056718826294,
      "learning_rate": 1.1913000000000001e-05,
      "loss": 0.05,
      "step": 3974
    },
    {
      "epoch": 7.334256694367498,
      "grad_norm": 0.6461535692214966,
      "learning_rate": 1.1916e-05,
      "loss": 0.0398,
      "step": 3975
    },
    {
      "epoch": 7.336103416435827,
      "grad_norm": 0.6558883786201477,
      "learning_rate": 1.1919e-05,
      "loss": 0.0392,
      "step": 3976
    },
    {
      "epoch": 7.337950138504155,
      "grad_norm": 0.6368681192398071,
      "learning_rate": 1.1922e-05,
      "loss": 0.0341,
      "step": 3977
    },
    {
      "epoch": 7.339796860572484,
      "grad_norm": 0.6736059784889221,
      "learning_rate": 1.1925e-05,
      "loss": 0.0378,
      "step": 3978
    },
    {
      "epoch": 7.341643582640812,
      "grad_norm": 0.7553744912147522,
      "learning_rate": 1.1928e-05,
      "loss": 0.0513,
      "step": 3979
    },
    {
      "epoch": 7.3434903047091415,
      "grad_norm": 0.7779793739318848,
      "learning_rate": 1.1931e-05,
      "loss": 0.0427,
      "step": 3980
    },
    {
      "epoch": 7.34533702677747,
      "grad_norm": 0.7781184315681458,
      "learning_rate": 1.1934e-05,
      "loss": 0.0525,
      "step": 3981
    },
    {
      "epoch": 7.347183748845799,
      "grad_norm": 0.5072903037071228,
      "learning_rate": 1.1937e-05,
      "loss": 0.0327,
      "step": 3982
    },
    {
      "epoch": 7.349030470914127,
      "grad_norm": 0.7815638780593872,
      "learning_rate": 1.1940000000000001e-05,
      "loss": 0.0421,
      "step": 3983
    },
    {
      "epoch": 7.350877192982456,
      "grad_norm": 0.5705381035804749,
      "learning_rate": 1.1943e-05,
      "loss": 0.0328,
      "step": 3984
    },
    {
      "epoch": 7.352723915050785,
      "grad_norm": 0.6040554642677307,
      "learning_rate": 1.1946e-05,
      "loss": 0.0374,
      "step": 3985
    },
    {
      "epoch": 7.354570637119114,
      "grad_norm": 0.7622888088226318,
      "learning_rate": 1.1949e-05,
      "loss": 0.0333,
      "step": 3986
    },
    {
      "epoch": 7.356417359187442,
      "grad_norm": 0.615376889705658,
      "learning_rate": 1.1952e-05,
      "loss": 0.0304,
      "step": 3987
    },
    {
      "epoch": 7.358264081255771,
      "grad_norm": 1.0284879207611084,
      "learning_rate": 1.1955000000000002e-05,
      "loss": 0.0381,
      "step": 3988
    },
    {
      "epoch": 7.360110803324099,
      "grad_norm": 0.5528692603111267,
      "learning_rate": 1.1958000000000001e-05,
      "loss": 0.0413,
      "step": 3989
    },
    {
      "epoch": 7.3619575253924285,
      "grad_norm": 0.6359366178512573,
      "learning_rate": 1.1961e-05,
      "loss": 0.0353,
      "step": 3990
    },
    {
      "epoch": 7.363804247460757,
      "grad_norm": 0.5964512825012207,
      "learning_rate": 1.1964e-05,
      "loss": 0.0295,
      "step": 3991
    },
    {
      "epoch": 7.365650969529086,
      "grad_norm": 0.7896857261657715,
      "learning_rate": 1.1966999999999999e-05,
      "loss": 0.0322,
      "step": 3992
    },
    {
      "epoch": 7.367497691597414,
      "grad_norm": 0.7088826298713684,
      "learning_rate": 1.197e-05,
      "loss": 0.0424,
      "step": 3993
    },
    {
      "epoch": 7.369344413665743,
      "grad_norm": 1.7857023477554321,
      "learning_rate": 1.1973e-05,
      "loss": 0.111,
      "step": 3994
    },
    {
      "epoch": 7.3711911357340725,
      "grad_norm": 1.8425383567810059,
      "learning_rate": 1.1976e-05,
      "loss": 0.3479,
      "step": 3995
    },
    {
      "epoch": 7.373037857802401,
      "grad_norm": 1.8757927417755127,
      "learning_rate": 1.1979e-05,
      "loss": 0.3627,
      "step": 3996
    },
    {
      "epoch": 7.374884579870729,
      "grad_norm": 1.019248366355896,
      "learning_rate": 1.1982e-05,
      "loss": 0.2404,
      "step": 3997
    },
    {
      "epoch": 7.376731301939058,
      "grad_norm": 0.9379916787147522,
      "learning_rate": 1.1985000000000001e-05,
      "loss": 0.2214,
      "step": 3998
    },
    {
      "epoch": 7.378578024007387,
      "grad_norm": 0.9834468364715576,
      "learning_rate": 1.1988000000000001e-05,
      "loss": 0.2044,
      "step": 3999
    },
    {
      "epoch": 7.3804247460757155,
      "grad_norm": 1.1186176538467407,
      "learning_rate": 1.1991000000000001e-05,
      "loss": 0.2837,
      "step": 4000
    },
    {
      "epoch": 7.3804247460757155,
      "eval_cer": 0.12484122663763382,
      "eval_loss": 0.3644801080226898,
      "eval_runtime": 15.1759,
      "eval_samples_per_second": 20.032,
      "eval_steps_per_second": 0.659,
      "eval_wer": 0.4376438986953185,
      "step": 4000
    },
    {
      "epoch": 7.382271468144045,
      "grad_norm": 1.1192575693130493,
      "learning_rate": 1.1994e-05,
      "loss": 0.1749,
      "step": 4001
    },
    {
      "epoch": 7.384118190212373,
      "grad_norm": 1.0478407144546509,
      "learning_rate": 1.1996999999999999e-05,
      "loss": 0.1652,
      "step": 4002
    },
    {
      "epoch": 7.385964912280702,
      "grad_norm": 0.9594413638114929,
      "learning_rate": 1.2e-05,
      "loss": 0.1445,
      "step": 4003
    },
    {
      "epoch": 7.38781163434903,
      "grad_norm": 0.8693169951438904,
      "learning_rate": 1.2003e-05,
      "loss": 0.1175,
      "step": 4004
    },
    {
      "epoch": 7.3896583564173595,
      "grad_norm": 0.9398826360702515,
      "learning_rate": 1.2006e-05,
      "loss": 0.1551,
      "step": 4005
    },
    {
      "epoch": 7.391505078485688,
      "grad_norm": 1.0473353862762451,
      "learning_rate": 1.2009e-05,
      "loss": 0.1708,
      "step": 4006
    },
    {
      "epoch": 7.393351800554017,
      "grad_norm": 0.5337854027748108,
      "learning_rate": 1.2012e-05,
      "loss": 0.0752,
      "step": 4007
    },
    {
      "epoch": 7.395198522622345,
      "grad_norm": 1.338207483291626,
      "learning_rate": 1.2015000000000001e-05,
      "loss": 0.1319,
      "step": 4008
    },
    {
      "epoch": 7.397045244690674,
      "grad_norm": 1.462694525718689,
      "learning_rate": 1.2018e-05,
      "loss": 0.0987,
      "step": 4009
    },
    {
      "epoch": 7.3988919667590025,
      "grad_norm": 0.9892147779464722,
      "learning_rate": 1.2021e-05,
      "loss": 0.0762,
      "step": 4010
    },
    {
      "epoch": 7.400738688827332,
      "grad_norm": 0.6074382066726685,
      "learning_rate": 1.2024e-05,
      "loss": 0.0722,
      "step": 4011
    },
    {
      "epoch": 7.40258541089566,
      "grad_norm": 0.8629452586174011,
      "learning_rate": 1.2027e-05,
      "loss": 0.0638,
      "step": 4012
    },
    {
      "epoch": 7.404432132963989,
      "grad_norm": 0.5274789333343506,
      "learning_rate": 1.2030000000000002e-05,
      "loss": 0.0608,
      "step": 4013
    },
    {
      "epoch": 7.406278855032317,
      "grad_norm": 0.6876819729804993,
      "learning_rate": 1.2033000000000002e-05,
      "loss": 0.059,
      "step": 4014
    },
    {
      "epoch": 7.4081255771006465,
      "grad_norm": 0.5091372728347778,
      "learning_rate": 1.2036e-05,
      "loss": 0.0488,
      "step": 4015
    },
    {
      "epoch": 7.409972299168975,
      "grad_norm": 0.47496551275253296,
      "learning_rate": 1.2039e-05,
      "loss": 0.0441,
      "step": 4016
    },
    {
      "epoch": 7.411819021237304,
      "grad_norm": 0.5922196507453918,
      "learning_rate": 1.2042e-05,
      "loss": 0.036,
      "step": 4017
    },
    {
      "epoch": 7.413665743305632,
      "grad_norm": 0.7583091855049133,
      "learning_rate": 1.2045e-05,
      "loss": 0.0383,
      "step": 4018
    },
    {
      "epoch": 7.415512465373961,
      "grad_norm": 0.6326509118080139,
      "learning_rate": 1.2048e-05,
      "loss": 0.0419,
      "step": 4019
    },
    {
      "epoch": 7.41735918744229,
      "grad_norm": 0.5933094024658203,
      "learning_rate": 1.2051e-05,
      "loss": 0.0331,
      "step": 4020
    },
    {
      "epoch": 7.419205909510619,
      "grad_norm": 0.6546366810798645,
      "learning_rate": 1.2054e-05,
      "loss": 0.0421,
      "step": 4021
    },
    {
      "epoch": 7.421052631578947,
      "grad_norm": 0.5232709646224976,
      "learning_rate": 1.2057e-05,
      "loss": 0.0411,
      "step": 4022
    },
    {
      "epoch": 7.422899353647276,
      "grad_norm": 0.48919686675071716,
      "learning_rate": 1.2060000000000001e-05,
      "loss": 0.0303,
      "step": 4023
    },
    {
      "epoch": 7.424746075715605,
      "grad_norm": 0.6700199842453003,
      "learning_rate": 1.2063000000000001e-05,
      "loss": 0.0301,
      "step": 4024
    },
    {
      "epoch": 7.426592797783933,
      "grad_norm": 0.749679446220398,
      "learning_rate": 1.2066000000000001e-05,
      "loss": 0.0335,
      "step": 4025
    },
    {
      "epoch": 7.428439519852263,
      "grad_norm": 0.957880437374115,
      "learning_rate": 1.2069e-05,
      "loss": 0.0577,
      "step": 4026
    },
    {
      "epoch": 7.430286241920591,
      "grad_norm": 0.9046592116355896,
      "learning_rate": 1.2071999999999999e-05,
      "loss": 0.0438,
      "step": 4027
    },
    {
      "epoch": 7.43213296398892,
      "grad_norm": 1.0317715406417847,
      "learning_rate": 1.2075e-05,
      "loss": 0.0437,
      "step": 4028
    },
    {
      "epoch": 7.433979686057248,
      "grad_norm": 0.8226898312568665,
      "learning_rate": 1.2078e-05,
      "loss": 0.036,
      "step": 4029
    },
    {
      "epoch": 7.435826408125577,
      "grad_norm": 0.6602367758750916,
      "learning_rate": 1.2081e-05,
      "loss": 0.0349,
      "step": 4030
    },
    {
      "epoch": 7.437673130193906,
      "grad_norm": 0.6261298656463623,
      "learning_rate": 1.2084e-05,
      "loss": 0.0381,
      "step": 4031
    },
    {
      "epoch": 7.439519852262235,
      "grad_norm": 2.0433335304260254,
      "learning_rate": 1.2087e-05,
      "loss": 0.0469,
      "step": 4032
    },
    {
      "epoch": 7.441366574330563,
      "grad_norm": 0.7569543123245239,
      "learning_rate": 1.2090000000000001e-05,
      "loss": 0.0352,
      "step": 4033
    },
    {
      "epoch": 7.443213296398892,
      "grad_norm": 0.6144489049911499,
      "learning_rate": 1.2093000000000001e-05,
      "loss": 0.0375,
      "step": 4034
    },
    {
      "epoch": 7.44506001846722,
      "grad_norm": 0.8899089694023132,
      "learning_rate": 1.2096e-05,
      "loss": 0.0439,
      "step": 4035
    },
    {
      "epoch": 7.44690674053555,
      "grad_norm": 0.8068799376487732,
      "learning_rate": 1.2099e-05,
      "loss": 0.0475,
      "step": 4036
    },
    {
      "epoch": 7.448753462603878,
      "grad_norm": 0.9589241743087769,
      "learning_rate": 1.2102e-05,
      "loss": 0.0353,
      "step": 4037
    },
    {
      "epoch": 7.450600184672207,
      "grad_norm": 1.7914063930511475,
      "learning_rate": 1.2105000000000002e-05,
      "loss": 0.0447,
      "step": 4038
    },
    {
      "epoch": 7.452446906740535,
      "grad_norm": 0.6683489084243774,
      "learning_rate": 1.2108e-05,
      "loss": 0.0395,
      "step": 4039
    },
    {
      "epoch": 7.454293628808864,
      "grad_norm": 1.6444193124771118,
      "learning_rate": 1.2111e-05,
      "loss": 0.0356,
      "step": 4040
    },
    {
      "epoch": 7.456140350877193,
      "grad_norm": 0.9469858407974243,
      "learning_rate": 1.2114e-05,
      "loss": 0.0649,
      "step": 4041
    },
    {
      "epoch": 7.457987072945522,
      "grad_norm": 0.9198259115219116,
      "learning_rate": 1.2117e-05,
      "loss": 0.0565,
      "step": 4042
    },
    {
      "epoch": 7.45983379501385,
      "grad_norm": 1.2903748750686646,
      "learning_rate": 1.2120000000000001e-05,
      "loss": 0.0419,
      "step": 4043
    },
    {
      "epoch": 7.461680517082179,
      "grad_norm": 1.6258609294891357,
      "learning_rate": 1.2123e-05,
      "loss": 0.0701,
      "step": 4044
    },
    {
      "epoch": 7.463527239150508,
      "grad_norm": 1.1336464881896973,
      "learning_rate": 1.2126e-05,
      "loss": 0.3336,
      "step": 4045
    },
    {
      "epoch": 7.465373961218837,
      "grad_norm": 1.0165140628814697,
      "learning_rate": 1.2129e-05,
      "loss": 0.2929,
      "step": 4046
    },
    {
      "epoch": 7.467220683287165,
      "grad_norm": 0.9984672665596008,
      "learning_rate": 1.2132e-05,
      "loss": 0.27,
      "step": 4047
    },
    {
      "epoch": 7.469067405355494,
      "grad_norm": 0.8502119779586792,
      "learning_rate": 1.2135000000000002e-05,
      "loss": 0.2118,
      "step": 4048
    },
    {
      "epoch": 7.470914127423823,
      "grad_norm": 1.1467849016189575,
      "learning_rate": 1.2138000000000001e-05,
      "loss": 0.2784,
      "step": 4049
    },
    {
      "epoch": 7.472760849492151,
      "grad_norm": 1.2587987184524536,
      "learning_rate": 1.2141000000000001e-05,
      "loss": 0.172,
      "step": 4050
    },
    {
      "epoch": 7.4746075715604805,
      "grad_norm": 1.3149694204330444,
      "learning_rate": 1.2144e-05,
      "loss": 0.2085,
      "step": 4051
    },
    {
      "epoch": 7.476454293628809,
      "grad_norm": 0.7081968784332275,
      "learning_rate": 1.2146999999999999e-05,
      "loss": 0.1349,
      "step": 4052
    },
    {
      "epoch": 7.478301015697138,
      "grad_norm": 1.1603951454162598,
      "learning_rate": 1.215e-05,
      "loss": 0.1923,
      "step": 4053
    },
    {
      "epoch": 7.480147737765466,
      "grad_norm": 1.0851069688796997,
      "learning_rate": 1.2153e-05,
      "loss": 0.2272,
      "step": 4054
    },
    {
      "epoch": 7.481994459833795,
      "grad_norm": 1.9162254333496094,
      "learning_rate": 1.2156e-05,
      "loss": 0.1324,
      "step": 4055
    },
    {
      "epoch": 7.4838411819021236,
      "grad_norm": 1.0658905506134033,
      "learning_rate": 1.2159e-05,
      "loss": 0.2051,
      "step": 4056
    },
    {
      "epoch": 7.485687903970453,
      "grad_norm": 0.8997984528541565,
      "learning_rate": 1.2162e-05,
      "loss": 0.1018,
      "step": 4057
    },
    {
      "epoch": 7.487534626038781,
      "grad_norm": 0.6505519151687622,
      "learning_rate": 1.2165000000000001e-05,
      "loss": 0.0812,
      "step": 4058
    },
    {
      "epoch": 7.48938134810711,
      "grad_norm": 0.5897863507270813,
      "learning_rate": 1.2168000000000001e-05,
      "loss": 0.0747,
      "step": 4059
    },
    {
      "epoch": 7.491228070175438,
      "grad_norm": 0.712464451789856,
      "learning_rate": 1.2171000000000001e-05,
      "loss": 0.0568,
      "step": 4060
    },
    {
      "epoch": 7.4930747922437675,
      "grad_norm": 0.5935330986976624,
      "learning_rate": 1.2174e-05,
      "loss": 0.0619,
      "step": 4061
    },
    {
      "epoch": 7.494921514312096,
      "grad_norm": 0.8959807753562927,
      "learning_rate": 1.2177e-05,
      "loss": 0.0852,
      "step": 4062
    },
    {
      "epoch": 7.496768236380425,
      "grad_norm": 0.7074034214019775,
      "learning_rate": 1.2180000000000002e-05,
      "loss": 0.0434,
      "step": 4063
    },
    {
      "epoch": 7.498614958448753,
      "grad_norm": 0.6812528371810913,
      "learning_rate": 1.2183e-05,
      "loss": 0.0431,
      "step": 4064
    },
    {
      "epoch": 7.500461680517082,
      "grad_norm": 0.4548470675945282,
      "learning_rate": 1.2186e-05,
      "loss": 0.0269,
      "step": 4065
    },
    {
      "epoch": 7.5023084025854105,
      "grad_norm": 0.8323044776916504,
      "learning_rate": 1.2189e-05,
      "loss": 0.0711,
      "step": 4066
    },
    {
      "epoch": 7.50415512465374,
      "grad_norm": 0.8252567052841187,
      "learning_rate": 1.2192e-05,
      "loss": 0.0321,
      "step": 4067
    },
    {
      "epoch": 7.506001846722068,
      "grad_norm": 0.6569534540176392,
      "learning_rate": 1.2195e-05,
      "loss": 0.0261,
      "step": 4068
    },
    {
      "epoch": 7.507848568790397,
      "grad_norm": 0.9114178419113159,
      "learning_rate": 1.2198e-05,
      "loss": 0.0442,
      "step": 4069
    },
    {
      "epoch": 7.509695290858726,
      "grad_norm": 0.7195739150047302,
      "learning_rate": 1.2201e-05,
      "loss": 0.0278,
      "step": 4070
    },
    {
      "epoch": 7.5115420129270545,
      "grad_norm": 1.6552071571350098,
      "learning_rate": 1.2204e-05,
      "loss": 0.0553,
      "step": 4071
    },
    {
      "epoch": 7.513388734995383,
      "grad_norm": 0.5468800067901611,
      "learning_rate": 1.2207e-05,
      "loss": 0.0267,
      "step": 4072
    },
    {
      "epoch": 7.515235457063712,
      "grad_norm": 0.7049787044525146,
      "learning_rate": 1.221e-05,
      "loss": 0.044,
      "step": 4073
    },
    {
      "epoch": 7.517082179132041,
      "grad_norm": 0.3411400020122528,
      "learning_rate": 1.2213000000000001e-05,
      "loss": 0.0176,
      "step": 4074
    },
    {
      "epoch": 7.518928901200369,
      "grad_norm": 1.0625396966934204,
      "learning_rate": 1.2216000000000001e-05,
      "loss": 0.0541,
      "step": 4075
    },
    {
      "epoch": 7.520775623268698,
      "grad_norm": 0.6516671776771545,
      "learning_rate": 1.2219e-05,
      "loss": 0.0417,
      "step": 4076
    },
    {
      "epoch": 7.522622345337027,
      "grad_norm": 0.6971566677093506,
      "learning_rate": 1.2222e-05,
      "loss": 0.0484,
      "step": 4077
    },
    {
      "epoch": 7.524469067405356,
      "grad_norm": 0.8688866496086121,
      "learning_rate": 1.2224999999999999e-05,
      "loss": 0.0487,
      "step": 4078
    },
    {
      "epoch": 7.526315789473684,
      "grad_norm": 0.49682992696762085,
      "learning_rate": 1.2228e-05,
      "loss": 0.0245,
      "step": 4079
    },
    {
      "epoch": 7.528162511542013,
      "grad_norm": 1.0755348205566406,
      "learning_rate": 1.2231e-05,
      "loss": 0.0438,
      "step": 4080
    },
    {
      "epoch": 7.5300092336103415,
      "grad_norm": 1.0635411739349365,
      "learning_rate": 1.2234e-05,
      "loss": 0.0455,
      "step": 4081
    },
    {
      "epoch": 7.531855955678671,
      "grad_norm": 0.7985247373580933,
      "learning_rate": 1.2237e-05,
      "loss": 0.0428,
      "step": 4082
    },
    {
      "epoch": 7.533702677746999,
      "grad_norm": 0.7823305726051331,
      "learning_rate": 1.224e-05,
      "loss": 0.0435,
      "step": 4083
    },
    {
      "epoch": 7.535549399815328,
      "grad_norm": 0.9674842953681946,
      "learning_rate": 1.2243000000000001e-05,
      "loss": 0.0624,
      "step": 4084
    },
    {
      "epoch": 7.537396121883656,
      "grad_norm": 0.6736524105072021,
      "learning_rate": 1.2246000000000001e-05,
      "loss": 0.0323,
      "step": 4085
    },
    {
      "epoch": 7.539242843951985,
      "grad_norm": 1.1502015590667725,
      "learning_rate": 1.2249e-05,
      "loss": 0.0457,
      "step": 4086
    },
    {
      "epoch": 7.541089566020314,
      "grad_norm": 0.727032482624054,
      "learning_rate": 1.2252e-05,
      "loss": 0.0485,
      "step": 4087
    },
    {
      "epoch": 7.542936288088643,
      "grad_norm": 0.7703151702880859,
      "learning_rate": 1.2254999999999999e-05,
      "loss": 0.0351,
      "step": 4088
    },
    {
      "epoch": 7.544783010156971,
      "grad_norm": 1.0239579677581787,
      "learning_rate": 1.2258e-05,
      "loss": 0.0526,
      "step": 4089
    },
    {
      "epoch": 7.5466297322253,
      "grad_norm": 0.6008153557777405,
      "learning_rate": 1.2261e-05,
      "loss": 0.0354,
      "step": 4090
    },
    {
      "epoch": 7.5484764542936285,
      "grad_norm": 0.7614055275917053,
      "learning_rate": 1.2264e-05,
      "loss": 0.0396,
      "step": 4091
    },
    {
      "epoch": 7.550323176361958,
      "grad_norm": 0.5776079297065735,
      "learning_rate": 1.2267e-05,
      "loss": 0.0426,
      "step": 4092
    },
    {
      "epoch": 7.552169898430286,
      "grad_norm": 0.7658137083053589,
      "learning_rate": 1.227e-05,
      "loss": 0.0488,
      "step": 4093
    },
    {
      "epoch": 7.554016620498615,
      "grad_norm": 1.04031240940094,
      "learning_rate": 1.2273000000000001e-05,
      "loss": 0.075,
      "step": 4094
    },
    {
      "epoch": 7.555863342566944,
      "grad_norm": 1.6620241403579712,
      "learning_rate": 1.2276e-05,
      "loss": 0.371,
      "step": 4095
    },
    {
      "epoch": 7.557710064635272,
      "grad_norm": 0.8936166763305664,
      "learning_rate": 1.2279e-05,
      "loss": 0.2259,
      "step": 4096
    },
    {
      "epoch": 7.559556786703601,
      "grad_norm": 0.7464591264724731,
      "learning_rate": 1.2282e-05,
      "loss": 0.231,
      "step": 4097
    },
    {
      "epoch": 7.56140350877193,
      "grad_norm": 1.0560076236724854,
      "learning_rate": 1.2285e-05,
      "loss": 0.2953,
      "step": 4098
    },
    {
      "epoch": 7.563250230840259,
      "grad_norm": 0.6979200839996338,
      "learning_rate": 1.2288000000000002e-05,
      "loss": 0.1647,
      "step": 4099
    },
    {
      "epoch": 7.565096952908587,
      "grad_norm": 0.9179125428199768,
      "learning_rate": 1.2291000000000001e-05,
      "loss": 0.2385,
      "step": 4100
    },
    {
      "epoch": 7.566943674976916,
      "grad_norm": 0.7582529783248901,
      "learning_rate": 1.2294e-05,
      "loss": 0.14,
      "step": 4101
    },
    {
      "epoch": 7.568790397045245,
      "grad_norm": 1.0879583358764648,
      "learning_rate": 1.2297e-05,
      "loss": 0.1912,
      "step": 4102
    },
    {
      "epoch": 7.570637119113574,
      "grad_norm": 0.9386312365531921,
      "learning_rate": 1.2299999999999999e-05,
      "loss": 0.1804,
      "step": 4103
    },
    {
      "epoch": 7.572483841181902,
      "grad_norm": 0.9516357779502869,
      "learning_rate": 1.2303e-05,
      "loss": 0.1823,
      "step": 4104
    },
    {
      "epoch": 7.574330563250231,
      "grad_norm": 0.9170127511024475,
      "learning_rate": 1.2306e-05,
      "loss": 0.1158,
      "step": 4105
    },
    {
      "epoch": 7.576177285318559,
      "grad_norm": 0.9759851098060608,
      "learning_rate": 1.2309e-05,
      "loss": 0.1446,
      "step": 4106
    },
    {
      "epoch": 7.5780240073868885,
      "grad_norm": 0.5087369084358215,
      "learning_rate": 1.2312e-05,
      "loss": 0.1097,
      "step": 4107
    },
    {
      "epoch": 7.579870729455217,
      "grad_norm": 0.6919082999229431,
      "learning_rate": 1.2315e-05,
      "loss": 0.1473,
      "step": 4108
    },
    {
      "epoch": 7.581717451523546,
      "grad_norm": 0.711525559425354,
      "learning_rate": 1.2318000000000001e-05,
      "loss": 0.0869,
      "step": 4109
    },
    {
      "epoch": 7.583564173591874,
      "grad_norm": 0.5739111304283142,
      "learning_rate": 1.2321000000000001e-05,
      "loss": 0.1035,
      "step": 4110
    },
    {
      "epoch": 7.585410895660203,
      "grad_norm": 0.7222297191619873,
      "learning_rate": 1.2324000000000001e-05,
      "loss": 0.0523,
      "step": 4111
    },
    {
      "epoch": 7.587257617728532,
      "grad_norm": 0.6934823989868164,
      "learning_rate": 1.2327e-05,
      "loss": 0.0441,
      "step": 4112
    },
    {
      "epoch": 7.589104339796861,
      "grad_norm": 0.6187019348144531,
      "learning_rate": 1.2329999999999999e-05,
      "loss": 0.0379,
      "step": 4113
    },
    {
      "epoch": 7.590951061865189,
      "grad_norm": 0.5205591320991516,
      "learning_rate": 1.2333e-05,
      "loss": 0.0589,
      "step": 4114
    },
    {
      "epoch": 7.592797783933518,
      "grad_norm": 0.7255731821060181,
      "learning_rate": 1.2336e-05,
      "loss": 0.0452,
      "step": 4115
    },
    {
      "epoch": 7.594644506001846,
      "grad_norm": 0.6000263690948486,
      "learning_rate": 1.2339e-05,
      "loss": 0.0625,
      "step": 4116
    },
    {
      "epoch": 7.5964912280701755,
      "grad_norm": 0.6422172784805298,
      "learning_rate": 1.2342e-05,
      "loss": 0.0465,
      "step": 4117
    },
    {
      "epoch": 7.598337950138504,
      "grad_norm": 0.5606541633605957,
      "learning_rate": 1.2345e-05,
      "loss": 0.0379,
      "step": 4118
    },
    {
      "epoch": 7.600184672206833,
      "grad_norm": 0.3826885521411896,
      "learning_rate": 1.2348000000000001e-05,
      "loss": 0.021,
      "step": 4119
    },
    {
      "epoch": 7.602031394275162,
      "grad_norm": 0.743736982345581,
      "learning_rate": 1.2351e-05,
      "loss": 0.054,
      "step": 4120
    },
    {
      "epoch": 7.60387811634349,
      "grad_norm": 0.5651935935020447,
      "learning_rate": 1.2354e-05,
      "loss": 0.0403,
      "step": 4121
    },
    {
      "epoch": 7.605724838411819,
      "grad_norm": 0.5326204895973206,
      "learning_rate": 1.2357e-05,
      "loss": 0.0312,
      "step": 4122
    },
    {
      "epoch": 7.607571560480148,
      "grad_norm": 0.6177520751953125,
      "learning_rate": 1.236e-05,
      "loss": 0.0365,
      "step": 4123
    },
    {
      "epoch": 7.609418282548477,
      "grad_norm": 1.1195601224899292,
      "learning_rate": 1.2363000000000002e-05,
      "loss": 0.099,
      "step": 4124
    },
    {
      "epoch": 7.611265004616805,
      "grad_norm": 0.590433657169342,
      "learning_rate": 1.2366e-05,
      "loss": 0.0362,
      "step": 4125
    },
    {
      "epoch": 7.613111726685134,
      "grad_norm": 0.7750802040100098,
      "learning_rate": 1.2369e-05,
      "loss": 0.0409,
      "step": 4126
    },
    {
      "epoch": 7.6149584487534625,
      "grad_norm": 0.581501841545105,
      "learning_rate": 1.2372e-05,
      "loss": 0.0392,
      "step": 4127
    },
    {
      "epoch": 7.616805170821792,
      "grad_norm": 0.3566802740097046,
      "learning_rate": 1.2375e-05,
      "loss": 0.0193,
      "step": 4128
    },
    {
      "epoch": 7.61865189289012,
      "grad_norm": 0.6354184746742249,
      "learning_rate": 1.2378e-05,
      "loss": 0.0352,
      "step": 4129
    },
    {
      "epoch": 7.620498614958449,
      "grad_norm": 0.6342035531997681,
      "learning_rate": 1.2381e-05,
      "loss": 0.0585,
      "step": 4130
    },
    {
      "epoch": 7.622345337026777,
      "grad_norm": 0.473110556602478,
      "learning_rate": 1.2384e-05,
      "loss": 0.0255,
      "step": 4131
    },
    {
      "epoch": 7.624192059095106,
      "grad_norm": 0.5356754064559937,
      "learning_rate": 1.2387e-05,
      "loss": 0.0334,
      "step": 4132
    },
    {
      "epoch": 7.626038781163435,
      "grad_norm": 0.5536307096481323,
      "learning_rate": 1.239e-05,
      "loss": 0.0506,
      "step": 4133
    },
    {
      "epoch": 7.627885503231764,
      "grad_norm": 0.7165328860282898,
      "learning_rate": 1.2393000000000001e-05,
      "loss": 0.0605,
      "step": 4134
    },
    {
      "epoch": 7.629732225300092,
      "grad_norm": 0.48536062240600586,
      "learning_rate": 1.2396000000000001e-05,
      "loss": 0.0277,
      "step": 4135
    },
    {
      "epoch": 7.631578947368421,
      "grad_norm": 0.6589899063110352,
      "learning_rate": 1.2399000000000001e-05,
      "loss": 0.0464,
      "step": 4136
    },
    {
      "epoch": 7.6334256694367495,
      "grad_norm": 0.8539356589317322,
      "learning_rate": 1.2402e-05,
      "loss": 0.0503,
      "step": 4137
    },
    {
      "epoch": 7.635272391505079,
      "grad_norm": 0.9269492626190186,
      "learning_rate": 1.2404999999999999e-05,
      "loss": 0.0487,
      "step": 4138
    },
    {
      "epoch": 7.637119113573407,
      "grad_norm": 0.7299609780311584,
      "learning_rate": 1.2408e-05,
      "loss": 0.0516,
      "step": 4139
    },
    {
      "epoch": 7.638965835641736,
      "grad_norm": 0.7203462719917297,
      "learning_rate": 1.2411e-05,
      "loss": 0.0255,
      "step": 4140
    },
    {
      "epoch": 7.640812557710064,
      "grad_norm": 0.5248554348945618,
      "learning_rate": 1.2414e-05,
      "loss": 0.0307,
      "step": 4141
    },
    {
      "epoch": 7.642659279778393,
      "grad_norm": 0.7557847499847412,
      "learning_rate": 1.2417e-05,
      "loss": 0.0451,
      "step": 4142
    },
    {
      "epoch": 7.644506001846722,
      "grad_norm": 0.8592972755432129,
      "learning_rate": 1.242e-05,
      "loss": 0.0482,
      "step": 4143
    },
    {
      "epoch": 7.646352723915051,
      "grad_norm": 0.8314827680587769,
      "learning_rate": 1.2423000000000001e-05,
      "loss": 0.0517,
      "step": 4144
    },
    {
      "epoch": 7.64819944598338,
      "grad_norm": 1.7957693338394165,
      "learning_rate": 1.2426000000000001e-05,
      "loss": 0.3635,
      "step": 4145
    },
    {
      "epoch": 7.650046168051708,
      "grad_norm": 1.1916546821594238,
      "learning_rate": 1.2429e-05,
      "loss": 0.297,
      "step": 4146
    },
    {
      "epoch": 7.6518928901200365,
      "grad_norm": 1.383188009262085,
      "learning_rate": 1.2432e-05,
      "loss": 0.2379,
      "step": 4147
    },
    {
      "epoch": 7.653739612188366,
      "grad_norm": 1.1948586702346802,
      "learning_rate": 1.2435e-05,
      "loss": 0.2576,
      "step": 4148
    },
    {
      "epoch": 7.655586334256695,
      "grad_norm": 0.7643679976463318,
      "learning_rate": 1.2438000000000002e-05,
      "loss": 0.1959,
      "step": 4149
    },
    {
      "epoch": 7.657433056325023,
      "grad_norm": 0.8329753875732422,
      "learning_rate": 1.2441e-05,
      "loss": 0.2095,
      "step": 4150
    },
    {
      "epoch": 7.659279778393352,
      "grad_norm": 0.8791571855545044,
      "learning_rate": 1.2444e-05,
      "loss": 0.169,
      "step": 4151
    },
    {
      "epoch": 7.66112650046168,
      "grad_norm": 0.766928493976593,
      "learning_rate": 1.2447e-05,
      "loss": 0.1511,
      "step": 4152
    },
    {
      "epoch": 7.66297322253001,
      "grad_norm": 0.9058430790901184,
      "learning_rate": 1.245e-05,
      "loss": 0.1512,
      "step": 4153
    },
    {
      "epoch": 7.664819944598338,
      "grad_norm": 0.6345282196998596,
      "learning_rate": 1.2453000000000001e-05,
      "loss": 0.1035,
      "step": 4154
    },
    {
      "epoch": 7.666666666666667,
      "grad_norm": 2.050560235977173,
      "learning_rate": 1.2456e-05,
      "loss": 0.1019,
      "step": 4155
    },
    {
      "epoch": 7.668513388734995,
      "grad_norm": 0.7613061666488647,
      "learning_rate": 1.2459e-05,
      "loss": 0.113,
      "step": 4156
    },
    {
      "epoch": 7.670360110803324,
      "grad_norm": 1.1732903718948364,
      "learning_rate": 1.2462e-05,
      "loss": 0.191,
      "step": 4157
    },
    {
      "epoch": 7.672206832871653,
      "grad_norm": 0.9498212933540344,
      "learning_rate": 1.2465e-05,
      "loss": 0.0967,
      "step": 4158
    },
    {
      "epoch": 7.674053554939982,
      "grad_norm": 0.49174702167510986,
      "learning_rate": 1.2468000000000002e-05,
      "loss": 0.0451,
      "step": 4159
    },
    {
      "epoch": 7.67590027700831,
      "grad_norm": 0.7871057391166687,
      "learning_rate": 1.2471000000000001e-05,
      "loss": 0.0774,
      "step": 4160
    },
    {
      "epoch": 7.677746999076639,
      "grad_norm": 0.4107961356639862,
      "learning_rate": 1.2474000000000001e-05,
      "loss": 0.0401,
      "step": 4161
    },
    {
      "epoch": 7.679593721144967,
      "grad_norm": 0.8008513450622559,
      "learning_rate": 1.2477e-05,
      "loss": 0.0526,
      "step": 4162
    },
    {
      "epoch": 7.6814404432132966,
      "grad_norm": 0.46558696031570435,
      "learning_rate": 1.2479999999999999e-05,
      "loss": 0.0307,
      "step": 4163
    },
    {
      "epoch": 7.683287165281625,
      "grad_norm": 0.44487065076828003,
      "learning_rate": 1.2483e-05,
      "loss": 0.036,
      "step": 4164
    },
    {
      "epoch": 7.685133887349954,
      "grad_norm": 0.5548285841941833,
      "learning_rate": 1.2486e-05,
      "loss": 0.0441,
      "step": 4165
    },
    {
      "epoch": 7.686980609418282,
      "grad_norm": 0.5006272196769714,
      "learning_rate": 1.2489e-05,
      "loss": 0.0331,
      "step": 4166
    },
    {
      "epoch": 7.688827331486611,
      "grad_norm": 0.3948770761489868,
      "learning_rate": 1.2492e-05,
      "loss": 0.0378,
      "step": 4167
    },
    {
      "epoch": 7.69067405355494,
      "grad_norm": 0.4485151767730713,
      "learning_rate": 1.2495e-05,
      "loss": 0.0309,
      "step": 4168
    },
    {
      "epoch": 7.692520775623269,
      "grad_norm": 0.5578030347824097,
      "learning_rate": 1.2498000000000001e-05,
      "loss": 0.0414,
      "step": 4169
    },
    {
      "epoch": 7.694367497691598,
      "grad_norm": 0.4920484721660614,
      "learning_rate": 1.2501000000000001e-05,
      "loss": 0.0289,
      "step": 4170
    },
    {
      "epoch": 7.696214219759926,
      "grad_norm": 0.7462189793586731,
      "learning_rate": 1.2504000000000001e-05,
      "loss": 0.0612,
      "step": 4171
    },
    {
      "epoch": 7.698060941828254,
      "grad_norm": 0.4768451452255249,
      "learning_rate": 1.2507e-05,
      "loss": 0.0297,
      "step": 4172
    },
    {
      "epoch": 7.6999076638965835,
      "grad_norm": 0.4360469877719879,
      "learning_rate": 1.251e-05,
      "loss": 0.0379,
      "step": 4173
    },
    {
      "epoch": 7.701754385964913,
      "grad_norm": 0.766214907169342,
      "learning_rate": 1.2513e-05,
      "loss": 0.0379,
      "step": 4174
    },
    {
      "epoch": 7.703601108033241,
      "grad_norm": 1.5813432931900024,
      "learning_rate": 1.2516e-05,
      "loss": 0.078,
      "step": 4175
    },
    {
      "epoch": 7.70544783010157,
      "grad_norm": 0.6080664396286011,
      "learning_rate": 1.2519e-05,
      "loss": 0.0409,
      "step": 4176
    },
    {
      "epoch": 7.707294552169898,
      "grad_norm": 1.1184157133102417,
      "learning_rate": 1.2522e-05,
      "loss": 0.0319,
      "step": 4177
    },
    {
      "epoch": 7.7091412742382275,
      "grad_norm": 0.6684539318084717,
      "learning_rate": 1.2525e-05,
      "loss": 0.0299,
      "step": 4178
    },
    {
      "epoch": 7.710987996306556,
      "grad_norm": 0.6261076331138611,
      "learning_rate": 1.2528000000000001e-05,
      "loss": 0.0342,
      "step": 4179
    },
    {
      "epoch": 7.712834718374885,
      "grad_norm": 0.5831961631774902,
      "learning_rate": 1.2531e-05,
      "loss": 0.0277,
      "step": 4180
    },
    {
      "epoch": 7.714681440443213,
      "grad_norm": 0.5599172711372375,
      "learning_rate": 1.2534e-05,
      "loss": 0.0453,
      "step": 4181
    },
    {
      "epoch": 7.716528162511542,
      "grad_norm": 0.5989073514938354,
      "learning_rate": 1.2537e-05,
      "loss": 0.0219,
      "step": 4182
    },
    {
      "epoch": 7.7183748845798705,
      "grad_norm": 0.48640766739845276,
      "learning_rate": 1.254e-05,
      "loss": 0.028,
      "step": 4183
    },
    {
      "epoch": 7.7202216066482,
      "grad_norm": 0.7419519424438477,
      "learning_rate": 1.2543000000000002e-05,
      "loss": 0.0479,
      "step": 4184
    },
    {
      "epoch": 7.722068328716528,
      "grad_norm": 0.4289011061191559,
      "learning_rate": 1.2546000000000002e-05,
      "loss": 0.0187,
      "step": 4185
    },
    {
      "epoch": 7.723915050784857,
      "grad_norm": 0.9601662158966064,
      "learning_rate": 1.2549000000000001e-05,
      "loss": 0.0284,
      "step": 4186
    },
    {
      "epoch": 7.725761772853185,
      "grad_norm": 0.7918846607208252,
      "learning_rate": 1.2552e-05,
      "loss": 0.0461,
      "step": 4187
    },
    {
      "epoch": 7.7276084949215145,
      "grad_norm": 0.7785574793815613,
      "learning_rate": 1.2555e-05,
      "loss": 0.0362,
      "step": 4188
    },
    {
      "epoch": 7.729455216989843,
      "grad_norm": 0.8537510633468628,
      "learning_rate": 1.2558e-05,
      "loss": 0.0385,
      "step": 4189
    },
    {
      "epoch": 7.731301939058172,
      "grad_norm": 1.2144277095794678,
      "learning_rate": 1.2561e-05,
      "loss": 0.0485,
      "step": 4190
    },
    {
      "epoch": 7.7331486611265,
      "grad_norm": 1.2213268280029297,
      "learning_rate": 1.2564e-05,
      "loss": 0.038,
      "step": 4191
    },
    {
      "epoch": 7.734995383194829,
      "grad_norm": 0.6018048524856567,
      "learning_rate": 1.2567e-05,
      "loss": 0.0365,
      "step": 4192
    },
    {
      "epoch": 7.7368421052631575,
      "grad_norm": 0.8807831406593323,
      "learning_rate": 1.257e-05,
      "loss": 0.0544,
      "step": 4193
    },
    {
      "epoch": 7.738688827331487,
      "grad_norm": 0.9523292183876038,
      "learning_rate": 1.2573e-05,
      "loss": 0.088,
      "step": 4194
    },
    {
      "epoch": 7.740535549399816,
      "grad_norm": 1.3522166013717651,
      "learning_rate": 1.2576000000000001e-05,
      "loss": 0.3798,
      "step": 4195
    },
    {
      "epoch": 7.742382271468144,
      "grad_norm": 1.0144202709197998,
      "learning_rate": 1.2579000000000001e-05,
      "loss": 0.3224,
      "step": 4196
    },
    {
      "epoch": 7.744228993536472,
      "grad_norm": 1.2468231916427612,
      "learning_rate": 1.2582e-05,
      "loss": 0.2713,
      "step": 4197
    },
    {
      "epoch": 7.7460757156048015,
      "grad_norm": 1.5040656328201294,
      "learning_rate": 1.2585e-05,
      "loss": 0.2684,
      "step": 4198
    },
    {
      "epoch": 7.747922437673131,
      "grad_norm": 1.2979474067687988,
      "learning_rate": 1.2587999999999999e-05,
      "loss": 0.2113,
      "step": 4199
    },
    {
      "epoch": 7.749769159741459,
      "grad_norm": 1.4345593452453613,
      "learning_rate": 1.2591e-05,
      "loss": 0.2385,
      "step": 4200
    },
    {
      "epoch": 7.751615881809788,
      "grad_norm": 1.5685144662857056,
      "learning_rate": 1.2594e-05,
      "loss": 0.2577,
      "step": 4201
    },
    {
      "epoch": 7.753462603878116,
      "grad_norm": 1.26348078250885,
      "learning_rate": 1.2597e-05,
      "loss": 0.1832,
      "step": 4202
    },
    {
      "epoch": 7.755309325946445,
      "grad_norm": 0.8908654451370239,
      "learning_rate": 1.26e-05,
      "loss": 0.1213,
      "step": 4203
    },
    {
      "epoch": 7.757156048014774,
      "grad_norm": 0.7878689169883728,
      "learning_rate": 1.2603e-05,
      "loss": 0.1387,
      "step": 4204
    },
    {
      "epoch": 7.759002770083103,
      "grad_norm": 0.7852795720100403,
      "learning_rate": 1.2606000000000001e-05,
      "loss": 0.147,
      "step": 4205
    },
    {
      "epoch": 7.760849492151431,
      "grad_norm": 0.8014175295829773,
      "learning_rate": 1.2609e-05,
      "loss": 0.1322,
      "step": 4206
    },
    {
      "epoch": 7.76269621421976,
      "grad_norm": 0.8985889554023743,
      "learning_rate": 1.2612e-05,
      "loss": 0.1557,
      "step": 4207
    },
    {
      "epoch": 7.7645429362880884,
      "grad_norm": 1.881499171257019,
      "learning_rate": 1.2615e-05,
      "loss": 0.0711,
      "step": 4208
    },
    {
      "epoch": 7.766389658356418,
      "grad_norm": 0.6447560787200928,
      "learning_rate": 1.2618e-05,
      "loss": 0.0909,
      "step": 4209
    },
    {
      "epoch": 7.768236380424746,
      "grad_norm": 0.5465224385261536,
      "learning_rate": 1.2621000000000002e-05,
      "loss": 0.0681,
      "step": 4210
    },
    {
      "epoch": 7.770083102493075,
      "grad_norm": 0.6470277905464172,
      "learning_rate": 1.2624e-05,
      "loss": 0.0669,
      "step": 4211
    },
    {
      "epoch": 7.771929824561403,
      "grad_norm": 0.58457350730896,
      "learning_rate": 1.2627e-05,
      "loss": 0.0585,
      "step": 4212
    },
    {
      "epoch": 7.773776546629732,
      "grad_norm": 0.6071618795394897,
      "learning_rate": 1.263e-05,
      "loss": 0.0395,
      "step": 4213
    },
    {
      "epoch": 7.775623268698061,
      "grad_norm": 0.676609456539154,
      "learning_rate": 1.2633e-05,
      "loss": 0.0525,
      "step": 4214
    },
    {
      "epoch": 7.77746999076639,
      "grad_norm": 0.3999547064304352,
      "learning_rate": 1.2636e-05,
      "loss": 0.037,
      "step": 4215
    },
    {
      "epoch": 7.779316712834718,
      "grad_norm": 0.5443705916404724,
      "learning_rate": 1.2639e-05,
      "loss": 0.0371,
      "step": 4216
    },
    {
      "epoch": 7.781163434903047,
      "grad_norm": 0.9816798567771912,
      "learning_rate": 1.2642e-05,
      "loss": 0.0419,
      "step": 4217
    },
    {
      "epoch": 7.783010156971375,
      "grad_norm": 0.6923866271972656,
      "learning_rate": 1.2645e-05,
      "loss": 0.0465,
      "step": 4218
    },
    {
      "epoch": 7.784856879039705,
      "grad_norm": 0.5095269083976746,
      "learning_rate": 1.2648e-05,
      "loss": 0.0343,
      "step": 4219
    },
    {
      "epoch": 7.786703601108034,
      "grad_norm": 0.6733057498931885,
      "learning_rate": 1.2651000000000001e-05,
      "loss": 0.0409,
      "step": 4220
    },
    {
      "epoch": 7.788550323176362,
      "grad_norm": 0.8574805855751038,
      "learning_rate": 1.2654000000000001e-05,
      "loss": 0.0374,
      "step": 4221
    },
    {
      "epoch": 7.79039704524469,
      "grad_norm": 0.6730523109436035,
      "learning_rate": 1.2657000000000001e-05,
      "loss": 0.0235,
      "step": 4222
    },
    {
      "epoch": 7.792243767313019,
      "grad_norm": 0.6429433226585388,
      "learning_rate": 1.2659999999999999e-05,
      "loss": 0.0508,
      "step": 4223
    },
    {
      "epoch": 7.7940904893813485,
      "grad_norm": 0.6499189138412476,
      "learning_rate": 1.2662999999999999e-05,
      "loss": 0.0388,
      "step": 4224
    },
    {
      "epoch": 7.795937211449677,
      "grad_norm": 0.8994177579879761,
      "learning_rate": 1.2666e-05,
      "loss": 0.0443,
      "step": 4225
    },
    {
      "epoch": 7.797783933518006,
      "grad_norm": 0.7503787279129028,
      "learning_rate": 1.2669e-05,
      "loss": 0.0268,
      "step": 4226
    },
    {
      "epoch": 7.799630655586334,
      "grad_norm": 0.5890717506408691,
      "learning_rate": 1.2672e-05,
      "loss": 0.0382,
      "step": 4227
    },
    {
      "epoch": 7.801477377654663,
      "grad_norm": 0.5628544092178345,
      "learning_rate": 1.2675e-05,
      "loss": 0.0344,
      "step": 4228
    },
    {
      "epoch": 7.803324099722992,
      "grad_norm": 0.4928682744503021,
      "learning_rate": 1.2678e-05,
      "loss": 0.0325,
      "step": 4229
    },
    {
      "epoch": 7.805170821791321,
      "grad_norm": 0.787543535232544,
      "learning_rate": 1.2681000000000001e-05,
      "loss": 0.0362,
      "step": 4230
    },
    {
      "epoch": 7.807017543859649,
      "grad_norm": 1.0132980346679688,
      "learning_rate": 1.2684000000000001e-05,
      "loss": 0.0364,
      "step": 4231
    },
    {
      "epoch": 7.808864265927978,
      "grad_norm": 0.5771153569221497,
      "learning_rate": 1.2687e-05,
      "loss": 0.0343,
      "step": 4232
    },
    {
      "epoch": 7.810710987996306,
      "grad_norm": 0.8791329860687256,
      "learning_rate": 1.269e-05,
      "loss": 0.0577,
      "step": 4233
    },
    {
      "epoch": 7.8125577100646355,
      "grad_norm": 0.7927799820899963,
      "learning_rate": 1.2693e-05,
      "loss": 0.0372,
      "step": 4234
    },
    {
      "epoch": 7.814404432132964,
      "grad_norm": 0.7587766647338867,
      "learning_rate": 1.2696000000000002e-05,
      "loss": 0.0323,
      "step": 4235
    },
    {
      "epoch": 7.816251154201293,
      "grad_norm": 1.1068307161331177,
      "learning_rate": 1.2699e-05,
      "loss": 0.0602,
      "step": 4236
    },
    {
      "epoch": 7.818097876269621,
      "grad_norm": 1.694675326347351,
      "learning_rate": 1.2702e-05,
      "loss": 0.0458,
      "step": 4237
    },
    {
      "epoch": 7.81994459833795,
      "grad_norm": 0.6220648288726807,
      "learning_rate": 1.2705e-05,
      "loss": 0.0459,
      "step": 4238
    },
    {
      "epoch": 7.821791320406279,
      "grad_norm": 0.6234679222106934,
      "learning_rate": 1.2708e-05,
      "loss": 0.0373,
      "step": 4239
    },
    {
      "epoch": 7.823638042474608,
      "grad_norm": 0.6579725742340088,
      "learning_rate": 1.2711e-05,
      "loss": 0.0335,
      "step": 4240
    },
    {
      "epoch": 7.825484764542936,
      "grad_norm": 0.7174988985061646,
      "learning_rate": 1.2714e-05,
      "loss": 0.0392,
      "step": 4241
    },
    {
      "epoch": 7.827331486611265,
      "grad_norm": 0.9258104562759399,
      "learning_rate": 1.2717e-05,
      "loss": 0.0305,
      "step": 4242
    },
    {
      "epoch": 7.829178208679593,
      "grad_norm": 0.7623401284217834,
      "learning_rate": 1.272e-05,
      "loss": 0.0434,
      "step": 4243
    },
    {
      "epoch": 7.8310249307479225,
      "grad_norm": 0.6249094009399414,
      "learning_rate": 1.2723e-05,
      "loss": 0.0507,
      "step": 4244
    },
    {
      "epoch": 7.832871652816252,
      "grad_norm": 1.315299153327942,
      "learning_rate": 1.2726000000000001e-05,
      "loss": 0.343,
      "step": 4245
    },
    {
      "epoch": 7.83471837488458,
      "grad_norm": 0.8846774101257324,
      "learning_rate": 1.2729000000000001e-05,
      "loss": 0.2324,
      "step": 4246
    },
    {
      "epoch": 7.836565096952908,
      "grad_norm": 0.7561314702033997,
      "learning_rate": 1.2732000000000001e-05,
      "loss": 0.2017,
      "step": 4247
    },
    {
      "epoch": 7.838411819021237,
      "grad_norm": 0.9137879014015198,
      "learning_rate": 1.2735e-05,
      "loss": 0.2314,
      "step": 4248
    },
    {
      "epoch": 7.840258541089566,
      "grad_norm": 0.7995381355285645,
      "learning_rate": 1.2737999999999999e-05,
      "loss": 0.2356,
      "step": 4249
    },
    {
      "epoch": 7.842105263157895,
      "grad_norm": 0.9145019054412842,
      "learning_rate": 1.2741e-05,
      "loss": 0.2342,
      "step": 4250
    },
    {
      "epoch": 7.843951985226224,
      "grad_norm": 0.6350741982460022,
      "learning_rate": 1.2744e-05,
      "loss": 0.1745,
      "step": 4251
    },
    {
      "epoch": 7.845798707294552,
      "grad_norm": 1.0012751817703247,
      "learning_rate": 1.2747e-05,
      "loss": 0.1851,
      "step": 4252
    },
    {
      "epoch": 7.847645429362881,
      "grad_norm": 1.0654654502868652,
      "learning_rate": 1.275e-05,
      "loss": 0.197,
      "step": 4253
    },
    {
      "epoch": 7.8494921514312095,
      "grad_norm": 0.7405928373336792,
      "learning_rate": 1.2753e-05,
      "loss": 0.1307,
      "step": 4254
    },
    {
      "epoch": 7.851338873499539,
      "grad_norm": 1.0535868406295776,
      "learning_rate": 1.2756000000000001e-05,
      "loss": 0.1201,
      "step": 4255
    },
    {
      "epoch": 7.853185595567867,
      "grad_norm": 0.6765692234039307,
      "learning_rate": 1.2759000000000001e-05,
      "loss": 0.0829,
      "step": 4256
    },
    {
      "epoch": 7.855032317636196,
      "grad_norm": 0.9713176488876343,
      "learning_rate": 1.2762e-05,
      "loss": 0.1442,
      "step": 4257
    },
    {
      "epoch": 7.856879039704524,
      "grad_norm": 0.7366648316383362,
      "learning_rate": 1.2765e-05,
      "loss": 0.0868,
      "step": 4258
    },
    {
      "epoch": 7.858725761772853,
      "grad_norm": 0.5480320453643799,
      "learning_rate": 1.2768e-05,
      "loss": 0.0614,
      "step": 4259
    },
    {
      "epoch": 7.860572483841182,
      "grad_norm": 0.5844116806983948,
      "learning_rate": 1.2771e-05,
      "loss": 0.0615,
      "step": 4260
    },
    {
      "epoch": 7.862419205909511,
      "grad_norm": 0.6884409785270691,
      "learning_rate": 1.2774e-05,
      "loss": 0.0648,
      "step": 4261
    },
    {
      "epoch": 7.864265927977839,
      "grad_norm": 0.7041671276092529,
      "learning_rate": 1.2777e-05,
      "loss": 0.0585,
      "step": 4262
    },
    {
      "epoch": 7.866112650046168,
      "grad_norm": 1.6187162399291992,
      "learning_rate": 1.278e-05,
      "loss": 0.0443,
      "step": 4263
    },
    {
      "epoch": 7.8679593721144965,
      "grad_norm": 0.5576724410057068,
      "learning_rate": 1.2783e-05,
      "loss": 0.0413,
      "step": 4264
    },
    {
      "epoch": 7.869806094182826,
      "grad_norm": 0.644615113735199,
      "learning_rate": 1.2786000000000001e-05,
      "loss": 0.0376,
      "step": 4265
    },
    {
      "epoch": 7.871652816251154,
      "grad_norm": 0.7323140501976013,
      "learning_rate": 1.2789e-05,
      "loss": 0.0435,
      "step": 4266
    },
    {
      "epoch": 7.873499538319483,
      "grad_norm": 0.9791684150695801,
      "learning_rate": 1.2792e-05,
      "loss": 0.0359,
      "step": 4267
    },
    {
      "epoch": 7.875346260387811,
      "grad_norm": 1.0607141256332397,
      "learning_rate": 1.2795e-05,
      "loss": 0.052,
      "step": 4268
    },
    {
      "epoch": 7.87719298245614,
      "grad_norm": 0.684356153011322,
      "learning_rate": 1.2798e-05,
      "loss": 0.0328,
      "step": 4269
    },
    {
      "epoch": 7.87903970452447,
      "grad_norm": 0.6425195932388306,
      "learning_rate": 1.2801000000000002e-05,
      "loss": 0.0327,
      "step": 4270
    },
    {
      "epoch": 7.880886426592798,
      "grad_norm": 0.7910063862800598,
      "learning_rate": 1.2804000000000001e-05,
      "loss": 0.0423,
      "step": 4271
    },
    {
      "epoch": 7.882733148661126,
      "grad_norm": 0.648790180683136,
      "learning_rate": 1.2807000000000001e-05,
      "loss": 0.061,
      "step": 4272
    },
    {
      "epoch": 7.884579870729455,
      "grad_norm": 0.8567507266998291,
      "learning_rate": 1.281e-05,
      "loss": 0.0444,
      "step": 4273
    },
    {
      "epoch": 7.886426592797784,
      "grad_norm": 0.44043436646461487,
      "learning_rate": 1.2812999999999999e-05,
      "loss": 0.029,
      "step": 4274
    },
    {
      "epoch": 7.888273314866113,
      "grad_norm": 0.7219623923301697,
      "learning_rate": 1.2816e-05,
      "loss": 0.0619,
      "step": 4275
    },
    {
      "epoch": 7.890120036934442,
      "grad_norm": 0.4809543192386627,
      "learning_rate": 1.2819e-05,
      "loss": 0.0396,
      "step": 4276
    },
    {
      "epoch": 7.89196675900277,
      "grad_norm": 0.48918649554252625,
      "learning_rate": 1.2822e-05,
      "loss": 0.0316,
      "step": 4277
    },
    {
      "epoch": 7.893813481071099,
      "grad_norm": 0.7103901505470276,
      "learning_rate": 1.2825e-05,
      "loss": 0.0436,
      "step": 4278
    },
    {
      "epoch": 7.895660203139427,
      "grad_norm": 1.1335327625274658,
      "learning_rate": 1.2828e-05,
      "loss": 0.042,
      "step": 4279
    },
    {
      "epoch": 7.8975069252077565,
      "grad_norm": 0.5287763476371765,
      "learning_rate": 1.2831000000000001e-05,
      "loss": 0.028,
      "step": 4280
    },
    {
      "epoch": 7.899353647276085,
      "grad_norm": 0.5864327549934387,
      "learning_rate": 1.2834000000000001e-05,
      "loss": 0.0476,
      "step": 4281
    },
    {
      "epoch": 7.901200369344414,
      "grad_norm": 0.7537019848823547,
      "learning_rate": 1.2837000000000001e-05,
      "loss": 0.0594,
      "step": 4282
    },
    {
      "epoch": 7.903047091412742,
      "grad_norm": 0.7733783721923828,
      "learning_rate": 1.284e-05,
      "loss": 0.0467,
      "step": 4283
    },
    {
      "epoch": 7.904893813481071,
      "grad_norm": 0.9843820929527283,
      "learning_rate": 1.2843e-05,
      "loss": 0.0616,
      "step": 4284
    },
    {
      "epoch": 7.9067405355494,
      "grad_norm": 0.5933932065963745,
      "learning_rate": 1.2846e-05,
      "loss": 0.0397,
      "step": 4285
    },
    {
      "epoch": 7.908587257617729,
      "grad_norm": 0.7289738059043884,
      "learning_rate": 1.2849e-05,
      "loss": 0.0563,
      "step": 4286
    },
    {
      "epoch": 7.910433979686057,
      "grad_norm": 0.6274788975715637,
      "learning_rate": 1.2852e-05,
      "loss": 0.0424,
      "step": 4287
    },
    {
      "epoch": 7.912280701754386,
      "grad_norm": 0.8971357941627502,
      "learning_rate": 1.2855e-05,
      "loss": 0.0607,
      "step": 4288
    },
    {
      "epoch": 7.914127423822714,
      "grad_norm": 0.6123123168945312,
      "learning_rate": 1.2858e-05,
      "loss": 0.0335,
      "step": 4289
    },
    {
      "epoch": 7.9159741458910435,
      "grad_norm": 0.6328318119049072,
      "learning_rate": 1.2861000000000001e-05,
      "loss": 0.0378,
      "step": 4290
    },
    {
      "epoch": 7.917820867959372,
      "grad_norm": 0.6868095993995667,
      "learning_rate": 1.2864e-05,
      "loss": 0.0554,
      "step": 4291
    },
    {
      "epoch": 7.919667590027701,
      "grad_norm": 0.9725096225738525,
      "learning_rate": 1.2867e-05,
      "loss": 0.0695,
      "step": 4292
    },
    {
      "epoch": 7.921514312096029,
      "grad_norm": 0.8192182183265686,
      "learning_rate": 1.287e-05,
      "loss": 0.0351,
      "step": 4293
    },
    {
      "epoch": 7.923361034164358,
      "grad_norm": 0.8341385722160339,
      "learning_rate": 1.2873e-05,
      "loss": 0.0802,
      "step": 4294
    },
    {
      "epoch": 7.9252077562326875,
      "grad_norm": 1.3568755388259888,
      "learning_rate": 1.2876000000000002e-05,
      "loss": 0.4823,
      "step": 4295
    },
    {
      "epoch": 7.927054478301016,
      "grad_norm": 1.1360949277877808,
      "learning_rate": 1.2879000000000002e-05,
      "loss": 0.3611,
      "step": 4296
    },
    {
      "epoch": 7.928901200369344,
      "grad_norm": 0.790391743183136,
      "learning_rate": 1.2882e-05,
      "loss": 0.1874,
      "step": 4297
    },
    {
      "epoch": 7.930747922437673,
      "grad_norm": 0.743624746799469,
      "learning_rate": 1.2885e-05,
      "loss": 0.172,
      "step": 4298
    },
    {
      "epoch": 7.932594644506002,
      "grad_norm": 0.6538652181625366,
      "learning_rate": 1.2888e-05,
      "loss": 0.1649,
      "step": 4299
    },
    {
      "epoch": 7.9344413665743305,
      "grad_norm": 0.8588427901268005,
      "learning_rate": 1.2891e-05,
      "loss": 0.1684,
      "step": 4300
    },
    {
      "epoch": 7.93628808864266,
      "grad_norm": 0.6219055652618408,
      "learning_rate": 1.2894e-05,
      "loss": 0.1309,
      "step": 4301
    },
    {
      "epoch": 7.938134810710988,
      "grad_norm": 0.6097257137298584,
      "learning_rate": 1.2897e-05,
      "loss": 0.1437,
      "step": 4302
    },
    {
      "epoch": 7.939981532779317,
      "grad_norm": 0.6845511794090271,
      "learning_rate": 1.29e-05,
      "loss": 0.148,
      "step": 4303
    },
    {
      "epoch": 7.941828254847645,
      "grad_norm": 0.8553453087806702,
      "learning_rate": 1.2903e-05,
      "loss": 0.1539,
      "step": 4304
    },
    {
      "epoch": 7.9436749769159745,
      "grad_norm": 0.8057902455329895,
      "learning_rate": 1.2906000000000001e-05,
      "loss": 0.1407,
      "step": 4305
    },
    {
      "epoch": 7.945521698984303,
      "grad_norm": 0.7886427640914917,
      "learning_rate": 1.2909000000000001e-05,
      "loss": 0.1285,
      "step": 4306
    },
    {
      "epoch": 7.947368421052632,
      "grad_norm": 0.5424556732177734,
      "learning_rate": 1.2912000000000001e-05,
      "loss": 0.0449,
      "step": 4307
    },
    {
      "epoch": 7.94921514312096,
      "grad_norm": 0.8618884086608887,
      "learning_rate": 1.2915000000000001e-05,
      "loss": 0.0979,
      "step": 4308
    },
    {
      "epoch": 7.951061865189289,
      "grad_norm": 0.5419620275497437,
      "learning_rate": 1.2917999999999999e-05,
      "loss": 0.0577,
      "step": 4309
    },
    {
      "epoch": 7.9529085872576175,
      "grad_norm": 0.9529350996017456,
      "learning_rate": 1.2921e-05,
      "loss": 0.0946,
      "step": 4310
    },
    {
      "epoch": 7.954755309325947,
      "grad_norm": 0.6171501278877258,
      "learning_rate": 1.2924e-05,
      "loss": 0.055,
      "step": 4311
    },
    {
      "epoch": 7.956602031394275,
      "grad_norm": 0.7650995850563049,
      "learning_rate": 1.2927e-05,
      "loss": 0.0598,
      "step": 4312
    },
    {
      "epoch": 7.958448753462604,
      "grad_norm": 0.5648701190948486,
      "learning_rate": 1.293e-05,
      "loss": 0.0758,
      "step": 4313
    },
    {
      "epoch": 7.960295475530932,
      "grad_norm": 1.0188759565353394,
      "learning_rate": 1.2933e-05,
      "loss": 0.0477,
      "step": 4314
    },
    {
      "epoch": 7.9621421975992615,
      "grad_norm": 0.4948999583721161,
      "learning_rate": 1.2936000000000001e-05,
      "loss": 0.0379,
      "step": 4315
    },
    {
      "epoch": 7.96398891966759,
      "grad_norm": 0.6018174290657043,
      "learning_rate": 1.2939000000000001e-05,
      "loss": 0.0388,
      "step": 4316
    },
    {
      "epoch": 7.965835641735919,
      "grad_norm": 1.477062463760376,
      "learning_rate": 1.2942e-05,
      "loss": 0.0588,
      "step": 4317
    },
    {
      "epoch": 7.967682363804247,
      "grad_norm": 0.666613757610321,
      "learning_rate": 1.2945e-05,
      "loss": 0.044,
      "step": 4318
    },
    {
      "epoch": 7.969529085872576,
      "grad_norm": 0.6991389989852905,
      "learning_rate": 1.2948e-05,
      "loss": 0.0346,
      "step": 4319
    },
    {
      "epoch": 7.971375807940905,
      "grad_norm": 0.8860199451446533,
      "learning_rate": 1.2951e-05,
      "loss": 0.0211,
      "step": 4320
    },
    {
      "epoch": 7.973222530009234,
      "grad_norm": 1.3285048007965088,
      "learning_rate": 1.2954000000000002e-05,
      "loss": 0.0375,
      "step": 4321
    },
    {
      "epoch": 7.975069252077562,
      "grad_norm": 0.5883099436759949,
      "learning_rate": 1.2957e-05,
      "loss": 0.034,
      "step": 4322
    },
    {
      "epoch": 7.976915974145891,
      "grad_norm": 0.7883331179618835,
      "learning_rate": 1.296e-05,
      "loss": 0.0376,
      "step": 4323
    },
    {
      "epoch": 7.97876269621422,
      "grad_norm": 1.2550441026687622,
      "learning_rate": 1.2963e-05,
      "loss": 0.0627,
      "step": 4324
    },
    {
      "epoch": 7.980609418282548,
      "grad_norm": 0.7064722180366516,
      "learning_rate": 1.2966e-05,
      "loss": 0.0626,
      "step": 4325
    },
    {
      "epoch": 7.982456140350877,
      "grad_norm": 0.9073736071586609,
      "learning_rate": 1.2969e-05,
      "loss": 0.066,
      "step": 4326
    },
    {
      "epoch": 7.984302862419206,
      "grad_norm": 0.7056010961532593,
      "learning_rate": 1.2972e-05,
      "loss": 0.0372,
      "step": 4327
    },
    {
      "epoch": 7.986149584487535,
      "grad_norm": 0.6290423274040222,
      "learning_rate": 1.2975e-05,
      "loss": 0.0393,
      "step": 4328
    },
    {
      "epoch": 7.987996306555863,
      "grad_norm": 1.1520283222198486,
      "learning_rate": 1.2978e-05,
      "loss": 0.0428,
      "step": 4329
    },
    {
      "epoch": 7.989843028624192,
      "grad_norm": 0.9760458469390869,
      "learning_rate": 1.2981e-05,
      "loss": 0.0663,
      "step": 4330
    },
    {
      "epoch": 7.991689750692521,
      "grad_norm": 0.6979522109031677,
      "learning_rate": 1.2984000000000001e-05,
      "loss": 0.0335,
      "step": 4331
    },
    {
      "epoch": 7.99353647276085,
      "grad_norm": 0.6798151731491089,
      "learning_rate": 1.2987000000000001e-05,
      "loss": 0.0329,
      "step": 4332
    },
    {
      "epoch": 7.995383194829178,
      "grad_norm": 0.6956556439399719,
      "learning_rate": 1.2990000000000001e-05,
      "loss": 0.0393,
      "step": 4333
    },
    {
      "epoch": 7.997229916897507,
      "grad_norm": 0.7906112670898438,
      "learning_rate": 1.2992999999999999e-05,
      "loss": 0.0387,
      "step": 4334
    },
    {
      "epoch": 7.999076638965835,
      "grad_norm": 1.0580320358276367,
      "learning_rate": 1.2995999999999999e-05,
      "loss": 0.0529,
      "step": 4335
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.4188879132270813,
      "learning_rate": 1.2999e-05,
      "loss": 0.0076,
      "step": 4336
    },
    {
      "epoch": 8.00184672206833,
      "grad_norm": 1.5326043367385864,
      "learning_rate": 1.3002e-05,
      "loss": 0.3817,
      "step": 4337
    },
    {
      "epoch": 8.003693444136658,
      "grad_norm": 1.1512473821640015,
      "learning_rate": 1.3005e-05,
      "loss": 0.3305,
      "step": 4338
    },
    {
      "epoch": 8.005540166204986,
      "grad_norm": 0.9963605999946594,
      "learning_rate": 1.3008e-05,
      "loss": 0.26,
      "step": 4339
    },
    {
      "epoch": 8.007386888273315,
      "grad_norm": 0.8791415691375732,
      "learning_rate": 1.3011e-05,
      "loss": 0.1898,
      "step": 4340
    },
    {
      "epoch": 8.009233610341644,
      "grad_norm": 1.1535075902938843,
      "learning_rate": 1.3014000000000001e-05,
      "loss": 0.2692,
      "step": 4341
    },
    {
      "epoch": 8.011080332409973,
      "grad_norm": 0.983793318271637,
      "learning_rate": 1.3017000000000001e-05,
      "loss": 0.2331,
      "step": 4342
    },
    {
      "epoch": 8.0129270544783,
      "grad_norm": 0.7149614095687866,
      "learning_rate": 1.302e-05,
      "loss": 0.183,
      "step": 4343
    },
    {
      "epoch": 8.01477377654663,
      "grad_norm": 0.6314206719398499,
      "learning_rate": 1.3023e-05,
      "loss": 0.1293,
      "step": 4344
    },
    {
      "epoch": 8.016620498614959,
      "grad_norm": 0.8021345734596252,
      "learning_rate": 1.3026e-05,
      "loss": 0.136,
      "step": 4345
    },
    {
      "epoch": 8.018467220683288,
      "grad_norm": 0.5998855233192444,
      "learning_rate": 1.3029e-05,
      "loss": 0.0968,
      "step": 4346
    },
    {
      "epoch": 8.020313942751615,
      "grad_norm": 0.6993886232376099,
      "learning_rate": 1.3032e-05,
      "loss": 0.1002,
      "step": 4347
    },
    {
      "epoch": 8.022160664819944,
      "grad_norm": 0.9774983525276184,
      "learning_rate": 1.3035e-05,
      "loss": 0.1381,
      "step": 4348
    },
    {
      "epoch": 8.024007386888274,
      "grad_norm": 0.7680040001869202,
      "learning_rate": 1.3038e-05,
      "loss": 0.0921,
      "step": 4349
    },
    {
      "epoch": 8.025854108956603,
      "grad_norm": 0.9355478882789612,
      "learning_rate": 1.3041e-05,
      "loss": 0.1065,
      "step": 4350
    },
    {
      "epoch": 8.02770083102493,
      "grad_norm": 0.7027696967124939,
      "learning_rate": 1.3044e-05,
      "loss": 0.0703,
      "step": 4351
    },
    {
      "epoch": 8.02954755309326,
      "grad_norm": 1.3707549571990967,
      "learning_rate": 1.3047e-05,
      "loss": 0.1136,
      "step": 4352
    },
    {
      "epoch": 8.031394275161588,
      "grad_norm": 0.7237825393676758,
      "learning_rate": 1.305e-05,
      "loss": 0.082,
      "step": 4353
    },
    {
      "epoch": 8.033240997229917,
      "grad_norm": 0.6745184659957886,
      "learning_rate": 1.3053e-05,
      "loss": 0.054,
      "step": 4354
    },
    {
      "epoch": 8.035087719298245,
      "grad_norm": 0.8653813600540161,
      "learning_rate": 1.3056e-05,
      "loss": 0.0645,
      "step": 4355
    },
    {
      "epoch": 8.036934441366574,
      "grad_norm": 0.7573667764663696,
      "learning_rate": 1.3059000000000002e-05,
      "loss": 0.0388,
      "step": 4356
    },
    {
      "epoch": 8.038781163434903,
      "grad_norm": 0.4688849151134491,
      "learning_rate": 1.3062000000000001e-05,
      "loss": 0.0368,
      "step": 4357
    },
    {
      "epoch": 8.040627885503232,
      "grad_norm": 0.8460974097251892,
      "learning_rate": 1.3065000000000001e-05,
      "loss": 0.051,
      "step": 4358
    },
    {
      "epoch": 8.04247460757156,
      "grad_norm": 0.6380207538604736,
      "learning_rate": 1.3068e-05,
      "loss": 0.0557,
      "step": 4359
    },
    {
      "epoch": 8.044321329639889,
      "grad_norm": 0.3948462903499603,
      "learning_rate": 1.3070999999999999e-05,
      "loss": 0.022,
      "step": 4360
    },
    {
      "epoch": 8.046168051708218,
      "grad_norm": 0.6261285543441772,
      "learning_rate": 1.3074e-05,
      "loss": 0.0437,
      "step": 4361
    },
    {
      "epoch": 8.048014773776547,
      "grad_norm": 0.6887944936752319,
      "learning_rate": 1.3077e-05,
      "loss": 0.0436,
      "step": 4362
    },
    {
      "epoch": 8.049861495844876,
      "grad_norm": 0.49890008568763733,
      "learning_rate": 1.308e-05,
      "loss": 0.0348,
      "step": 4363
    },
    {
      "epoch": 8.051708217913204,
      "grad_norm": 0.5046862363815308,
      "learning_rate": 1.3083e-05,
      "loss": 0.0288,
      "step": 4364
    },
    {
      "epoch": 8.053554939981533,
      "grad_norm": 0.6013908386230469,
      "learning_rate": 1.3086e-05,
      "loss": 0.0362,
      "step": 4365
    },
    {
      "epoch": 8.055401662049862,
      "grad_norm": 1.3356207609176636,
      "learning_rate": 1.3089000000000001e-05,
      "loss": 0.0907,
      "step": 4366
    },
    {
      "epoch": 8.057248384118191,
      "grad_norm": 0.937406599521637,
      "learning_rate": 1.3092000000000001e-05,
      "loss": 0.0265,
      "step": 4367
    },
    {
      "epoch": 8.059095106186518,
      "grad_norm": 0.48784348368644714,
      "learning_rate": 1.3095e-05,
      "loss": 0.0651,
      "step": 4368
    },
    {
      "epoch": 8.060941828254848,
      "grad_norm": 0.6230420470237732,
      "learning_rate": 1.3098e-05,
      "loss": 0.0523,
      "step": 4369
    },
    {
      "epoch": 8.062788550323177,
      "grad_norm": 0.4827072024345398,
      "learning_rate": 1.3101e-05,
      "loss": 0.023,
      "step": 4370
    },
    {
      "epoch": 8.064635272391506,
      "grad_norm": 0.40178000926971436,
      "learning_rate": 1.3104e-05,
      "loss": 0.0281,
      "step": 4371
    },
    {
      "epoch": 8.066481994459833,
      "grad_norm": 0.5730307102203369,
      "learning_rate": 1.3107e-05,
      "loss": 0.0431,
      "step": 4372
    },
    {
      "epoch": 8.068328716528162,
      "grad_norm": 0.5679459571838379,
      "learning_rate": 1.311e-05,
      "loss": 0.0207,
      "step": 4373
    },
    {
      "epoch": 8.070175438596491,
      "grad_norm": 0.4760940372943878,
      "learning_rate": 1.3113e-05,
      "loss": 0.0386,
      "step": 4374
    },
    {
      "epoch": 8.07202216066482,
      "grad_norm": 0.8767454028129578,
      "learning_rate": 1.3116e-05,
      "loss": 0.0333,
      "step": 4375
    },
    {
      "epoch": 8.073868882733148,
      "grad_norm": 1.0410170555114746,
      "learning_rate": 1.3119000000000001e-05,
      "loss": 0.0701,
      "step": 4376
    },
    {
      "epoch": 8.075715604801477,
      "grad_norm": 0.7946456670761108,
      "learning_rate": 1.3122e-05,
      "loss": 0.0537,
      "step": 4377
    },
    {
      "epoch": 8.077562326869806,
      "grad_norm": 0.7042535543441772,
      "learning_rate": 1.3125e-05,
      "loss": 0.0382,
      "step": 4378
    },
    {
      "epoch": 8.079409048938135,
      "grad_norm": 0.6931938529014587,
      "learning_rate": 1.3128e-05,
      "loss": 0.0422,
      "step": 4379
    },
    {
      "epoch": 8.081255771006463,
      "grad_norm": 0.4324564039707184,
      "learning_rate": 1.3131e-05,
      "loss": 0.0203,
      "step": 4380
    },
    {
      "epoch": 8.083102493074792,
      "grad_norm": 0.7083581686019897,
      "learning_rate": 1.3134000000000002e-05,
      "loss": 0.0372,
      "step": 4381
    },
    {
      "epoch": 8.084949215143121,
      "grad_norm": 0.7215295433998108,
      "learning_rate": 1.3137000000000001e-05,
      "loss": 0.0263,
      "step": 4382
    },
    {
      "epoch": 8.08679593721145,
      "grad_norm": 0.36402127146720886,
      "learning_rate": 1.314e-05,
      "loss": 0.0205,
      "step": 4383
    },
    {
      "epoch": 8.088642659279778,
      "grad_norm": 0.4735167920589447,
      "learning_rate": 1.3143e-05,
      "loss": 0.0366,
      "step": 4384
    },
    {
      "epoch": 8.090489381348107,
      "grad_norm": 0.7522518038749695,
      "learning_rate": 1.3146e-05,
      "loss": 0.0374,
      "step": 4385
    },
    {
      "epoch": 8.092336103416436,
      "grad_norm": 2.186006784439087,
      "learning_rate": 1.3149e-05,
      "loss": 0.0495,
      "step": 4386
    },
    {
      "epoch": 8.094182825484765,
      "grad_norm": 1.0902612209320068,
      "learning_rate": 1.3152e-05,
      "loss": 0.284,
      "step": 4387
    },
    {
      "epoch": 8.096029547553094,
      "grad_norm": 0.8116087913513184,
      "learning_rate": 1.3155e-05,
      "loss": 0.2475,
      "step": 4388
    },
    {
      "epoch": 8.097876269621421,
      "grad_norm": 0.6619287133216858,
      "learning_rate": 1.3158e-05,
      "loss": 0.2024,
      "step": 4389
    },
    {
      "epoch": 8.09972299168975,
      "grad_norm": 0.7266683578491211,
      "learning_rate": 1.3161e-05,
      "loss": 0.1673,
      "step": 4390
    },
    {
      "epoch": 8.10156971375808,
      "grad_norm": 1.3175115585327148,
      "learning_rate": 1.3164000000000001e-05,
      "loss": 0.161,
      "step": 4391
    },
    {
      "epoch": 8.103416435826409,
      "grad_norm": 1.0672887563705444,
      "learning_rate": 1.3167000000000001e-05,
      "loss": 0.2241,
      "step": 4392
    },
    {
      "epoch": 8.105263157894736,
      "grad_norm": 0.9696524143218994,
      "learning_rate": 1.3170000000000001e-05,
      "loss": 0.1349,
      "step": 4393
    },
    {
      "epoch": 8.107109879963065,
      "grad_norm": 0.6774873733520508,
      "learning_rate": 1.3173e-05,
      "loss": 0.1617,
      "step": 4394
    },
    {
      "epoch": 8.108956602031395,
      "grad_norm": 1.2551002502441406,
      "learning_rate": 1.3175999999999999e-05,
      "loss": 0.2438,
      "step": 4395
    },
    {
      "epoch": 8.110803324099724,
      "grad_norm": 0.8130558133125305,
      "learning_rate": 1.3179e-05,
      "loss": 0.136,
      "step": 4396
    },
    {
      "epoch": 8.112650046168051,
      "grad_norm": 0.7603598833084106,
      "learning_rate": 1.3182e-05,
      "loss": 0.1235,
      "step": 4397
    },
    {
      "epoch": 8.11449676823638,
      "grad_norm": 0.7122944593429565,
      "learning_rate": 1.3185e-05,
      "loss": 0.1325,
      "step": 4398
    },
    {
      "epoch": 8.11634349030471,
      "grad_norm": 0.7684652805328369,
      "learning_rate": 1.3188e-05,
      "loss": 0.1338,
      "step": 4399
    },
    {
      "epoch": 8.118190212373039,
      "grad_norm": 0.7754552960395813,
      "learning_rate": 1.3191e-05,
      "loss": 0.1212,
      "step": 4400
    },
    {
      "epoch": 8.120036934441366,
      "grad_norm": 0.6043549180030823,
      "learning_rate": 1.3194000000000001e-05,
      "loss": 0.0441,
      "step": 4401
    },
    {
      "epoch": 8.121883656509695,
      "grad_norm": 0.5200878977775574,
      "learning_rate": 1.3197000000000001e-05,
      "loss": 0.0658,
      "step": 4402
    },
    {
      "epoch": 8.123730378578024,
      "grad_norm": 0.8166391253471375,
      "learning_rate": 1.32e-05,
      "loss": 0.0747,
      "step": 4403
    },
    {
      "epoch": 8.125577100646353,
      "grad_norm": 0.5633155107498169,
      "learning_rate": 1.3203e-05,
      "loss": 0.0467,
      "step": 4404
    },
    {
      "epoch": 8.12742382271468,
      "grad_norm": 0.66031813621521,
      "learning_rate": 1.3206e-05,
      "loss": 0.0333,
      "step": 4405
    },
    {
      "epoch": 8.12927054478301,
      "grad_norm": 1.0485261678695679,
      "learning_rate": 1.3209000000000002e-05,
      "loss": 0.0641,
      "step": 4406
    },
    {
      "epoch": 8.131117266851339,
      "grad_norm": 1.0621566772460938,
      "learning_rate": 1.3212000000000002e-05,
      "loss": 0.0357,
      "step": 4407
    },
    {
      "epoch": 8.132963988919668,
      "grad_norm": 0.6026491522789001,
      "learning_rate": 1.3215e-05,
      "loss": 0.0408,
      "step": 4408
    },
    {
      "epoch": 8.134810710987995,
      "grad_norm": 0.8135718703269958,
      "learning_rate": 1.3218e-05,
      "loss": 0.0391,
      "step": 4409
    },
    {
      "epoch": 8.136657433056325,
      "grad_norm": 0.5995201468467712,
      "learning_rate": 1.3221e-05,
      "loss": 0.0317,
      "step": 4410
    },
    {
      "epoch": 8.138504155124654,
      "grad_norm": 0.5987685322761536,
      "learning_rate": 1.3224e-05,
      "loss": 0.0354,
      "step": 4411
    },
    {
      "epoch": 8.140350877192983,
      "grad_norm": 0.5825679302215576,
      "learning_rate": 1.3227e-05,
      "loss": 0.0334,
      "step": 4412
    },
    {
      "epoch": 8.142197599261312,
      "grad_norm": 0.8200697898864746,
      "learning_rate": 1.323e-05,
      "loss": 0.0384,
      "step": 4413
    },
    {
      "epoch": 8.14404432132964,
      "grad_norm": 0.46164023876190186,
      "learning_rate": 1.3233e-05,
      "loss": 0.0259,
      "step": 4414
    },
    {
      "epoch": 8.145891043397969,
      "grad_norm": 0.7400566935539246,
      "learning_rate": 1.3236e-05,
      "loss": 0.0422,
      "step": 4415
    },
    {
      "epoch": 8.147737765466298,
      "grad_norm": 0.600922703742981,
      "learning_rate": 1.3239000000000001e-05,
      "loss": 0.0329,
      "step": 4416
    },
    {
      "epoch": 8.149584487534627,
      "grad_norm": 0.5295440554618835,
      "learning_rate": 1.3242000000000001e-05,
      "loss": 0.0308,
      "step": 4417
    },
    {
      "epoch": 8.151431209602954,
      "grad_norm": 0.8074960112571716,
      "learning_rate": 1.3245000000000001e-05,
      "loss": 0.0493,
      "step": 4418
    },
    {
      "epoch": 8.153277931671283,
      "grad_norm": 0.8136438131332397,
      "learning_rate": 1.3248000000000001e-05,
      "loss": 0.0399,
      "step": 4419
    },
    {
      "epoch": 8.155124653739612,
      "grad_norm": 0.4094987213611603,
      "learning_rate": 1.3250999999999999e-05,
      "loss": 0.031,
      "step": 4420
    },
    {
      "epoch": 8.156971375807942,
      "grad_norm": 0.5699203014373779,
      "learning_rate": 1.3254e-05,
      "loss": 0.0375,
      "step": 4421
    },
    {
      "epoch": 8.158818097876269,
      "grad_norm": 2.046877861022949,
      "learning_rate": 1.3257e-05,
      "loss": 0.0254,
      "step": 4422
    },
    {
      "epoch": 8.160664819944598,
      "grad_norm": 0.6146877408027649,
      "learning_rate": 1.326e-05,
      "loss": 0.0315,
      "step": 4423
    },
    {
      "epoch": 8.162511542012927,
      "grad_norm": 0.7275285124778748,
      "learning_rate": 1.3263e-05,
      "loss": 0.0254,
      "step": 4424
    },
    {
      "epoch": 8.164358264081256,
      "grad_norm": 0.54784095287323,
      "learning_rate": 1.3266e-05,
      "loss": 0.0245,
      "step": 4425
    },
    {
      "epoch": 8.166204986149584,
      "grad_norm": 0.7300387024879456,
      "learning_rate": 1.3269000000000001e-05,
      "loss": 0.0336,
      "step": 4426
    },
    {
      "epoch": 8.168051708217913,
      "grad_norm": 0.9745352864265442,
      "learning_rate": 1.3272000000000001e-05,
      "loss": 0.0493,
      "step": 4427
    },
    {
      "epoch": 8.169898430286242,
      "grad_norm": 1.2098708152770996,
      "learning_rate": 1.3275e-05,
      "loss": 0.0445,
      "step": 4428
    },
    {
      "epoch": 8.171745152354571,
      "grad_norm": 0.39160364866256714,
      "learning_rate": 1.3278e-05,
      "loss": 0.0188,
      "step": 4429
    },
    {
      "epoch": 8.173591874422899,
      "grad_norm": 0.7158820033073425,
      "learning_rate": 1.3281e-05,
      "loss": 0.0272,
      "step": 4430
    },
    {
      "epoch": 8.175438596491228,
      "grad_norm": 0.6653376817703247,
      "learning_rate": 1.3284000000000002e-05,
      "loss": 0.0292,
      "step": 4431
    },
    {
      "epoch": 8.177285318559557,
      "grad_norm": 0.5458754301071167,
      "learning_rate": 1.3287e-05,
      "loss": 0.0275,
      "step": 4432
    },
    {
      "epoch": 8.179132040627886,
      "grad_norm": 1.1866546869277954,
      "learning_rate": 1.329e-05,
      "loss": 0.0475,
      "step": 4433
    },
    {
      "epoch": 8.180978762696213,
      "grad_norm": 0.9386422634124756,
      "learning_rate": 1.3293e-05,
      "loss": 0.0426,
      "step": 4434
    },
    {
      "epoch": 8.182825484764543,
      "grad_norm": 1.0976414680480957,
      "learning_rate": 1.3296e-05,
      "loss": 0.0386,
      "step": 4435
    },
    {
      "epoch": 8.184672206832872,
      "grad_norm": 0.8207219243049622,
      "learning_rate": 1.3299000000000001e-05,
      "loss": 0.0349,
      "step": 4436
    },
    {
      "epoch": 8.1865189289012,
      "grad_norm": 1.59428870677948,
      "learning_rate": 1.3302e-05,
      "loss": 0.3254,
      "step": 4437
    },
    {
      "epoch": 8.18836565096953,
      "grad_norm": 1.0320184230804443,
      "learning_rate": 1.3305e-05,
      "loss": 0.2136,
      "step": 4438
    },
    {
      "epoch": 8.190212373037857,
      "grad_norm": 0.8654654622077942,
      "learning_rate": 1.3308e-05,
      "loss": 0.1964,
      "step": 4439
    },
    {
      "epoch": 8.192059095106186,
      "grad_norm": 0.7566434741020203,
      "learning_rate": 1.3311e-05,
      "loss": 0.2189,
      "step": 4440
    },
    {
      "epoch": 8.193905817174516,
      "grad_norm": 0.9181260466575623,
      "learning_rate": 1.3314e-05,
      "loss": 0.269,
      "step": 4441
    },
    {
      "epoch": 8.195752539242845,
      "grad_norm": 0.7232855558395386,
      "learning_rate": 1.3317000000000001e-05,
      "loss": 0.1816,
      "step": 4442
    },
    {
      "epoch": 8.197599261311172,
      "grad_norm": 0.8786792159080505,
      "learning_rate": 1.3320000000000001e-05,
      "loss": 0.1963,
      "step": 4443
    },
    {
      "epoch": 8.199445983379501,
      "grad_norm": 0.8025330901145935,
      "learning_rate": 1.3323000000000001e-05,
      "loss": 0.1538,
      "step": 4444
    },
    {
      "epoch": 8.20129270544783,
      "grad_norm": 1.081365704536438,
      "learning_rate": 1.3325999999999999e-05,
      "loss": 0.1918,
      "step": 4445
    },
    {
      "epoch": 8.20313942751616,
      "grad_norm": 0.6874704360961914,
      "learning_rate": 1.3328999999999999e-05,
      "loss": 0.0882,
      "step": 4446
    },
    {
      "epoch": 8.204986149584487,
      "grad_norm": 0.6587651968002319,
      "learning_rate": 1.3332e-05,
      "loss": 0.1124,
      "step": 4447
    },
    {
      "epoch": 8.206832871652816,
      "grad_norm": 0.6518075466156006,
      "learning_rate": 1.3335e-05,
      "loss": 0.1039,
      "step": 4448
    },
    {
      "epoch": 8.208679593721145,
      "grad_norm": 0.8533167243003845,
      "learning_rate": 1.3338e-05,
      "loss": 0.1805,
      "step": 4449
    },
    {
      "epoch": 8.210526315789474,
      "grad_norm": 0.5440764427185059,
      "learning_rate": 1.3341e-05,
      "loss": 0.0368,
      "step": 4450
    },
    {
      "epoch": 8.212373037857802,
      "grad_norm": 0.7944900393486023,
      "learning_rate": 1.3344e-05,
      "loss": 0.1022,
      "step": 4451
    },
    {
      "epoch": 8.21421975992613,
      "grad_norm": 0.55717933177948,
      "learning_rate": 1.3347000000000001e-05,
      "loss": 0.0471,
      "step": 4452
    },
    {
      "epoch": 8.21606648199446,
      "grad_norm": 0.6322590112686157,
      "learning_rate": 1.3350000000000001e-05,
      "loss": 0.0747,
      "step": 4453
    },
    {
      "epoch": 8.21791320406279,
      "grad_norm": 0.7431051135063171,
      "learning_rate": 1.3353e-05,
      "loss": 0.0504,
      "step": 4454
    },
    {
      "epoch": 8.219759926131117,
      "grad_norm": 0.4595365524291992,
      "learning_rate": 1.3356e-05,
      "loss": 0.0277,
      "step": 4455
    },
    {
      "epoch": 8.221606648199446,
      "grad_norm": 0.5381754040718079,
      "learning_rate": 1.3359e-05,
      "loss": 0.0631,
      "step": 4456
    },
    {
      "epoch": 8.223453370267775,
      "grad_norm": 0.5509288311004639,
      "learning_rate": 1.3362e-05,
      "loss": 0.0244,
      "step": 4457
    },
    {
      "epoch": 8.225300092336104,
      "grad_norm": 0.733174204826355,
      "learning_rate": 1.3365e-05,
      "loss": 0.0437,
      "step": 4458
    },
    {
      "epoch": 8.227146814404431,
      "grad_norm": 0.7149167656898499,
      "learning_rate": 1.3368e-05,
      "loss": 0.0451,
      "step": 4459
    },
    {
      "epoch": 8.22899353647276,
      "grad_norm": 0.6295753121376038,
      "learning_rate": 1.3371e-05,
      "loss": 0.0375,
      "step": 4460
    },
    {
      "epoch": 8.23084025854109,
      "grad_norm": 0.4743589460849762,
      "learning_rate": 1.3374e-05,
      "loss": 0.0344,
      "step": 4461
    },
    {
      "epoch": 8.232686980609419,
      "grad_norm": 0.7781702280044556,
      "learning_rate": 1.3377e-05,
      "loss": 0.0464,
      "step": 4462
    },
    {
      "epoch": 8.234533702677748,
      "grad_norm": 0.48906221985816956,
      "learning_rate": 1.338e-05,
      "loss": 0.0272,
      "step": 4463
    },
    {
      "epoch": 8.236380424746075,
      "grad_norm": 0.4372539818286896,
      "learning_rate": 1.3383e-05,
      "loss": 0.0324,
      "step": 4464
    },
    {
      "epoch": 8.238227146814404,
      "grad_norm": 0.47455981373786926,
      "learning_rate": 1.3386e-05,
      "loss": 0.0229,
      "step": 4465
    },
    {
      "epoch": 8.240073868882734,
      "grad_norm": 0.5755661725997925,
      "learning_rate": 1.3389e-05,
      "loss": 0.0272,
      "step": 4466
    },
    {
      "epoch": 8.241920590951063,
      "grad_norm": 0.5248891115188599,
      "learning_rate": 1.3392000000000002e-05,
      "loss": 0.0295,
      "step": 4467
    },
    {
      "epoch": 8.24376731301939,
      "grad_norm": 0.47399458289146423,
      "learning_rate": 1.3395000000000001e-05,
      "loss": 0.0272,
      "step": 4468
    },
    {
      "epoch": 8.24561403508772,
      "grad_norm": 0.6420528292655945,
      "learning_rate": 1.3398e-05,
      "loss": 0.0317,
      "step": 4469
    },
    {
      "epoch": 8.247460757156048,
      "grad_norm": 0.5973543524742126,
      "learning_rate": 1.3401e-05,
      "loss": 0.028,
      "step": 4470
    },
    {
      "epoch": 8.249307479224377,
      "grad_norm": 2.3570241928100586,
      "learning_rate": 1.3403999999999999e-05,
      "loss": 0.04,
      "step": 4471
    },
    {
      "epoch": 8.251154201292705,
      "grad_norm": 1.1083556413650513,
      "learning_rate": 1.3407e-05,
      "loss": 0.0303,
      "step": 4472
    },
    {
      "epoch": 8.253000923361034,
      "grad_norm": 1.0114208459854126,
      "learning_rate": 1.341e-05,
      "loss": 0.0331,
      "step": 4473
    },
    {
      "epoch": 8.254847645429363,
      "grad_norm": 0.3611949682235718,
      "learning_rate": 1.3413e-05,
      "loss": 0.0149,
      "step": 4474
    },
    {
      "epoch": 8.256694367497692,
      "grad_norm": 0.7732037901878357,
      "learning_rate": 1.3416e-05,
      "loss": 0.044,
      "step": 4475
    },
    {
      "epoch": 8.25854108956602,
      "grad_norm": 0.5984740257263184,
      "learning_rate": 1.3419e-05,
      "loss": 0.0373,
      "step": 4476
    },
    {
      "epoch": 8.260387811634349,
      "grad_norm": 0.6929849982261658,
      "learning_rate": 1.3422000000000001e-05,
      "loss": 0.0363,
      "step": 4477
    },
    {
      "epoch": 8.262234533702678,
      "grad_norm": 0.8012136816978455,
      "learning_rate": 1.3425000000000001e-05,
      "loss": 0.0339,
      "step": 4478
    },
    {
      "epoch": 8.264081255771007,
      "grad_norm": 1.2871159315109253,
      "learning_rate": 1.3428000000000001e-05,
      "loss": 0.0359,
      "step": 4479
    },
    {
      "epoch": 8.265927977839334,
      "grad_norm": 0.7725412845611572,
      "learning_rate": 1.3431e-05,
      "loss": 0.0367,
      "step": 4480
    },
    {
      "epoch": 8.267774699907664,
      "grad_norm": 0.8588116765022278,
      "learning_rate": 1.3433999999999999e-05,
      "loss": 0.0357,
      "step": 4481
    },
    {
      "epoch": 8.269621421975993,
      "grad_norm": 0.3712838888168335,
      "learning_rate": 1.3437e-05,
      "loss": 0.0265,
      "step": 4482
    },
    {
      "epoch": 8.271468144044322,
      "grad_norm": 0.7679174542427063,
      "learning_rate": 1.344e-05,
      "loss": 0.0434,
      "step": 4483
    },
    {
      "epoch": 8.27331486611265,
      "grad_norm": 0.5691342949867249,
      "learning_rate": 1.3443e-05,
      "loss": 0.0313,
      "step": 4484
    },
    {
      "epoch": 8.275161588180978,
      "grad_norm": 0.9302529692649841,
      "learning_rate": 1.3446e-05,
      "loss": 0.0326,
      "step": 4485
    },
    {
      "epoch": 8.277008310249307,
      "grad_norm": 0.6226106286048889,
      "learning_rate": 1.3449e-05,
      "loss": 0.0753,
      "step": 4486
    },
    {
      "epoch": 8.278855032317637,
      "grad_norm": 1.1870638132095337,
      "learning_rate": 1.3452000000000001e-05,
      "loss": 0.3242,
      "step": 4487
    },
    {
      "epoch": 8.280701754385966,
      "grad_norm": 1.0860295295715332,
      "learning_rate": 1.3455e-05,
      "loss": 0.3255,
      "step": 4488
    },
    {
      "epoch": 8.282548476454293,
      "grad_norm": 1.1221368312835693,
      "learning_rate": 1.3458e-05,
      "loss": 0.2718,
      "step": 4489
    },
    {
      "epoch": 8.284395198522622,
      "grad_norm": 0.7170649170875549,
      "learning_rate": 1.3461e-05,
      "loss": 0.1995,
      "step": 4490
    },
    {
      "epoch": 8.286241920590951,
      "grad_norm": 0.837647020816803,
      "learning_rate": 1.3464e-05,
      "loss": 0.2252,
      "step": 4491
    },
    {
      "epoch": 8.28808864265928,
      "grad_norm": 0.7650190591812134,
      "learning_rate": 1.3467000000000002e-05,
      "loss": 0.1505,
      "step": 4492
    },
    {
      "epoch": 8.289935364727608,
      "grad_norm": 1.1400846242904663,
      "learning_rate": 1.3470000000000001e-05,
      "loss": 0.1871,
      "step": 4493
    },
    {
      "epoch": 8.291782086795937,
      "grad_norm": 0.7779901623725891,
      "learning_rate": 1.3473e-05,
      "loss": 0.1292,
      "step": 4494
    },
    {
      "epoch": 8.293628808864266,
      "grad_norm": 0.8995106220245361,
      "learning_rate": 1.3476e-05,
      "loss": 0.1429,
      "step": 4495
    },
    {
      "epoch": 8.295475530932595,
      "grad_norm": 1.1236376762390137,
      "learning_rate": 1.3479e-05,
      "loss": 0.092,
      "step": 4496
    },
    {
      "epoch": 8.297322253000923,
      "grad_norm": 1.010858178138733,
      "learning_rate": 1.3482e-05,
      "loss": 0.1395,
      "step": 4497
    },
    {
      "epoch": 8.299168975069252,
      "grad_norm": 0.7352631092071533,
      "learning_rate": 1.3485e-05,
      "loss": 0.0927,
      "step": 4498
    },
    {
      "epoch": 8.301015697137581,
      "grad_norm": 0.6601068377494812,
      "learning_rate": 1.3488e-05,
      "loss": 0.0438,
      "step": 4499
    },
    {
      "epoch": 8.30286241920591,
      "grad_norm": 0.5581169724464417,
      "learning_rate": 1.3491e-05,
      "loss": 0.0904,
      "step": 4500
    },
    {
      "epoch": 8.304709141274238,
      "grad_norm": 0.9757201671600342,
      "learning_rate": 1.3494e-05,
      "loss": 0.0447,
      "step": 4501
    },
    {
      "epoch": 8.306555863342567,
      "grad_norm": 0.38367536664009094,
      "learning_rate": 1.3497000000000001e-05,
      "loss": 0.0359,
      "step": 4502
    },
    {
      "epoch": 8.308402585410896,
      "grad_norm": 0.3730805516242981,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.0398,
      "step": 4503
    },
    {
      "epoch": 8.310249307479225,
      "grad_norm": 0.6186391115188599,
      "learning_rate": 1.3503000000000001e-05,
      "loss": 0.0255,
      "step": 4504
    },
    {
      "epoch": 8.312096029547552,
      "grad_norm": 0.5357188582420349,
      "learning_rate": 1.3506e-05,
      "loss": 0.0438,
      "step": 4505
    },
    {
      "epoch": 8.313942751615881,
      "grad_norm": 0.9559025764465332,
      "learning_rate": 1.3508999999999999e-05,
      "loss": 0.0319,
      "step": 4506
    },
    {
      "epoch": 8.31578947368421,
      "grad_norm": 0.9470544457435608,
      "learning_rate": 1.3512e-05,
      "loss": 0.087,
      "step": 4507
    },
    {
      "epoch": 8.31763619575254,
      "grad_norm": 0.44523105025291443,
      "learning_rate": 1.3515e-05,
      "loss": 0.031,
      "step": 4508
    },
    {
      "epoch": 8.319482917820867,
      "grad_norm": 0.49122002720832825,
      "learning_rate": 1.3518e-05,
      "loss": 0.0325,
      "step": 4509
    },
    {
      "epoch": 8.321329639889196,
      "grad_norm": 0.5576708316802979,
      "learning_rate": 1.3521e-05,
      "loss": 0.0338,
      "step": 4510
    },
    {
      "epoch": 8.323176361957525,
      "grad_norm": 0.8101223707199097,
      "learning_rate": 1.3524e-05,
      "loss": 0.0413,
      "step": 4511
    },
    {
      "epoch": 8.325023084025855,
      "grad_norm": 0.5844168066978455,
      "learning_rate": 1.3527000000000001e-05,
      "loss": 0.0344,
      "step": 4512
    },
    {
      "epoch": 8.326869806094184,
      "grad_norm": 0.4756903350353241,
      "learning_rate": 1.3530000000000001e-05,
      "loss": 0.0314,
      "step": 4513
    },
    {
      "epoch": 8.328716528162511,
      "grad_norm": 0.48679304122924805,
      "learning_rate": 1.3533e-05,
      "loss": 0.0369,
      "step": 4514
    },
    {
      "epoch": 8.33056325023084,
      "grad_norm": 1.2075324058532715,
      "learning_rate": 1.3536e-05,
      "loss": 0.0439,
      "step": 4515
    },
    {
      "epoch": 8.33240997229917,
      "grad_norm": 1.6481680870056152,
      "learning_rate": 1.3539e-05,
      "loss": 0.0377,
      "step": 4516
    },
    {
      "epoch": 8.334256694367498,
      "grad_norm": 0.4778039753437042,
      "learning_rate": 1.3542000000000002e-05,
      "loss": 0.0359,
      "step": 4517
    },
    {
      "epoch": 8.336103416435826,
      "grad_norm": 0.3544483482837677,
      "learning_rate": 1.3545e-05,
      "loss": 0.0168,
      "step": 4518
    },
    {
      "epoch": 8.337950138504155,
      "grad_norm": 0.9810410141944885,
      "learning_rate": 1.3548e-05,
      "loss": 0.0516,
      "step": 4519
    },
    {
      "epoch": 8.339796860572484,
      "grad_norm": 0.5142959356307983,
      "learning_rate": 1.3551e-05,
      "loss": 0.0326,
      "step": 4520
    },
    {
      "epoch": 8.341643582640813,
      "grad_norm": 0.7074347734451294,
      "learning_rate": 1.3554e-05,
      "loss": 0.0432,
      "step": 4521
    },
    {
      "epoch": 8.34349030470914,
      "grad_norm": 0.4334162473678589,
      "learning_rate": 1.3557e-05,
      "loss": 0.0162,
      "step": 4522
    },
    {
      "epoch": 8.34533702677747,
      "grad_norm": 0.6149067878723145,
      "learning_rate": 1.356e-05,
      "loss": 0.0588,
      "step": 4523
    },
    {
      "epoch": 8.347183748845799,
      "grad_norm": 0.4578056037425995,
      "learning_rate": 1.3563e-05,
      "loss": 0.0198,
      "step": 4524
    },
    {
      "epoch": 8.349030470914128,
      "grad_norm": 0.6646698117256165,
      "learning_rate": 1.3566e-05,
      "loss": 0.035,
      "step": 4525
    },
    {
      "epoch": 8.350877192982455,
      "grad_norm": 0.9544262886047363,
      "learning_rate": 1.3569e-05,
      "loss": 0.0309,
      "step": 4526
    },
    {
      "epoch": 8.352723915050785,
      "grad_norm": 0.9538107514381409,
      "learning_rate": 1.3572000000000002e-05,
      "loss": 0.0265,
      "step": 4527
    },
    {
      "epoch": 8.354570637119114,
      "grad_norm": 0.7917845845222473,
      "learning_rate": 1.3575000000000001e-05,
      "loss": 0.0433,
      "step": 4528
    },
    {
      "epoch": 8.356417359187443,
      "grad_norm": 0.6048826575279236,
      "learning_rate": 1.3578000000000001e-05,
      "loss": 0.0415,
      "step": 4529
    },
    {
      "epoch": 8.35826408125577,
      "grad_norm": 0.813008725643158,
      "learning_rate": 1.3581000000000001e-05,
      "loss": 0.0437,
      "step": 4530
    },
    {
      "epoch": 8.3601108033241,
      "grad_norm": 0.7485079169273376,
      "learning_rate": 1.3583999999999999e-05,
      "loss": 0.0411,
      "step": 4531
    },
    {
      "epoch": 8.361957525392429,
      "grad_norm": 0.6766875982284546,
      "learning_rate": 1.3587e-05,
      "loss": 0.0285,
      "step": 4532
    },
    {
      "epoch": 8.363804247460758,
      "grad_norm": 0.8954643607139587,
      "learning_rate": 1.359e-05,
      "loss": 0.0436,
      "step": 4533
    },
    {
      "epoch": 8.365650969529085,
      "grad_norm": 0.9389119148254395,
      "learning_rate": 1.3593e-05,
      "loss": 0.033,
      "step": 4534
    },
    {
      "epoch": 8.367497691597414,
      "grad_norm": 1.0599180459976196,
      "learning_rate": 1.3596e-05,
      "loss": 0.0418,
      "step": 4535
    },
    {
      "epoch": 8.369344413665743,
      "grad_norm": 1.466378092765808,
      "learning_rate": 1.3599e-05,
      "loss": 0.0693,
      "step": 4536
    },
    {
      "epoch": 8.371191135734072,
      "grad_norm": 1.0018821954727173,
      "learning_rate": 1.3602000000000001e-05,
      "loss": 0.2913,
      "step": 4537
    },
    {
      "epoch": 8.373037857802402,
      "grad_norm": 1.0712130069732666,
      "learning_rate": 1.3605000000000001e-05,
      "loss": 0.275,
      "step": 4538
    },
    {
      "epoch": 8.374884579870729,
      "grad_norm": 0.7662477493286133,
      "learning_rate": 1.3608e-05,
      "loss": 0.266,
      "step": 4539
    },
    {
      "epoch": 8.376731301939058,
      "grad_norm": 0.9335657358169556,
      "learning_rate": 1.3611e-05,
      "loss": 0.232,
      "step": 4540
    },
    {
      "epoch": 8.378578024007387,
      "grad_norm": 0.8165726661682129,
      "learning_rate": 1.3614e-05,
      "loss": 0.186,
      "step": 4541
    },
    {
      "epoch": 8.380424746075716,
      "grad_norm": 0.6568682193756104,
      "learning_rate": 1.3617000000000002e-05,
      "loss": 0.1721,
      "step": 4542
    },
    {
      "epoch": 8.382271468144044,
      "grad_norm": 0.779513955116272,
      "learning_rate": 1.362e-05,
      "loss": 0.1871,
      "step": 4543
    },
    {
      "epoch": 8.384118190212373,
      "grad_norm": 0.8209591507911682,
      "learning_rate": 1.3623e-05,
      "loss": 0.1078,
      "step": 4544
    },
    {
      "epoch": 8.385964912280702,
      "grad_norm": 0.8797643184661865,
      "learning_rate": 1.3626e-05,
      "loss": 0.1393,
      "step": 4545
    },
    {
      "epoch": 8.387811634349031,
      "grad_norm": 0.8661841154098511,
      "learning_rate": 1.3629e-05,
      "loss": 0.1451,
      "step": 4546
    },
    {
      "epoch": 8.389658356417359,
      "grad_norm": 0.5946820974349976,
      "learning_rate": 1.3632000000000001e-05,
      "loss": 0.081,
      "step": 4547
    },
    {
      "epoch": 8.391505078485688,
      "grad_norm": 1.008434534072876,
      "learning_rate": 1.3635e-05,
      "loss": 0.1038,
      "step": 4548
    },
    {
      "epoch": 8.393351800554017,
      "grad_norm": 0.6491675972938538,
      "learning_rate": 1.3638e-05,
      "loss": 0.075,
      "step": 4549
    },
    {
      "epoch": 8.395198522622346,
      "grad_norm": 0.585092306137085,
      "learning_rate": 1.3641e-05,
      "loss": 0.0561,
      "step": 4550
    },
    {
      "epoch": 8.397045244690673,
      "grad_norm": 0.5089346170425415,
      "learning_rate": 1.3644e-05,
      "loss": 0.0644,
      "step": 4551
    },
    {
      "epoch": 8.398891966759003,
      "grad_norm": 0.616301953792572,
      "learning_rate": 1.3647000000000002e-05,
      "loss": 0.0416,
      "step": 4552
    },
    {
      "epoch": 8.400738688827332,
      "grad_norm": 0.7324120998382568,
      "learning_rate": 1.3650000000000001e-05,
      "loss": 0.0344,
      "step": 4553
    },
    {
      "epoch": 8.40258541089566,
      "grad_norm": 0.462205171585083,
      "learning_rate": 1.3653000000000001e-05,
      "loss": 0.0295,
      "step": 4554
    },
    {
      "epoch": 8.404432132963988,
      "grad_norm": 1.396769404411316,
      "learning_rate": 1.3656e-05,
      "loss": 0.0589,
      "step": 4555
    },
    {
      "epoch": 8.406278855032317,
      "grad_norm": 0.5556825399398804,
      "learning_rate": 1.3659e-05,
      "loss": 0.0327,
      "step": 4556
    },
    {
      "epoch": 8.408125577100646,
      "grad_norm": 0.4635709226131439,
      "learning_rate": 1.3662e-05,
      "loss": 0.0248,
      "step": 4557
    },
    {
      "epoch": 8.409972299168976,
      "grad_norm": 0.39975813031196594,
      "learning_rate": 1.3665e-05,
      "loss": 0.0459,
      "step": 4558
    },
    {
      "epoch": 8.411819021237303,
      "grad_norm": 0.8045360445976257,
      "learning_rate": 1.3668e-05,
      "loss": 0.0371,
      "step": 4559
    },
    {
      "epoch": 8.413665743305632,
      "grad_norm": 0.6244034767150879,
      "learning_rate": 1.3671e-05,
      "loss": 0.0435,
      "step": 4560
    },
    {
      "epoch": 8.415512465373961,
      "grad_norm": 0.7523409724235535,
      "learning_rate": 1.3674e-05,
      "loss": 0.0568,
      "step": 4561
    },
    {
      "epoch": 8.41735918744229,
      "grad_norm": 1.0529162883758545,
      "learning_rate": 1.3677000000000001e-05,
      "loss": 0.0473,
      "step": 4562
    },
    {
      "epoch": 8.41920590951062,
      "grad_norm": 0.44346120953559875,
      "learning_rate": 1.3680000000000001e-05,
      "loss": 0.0243,
      "step": 4563
    },
    {
      "epoch": 8.421052631578947,
      "grad_norm": 0.27785372734069824,
      "learning_rate": 1.3683000000000001e-05,
      "loss": 0.0116,
      "step": 4564
    },
    {
      "epoch": 8.422899353647276,
      "grad_norm": 0.48640650510787964,
      "learning_rate": 1.3686e-05,
      "loss": 0.0312,
      "step": 4565
    },
    {
      "epoch": 8.424746075715605,
      "grad_norm": 0.8661382794380188,
      "learning_rate": 1.3689e-05,
      "loss": 0.096,
      "step": 4566
    },
    {
      "epoch": 8.426592797783934,
      "grad_norm": 0.6696983575820923,
      "learning_rate": 1.3691999999999999e-05,
      "loss": 0.0402,
      "step": 4567
    },
    {
      "epoch": 8.428439519852262,
      "grad_norm": 0.47121769189834595,
      "learning_rate": 1.3695e-05,
      "loss": 0.0309,
      "step": 4568
    },
    {
      "epoch": 8.43028624192059,
      "grad_norm": 0.5439361929893494,
      "learning_rate": 1.3698e-05,
      "loss": 0.0376,
      "step": 4569
    },
    {
      "epoch": 8.43213296398892,
      "grad_norm": 1.22900390625,
      "learning_rate": 1.3701e-05,
      "loss": 0.0461,
      "step": 4570
    },
    {
      "epoch": 8.43397968605725,
      "grad_norm": 0.6019194722175598,
      "learning_rate": 1.3704e-05,
      "loss": 0.0366,
      "step": 4571
    },
    {
      "epoch": 8.435826408125576,
      "grad_norm": 0.981429398059845,
      "learning_rate": 1.3707e-05,
      "loss": 0.0447,
      "step": 4572
    },
    {
      "epoch": 8.437673130193906,
      "grad_norm": 0.621160089969635,
      "learning_rate": 1.3710000000000001e-05,
      "loss": 0.0466,
      "step": 4573
    },
    {
      "epoch": 8.439519852262235,
      "grad_norm": 0.46072280406951904,
      "learning_rate": 1.3713e-05,
      "loss": 0.0297,
      "step": 4574
    },
    {
      "epoch": 8.441366574330564,
      "grad_norm": 0.8340686559677124,
      "learning_rate": 1.3716e-05,
      "loss": 0.0386,
      "step": 4575
    },
    {
      "epoch": 8.443213296398891,
      "grad_norm": 0.5266955494880676,
      "learning_rate": 1.3719e-05,
      "loss": 0.0314,
      "step": 4576
    },
    {
      "epoch": 8.44506001846722,
      "grad_norm": 0.485097199678421,
      "learning_rate": 1.3722e-05,
      "loss": 0.0282,
      "step": 4577
    },
    {
      "epoch": 8.44690674053555,
      "grad_norm": 0.617648184299469,
      "learning_rate": 1.3725000000000002e-05,
      "loss": 0.0341,
      "step": 4578
    },
    {
      "epoch": 8.448753462603879,
      "grad_norm": 0.7114557027816772,
      "learning_rate": 1.3728000000000001e-05,
      "loss": 0.037,
      "step": 4579
    },
    {
      "epoch": 8.450600184672206,
      "grad_norm": 0.8123380541801453,
      "learning_rate": 1.3731e-05,
      "loss": 0.0243,
      "step": 4580
    },
    {
      "epoch": 8.452446906740535,
      "grad_norm": 0.40049365162849426,
      "learning_rate": 1.3734e-05,
      "loss": 0.0187,
      "step": 4581
    },
    {
      "epoch": 8.454293628808864,
      "grad_norm": 0.8814468383789062,
      "learning_rate": 1.3736999999999999e-05,
      "loss": 0.0387,
      "step": 4582
    },
    {
      "epoch": 8.456140350877194,
      "grad_norm": 0.885870099067688,
      "learning_rate": 1.374e-05,
      "loss": 0.042,
      "step": 4583
    },
    {
      "epoch": 8.45798707294552,
      "grad_norm": 0.9987731575965881,
      "learning_rate": 1.3743e-05,
      "loss": 0.0392,
      "step": 4584
    },
    {
      "epoch": 8.45983379501385,
      "grad_norm": 0.6637702584266663,
      "learning_rate": 1.3746e-05,
      "loss": 0.037,
      "step": 4585
    },
    {
      "epoch": 8.46168051708218,
      "grad_norm": 0.9246504306793213,
      "learning_rate": 1.3749e-05,
      "loss": 0.0537,
      "step": 4586
    },
    {
      "epoch": 8.463527239150508,
      "grad_norm": 1.5685744285583496,
      "learning_rate": 1.3752e-05,
      "loss": 0.274,
      "step": 4587
    },
    {
      "epoch": 8.465373961218837,
      "grad_norm": 0.9505859017372131,
      "learning_rate": 1.3755000000000001e-05,
      "loss": 0.2991,
      "step": 4588
    },
    {
      "epoch": 8.467220683287165,
      "grad_norm": 0.8866581320762634,
      "learning_rate": 1.3758000000000001e-05,
      "loss": 0.2452,
      "step": 4589
    },
    {
      "epoch": 8.469067405355494,
      "grad_norm": 0.9090424180030823,
      "learning_rate": 1.3761000000000001e-05,
      "loss": 0.2359,
      "step": 4590
    },
    {
      "epoch": 8.470914127423823,
      "grad_norm": 0.9789994955062866,
      "learning_rate": 1.3764e-05,
      "loss": 0.2069,
      "step": 4591
    },
    {
      "epoch": 8.472760849492152,
      "grad_norm": 0.6089072227478027,
      "learning_rate": 1.3766999999999999e-05,
      "loss": 0.1521,
      "step": 4592
    },
    {
      "epoch": 8.47460757156048,
      "grad_norm": 0.5650318264961243,
      "learning_rate": 1.377e-05,
      "loss": 0.1246,
      "step": 4593
    },
    {
      "epoch": 8.476454293628809,
      "grad_norm": 0.7467601895332336,
      "learning_rate": 1.3773e-05,
      "loss": 0.157,
      "step": 4594
    },
    {
      "epoch": 8.478301015697138,
      "grad_norm": 1.8868144750595093,
      "learning_rate": 1.3776e-05,
      "loss": 0.1856,
      "step": 4595
    },
    {
      "epoch": 8.480147737765467,
      "grad_norm": 1.5848222970962524,
      "learning_rate": 1.3779e-05,
      "loss": 0.276,
      "step": 4596
    },
    {
      "epoch": 8.481994459833794,
      "grad_norm": 1.11818528175354,
      "learning_rate": 1.3782e-05,
      "loss": 0.0852,
      "step": 4597
    },
    {
      "epoch": 8.483841181902124,
      "grad_norm": 0.7371710538864136,
      "learning_rate": 1.3785000000000001e-05,
      "loss": 0.1073,
      "step": 4598
    },
    {
      "epoch": 8.485687903970453,
      "grad_norm": 0.9325425624847412,
      "learning_rate": 1.3788e-05,
      "loss": 0.1418,
      "step": 4599
    },
    {
      "epoch": 8.487534626038782,
      "grad_norm": 0.6657596826553345,
      "learning_rate": 1.3791e-05,
      "loss": 0.0736,
      "step": 4600
    },
    {
      "epoch": 8.48938134810711,
      "grad_norm": 0.5805692672729492,
      "learning_rate": 1.3794e-05,
      "loss": 0.0964,
      "step": 4601
    },
    {
      "epoch": 8.491228070175438,
      "grad_norm": 0.7410169243812561,
      "learning_rate": 1.3797e-05,
      "loss": 0.0827,
      "step": 4602
    },
    {
      "epoch": 8.493074792243767,
      "grad_norm": 0.5074139833450317,
      "learning_rate": 1.3800000000000002e-05,
      "loss": 0.042,
      "step": 4603
    },
    {
      "epoch": 8.494921514312097,
      "grad_norm": 0.4896104633808136,
      "learning_rate": 1.3803e-05,
      "loss": 0.0373,
      "step": 4604
    },
    {
      "epoch": 8.496768236380424,
      "grad_norm": 0.4625118374824524,
      "learning_rate": 1.3806e-05,
      "loss": 0.0454,
      "step": 4605
    },
    {
      "epoch": 8.498614958448753,
      "grad_norm": 0.5676958560943604,
      "learning_rate": 1.3809e-05,
      "loss": 0.0379,
      "step": 4606
    },
    {
      "epoch": 8.500461680517082,
      "grad_norm": 0.3820827901363373,
      "learning_rate": 1.3812e-05,
      "loss": 0.0202,
      "step": 4607
    },
    {
      "epoch": 8.502308402585411,
      "grad_norm": 1.2073335647583008,
      "learning_rate": 1.3815e-05,
      "loss": 0.0194,
      "step": 4608
    },
    {
      "epoch": 8.504155124653739,
      "grad_norm": 0.7418472170829773,
      "learning_rate": 1.3818e-05,
      "loss": 0.0414,
      "step": 4609
    },
    {
      "epoch": 8.506001846722068,
      "grad_norm": 0.5792127251625061,
      "learning_rate": 1.3821e-05,
      "loss": 0.0418,
      "step": 4610
    },
    {
      "epoch": 8.507848568790397,
      "grad_norm": 0.5523576736450195,
      "learning_rate": 1.3824e-05,
      "loss": 0.0542,
      "step": 4611
    },
    {
      "epoch": 8.509695290858726,
      "grad_norm": 0.4889757037162781,
      "learning_rate": 1.3827e-05,
      "loss": 0.0219,
      "step": 4612
    },
    {
      "epoch": 8.511542012927055,
      "grad_norm": 1.076654314994812,
      "learning_rate": 1.3830000000000001e-05,
      "loss": 0.0342,
      "step": 4613
    },
    {
      "epoch": 8.513388734995383,
      "grad_norm": 0.5193042755126953,
      "learning_rate": 1.3833000000000001e-05,
      "loss": 0.0266,
      "step": 4614
    },
    {
      "epoch": 8.515235457063712,
      "grad_norm": 0.437115341424942,
      "learning_rate": 1.3836000000000001e-05,
      "loss": 0.0335,
      "step": 4615
    },
    {
      "epoch": 8.517082179132041,
      "grad_norm": 0.41590920090675354,
      "learning_rate": 1.3839e-05,
      "loss": 0.0266,
      "step": 4616
    },
    {
      "epoch": 8.51892890120037,
      "grad_norm": 0.4554344117641449,
      "learning_rate": 1.3841999999999999e-05,
      "loss": 0.0204,
      "step": 4617
    },
    {
      "epoch": 8.520775623268698,
      "grad_norm": 0.5499839782714844,
      "learning_rate": 1.3845e-05,
      "loss": 0.0224,
      "step": 4618
    },
    {
      "epoch": 8.522622345337027,
      "grad_norm": 0.9251109957695007,
      "learning_rate": 1.3848e-05,
      "loss": 0.0334,
      "step": 4619
    },
    {
      "epoch": 8.524469067405356,
      "grad_norm": 0.5408797264099121,
      "learning_rate": 1.3851e-05,
      "loss": 0.0212,
      "step": 4620
    },
    {
      "epoch": 8.526315789473685,
      "grad_norm": 1.166687250137329,
      "learning_rate": 1.3854e-05,
      "loss": 0.0365,
      "step": 4621
    },
    {
      "epoch": 8.528162511542012,
      "grad_norm": 0.9165891408920288,
      "learning_rate": 1.3857e-05,
      "loss": 0.0382,
      "step": 4622
    },
    {
      "epoch": 8.530009233610341,
      "grad_norm": 1.4344496726989746,
      "learning_rate": 1.3860000000000001e-05,
      "loss": 0.0513,
      "step": 4623
    },
    {
      "epoch": 8.53185595567867,
      "grad_norm": 0.6101006865501404,
      "learning_rate": 1.3863000000000001e-05,
      "loss": 0.0363,
      "step": 4624
    },
    {
      "epoch": 8.533702677747,
      "grad_norm": 1.1394438743591309,
      "learning_rate": 1.3866e-05,
      "loss": 0.0316,
      "step": 4625
    },
    {
      "epoch": 8.535549399815327,
      "grad_norm": 0.918218195438385,
      "learning_rate": 1.3869e-05,
      "loss": 0.0425,
      "step": 4626
    },
    {
      "epoch": 8.537396121883656,
      "grad_norm": 0.9316399693489075,
      "learning_rate": 1.3872e-05,
      "loss": 0.0463,
      "step": 4627
    },
    {
      "epoch": 8.539242843951985,
      "grad_norm": 0.7699198722839355,
      "learning_rate": 1.3875000000000002e-05,
      "loss": 0.0452,
      "step": 4628
    },
    {
      "epoch": 8.541089566020315,
      "grad_norm": 0.7271835803985596,
      "learning_rate": 1.3878e-05,
      "loss": 0.0234,
      "step": 4629
    },
    {
      "epoch": 8.542936288088642,
      "grad_norm": 0.638312041759491,
      "learning_rate": 1.3881e-05,
      "loss": 0.0331,
      "step": 4630
    },
    {
      "epoch": 8.544783010156971,
      "grad_norm": 0.8469995856285095,
      "learning_rate": 1.3884e-05,
      "loss": 0.0491,
      "step": 4631
    },
    {
      "epoch": 8.5466297322253,
      "grad_norm": 0.7678780555725098,
      "learning_rate": 1.3887e-05,
      "loss": 0.0373,
      "step": 4632
    },
    {
      "epoch": 8.54847645429363,
      "grad_norm": 0.8009999990463257,
      "learning_rate": 1.389e-05,
      "loss": 0.0336,
      "step": 4633
    },
    {
      "epoch": 8.550323176361957,
      "grad_norm": 0.6101869940757751,
      "learning_rate": 1.3893e-05,
      "loss": 0.0557,
      "step": 4634
    },
    {
      "epoch": 8.552169898430286,
      "grad_norm": 0.7427794933319092,
      "learning_rate": 1.3896e-05,
      "loss": 0.0477,
      "step": 4635
    },
    {
      "epoch": 8.554016620498615,
      "grad_norm": 0.9742395281791687,
      "learning_rate": 1.3899e-05,
      "loss": 0.0373,
      "step": 4636
    },
    {
      "epoch": 8.555863342566944,
      "grad_norm": 1.3674216270446777,
      "learning_rate": 1.3902e-05,
      "loss": 0.2994,
      "step": 4637
    },
    {
      "epoch": 8.557710064635273,
      "grad_norm": 1.1344772577285767,
      "learning_rate": 1.3905000000000002e-05,
      "loss": 0.2497,
      "step": 4638
    },
    {
      "epoch": 8.5595567867036,
      "grad_norm": 1.1575214862823486,
      "learning_rate": 1.3908000000000001e-05,
      "loss": 0.2529,
      "step": 4639
    },
    {
      "epoch": 8.56140350877193,
      "grad_norm": 0.8929944038391113,
      "learning_rate": 1.3911000000000001e-05,
      "loss": 0.2333,
      "step": 4640
    },
    {
      "epoch": 8.563250230840259,
      "grad_norm": 0.8854674100875854,
      "learning_rate": 1.3914e-05,
      "loss": 0.1999,
      "step": 4641
    },
    {
      "epoch": 8.565096952908588,
      "grad_norm": 0.9652591347694397,
      "learning_rate": 1.3916999999999999e-05,
      "loss": 0.193,
      "step": 4642
    },
    {
      "epoch": 8.566943674976915,
      "grad_norm": 0.685058057308197,
      "learning_rate": 1.392e-05,
      "loss": 0.1591,
      "step": 4643
    },
    {
      "epoch": 8.568790397045245,
      "grad_norm": 0.6758031845092773,
      "learning_rate": 1.3923e-05,
      "loss": 0.1499,
      "step": 4644
    },
    {
      "epoch": 8.570637119113574,
      "grad_norm": 0.7304430603981018,
      "learning_rate": 1.3926e-05,
      "loss": 0.1686,
      "step": 4645
    },
    {
      "epoch": 8.572483841181903,
      "grad_norm": 0.6966080069541931,
      "learning_rate": 1.3929e-05,
      "loss": 0.1095,
      "step": 4646
    },
    {
      "epoch": 8.57433056325023,
      "grad_norm": 0.9024938344955444,
      "learning_rate": 1.3932e-05,
      "loss": 0.1383,
      "step": 4647
    },
    {
      "epoch": 8.57617728531856,
      "grad_norm": 0.5872875452041626,
      "learning_rate": 1.3935000000000001e-05,
      "loss": 0.1028,
      "step": 4648
    },
    {
      "epoch": 8.578024007386889,
      "grad_norm": 0.976383626461029,
      "learning_rate": 1.3938000000000001e-05,
      "loss": 0.075,
      "step": 4649
    },
    {
      "epoch": 8.579870729455218,
      "grad_norm": 0.7842995524406433,
      "learning_rate": 1.3941000000000001e-05,
      "loss": 0.1423,
      "step": 4650
    },
    {
      "epoch": 8.581717451523545,
      "grad_norm": 0.5742107629776001,
      "learning_rate": 1.3944e-05,
      "loss": 0.0525,
      "step": 4651
    },
    {
      "epoch": 8.583564173591874,
      "grad_norm": 0.430160790681839,
      "learning_rate": 1.3947e-05,
      "loss": 0.038,
      "step": 4652
    },
    {
      "epoch": 8.585410895660203,
      "grad_norm": 0.6016567349433899,
      "learning_rate": 1.395e-05,
      "loss": 0.0432,
      "step": 4653
    },
    {
      "epoch": 8.587257617728532,
      "grad_norm": 0.45133915543556213,
      "learning_rate": 1.3953e-05,
      "loss": 0.053,
      "step": 4654
    },
    {
      "epoch": 8.58910433979686,
      "grad_norm": 0.6579575538635254,
      "learning_rate": 1.3956e-05,
      "loss": 0.0683,
      "step": 4655
    },
    {
      "epoch": 8.590951061865189,
      "grad_norm": 0.9302634000778198,
      "learning_rate": 1.3959e-05,
      "loss": 0.0924,
      "step": 4656
    },
    {
      "epoch": 8.592797783933518,
      "grad_norm": 0.6408567428588867,
      "learning_rate": 1.3962e-05,
      "loss": 0.0325,
      "step": 4657
    },
    {
      "epoch": 8.594644506001847,
      "grad_norm": 0.5441372394561768,
      "learning_rate": 1.3965000000000001e-05,
      "loss": 0.0242,
      "step": 4658
    },
    {
      "epoch": 8.596491228070175,
      "grad_norm": 0.7796469926834106,
      "learning_rate": 1.3968e-05,
      "loss": 0.055,
      "step": 4659
    },
    {
      "epoch": 8.598337950138504,
      "grad_norm": 0.46276140213012695,
      "learning_rate": 1.3971e-05,
      "loss": 0.0349,
      "step": 4660
    },
    {
      "epoch": 8.600184672206833,
      "grad_norm": 0.4005415439605713,
      "learning_rate": 1.3974e-05,
      "loss": 0.0215,
      "step": 4661
    },
    {
      "epoch": 8.602031394275162,
      "grad_norm": 0.478162944316864,
      "learning_rate": 1.3977e-05,
      "loss": 0.0313,
      "step": 4662
    },
    {
      "epoch": 8.603878116343491,
      "grad_norm": 0.6175740361213684,
      "learning_rate": 1.3980000000000002e-05,
      "loss": 0.046,
      "step": 4663
    },
    {
      "epoch": 8.605724838411819,
      "grad_norm": 0.43752771615982056,
      "learning_rate": 1.3983000000000001e-05,
      "loss": 0.0263,
      "step": 4664
    },
    {
      "epoch": 8.607571560480148,
      "grad_norm": 0.5159079432487488,
      "learning_rate": 1.3986000000000001e-05,
      "loss": 0.064,
      "step": 4665
    },
    {
      "epoch": 8.609418282548477,
      "grad_norm": 0.44390735030174255,
      "learning_rate": 1.3989e-05,
      "loss": 0.0253,
      "step": 4666
    },
    {
      "epoch": 8.611265004616806,
      "grad_norm": 0.9245308637619019,
      "learning_rate": 1.3992e-05,
      "loss": 0.0375,
      "step": 4667
    },
    {
      "epoch": 8.613111726685133,
      "grad_norm": 0.644258975982666,
      "learning_rate": 1.3995e-05,
      "loss": 0.0519,
      "step": 4668
    },
    {
      "epoch": 8.614958448753463,
      "grad_norm": 0.6799528002738953,
      "learning_rate": 1.3998e-05,
      "loss": 0.0317,
      "step": 4669
    },
    {
      "epoch": 8.616805170821792,
      "grad_norm": 0.6856921315193176,
      "learning_rate": 1.4001e-05,
      "loss": 0.0274,
      "step": 4670
    },
    {
      "epoch": 8.61865189289012,
      "grad_norm": 0.5836756825447083,
      "learning_rate": 1.4004e-05,
      "loss": 0.0345,
      "step": 4671
    },
    {
      "epoch": 8.620498614958448,
      "grad_norm": 0.94598388671875,
      "learning_rate": 1.4007e-05,
      "loss": 0.0319,
      "step": 4672
    },
    {
      "epoch": 8.622345337026777,
      "grad_norm": 0.6048069000244141,
      "learning_rate": 1.4010000000000001e-05,
      "loss": 0.0371,
      "step": 4673
    },
    {
      "epoch": 8.624192059095106,
      "grad_norm": 0.6870465278625488,
      "learning_rate": 1.4013000000000001e-05,
      "loss": 0.0268,
      "step": 4674
    },
    {
      "epoch": 8.626038781163436,
      "grad_norm": 0.5122883319854736,
      "learning_rate": 1.4016000000000001e-05,
      "loss": 0.032,
      "step": 4675
    },
    {
      "epoch": 8.627885503231763,
      "grad_norm": 0.653048038482666,
      "learning_rate": 1.4019e-05,
      "loss": 0.0393,
      "step": 4676
    },
    {
      "epoch": 8.629732225300092,
      "grad_norm": 0.5902331471443176,
      "learning_rate": 1.4022e-05,
      "loss": 0.0383,
      "step": 4677
    },
    {
      "epoch": 8.631578947368421,
      "grad_norm": 0.6037182807922363,
      "learning_rate": 1.4025e-05,
      "loss": 0.0329,
      "step": 4678
    },
    {
      "epoch": 8.63342566943675,
      "grad_norm": 0.9846788644790649,
      "learning_rate": 1.4028e-05,
      "loss": 0.0627,
      "step": 4679
    },
    {
      "epoch": 8.635272391505078,
      "grad_norm": 0.4753408432006836,
      "learning_rate": 1.4031e-05,
      "loss": 0.0253,
      "step": 4680
    },
    {
      "epoch": 8.637119113573407,
      "grad_norm": 0.5847453474998474,
      "learning_rate": 1.4034e-05,
      "loss": 0.0326,
      "step": 4681
    },
    {
      "epoch": 8.638965835641736,
      "grad_norm": 0.6286370158195496,
      "learning_rate": 1.4037e-05,
      "loss": 0.0384,
      "step": 4682
    },
    {
      "epoch": 8.640812557710065,
      "grad_norm": 0.6938286423683167,
      "learning_rate": 1.4040000000000001e-05,
      "loss": 0.042,
      "step": 4683
    },
    {
      "epoch": 8.642659279778393,
      "grad_norm": 0.8416529297828674,
      "learning_rate": 1.4043000000000001e-05,
      "loss": 0.0444,
      "step": 4684
    },
    {
      "epoch": 8.644506001846722,
      "grad_norm": 0.6412912607192993,
      "learning_rate": 1.4046e-05,
      "loss": 0.0261,
      "step": 4685
    },
    {
      "epoch": 8.64635272391505,
      "grad_norm": 0.8336824178695679,
      "learning_rate": 1.4049e-05,
      "loss": 0.046,
      "step": 4686
    },
    {
      "epoch": 8.64819944598338,
      "grad_norm": 1.7124323844909668,
      "learning_rate": 1.4052e-05,
      "loss": 0.3239,
      "step": 4687
    },
    {
      "epoch": 8.65004616805171,
      "grad_norm": 0.948986828327179,
      "learning_rate": 1.4055000000000002e-05,
      "loss": 0.2906,
      "step": 4688
    },
    {
      "epoch": 8.651892890120036,
      "grad_norm": 1.0201292037963867,
      "learning_rate": 1.4058000000000002e-05,
      "loss": 0.2812,
      "step": 4689
    },
    {
      "epoch": 8.653739612188366,
      "grad_norm": 0.8282414674758911,
      "learning_rate": 1.4061e-05,
      "loss": 0.2266,
      "step": 4690
    },
    {
      "epoch": 8.655586334256695,
      "grad_norm": 0.8216656446456909,
      "learning_rate": 1.4064e-05,
      "loss": 0.1907,
      "step": 4691
    },
    {
      "epoch": 8.657433056325024,
      "grad_norm": 0.783562421798706,
      "learning_rate": 1.4067e-05,
      "loss": 0.1587,
      "step": 4692
    },
    {
      "epoch": 8.659279778393351,
      "grad_norm": 1.0356827974319458,
      "learning_rate": 1.4069999999999999e-05,
      "loss": 0.1776,
      "step": 4693
    },
    {
      "epoch": 8.66112650046168,
      "grad_norm": 0.6998425126075745,
      "learning_rate": 1.4073e-05,
      "loss": 0.1448,
      "step": 4694
    },
    {
      "epoch": 8.66297322253001,
      "grad_norm": 1.0429692268371582,
      "learning_rate": 1.4076e-05,
      "loss": 0.1745,
      "step": 4695
    },
    {
      "epoch": 8.664819944598339,
      "grad_norm": 1.0342075824737549,
      "learning_rate": 1.4079e-05,
      "loss": 0.1288,
      "step": 4696
    },
    {
      "epoch": 8.666666666666666,
      "grad_norm": 0.6225987076759338,
      "learning_rate": 1.4082e-05,
      "loss": 0.1283,
      "step": 4697
    },
    {
      "epoch": 8.668513388734995,
      "grad_norm": 0.8024329543113708,
      "learning_rate": 1.4085e-05,
      "loss": 0.1496,
      "step": 4698
    },
    {
      "epoch": 8.670360110803324,
      "grad_norm": 1.1623533964157104,
      "learning_rate": 1.4088000000000001e-05,
      "loss": 0.1742,
      "step": 4699
    },
    {
      "epoch": 8.672206832871654,
      "grad_norm": 0.6703691482543945,
      "learning_rate": 1.4091000000000001e-05,
      "loss": 0.0818,
      "step": 4700
    },
    {
      "epoch": 8.67405355493998,
      "grad_norm": 0.5389675498008728,
      "learning_rate": 1.4094000000000001e-05,
      "loss": 0.0464,
      "step": 4701
    },
    {
      "epoch": 8.67590027700831,
      "grad_norm": 0.38047945499420166,
      "learning_rate": 1.4097e-05,
      "loss": 0.039,
      "step": 4702
    },
    {
      "epoch": 8.67774699907664,
      "grad_norm": 0.5646745562553406,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 0.0392,
      "step": 4703
    },
    {
      "epoch": 8.679593721144968,
      "grad_norm": 0.8179369568824768,
      "learning_rate": 1.4103e-05,
      "loss": 0.0522,
      "step": 4704
    },
    {
      "epoch": 8.681440443213296,
      "grad_norm": 0.46813997626304626,
      "learning_rate": 1.4106e-05,
      "loss": 0.0409,
      "step": 4705
    },
    {
      "epoch": 8.683287165281625,
      "grad_norm": 0.8852002024650574,
      "learning_rate": 1.4109e-05,
      "loss": 0.0412,
      "step": 4706
    },
    {
      "epoch": 8.685133887349954,
      "grad_norm": 0.3803097605705261,
      "learning_rate": 1.4112e-05,
      "loss": 0.0348,
      "step": 4707
    },
    {
      "epoch": 8.686980609418283,
      "grad_norm": 0.858212947845459,
      "learning_rate": 1.4115e-05,
      "loss": 0.0358,
      "step": 4708
    },
    {
      "epoch": 8.68882733148661,
      "grad_norm": 0.5910655856132507,
      "learning_rate": 1.4118000000000001e-05,
      "loss": 0.0353,
      "step": 4709
    },
    {
      "epoch": 8.69067405355494,
      "grad_norm": 0.4760793149471283,
      "learning_rate": 1.4121e-05,
      "loss": 0.0341,
      "step": 4710
    },
    {
      "epoch": 8.692520775623269,
      "grad_norm": 0.3494793176651001,
      "learning_rate": 1.4124e-05,
      "loss": 0.0199,
      "step": 4711
    },
    {
      "epoch": 8.694367497691598,
      "grad_norm": 0.6482746601104736,
      "learning_rate": 1.4127e-05,
      "loss": 0.0303,
      "step": 4712
    },
    {
      "epoch": 8.696214219759927,
      "grad_norm": 0.547419548034668,
      "learning_rate": 1.413e-05,
      "loss": 0.0254,
      "step": 4713
    },
    {
      "epoch": 8.698060941828254,
      "grad_norm": 0.493529736995697,
      "learning_rate": 1.4133000000000002e-05,
      "loss": 0.0318,
      "step": 4714
    },
    {
      "epoch": 8.699907663896584,
      "grad_norm": 0.4994904696941376,
      "learning_rate": 1.4136e-05,
      "loss": 0.0191,
      "step": 4715
    },
    {
      "epoch": 8.701754385964913,
      "grad_norm": 0.3659327030181885,
      "learning_rate": 1.4139e-05,
      "loss": 0.0182,
      "step": 4716
    },
    {
      "epoch": 8.703601108033242,
      "grad_norm": 0.5548020601272583,
      "learning_rate": 1.4142e-05,
      "loss": 0.0366,
      "step": 4717
    },
    {
      "epoch": 8.70544783010157,
      "grad_norm": 0.8539415597915649,
      "learning_rate": 1.4145e-05,
      "loss": 0.0229,
      "step": 4718
    },
    {
      "epoch": 8.707294552169898,
      "grad_norm": 0.7660424709320068,
      "learning_rate": 1.4148e-05,
      "loss": 0.0378,
      "step": 4719
    },
    {
      "epoch": 8.709141274238227,
      "grad_norm": 0.6833702325820923,
      "learning_rate": 1.4151e-05,
      "loss": 0.0251,
      "step": 4720
    },
    {
      "epoch": 8.710987996306557,
      "grad_norm": 0.6271822452545166,
      "learning_rate": 1.4154e-05,
      "loss": 0.0209,
      "step": 4721
    },
    {
      "epoch": 8.712834718374884,
      "grad_norm": 0.5643937587738037,
      "learning_rate": 1.4157e-05,
      "loss": 0.0169,
      "step": 4722
    },
    {
      "epoch": 8.714681440443213,
      "grad_norm": 0.5487121939659119,
      "learning_rate": 1.416e-05,
      "loss": 0.0333,
      "step": 4723
    },
    {
      "epoch": 8.716528162511542,
      "grad_norm": 0.5783451199531555,
      "learning_rate": 1.4163000000000001e-05,
      "loss": 0.0316,
      "step": 4724
    },
    {
      "epoch": 8.718374884579871,
      "grad_norm": 0.7464359402656555,
      "learning_rate": 1.4166000000000001e-05,
      "loss": 0.0305,
      "step": 4725
    },
    {
      "epoch": 8.720221606648199,
      "grad_norm": 0.9655202627182007,
      "learning_rate": 1.4169000000000001e-05,
      "loss": 0.029,
      "step": 4726
    },
    {
      "epoch": 8.722068328716528,
      "grad_norm": 0.885925829410553,
      "learning_rate": 1.4172e-05,
      "loss": 0.0289,
      "step": 4727
    },
    {
      "epoch": 8.723915050784857,
      "grad_norm": 0.8719249367713928,
      "learning_rate": 1.4174999999999999e-05,
      "loss": 0.0319,
      "step": 4728
    },
    {
      "epoch": 8.725761772853186,
      "grad_norm": 0.9360407590866089,
      "learning_rate": 1.4178e-05,
      "loss": 0.047,
      "step": 4729
    },
    {
      "epoch": 8.727608494921514,
      "grad_norm": 0.672272264957428,
      "learning_rate": 1.4181e-05,
      "loss": 0.051,
      "step": 4730
    },
    {
      "epoch": 8.729455216989843,
      "grad_norm": 0.5661531090736389,
      "learning_rate": 1.4184e-05,
      "loss": 0.0237,
      "step": 4731
    },
    {
      "epoch": 8.731301939058172,
      "grad_norm": 0.8892865777015686,
      "learning_rate": 1.4187e-05,
      "loss": 0.0275,
      "step": 4732
    },
    {
      "epoch": 8.733148661126501,
      "grad_norm": 0.5317668318748474,
      "learning_rate": 1.419e-05,
      "loss": 0.0258,
      "step": 4733
    },
    {
      "epoch": 8.734995383194828,
      "grad_norm": 0.7972832322120667,
      "learning_rate": 1.4193000000000001e-05,
      "loss": 0.0391,
      "step": 4734
    },
    {
      "epoch": 8.736842105263158,
      "grad_norm": 0.6717183589935303,
      "learning_rate": 1.4196000000000001e-05,
      "loss": 0.029,
      "step": 4735
    },
    {
      "epoch": 8.738688827331487,
      "grad_norm": 0.8849234580993652,
      "learning_rate": 1.4199e-05,
      "loss": 0.0526,
      "step": 4736
    },
    {
      "epoch": 8.740535549399816,
      "grad_norm": 1.2492313385009766,
      "learning_rate": 1.4202e-05,
      "loss": 0.3444,
      "step": 4737
    },
    {
      "epoch": 8.742382271468145,
      "grad_norm": 0.9331575036048889,
      "learning_rate": 1.4205e-05,
      "loss": 0.2953,
      "step": 4738
    },
    {
      "epoch": 8.744228993536472,
      "grad_norm": 0.8331683278083801,
      "learning_rate": 1.4208e-05,
      "loss": 0.2245,
      "step": 4739
    },
    {
      "epoch": 8.746075715604801,
      "grad_norm": 0.6677910089492798,
      "learning_rate": 1.4211e-05,
      "loss": 0.1566,
      "step": 4740
    },
    {
      "epoch": 8.74792243767313,
      "grad_norm": 0.9906908273696899,
      "learning_rate": 1.4214e-05,
      "loss": 0.2517,
      "step": 4741
    },
    {
      "epoch": 8.749769159741458,
      "grad_norm": 0.6925485730171204,
      "learning_rate": 1.4217e-05,
      "loss": 0.1647,
      "step": 4742
    },
    {
      "epoch": 8.751615881809787,
      "grad_norm": 0.5896223783493042,
      "learning_rate": 1.422e-05,
      "loss": 0.1449,
      "step": 4743
    },
    {
      "epoch": 8.753462603878116,
      "grad_norm": 0.5901791453361511,
      "learning_rate": 1.4223000000000001e-05,
      "loss": 0.1015,
      "step": 4744
    },
    {
      "epoch": 8.755309325946445,
      "grad_norm": 0.6510495543479919,
      "learning_rate": 1.4226e-05,
      "loss": 0.1121,
      "step": 4745
    },
    {
      "epoch": 8.757156048014775,
      "grad_norm": 0.635270893573761,
      "learning_rate": 1.4229e-05,
      "loss": 0.105,
      "step": 4746
    },
    {
      "epoch": 8.759002770083102,
      "grad_norm": 0.5553750395774841,
      "learning_rate": 1.4232e-05,
      "loss": 0.0773,
      "step": 4747
    },
    {
      "epoch": 8.760849492151431,
      "grad_norm": 0.9713811278343201,
      "learning_rate": 1.4235e-05,
      "loss": 0.1125,
      "step": 4748
    },
    {
      "epoch": 8.76269621421976,
      "grad_norm": 0.6695632934570312,
      "learning_rate": 1.4238000000000002e-05,
      "loss": 0.099,
      "step": 4749
    },
    {
      "epoch": 8.76454293628809,
      "grad_norm": 0.6211198568344116,
      "learning_rate": 1.4241000000000001e-05,
      "loss": 0.0644,
      "step": 4750
    },
    {
      "epoch": 8.766389658356417,
      "grad_norm": 0.8161517977714539,
      "learning_rate": 1.4244000000000001e-05,
      "loss": 0.0644,
      "step": 4751
    },
    {
      "epoch": 8.768236380424746,
      "grad_norm": 0.4428170323371887,
      "learning_rate": 1.4247e-05,
      "loss": 0.0423,
      "step": 4752
    },
    {
      "epoch": 8.770083102493075,
      "grad_norm": 0.5940005779266357,
      "learning_rate": 1.4249999999999999e-05,
      "loss": 0.0437,
      "step": 4753
    },
    {
      "epoch": 8.771929824561404,
      "grad_norm": 0.4171900749206543,
      "learning_rate": 1.4253e-05,
      "loss": 0.0327,
      "step": 4754
    },
    {
      "epoch": 8.773776546629731,
      "grad_norm": 0.3937453627586365,
      "learning_rate": 1.4256e-05,
      "loss": 0.0255,
      "step": 4755
    },
    {
      "epoch": 8.77562326869806,
      "grad_norm": 0.6645901799201965,
      "learning_rate": 1.4259e-05,
      "loss": 0.0366,
      "step": 4756
    },
    {
      "epoch": 8.77746999076639,
      "grad_norm": 1.1906750202178955,
      "learning_rate": 1.4262e-05,
      "loss": 0.0391,
      "step": 4757
    },
    {
      "epoch": 8.779316712834719,
      "grad_norm": 0.592204213142395,
      "learning_rate": 1.4265e-05,
      "loss": 0.0487,
      "step": 4758
    },
    {
      "epoch": 8.781163434903046,
      "grad_norm": 0.6466925144195557,
      "learning_rate": 1.4268000000000001e-05,
      "loss": 0.0395,
      "step": 4759
    },
    {
      "epoch": 8.783010156971375,
      "grad_norm": 0.531396210193634,
      "learning_rate": 1.4271000000000001e-05,
      "loss": 0.0301,
      "step": 4760
    },
    {
      "epoch": 8.784856879039705,
      "grad_norm": 0.8401514887809753,
      "learning_rate": 1.4274000000000001e-05,
      "loss": 0.0415,
      "step": 4761
    },
    {
      "epoch": 8.786703601108034,
      "grad_norm": 0.6957353353500366,
      "learning_rate": 1.4277e-05,
      "loss": 0.0299,
      "step": 4762
    },
    {
      "epoch": 8.788550323176363,
      "grad_norm": 0.6141769289970398,
      "learning_rate": 1.428e-05,
      "loss": 0.0373,
      "step": 4763
    },
    {
      "epoch": 8.79039704524469,
      "grad_norm": 0.6922504901885986,
      "learning_rate": 1.4283e-05,
      "loss": 0.0712,
      "step": 4764
    },
    {
      "epoch": 8.79224376731302,
      "grad_norm": 0.7587156891822815,
      "learning_rate": 1.4286e-05,
      "loss": 0.025,
      "step": 4765
    },
    {
      "epoch": 8.794090489381349,
      "grad_norm": 0.4232429265975952,
      "learning_rate": 1.4289e-05,
      "loss": 0.0319,
      "step": 4766
    },
    {
      "epoch": 8.795937211449676,
      "grad_norm": 0.6227977871894836,
      "learning_rate": 1.4292e-05,
      "loss": 0.0451,
      "step": 4767
    },
    {
      "epoch": 8.797783933518005,
      "grad_norm": 0.5408965945243835,
      "learning_rate": 1.4295e-05,
      "loss": 0.0369,
      "step": 4768
    },
    {
      "epoch": 8.799630655586334,
      "grad_norm": 0.45448073744773865,
      "learning_rate": 1.4298000000000001e-05,
      "loss": 0.023,
      "step": 4769
    },
    {
      "epoch": 8.801477377654663,
      "grad_norm": 0.42481276392936707,
      "learning_rate": 1.4301e-05,
      "loss": 0.0174,
      "step": 4770
    },
    {
      "epoch": 8.803324099722992,
      "grad_norm": 0.4406210482120514,
      "learning_rate": 1.4304e-05,
      "loss": 0.0232,
      "step": 4771
    },
    {
      "epoch": 8.80517082179132,
      "grad_norm": 0.7529833316802979,
      "learning_rate": 1.4307e-05,
      "loss": 0.028,
      "step": 4772
    },
    {
      "epoch": 8.807017543859649,
      "grad_norm": 0.8267642855644226,
      "learning_rate": 1.431e-05,
      "loss": 0.0386,
      "step": 4773
    },
    {
      "epoch": 8.808864265927978,
      "grad_norm": 0.7792585492134094,
      "learning_rate": 1.4313000000000002e-05,
      "loss": 0.0436,
      "step": 4774
    },
    {
      "epoch": 8.810710987996307,
      "grad_norm": 0.6700918674468994,
      "learning_rate": 1.4316000000000002e-05,
      "loss": 0.0339,
      "step": 4775
    },
    {
      "epoch": 8.812557710064635,
      "grad_norm": 0.4729313552379608,
      "learning_rate": 1.4319e-05,
      "loss": 0.0412,
      "step": 4776
    },
    {
      "epoch": 8.814404432132964,
      "grad_norm": 0.7246980667114258,
      "learning_rate": 1.4322e-05,
      "loss": 0.0492,
      "step": 4777
    },
    {
      "epoch": 8.816251154201293,
      "grad_norm": 0.5295450091362,
      "learning_rate": 1.4325e-05,
      "loss": 0.0213,
      "step": 4778
    },
    {
      "epoch": 8.818097876269622,
      "grad_norm": 0.5199748873710632,
      "learning_rate": 1.4328e-05,
      "loss": 0.0266,
      "step": 4779
    },
    {
      "epoch": 8.81994459833795,
      "grad_norm": 0.5751835703849792,
      "learning_rate": 1.4331e-05,
      "loss": 0.0293,
      "step": 4780
    },
    {
      "epoch": 8.821791320406279,
      "grad_norm": 0.6247977614402771,
      "learning_rate": 1.4334e-05,
      "loss": 0.04,
      "step": 4781
    },
    {
      "epoch": 8.823638042474608,
      "grad_norm": 0.44509533047676086,
      "learning_rate": 1.4337e-05,
      "loss": 0.019,
      "step": 4782
    },
    {
      "epoch": 8.825484764542937,
      "grad_norm": 1.0526492595672607,
      "learning_rate": 1.434e-05,
      "loss": 0.036,
      "step": 4783
    },
    {
      "epoch": 8.827331486611264,
      "grad_norm": 0.4944753050804138,
      "learning_rate": 1.4343000000000001e-05,
      "loss": 0.0313,
      "step": 4784
    },
    {
      "epoch": 8.829178208679593,
      "grad_norm": 0.8059141635894775,
      "learning_rate": 1.4346000000000001e-05,
      "loss": 0.0512,
      "step": 4785
    },
    {
      "epoch": 8.831024930747922,
      "grad_norm": 0.8032324314117432,
      "learning_rate": 1.4349000000000001e-05,
      "loss": 0.0417,
      "step": 4786
    },
    {
      "epoch": 8.832871652816252,
      "grad_norm": 1.1484620571136475,
      "learning_rate": 1.4352e-05,
      "loss": 0.2946,
      "step": 4787
    },
    {
      "epoch": 8.83471837488458,
      "grad_norm": 0.8292638063430786,
      "learning_rate": 1.4355e-05,
      "loss": 0.2373,
      "step": 4788
    },
    {
      "epoch": 8.836565096952908,
      "grad_norm": 0.7926152944564819,
      "learning_rate": 1.4358e-05,
      "loss": 0.2248,
      "step": 4789
    },
    {
      "epoch": 8.838411819021237,
      "grad_norm": 0.909888744354248,
      "learning_rate": 1.4361e-05,
      "loss": 0.1743,
      "step": 4790
    },
    {
      "epoch": 8.840258541089566,
      "grad_norm": 1.0106759071350098,
      "learning_rate": 1.4364e-05,
      "loss": 0.2385,
      "step": 4791
    },
    {
      "epoch": 8.842105263157894,
      "grad_norm": 0.668520450592041,
      "learning_rate": 1.4367e-05,
      "loss": 0.1138,
      "step": 4792
    },
    {
      "epoch": 8.843951985226223,
      "grad_norm": 0.7502413988113403,
      "learning_rate": 1.437e-05,
      "loss": 0.1626,
      "step": 4793
    },
    {
      "epoch": 8.845798707294552,
      "grad_norm": 0.8197979927062988,
      "learning_rate": 1.4373000000000001e-05,
      "loss": 0.1745,
      "step": 4794
    },
    {
      "epoch": 8.847645429362881,
      "grad_norm": 0.7981426119804382,
      "learning_rate": 1.4376000000000001e-05,
      "loss": 0.1643,
      "step": 4795
    },
    {
      "epoch": 8.84949215143121,
      "grad_norm": 0.7516316175460815,
      "learning_rate": 1.4379e-05,
      "loss": 0.1304,
      "step": 4796
    },
    {
      "epoch": 8.851338873499538,
      "grad_norm": 0.6946542263031006,
      "learning_rate": 1.4382e-05,
      "loss": 0.1273,
      "step": 4797
    },
    {
      "epoch": 8.853185595567867,
      "grad_norm": 0.6469597816467285,
      "learning_rate": 1.4385e-05,
      "loss": 0.1079,
      "step": 4798
    },
    {
      "epoch": 8.855032317636196,
      "grad_norm": 0.7252698540687561,
      "learning_rate": 1.4388000000000002e-05,
      "loss": 0.1114,
      "step": 4799
    },
    {
      "epoch": 8.856879039704525,
      "grad_norm": 0.569025456905365,
      "learning_rate": 1.4391000000000002e-05,
      "loss": 0.0745,
      "step": 4800
    },
    {
      "epoch": 8.858725761772853,
      "grad_norm": 0.9024773836135864,
      "learning_rate": 1.4394e-05,
      "loss": 0.078,
      "step": 4801
    },
    {
      "epoch": 8.860572483841182,
      "grad_norm": 0.5917884111404419,
      "learning_rate": 1.4397e-05,
      "loss": 0.0585,
      "step": 4802
    },
    {
      "epoch": 8.86241920590951,
      "grad_norm": 0.6219111680984497,
      "learning_rate": 1.44e-05,
      "loss": 0.0345,
      "step": 4803
    },
    {
      "epoch": 8.86426592797784,
      "grad_norm": 0.46978530287742615,
      "learning_rate": 1.4403e-05,
      "loss": 0.0642,
      "step": 4804
    },
    {
      "epoch": 8.866112650046167,
      "grad_norm": 0.7603000998497009,
      "learning_rate": 1.4406e-05,
      "loss": 0.0584,
      "step": 4805
    },
    {
      "epoch": 8.867959372114496,
      "grad_norm": 0.9757034182548523,
      "learning_rate": 1.4409e-05,
      "loss": 0.0394,
      "step": 4806
    },
    {
      "epoch": 8.869806094182826,
      "grad_norm": 0.5145926475524902,
      "learning_rate": 1.4412e-05,
      "loss": 0.0591,
      "step": 4807
    },
    {
      "epoch": 8.871652816251155,
      "grad_norm": 0.40463876724243164,
      "learning_rate": 1.4415e-05,
      "loss": 0.0198,
      "step": 4808
    },
    {
      "epoch": 8.873499538319482,
      "grad_norm": 0.4177763760089874,
      "learning_rate": 1.4418000000000002e-05,
      "loss": 0.0331,
      "step": 4809
    },
    {
      "epoch": 8.875346260387811,
      "grad_norm": 0.4561941921710968,
      "learning_rate": 1.4421000000000001e-05,
      "loss": 0.0333,
      "step": 4810
    },
    {
      "epoch": 8.87719298245614,
      "grad_norm": 0.36311978101730347,
      "learning_rate": 1.4424000000000001e-05,
      "loss": 0.0231,
      "step": 4811
    },
    {
      "epoch": 8.87903970452447,
      "grad_norm": 0.6601563096046448,
      "learning_rate": 1.4427000000000001e-05,
      "loss": 0.0263,
      "step": 4812
    },
    {
      "epoch": 8.880886426592799,
      "grad_norm": 0.49798139929771423,
      "learning_rate": 1.4429999999999999e-05,
      "loss": 0.0271,
      "step": 4813
    },
    {
      "epoch": 8.882733148661126,
      "grad_norm": 0.5754868388175964,
      "learning_rate": 1.4433e-05,
      "loss": 0.0381,
      "step": 4814
    },
    {
      "epoch": 8.884579870729455,
      "grad_norm": 0.6019905209541321,
      "learning_rate": 1.4436e-05,
      "loss": 0.0344,
      "step": 4815
    },
    {
      "epoch": 8.886426592797784,
      "grad_norm": 0.39315265417099,
      "learning_rate": 1.4439e-05,
      "loss": 0.0229,
      "step": 4816
    },
    {
      "epoch": 8.888273314866112,
      "grad_norm": 0.4014921486377716,
      "learning_rate": 1.4442e-05,
      "loss": 0.018,
      "step": 4817
    },
    {
      "epoch": 8.89012003693444,
      "grad_norm": 0.5342328548431396,
      "learning_rate": 1.4445e-05,
      "loss": 0.0272,
      "step": 4818
    },
    {
      "epoch": 8.89196675900277,
      "grad_norm": 0.5129212141036987,
      "learning_rate": 1.4448e-05,
      "loss": 0.0189,
      "step": 4819
    },
    {
      "epoch": 8.8938134810711,
      "grad_norm": 0.4567769169807434,
      "learning_rate": 1.4451000000000001e-05,
      "loss": 0.0205,
      "step": 4820
    },
    {
      "epoch": 8.895660203139428,
      "grad_norm": 0.5290626287460327,
      "learning_rate": 1.4454000000000001e-05,
      "loss": 0.0312,
      "step": 4821
    },
    {
      "epoch": 8.897506925207756,
      "grad_norm": 0.5846696496009827,
      "learning_rate": 1.4457e-05,
      "loss": 0.0266,
      "step": 4822
    },
    {
      "epoch": 8.899353647276085,
      "grad_norm": 0.9421910643577576,
      "learning_rate": 1.446e-05,
      "loss": 0.0623,
      "step": 4823
    },
    {
      "epoch": 8.901200369344414,
      "grad_norm": 0.532992422580719,
      "learning_rate": 1.4463e-05,
      "loss": 0.0336,
      "step": 4824
    },
    {
      "epoch": 8.903047091412743,
      "grad_norm": 0.4452487826347351,
      "learning_rate": 1.4466e-05,
      "loss": 0.0262,
      "step": 4825
    },
    {
      "epoch": 8.90489381348107,
      "grad_norm": 0.6956632733345032,
      "learning_rate": 1.4469e-05,
      "loss": 0.0392,
      "step": 4826
    },
    {
      "epoch": 8.9067405355494,
      "grad_norm": 0.6956707835197449,
      "learning_rate": 1.4472e-05,
      "loss": 0.0369,
      "step": 4827
    },
    {
      "epoch": 8.908587257617729,
      "grad_norm": 0.5482403635978699,
      "learning_rate": 1.4475e-05,
      "loss": 0.0269,
      "step": 4828
    },
    {
      "epoch": 8.910433979686058,
      "grad_norm": 0.5756843686103821,
      "learning_rate": 1.4478e-05,
      "loss": 0.0346,
      "step": 4829
    },
    {
      "epoch": 8.912280701754385,
      "grad_norm": 0.3994274437427521,
      "learning_rate": 1.4481e-05,
      "loss": 0.0266,
      "step": 4830
    },
    {
      "epoch": 8.914127423822714,
      "grad_norm": 0.5239002108573914,
      "learning_rate": 1.4484e-05,
      "loss": 0.0301,
      "step": 4831
    },
    {
      "epoch": 8.915974145891044,
      "grad_norm": 0.5991451740264893,
      "learning_rate": 1.4487e-05,
      "loss": 0.0362,
      "step": 4832
    },
    {
      "epoch": 8.917820867959373,
      "grad_norm": 0.5694364905357361,
      "learning_rate": 1.449e-05,
      "loss": 0.0286,
      "step": 4833
    },
    {
      "epoch": 8.9196675900277,
      "grad_norm": 0.8278652429580688,
      "learning_rate": 1.4493e-05,
      "loss": 0.0487,
      "step": 4834
    },
    {
      "epoch": 8.92151431209603,
      "grad_norm": 0.6823614835739136,
      "learning_rate": 1.4496000000000001e-05,
      "loss": 0.0332,
      "step": 4835
    },
    {
      "epoch": 8.923361034164358,
      "grad_norm": 1.0626842975616455,
      "learning_rate": 1.4499000000000001e-05,
      "loss": 0.0489,
      "step": 4836
    },
    {
      "epoch": 8.925207756232687,
      "grad_norm": 1.330648422241211,
      "learning_rate": 1.4502000000000001e-05,
      "loss": 0.295,
      "step": 4837
    },
    {
      "epoch": 8.927054478301017,
      "grad_norm": 0.8124020099639893,
      "learning_rate": 1.4505e-05,
      "loss": 0.2178,
      "step": 4838
    },
    {
      "epoch": 8.928901200369344,
      "grad_norm": 1.0896893739700317,
      "learning_rate": 1.4507999999999999e-05,
      "loss": 0.2638,
      "step": 4839
    },
    {
      "epoch": 8.930747922437673,
      "grad_norm": 1.3501861095428467,
      "learning_rate": 1.4511e-05,
      "loss": 0.2023,
      "step": 4840
    },
    {
      "epoch": 8.932594644506002,
      "grad_norm": 0.8362381458282471,
      "learning_rate": 1.4514e-05,
      "loss": 0.142,
      "step": 4841
    },
    {
      "epoch": 8.93444136657433,
      "grad_norm": 0.6680560111999512,
      "learning_rate": 1.4517e-05,
      "loss": 0.1191,
      "step": 4842
    },
    {
      "epoch": 8.936288088642659,
      "grad_norm": 0.6559258103370667,
      "learning_rate": 1.452e-05,
      "loss": 0.1294,
      "step": 4843
    },
    {
      "epoch": 8.938134810710988,
      "grad_norm": 1.0417401790618896,
      "learning_rate": 1.4523e-05,
      "loss": 0.143,
      "step": 4844
    },
    {
      "epoch": 8.939981532779317,
      "grad_norm": 1.0837085247039795,
      "learning_rate": 1.4526000000000001e-05,
      "loss": 0.1465,
      "step": 4845
    },
    {
      "epoch": 8.941828254847646,
      "grad_norm": 0.5968208312988281,
      "learning_rate": 1.4529000000000001e-05,
      "loss": 0.0799,
      "step": 4846
    },
    {
      "epoch": 8.943674976915974,
      "grad_norm": 0.495822012424469,
      "learning_rate": 1.4532e-05,
      "loss": 0.0643,
      "step": 4847
    },
    {
      "epoch": 8.945521698984303,
      "grad_norm": 0.7860730886459351,
      "learning_rate": 1.4535e-05,
      "loss": 0.0953,
      "step": 4848
    },
    {
      "epoch": 8.947368421052632,
      "grad_norm": 0.5441517233848572,
      "learning_rate": 1.4538e-05,
      "loss": 0.0703,
      "step": 4849
    },
    {
      "epoch": 8.949215143120961,
      "grad_norm": 0.5062694549560547,
      "learning_rate": 1.4541e-05,
      "loss": 0.039,
      "step": 4850
    },
    {
      "epoch": 8.951061865189288,
      "grad_norm": 0.6828703880310059,
      "learning_rate": 1.4544e-05,
      "loss": 0.0599,
      "step": 4851
    },
    {
      "epoch": 8.952908587257618,
      "grad_norm": 0.5830159187316895,
      "learning_rate": 1.4547e-05,
      "loss": 0.0811,
      "step": 4852
    },
    {
      "epoch": 8.954755309325947,
      "grad_norm": 0.44063887000083923,
      "learning_rate": 1.455e-05,
      "loss": 0.0283,
      "step": 4853
    },
    {
      "epoch": 8.956602031394276,
      "grad_norm": 0.5686138868331909,
      "learning_rate": 1.4553e-05,
      "loss": 0.0335,
      "step": 4854
    },
    {
      "epoch": 8.958448753462603,
      "grad_norm": 0.6429723501205444,
      "learning_rate": 1.4556000000000001e-05,
      "loss": 0.0349,
      "step": 4855
    },
    {
      "epoch": 8.960295475530932,
      "grad_norm": 0.43935972452163696,
      "learning_rate": 1.4559e-05,
      "loss": 0.0238,
      "step": 4856
    },
    {
      "epoch": 8.962142197599261,
      "grad_norm": 0.5402309894561768,
      "learning_rate": 1.4562e-05,
      "loss": 0.029,
      "step": 4857
    },
    {
      "epoch": 8.96398891966759,
      "grad_norm": 0.7583809494972229,
      "learning_rate": 1.4565e-05,
      "loss": 0.0632,
      "step": 4858
    },
    {
      "epoch": 8.965835641735918,
      "grad_norm": 0.5806187391281128,
      "learning_rate": 1.4568e-05,
      "loss": 0.0397,
      "step": 4859
    },
    {
      "epoch": 8.967682363804247,
      "grad_norm": 0.4688240587711334,
      "learning_rate": 1.4571000000000002e-05,
      "loss": 0.0358,
      "step": 4860
    },
    {
      "epoch": 8.969529085872576,
      "grad_norm": 0.6508371829986572,
      "learning_rate": 1.4574000000000001e-05,
      "loss": 0.0313,
      "step": 4861
    },
    {
      "epoch": 8.971375807940905,
      "grad_norm": 0.5394099950790405,
      "learning_rate": 1.4577e-05,
      "loss": 0.0347,
      "step": 4862
    },
    {
      "epoch": 8.973222530009235,
      "grad_norm": 0.2626785635948181,
      "learning_rate": 1.458e-05,
      "loss": 0.014,
      "step": 4863
    },
    {
      "epoch": 8.975069252077562,
      "grad_norm": 0.7136227488517761,
      "learning_rate": 1.4582999999999999e-05,
      "loss": 0.0326,
      "step": 4864
    },
    {
      "epoch": 8.976915974145891,
      "grad_norm": 0.5395230054855347,
      "learning_rate": 1.4586e-05,
      "loss": 0.024,
      "step": 4865
    },
    {
      "epoch": 8.97876269621422,
      "grad_norm": 0.5298368334770203,
      "learning_rate": 1.4589e-05,
      "loss": 0.038,
      "step": 4866
    },
    {
      "epoch": 8.980609418282548,
      "grad_norm": 0.6000943779945374,
      "learning_rate": 1.4592e-05,
      "loss": 0.0362,
      "step": 4867
    },
    {
      "epoch": 8.982456140350877,
      "grad_norm": 0.5490943193435669,
      "learning_rate": 1.4595e-05,
      "loss": 0.0466,
      "step": 4868
    },
    {
      "epoch": 8.984302862419206,
      "grad_norm": 0.5911248326301575,
      "learning_rate": 1.4598e-05,
      "loss": 0.035,
      "step": 4869
    },
    {
      "epoch": 8.986149584487535,
      "grad_norm": 0.5520859956741333,
      "learning_rate": 1.4601000000000001e-05,
      "loss": 0.0173,
      "step": 4870
    },
    {
      "epoch": 8.987996306555864,
      "grad_norm": 0.45630595088005066,
      "learning_rate": 1.4604000000000001e-05,
      "loss": 0.0276,
      "step": 4871
    },
    {
      "epoch": 8.989843028624191,
      "grad_norm": 0.4523727297782898,
      "learning_rate": 1.4607000000000001e-05,
      "loss": 0.0223,
      "step": 4872
    },
    {
      "epoch": 8.99168975069252,
      "grad_norm": 0.5295506119728088,
      "learning_rate": 1.461e-05,
      "loss": 0.0301,
      "step": 4873
    },
    {
      "epoch": 8.99353647276085,
      "grad_norm": 0.6663303375244141,
      "learning_rate": 1.4613e-05,
      "loss": 0.0329,
      "step": 4874
    },
    {
      "epoch": 8.995383194829179,
      "grad_norm": 0.7034157514572144,
      "learning_rate": 1.4616e-05,
      "loss": 0.0316,
      "step": 4875
    },
    {
      "epoch": 8.997229916897506,
      "grad_norm": 0.5341060161590576,
      "learning_rate": 1.4619e-05,
      "loss": 0.0348,
      "step": 4876
    },
    {
      "epoch": 8.999076638965835,
      "grad_norm": 0.8159637451171875,
      "learning_rate": 1.4622e-05,
      "loss": 0.0454,
      "step": 4877
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.6270403265953064,
      "learning_rate": 1.4625e-05,
      "loss": 0.0281,
      "step": 4878
    },
    {
      "epoch": 9.00184672206833,
      "grad_norm": 0.7949193120002747,
      "learning_rate": 1.4628e-05,
      "loss": 0.2474,
      "step": 4879
    },
    {
      "epoch": 9.003693444136658,
      "grad_norm": 0.7404318451881409,
      "learning_rate": 1.4631000000000001e-05,
      "loss": 0.2327,
      "step": 4880
    },
    {
      "epoch": 9.005540166204986,
      "grad_norm": 0.704534649848938,
      "learning_rate": 1.4634e-05,
      "loss": 0.2012,
      "step": 4881
    },
    {
      "epoch": 9.007386888273315,
      "grad_norm": 0.7474093437194824,
      "learning_rate": 1.4637e-05,
      "loss": 0.187,
      "step": 4882
    },
    {
      "epoch": 9.009233610341644,
      "grad_norm": 0.8021607995033264,
      "learning_rate": 1.464e-05,
      "loss": 0.2167,
      "step": 4883
    },
    {
      "epoch": 9.011080332409973,
      "grad_norm": 0.6698554754257202,
      "learning_rate": 1.4643e-05,
      "loss": 0.184,
      "step": 4884
    },
    {
      "epoch": 9.0129270544783,
      "grad_norm": 0.5793368220329285,
      "learning_rate": 1.4646000000000002e-05,
      "loss": 0.0935,
      "step": 4885
    },
    {
      "epoch": 9.01477377654663,
      "grad_norm": 0.963193416595459,
      "learning_rate": 1.4649000000000002e-05,
      "loss": 0.1459,
      "step": 4886
    },
    {
      "epoch": 9.016620498614959,
      "grad_norm": 0.5509797930717468,
      "learning_rate": 1.4652e-05,
      "loss": 0.0893,
      "step": 4887
    },
    {
      "epoch": 9.018467220683288,
      "grad_norm": 0.5855543613433838,
      "learning_rate": 1.4655e-05,
      "loss": 0.1437,
      "step": 4888
    },
    {
      "epoch": 9.020313942751615,
      "grad_norm": 0.5491319894790649,
      "learning_rate": 1.4658e-05,
      "loss": 0.1161,
      "step": 4889
    },
    {
      "epoch": 9.022160664819944,
      "grad_norm": 0.6451462507247925,
      "learning_rate": 1.4661e-05,
      "loss": 0.0724,
      "step": 4890
    },
    {
      "epoch": 9.024007386888274,
      "grad_norm": 0.5323352217674255,
      "learning_rate": 1.4664e-05,
      "loss": 0.0917,
      "step": 4891
    },
    {
      "epoch": 9.025854108956603,
      "grad_norm": 0.5460320115089417,
      "learning_rate": 1.4667e-05,
      "loss": 0.0553,
      "step": 4892
    },
    {
      "epoch": 9.02770083102493,
      "grad_norm": 0.6950182914733887,
      "learning_rate": 1.467e-05,
      "loss": 0.1156,
      "step": 4893
    },
    {
      "epoch": 9.02954755309326,
      "grad_norm": 0.9646598696708679,
      "learning_rate": 1.4673e-05,
      "loss": 0.0439,
      "step": 4894
    },
    {
      "epoch": 9.031394275161588,
      "grad_norm": 0.4271436929702759,
      "learning_rate": 1.4676000000000001e-05,
      "loss": 0.0377,
      "step": 4895
    },
    {
      "epoch": 9.033240997229917,
      "grad_norm": 0.6988269686698914,
      "learning_rate": 1.4679000000000001e-05,
      "loss": 0.0372,
      "step": 4896
    },
    {
      "epoch": 9.035087719298245,
      "grad_norm": 0.4358237683773041,
      "learning_rate": 1.4682000000000001e-05,
      "loss": 0.0305,
      "step": 4897
    },
    {
      "epoch": 9.036934441366574,
      "grad_norm": 0.47687071561813354,
      "learning_rate": 1.4685000000000001e-05,
      "loss": 0.0283,
      "step": 4898
    },
    {
      "epoch": 9.038781163434903,
      "grad_norm": 0.589789628982544,
      "learning_rate": 1.4687999999999999e-05,
      "loss": 0.0187,
      "step": 4899
    },
    {
      "epoch": 9.040627885503232,
      "grad_norm": 0.7305690050125122,
      "learning_rate": 1.4691e-05,
      "loss": 0.05,
      "step": 4900
    },
    {
      "epoch": 9.04247460757156,
      "grad_norm": 0.3432635962963104,
      "learning_rate": 1.4694e-05,
      "loss": 0.0259,
      "step": 4901
    },
    {
      "epoch": 9.044321329639889,
      "grad_norm": 0.3739214837551117,
      "learning_rate": 1.4697e-05,
      "loss": 0.03,
      "step": 4902
    },
    {
      "epoch": 9.046168051708218,
      "grad_norm": 0.5927174091339111,
      "learning_rate": 1.47e-05,
      "loss": 0.037,
      "step": 4903
    },
    {
      "epoch": 9.048014773776547,
      "grad_norm": 0.7059500813484192,
      "learning_rate": 1.4703e-05,
      "loss": 0.0446,
      "step": 4904
    },
    {
      "epoch": 9.049861495844876,
      "grad_norm": 0.37194669246673584,
      "learning_rate": 1.4706000000000001e-05,
      "loss": 0.0214,
      "step": 4905
    },
    {
      "epoch": 9.051708217913204,
      "grad_norm": 1.4105937480926514,
      "learning_rate": 1.4709000000000001e-05,
      "loss": 0.056,
      "step": 4906
    },
    {
      "epoch": 9.053554939981533,
      "grad_norm": 0.6077854037284851,
      "learning_rate": 1.4712e-05,
      "loss": 0.029,
      "step": 4907
    },
    {
      "epoch": 9.055401662049862,
      "grad_norm": 0.4426223337650299,
      "learning_rate": 1.4715e-05,
      "loss": 0.0204,
      "step": 4908
    },
    {
      "epoch": 9.057248384118191,
      "grad_norm": 1.0114620923995972,
      "learning_rate": 1.4718e-05,
      "loss": 0.0545,
      "step": 4909
    },
    {
      "epoch": 9.059095106186518,
      "grad_norm": 0.5293107032775879,
      "learning_rate": 1.4721000000000002e-05,
      "loss": 0.0183,
      "step": 4910
    },
    {
      "epoch": 9.060941828254848,
      "grad_norm": 0.7033582329750061,
      "learning_rate": 1.4724e-05,
      "loss": 0.0342,
      "step": 4911
    },
    {
      "epoch": 9.062788550323177,
      "grad_norm": 0.4826132655143738,
      "learning_rate": 1.4727e-05,
      "loss": 0.0257,
      "step": 4912
    },
    {
      "epoch": 9.064635272391506,
      "grad_norm": 0.5029767155647278,
      "learning_rate": 1.473e-05,
      "loss": 0.0249,
      "step": 4913
    },
    {
      "epoch": 9.066481994459833,
      "grad_norm": 0.35271936655044556,
      "learning_rate": 1.4733e-05,
      "loss": 0.0197,
      "step": 4914
    },
    {
      "epoch": 9.068328716528162,
      "grad_norm": 0.6118987202644348,
      "learning_rate": 1.4736000000000001e-05,
      "loss": 0.017,
      "step": 4915
    },
    {
      "epoch": 9.070175438596491,
      "grad_norm": 0.4268760681152344,
      "learning_rate": 1.4739e-05,
      "loss": 0.0292,
      "step": 4916
    },
    {
      "epoch": 9.07202216066482,
      "grad_norm": 0.6673272252082825,
      "learning_rate": 1.4742e-05,
      "loss": 0.037,
      "step": 4917
    },
    {
      "epoch": 9.073868882733148,
      "grad_norm": 0.5449793934822083,
      "learning_rate": 1.4745e-05,
      "loss": 0.0386,
      "step": 4918
    },
    {
      "epoch": 9.075715604801477,
      "grad_norm": 0.45347753167152405,
      "learning_rate": 1.4748e-05,
      "loss": 0.0207,
      "step": 4919
    },
    {
      "epoch": 9.077562326869806,
      "grad_norm": 0.36634916067123413,
      "learning_rate": 1.4751000000000002e-05,
      "loss": 0.0185,
      "step": 4920
    },
    {
      "epoch": 9.079409048938135,
      "grad_norm": 0.9379186034202576,
      "learning_rate": 1.4754000000000001e-05,
      "loss": 0.0508,
      "step": 4921
    },
    {
      "epoch": 9.081255771006463,
      "grad_norm": 0.5121855139732361,
      "learning_rate": 1.4757000000000001e-05,
      "loss": 0.0271,
      "step": 4922
    },
    {
      "epoch": 9.083102493074792,
      "grad_norm": 1.2979964017868042,
      "learning_rate": 1.4760000000000001e-05,
      "loss": 0.0269,
      "step": 4923
    },
    {
      "epoch": 9.084949215143121,
      "grad_norm": 0.4717375934123993,
      "learning_rate": 1.4762999999999999e-05,
      "loss": 0.0234,
      "step": 4924
    },
    {
      "epoch": 9.08679593721145,
      "grad_norm": 1.4416847229003906,
      "learning_rate": 1.4766e-05,
      "loss": 0.0359,
      "step": 4925
    },
    {
      "epoch": 9.088642659279778,
      "grad_norm": 0.8894223570823669,
      "learning_rate": 1.4769e-05,
      "loss": 0.0437,
      "step": 4926
    },
    {
      "epoch": 9.090489381348107,
      "grad_norm": 0.4688844382762909,
      "learning_rate": 1.4772e-05,
      "loss": 0.0261,
      "step": 4927
    },
    {
      "epoch": 9.092336103416436,
      "grad_norm": 0.5282978415489197,
      "learning_rate": 1.4775e-05,
      "loss": 0.0194,
      "step": 4928
    },
    {
      "epoch": 9.094182825484765,
      "grad_norm": 0.9984002113342285,
      "learning_rate": 1.4778e-05,
      "loss": 0.3795,
      "step": 4929
    },
    {
      "epoch": 9.096029547553094,
      "grad_norm": 0.9583333134651184,
      "learning_rate": 1.4781000000000001e-05,
      "loss": 0.2978,
      "step": 4930
    },
    {
      "epoch": 9.097876269621421,
      "grad_norm": 0.9209781885147095,
      "learning_rate": 1.4784000000000001e-05,
      "loss": 0.2958,
      "step": 4931
    },
    {
      "epoch": 9.09972299168975,
      "grad_norm": 0.6429269313812256,
      "learning_rate": 1.4787000000000001e-05,
      "loss": 0.1974,
      "step": 4932
    },
    {
      "epoch": 9.10156971375808,
      "grad_norm": 0.8534770607948303,
      "learning_rate": 1.479e-05,
      "loss": 0.2142,
      "step": 4933
    },
    {
      "epoch": 9.103416435826409,
      "grad_norm": 0.6178396344184875,
      "learning_rate": 1.4793e-05,
      "loss": 0.1418,
      "step": 4934
    },
    {
      "epoch": 9.105263157894736,
      "grad_norm": 0.5363320708274841,
      "learning_rate": 1.4796000000000002e-05,
      "loss": 0.0966,
      "step": 4935
    },
    {
      "epoch": 9.107109879963065,
      "grad_norm": 0.754754364490509,
      "learning_rate": 1.4799e-05,
      "loss": 0.1319,
      "step": 4936
    },
    {
      "epoch": 9.108956602031395,
      "grad_norm": 0.7564411759376526,
      "learning_rate": 1.4802e-05,
      "loss": 0.1429,
      "step": 4937
    },
    {
      "epoch": 9.110803324099724,
      "grad_norm": 1.0024757385253906,
      "learning_rate": 1.4805e-05,
      "loss": 0.1361,
      "step": 4938
    },
    {
      "epoch": 9.112650046168051,
      "grad_norm": 1.137149691581726,
      "learning_rate": 1.4808e-05,
      "loss": 0.0977,
      "step": 4939
    },
    {
      "epoch": 9.11449676823638,
      "grad_norm": 0.846430242061615,
      "learning_rate": 1.4811000000000001e-05,
      "loss": 0.104,
      "step": 4940
    },
    {
      "epoch": 9.11634349030471,
      "grad_norm": 0.9774792194366455,
      "learning_rate": 1.4814e-05,
      "loss": 0.0959,
      "step": 4941
    },
    {
      "epoch": 9.118190212373039,
      "grad_norm": 0.5531234741210938,
      "learning_rate": 1.4817e-05,
      "loss": 0.0644,
      "step": 4942
    },
    {
      "epoch": 9.120036934441366,
      "grad_norm": 0.7496355175971985,
      "learning_rate": 1.482e-05,
      "loss": 0.0717,
      "step": 4943
    },
    {
      "epoch": 9.121883656509695,
      "grad_norm": 0.5128370523452759,
      "learning_rate": 1.4823e-05,
      "loss": 0.0551,
      "step": 4944
    },
    {
      "epoch": 9.123730378578024,
      "grad_norm": 0.3992457389831543,
      "learning_rate": 1.4826e-05,
      "loss": 0.0504,
      "step": 4945
    },
    {
      "epoch": 9.125577100646353,
      "grad_norm": 0.5101442337036133,
      "learning_rate": 1.4829000000000002e-05,
      "loss": 0.0348,
      "step": 4946
    },
    {
      "epoch": 9.12742382271468,
      "grad_norm": 0.439591646194458,
      "learning_rate": 1.4832000000000001e-05,
      "loss": 0.0311,
      "step": 4947
    },
    {
      "epoch": 9.12927054478301,
      "grad_norm": 0.38307738304138184,
      "learning_rate": 1.4835e-05,
      "loss": 0.0277,
      "step": 4948
    },
    {
      "epoch": 9.131117266851339,
      "grad_norm": 0.49682798981666565,
      "learning_rate": 1.4838e-05,
      "loss": 0.032,
      "step": 4949
    },
    {
      "epoch": 9.132963988919668,
      "grad_norm": 0.4077721834182739,
      "learning_rate": 1.4840999999999999e-05,
      "loss": 0.0248,
      "step": 4950
    },
    {
      "epoch": 9.134810710987995,
      "grad_norm": 0.5288215279579163,
      "learning_rate": 1.4844e-05,
      "loss": 0.039,
      "step": 4951
    },
    {
      "epoch": 9.136657433056325,
      "grad_norm": 0.49088090658187866,
      "learning_rate": 1.4847e-05,
      "loss": 0.0323,
      "step": 4952
    },
    {
      "epoch": 9.138504155124654,
      "grad_norm": 0.45848211646080017,
      "learning_rate": 1.485e-05,
      "loss": 0.0203,
      "step": 4953
    },
    {
      "epoch": 9.140350877192983,
      "grad_norm": 0.7101696133613586,
      "learning_rate": 1.4853e-05,
      "loss": 0.027,
      "step": 4954
    },
    {
      "epoch": 9.142197599261312,
      "grad_norm": 0.31752264499664307,
      "learning_rate": 1.4856e-05,
      "loss": 0.0163,
      "step": 4955
    },
    {
      "epoch": 9.14404432132964,
      "grad_norm": 0.41779816150665283,
      "learning_rate": 1.4859000000000001e-05,
      "loss": 0.0195,
      "step": 4956
    },
    {
      "epoch": 9.145891043397969,
      "grad_norm": 0.36568403244018555,
      "learning_rate": 1.4862000000000001e-05,
      "loss": 0.0212,
      "step": 4957
    },
    {
      "epoch": 9.147737765466298,
      "grad_norm": 0.7210214734077454,
      "learning_rate": 1.4865e-05,
      "loss": 0.0293,
      "step": 4958
    },
    {
      "epoch": 9.149584487534627,
      "grad_norm": 0.30898305773735046,
      "learning_rate": 1.4868e-05,
      "loss": 0.0164,
      "step": 4959
    },
    {
      "epoch": 9.151431209602954,
      "grad_norm": 0.46881088614463806,
      "learning_rate": 1.4871e-05,
      "loss": 0.0285,
      "step": 4960
    },
    {
      "epoch": 9.153277931671283,
      "grad_norm": 0.4806945025920868,
      "learning_rate": 1.4874e-05,
      "loss": 0.0196,
      "step": 4961
    },
    {
      "epoch": 9.155124653739612,
      "grad_norm": 0.7896641492843628,
      "learning_rate": 1.4877e-05,
      "loss": 0.0395,
      "step": 4962
    },
    {
      "epoch": 9.156971375807942,
      "grad_norm": 0.4346265196800232,
      "learning_rate": 1.488e-05,
      "loss": 0.0285,
      "step": 4963
    },
    {
      "epoch": 9.158818097876269,
      "grad_norm": 0.5687569379806519,
      "learning_rate": 1.4883e-05,
      "loss": 0.0194,
      "step": 4964
    },
    {
      "epoch": 9.160664819944598,
      "grad_norm": 0.6780779361724854,
      "learning_rate": 1.4886e-05,
      "loss": 0.034,
      "step": 4965
    },
    {
      "epoch": 9.162511542012927,
      "grad_norm": 0.4466516673564911,
      "learning_rate": 1.4889000000000001e-05,
      "loss": 0.0194,
      "step": 4966
    },
    {
      "epoch": 9.164358264081256,
      "grad_norm": 0.8937951922416687,
      "learning_rate": 1.4892e-05,
      "loss": 0.0461,
      "step": 4967
    },
    {
      "epoch": 9.166204986149584,
      "grad_norm": 0.43590834736824036,
      "learning_rate": 1.4895e-05,
      "loss": 0.0257,
      "step": 4968
    },
    {
      "epoch": 9.168051708217913,
      "grad_norm": 0.6083908677101135,
      "learning_rate": 1.4898e-05,
      "loss": 0.0324,
      "step": 4969
    },
    {
      "epoch": 9.169898430286242,
      "grad_norm": 0.36719781160354614,
      "learning_rate": 1.4901e-05,
      "loss": 0.0204,
      "step": 4970
    },
    {
      "epoch": 9.171745152354571,
      "grad_norm": 0.6315568089485168,
      "learning_rate": 1.4904000000000002e-05,
      "loss": 0.0361,
      "step": 4971
    },
    {
      "epoch": 9.173591874422899,
      "grad_norm": 0.48444539308547974,
      "learning_rate": 1.4907000000000001e-05,
      "loss": 0.0197,
      "step": 4972
    },
    {
      "epoch": 9.175438596491228,
      "grad_norm": 0.7261915802955627,
      "learning_rate": 1.491e-05,
      "loss": 0.0218,
      "step": 4973
    },
    {
      "epoch": 9.177285318559557,
      "grad_norm": 0.6012871861457825,
      "learning_rate": 1.4913e-05,
      "loss": 0.0241,
      "step": 4974
    },
    {
      "epoch": 9.179132040627886,
      "grad_norm": 0.42087167501449585,
      "learning_rate": 1.4915999999999999e-05,
      "loss": 0.0163,
      "step": 4975
    },
    {
      "epoch": 9.180978762696213,
      "grad_norm": 0.5307780504226685,
      "learning_rate": 1.4919e-05,
      "loss": 0.0207,
      "step": 4976
    },
    {
      "epoch": 9.182825484764543,
      "grad_norm": 0.5719494819641113,
      "learning_rate": 1.4922e-05,
      "loss": 0.0255,
      "step": 4977
    },
    {
      "epoch": 9.184672206832872,
      "grad_norm": 0.8112320303916931,
      "learning_rate": 1.4925e-05,
      "loss": 0.0259,
      "step": 4978
    },
    {
      "epoch": 9.1865189289012,
      "grad_norm": 0.9635986685752869,
      "learning_rate": 1.4928e-05,
      "loss": 0.2401,
      "step": 4979
    },
    {
      "epoch": 9.18836565096953,
      "grad_norm": 0.8502519726753235,
      "learning_rate": 1.4931e-05,
      "loss": 0.247,
      "step": 4980
    },
    {
      "epoch": 9.190212373037857,
      "grad_norm": 0.8046050667762756,
      "learning_rate": 1.4934000000000001e-05,
      "loss": 0.2011,
      "step": 4981
    },
    {
      "epoch": 9.192059095106186,
      "grad_norm": 0.6960356831550598,
      "learning_rate": 1.4937000000000001e-05,
      "loss": 0.1636,
      "step": 4982
    },
    {
      "epoch": 9.193905817174516,
      "grad_norm": 0.73581862449646,
      "learning_rate": 1.4940000000000001e-05,
      "loss": 0.144,
      "step": 4983
    },
    {
      "epoch": 9.195752539242845,
      "grad_norm": 0.6557340621948242,
      "learning_rate": 1.4943e-05,
      "loss": 0.1418,
      "step": 4984
    },
    {
      "epoch": 9.197599261311172,
      "grad_norm": 0.6575548648834229,
      "learning_rate": 1.4945999999999999e-05,
      "loss": 0.1428,
      "step": 4985
    },
    {
      "epoch": 9.199445983379501,
      "grad_norm": 0.5466626882553101,
      "learning_rate": 1.4949e-05,
      "loss": 0.1301,
      "step": 4986
    },
    {
      "epoch": 9.20129270544783,
      "grad_norm": 0.5372762084007263,
      "learning_rate": 1.4952e-05,
      "loss": 0.112,
      "step": 4987
    },
    {
      "epoch": 9.20313942751616,
      "grad_norm": 0.9452289938926697,
      "learning_rate": 1.4955e-05,
      "loss": 0.1621,
      "step": 4988
    },
    {
      "epoch": 9.204986149584487,
      "grad_norm": 0.7675982713699341,
      "learning_rate": 1.4958e-05,
      "loss": 0.0883,
      "step": 4989
    },
    {
      "epoch": 9.206832871652816,
      "grad_norm": 0.41777050495147705,
      "learning_rate": 1.4961e-05,
      "loss": 0.0768,
      "step": 4990
    },
    {
      "epoch": 9.208679593721145,
      "grad_norm": 0.6762139797210693,
      "learning_rate": 1.4964000000000001e-05,
      "loss": 0.0669,
      "step": 4991
    },
    {
      "epoch": 9.210526315789474,
      "grad_norm": 0.6928996443748474,
      "learning_rate": 1.4967000000000001e-05,
      "loss": 0.0477,
      "step": 4992
    },
    {
      "epoch": 9.212373037857802,
      "grad_norm": 0.4634210169315338,
      "learning_rate": 1.497e-05,
      "loss": 0.055,
      "step": 4993
    },
    {
      "epoch": 9.21421975992613,
      "grad_norm": 0.3059404492378235,
      "learning_rate": 1.4973e-05,
      "loss": 0.0285,
      "step": 4994
    },
    {
      "epoch": 9.21606648199446,
      "grad_norm": 0.42111411690711975,
      "learning_rate": 1.4976e-05,
      "loss": 0.052,
      "step": 4995
    },
    {
      "epoch": 9.21791320406279,
      "grad_norm": 0.8493423461914062,
      "learning_rate": 1.4979000000000002e-05,
      "loss": 0.0459,
      "step": 4996
    },
    {
      "epoch": 9.219759926131117,
      "grad_norm": 0.6417502164840698,
      "learning_rate": 1.4982e-05,
      "loss": 0.0637,
      "step": 4997
    },
    {
      "epoch": 9.221606648199446,
      "grad_norm": 0.4340676963329315,
      "learning_rate": 1.4985e-05,
      "loss": 0.0339,
      "step": 4998
    },
    {
      "epoch": 9.223453370267775,
      "grad_norm": 0.47145336866378784,
      "learning_rate": 1.4988e-05,
      "loss": 0.0307,
      "step": 4999
    },
    {
      "epoch": 9.225300092336104,
      "grad_norm": 0.5584102869033813,
      "learning_rate": 1.4991e-05,
      "loss": 0.0217,
      "step": 5000
    },
    {
      "epoch": 9.225300092336104,
      "eval_cer": 0.12315628483293153,
      "eval_loss": 0.34040430188179016,
      "eval_runtime": 16.1361,
      "eval_samples_per_second": 18.84,
      "eval_steps_per_second": 0.62,
      "eval_wer": 0.4468534151957022,
      "step": 5000
    },
    {
      "epoch": 9.227146814404431,
      "grad_norm": 0.4516844153404236,
      "learning_rate": 1.4994e-05,
      "loss": 0.0291,
      "step": 5001
    },
    {
      "epoch": 9.22899353647276,
      "grad_norm": 0.40577274560928345,
      "learning_rate": 1.4997e-05,
      "loss": 0.0299,
      "step": 5002
    },
    {
      "epoch": 9.23084025854109,
      "grad_norm": 0.6855767965316772,
      "learning_rate": 1.5e-05,
      "loss": 0.0643,
      "step": 5003
    },
    {
      "epoch": 9.232686980609419,
      "grad_norm": 0.4035259187221527,
      "learning_rate": 1.5003e-05,
      "loss": 0.0304,
      "step": 5004
    },
    {
      "epoch": 9.234533702677748,
      "grad_norm": 0.4719028174877167,
      "learning_rate": 1.5006e-05,
      "loss": 0.0247,
      "step": 5005
    },
    {
      "epoch": 9.236380424746075,
      "grad_norm": 0.4207598865032196,
      "learning_rate": 1.5009e-05,
      "loss": 0.02,
      "step": 5006
    },
    {
      "epoch": 9.238227146814404,
      "grad_norm": 0.34324145317077637,
      "learning_rate": 1.5012e-05,
      "loss": 0.0279,
      "step": 5007
    },
    {
      "epoch": 9.240073868882734,
      "grad_norm": 0.6258143782615662,
      "learning_rate": 1.5015e-05,
      "loss": 0.0685,
      "step": 5008
    },
    {
      "epoch": 9.241920590951063,
      "grad_norm": 0.6973277926445007,
      "learning_rate": 1.5018000000000001e-05,
      "loss": 0.0403,
      "step": 5009
    },
    {
      "epoch": 9.24376731301939,
      "grad_norm": 0.4057953655719757,
      "learning_rate": 1.5021e-05,
      "loss": 0.0283,
      "step": 5010
    },
    {
      "epoch": 9.24561403508772,
      "grad_norm": 0.42994633316993713,
      "learning_rate": 1.5024e-05,
      "loss": 0.0214,
      "step": 5011
    },
    {
      "epoch": 9.247460757156048,
      "grad_norm": 0.5259776711463928,
      "learning_rate": 1.5027e-05,
      "loss": 0.0324,
      "step": 5012
    },
    {
      "epoch": 9.249307479224377,
      "grad_norm": 0.5950026512145996,
      "learning_rate": 1.503e-05,
      "loss": 0.0259,
      "step": 5013
    },
    {
      "epoch": 9.251154201292705,
      "grad_norm": 0.3533867597579956,
      "learning_rate": 1.5033e-05,
      "loss": 0.0127,
      "step": 5014
    },
    {
      "epoch": 9.253000923361034,
      "grad_norm": 0.5182950496673584,
      "learning_rate": 1.5036e-05,
      "loss": 0.025,
      "step": 5015
    },
    {
      "epoch": 9.254847645429363,
      "grad_norm": 0.9183833599090576,
      "learning_rate": 1.5039e-05,
      "loss": 0.0226,
      "step": 5016
    },
    {
      "epoch": 9.256694367497692,
      "grad_norm": 0.35641777515411377,
      "learning_rate": 1.5042e-05,
      "loss": 0.0186,
      "step": 5017
    },
    {
      "epoch": 9.25854108956602,
      "grad_norm": 0.6755897402763367,
      "learning_rate": 1.5044999999999999e-05,
      "loss": 0.0291,
      "step": 5018
    },
    {
      "epoch": 9.260387811634349,
      "grad_norm": 0.5879606008529663,
      "learning_rate": 1.5048000000000002e-05,
      "loss": 0.0334,
      "step": 5019
    },
    {
      "epoch": 9.262234533702678,
      "grad_norm": 0.5946062207221985,
      "learning_rate": 1.5051000000000002e-05,
      "loss": 0.0323,
      "step": 5020
    },
    {
      "epoch": 9.264081255771007,
      "grad_norm": 0.8611308932304382,
      "learning_rate": 1.5054000000000002e-05,
      "loss": 0.0325,
      "step": 5021
    },
    {
      "epoch": 9.265927977839334,
      "grad_norm": 0.6337459683418274,
      "learning_rate": 1.5057e-05,
      "loss": 0.0278,
      "step": 5022
    },
    {
      "epoch": 9.267774699907664,
      "grad_norm": 0.7506858706474304,
      "learning_rate": 1.506e-05,
      "loss": 0.0377,
      "step": 5023
    },
    {
      "epoch": 9.269621421975993,
      "grad_norm": 0.7468025088310242,
      "learning_rate": 1.5063e-05,
      "loss": 0.036,
      "step": 5024
    },
    {
      "epoch": 9.271468144044322,
      "grad_norm": 0.5505267977714539,
      "learning_rate": 1.5066e-05,
      "loss": 0.0298,
      "step": 5025
    },
    {
      "epoch": 9.27331486611265,
      "grad_norm": 0.43939507007598877,
      "learning_rate": 1.5069e-05,
      "loss": 0.0177,
      "step": 5026
    },
    {
      "epoch": 9.275161588180978,
      "grad_norm": 1.4087032079696655,
      "learning_rate": 1.5071999999999999e-05,
      "loss": 0.0352,
      "step": 5027
    },
    {
      "epoch": 9.277008310249307,
      "grad_norm": 0.6867530345916748,
      "learning_rate": 1.5074999999999999e-05,
      "loss": 0.0324,
      "step": 5028
    },
    {
      "epoch": 9.278855032317637,
      "grad_norm": 1.1742939949035645,
      "learning_rate": 1.5078000000000002e-05,
      "loss": 0.3063,
      "step": 5029
    },
    {
      "epoch": 9.280701754385966,
      "grad_norm": 0.9454092979431152,
      "learning_rate": 1.5081000000000002e-05,
      "loss": 0.2511,
      "step": 5030
    },
    {
      "epoch": 9.282548476454293,
      "grad_norm": 1.0500773191452026,
      "learning_rate": 1.5084000000000002e-05,
      "loss": 0.2473,
      "step": 5031
    },
    {
      "epoch": 9.284395198522622,
      "grad_norm": 1.6689544916152954,
      "learning_rate": 1.5087000000000001e-05,
      "loss": 0.2361,
      "step": 5032
    },
    {
      "epoch": 9.286241920590951,
      "grad_norm": 0.8237836360931396,
      "learning_rate": 1.5090000000000001e-05,
      "loss": 0.1845,
      "step": 5033
    },
    {
      "epoch": 9.28808864265928,
      "grad_norm": 0.7189024686813354,
      "learning_rate": 1.5093e-05,
      "loss": 0.1309,
      "step": 5034
    },
    {
      "epoch": 9.289935364727608,
      "grad_norm": 0.8112640976905823,
      "learning_rate": 1.5095999999999999e-05,
      "loss": 0.1582,
      "step": 5035
    },
    {
      "epoch": 9.291782086795937,
      "grad_norm": 0.6921137571334839,
      "learning_rate": 1.5098999999999999e-05,
      "loss": 0.1338,
      "step": 5036
    },
    {
      "epoch": 9.293628808864266,
      "grad_norm": 0.7306265830993652,
      "learning_rate": 1.5101999999999999e-05,
      "loss": 0.148,
      "step": 5037
    },
    {
      "epoch": 9.295475530932595,
      "grad_norm": 0.6113399863243103,
      "learning_rate": 1.5104999999999999e-05,
      "loss": 0.0967,
      "step": 5038
    },
    {
      "epoch": 9.297322253000923,
      "grad_norm": 0.5886443853378296,
      "learning_rate": 1.5108000000000002e-05,
      "loss": 0.1014,
      "step": 5039
    },
    {
      "epoch": 9.299168975069252,
      "grad_norm": 0.617939293384552,
      "learning_rate": 1.5111000000000002e-05,
      "loss": 0.1227,
      "step": 5040
    },
    {
      "epoch": 9.301015697137581,
      "grad_norm": 0.7025123238563538,
      "learning_rate": 1.5114000000000001e-05,
      "loss": 0.0787,
      "step": 5041
    },
    {
      "epoch": 9.30286241920591,
      "grad_norm": 0.34799450635910034,
      "learning_rate": 1.5117000000000001e-05,
      "loss": 0.0315,
      "step": 5042
    },
    {
      "epoch": 9.304709141274238,
      "grad_norm": 0.5829216241836548,
      "learning_rate": 1.5120000000000001e-05,
      "loss": 0.0549,
      "step": 5043
    },
    {
      "epoch": 9.306555863342567,
      "grad_norm": 0.630823016166687,
      "learning_rate": 1.5123e-05,
      "loss": 0.0816,
      "step": 5044
    },
    {
      "epoch": 9.308402585410896,
      "grad_norm": 0.6002022624015808,
      "learning_rate": 1.5126e-05,
      "loss": 0.0489,
      "step": 5045
    },
    {
      "epoch": 9.310249307479225,
      "grad_norm": 0.6191938519477844,
      "learning_rate": 1.5129e-05,
      "loss": 0.0354,
      "step": 5046
    },
    {
      "epoch": 9.312096029547552,
      "grad_norm": 0.45328018069267273,
      "learning_rate": 1.5131999999999998e-05,
      "loss": 0.039,
      "step": 5047
    },
    {
      "epoch": 9.313942751615881,
      "grad_norm": 0.8805258274078369,
      "learning_rate": 1.5134999999999998e-05,
      "loss": 0.0366,
      "step": 5048
    },
    {
      "epoch": 9.31578947368421,
      "grad_norm": 0.7245248556137085,
      "learning_rate": 1.5138000000000001e-05,
      "loss": 0.0715,
      "step": 5049
    },
    {
      "epoch": 9.31763619575254,
      "grad_norm": 0.5289453864097595,
      "learning_rate": 1.5141000000000001e-05,
      "loss": 0.0369,
      "step": 5050
    },
    {
      "epoch": 9.319482917820867,
      "grad_norm": 0.6848105192184448,
      "learning_rate": 1.5144000000000001e-05,
      "loss": 0.0288,
      "step": 5051
    },
    {
      "epoch": 9.321329639889196,
      "grad_norm": 0.41027796268463135,
      "learning_rate": 1.5147e-05,
      "loss": 0.0156,
      "step": 5052
    },
    {
      "epoch": 9.323176361957525,
      "grad_norm": 0.36490383744239807,
      "learning_rate": 1.515e-05,
      "loss": 0.0261,
      "step": 5053
    },
    {
      "epoch": 9.325023084025855,
      "grad_norm": 0.34889253973960876,
      "learning_rate": 1.5153e-05,
      "loss": 0.0243,
      "step": 5054
    },
    {
      "epoch": 9.326869806094184,
      "grad_norm": 0.6471371054649353,
      "learning_rate": 1.5156e-05,
      "loss": 0.0477,
      "step": 5055
    },
    {
      "epoch": 9.328716528162511,
      "grad_norm": 0.4622030556201935,
      "learning_rate": 1.5159e-05,
      "loss": 0.0277,
      "step": 5056
    },
    {
      "epoch": 9.33056325023084,
      "grad_norm": 0.5205875635147095,
      "learning_rate": 1.5162e-05,
      "loss": 0.0272,
      "step": 5057
    },
    {
      "epoch": 9.33240997229917,
      "grad_norm": 0.35904061794281006,
      "learning_rate": 1.5165e-05,
      "loss": 0.0163,
      "step": 5058
    },
    {
      "epoch": 9.334256694367498,
      "grad_norm": 0.6207438707351685,
      "learning_rate": 1.5168000000000001e-05,
      "loss": 0.0328,
      "step": 5059
    },
    {
      "epoch": 9.336103416435826,
      "grad_norm": 0.7448795437812805,
      "learning_rate": 1.5171000000000001e-05,
      "loss": 0.0255,
      "step": 5060
    },
    {
      "epoch": 9.337950138504155,
      "grad_norm": 0.5210355520248413,
      "learning_rate": 1.5174e-05,
      "loss": 0.0203,
      "step": 5061
    },
    {
      "epoch": 9.339796860572484,
      "grad_norm": 0.474969744682312,
      "learning_rate": 1.5177e-05,
      "loss": 0.0191,
      "step": 5062
    },
    {
      "epoch": 9.341643582640813,
      "grad_norm": 0.40252211689949036,
      "learning_rate": 1.518e-05,
      "loss": 0.0244,
      "step": 5063
    },
    {
      "epoch": 9.34349030470914,
      "grad_norm": 0.7463494539260864,
      "learning_rate": 1.5183e-05,
      "loss": 0.0339,
      "step": 5064
    },
    {
      "epoch": 9.34533702677747,
      "grad_norm": 0.6289669275283813,
      "learning_rate": 1.5186e-05,
      "loss": 0.0395,
      "step": 5065
    },
    {
      "epoch": 9.347183748845799,
      "grad_norm": 0.432212769985199,
      "learning_rate": 1.5189e-05,
      "loss": 0.0272,
      "step": 5066
    },
    {
      "epoch": 9.349030470914128,
      "grad_norm": 0.6764055490493774,
      "learning_rate": 1.5192e-05,
      "loss": 0.0319,
      "step": 5067
    },
    {
      "epoch": 9.350877192982455,
      "grad_norm": 0.6974257230758667,
      "learning_rate": 1.5195e-05,
      "loss": 0.0491,
      "step": 5068
    },
    {
      "epoch": 9.352723915050785,
      "grad_norm": 0.46298375725746155,
      "learning_rate": 1.5198000000000003e-05,
      "loss": 0.0181,
      "step": 5069
    },
    {
      "epoch": 9.354570637119114,
      "grad_norm": 0.5116629004478455,
      "learning_rate": 1.5201000000000002e-05,
      "loss": 0.0245,
      "step": 5070
    },
    {
      "epoch": 9.356417359187443,
      "grad_norm": 0.6362098455429077,
      "learning_rate": 1.5204e-05,
      "loss": 0.0248,
      "step": 5071
    },
    {
      "epoch": 9.35826408125577,
      "grad_norm": 0.768278181552887,
      "learning_rate": 1.5207e-05,
      "loss": 0.0259,
      "step": 5072
    },
    {
      "epoch": 9.3601108033241,
      "grad_norm": 0.42540857195854187,
      "learning_rate": 1.521e-05,
      "loss": 0.0279,
      "step": 5073
    },
    {
      "epoch": 9.361957525392429,
      "grad_norm": 0.5822314023971558,
      "learning_rate": 1.5213e-05,
      "loss": 0.0318,
      "step": 5074
    },
    {
      "epoch": 9.363804247460758,
      "grad_norm": 0.5463801026344299,
      "learning_rate": 1.5216e-05,
      "loss": 0.032,
      "step": 5075
    },
    {
      "epoch": 9.365650969529085,
      "grad_norm": 0.7111003994941711,
      "learning_rate": 1.5219e-05,
      "loss": 0.0346,
      "step": 5076
    },
    {
      "epoch": 9.367497691597414,
      "grad_norm": 0.8963141441345215,
      "learning_rate": 1.5222e-05,
      "loss": 0.0408,
      "step": 5077
    },
    {
      "epoch": 9.369344413665743,
      "grad_norm": 1.1584186553955078,
      "learning_rate": 1.5224999999999999e-05,
      "loss": 0.0367,
      "step": 5078
    },
    {
      "epoch": 9.371191135734072,
      "grad_norm": 1.1298359632492065,
      "learning_rate": 1.5228000000000002e-05,
      "loss": 0.3049,
      "step": 5079
    },
    {
      "epoch": 9.373037857802402,
      "grad_norm": 1.229810357093811,
      "learning_rate": 1.5231000000000002e-05,
      "loss": 0.3381,
      "step": 5080
    },
    {
      "epoch": 9.374884579870729,
      "grad_norm": 0.6872143745422363,
      "learning_rate": 1.5234000000000002e-05,
      "loss": 0.1857,
      "step": 5081
    },
    {
      "epoch": 9.376731301939058,
      "grad_norm": 0.7117223143577576,
      "learning_rate": 1.5237000000000002e-05,
      "loss": 0.1947,
      "step": 5082
    },
    {
      "epoch": 9.378578024007387,
      "grad_norm": 0.6277230978012085,
      "learning_rate": 1.524e-05,
      "loss": 0.1643,
      "step": 5083
    },
    {
      "epoch": 9.380424746075716,
      "grad_norm": 0.951104998588562,
      "learning_rate": 1.5243e-05,
      "loss": 0.2928,
      "step": 5084
    },
    {
      "epoch": 9.382271468144044,
      "grad_norm": 1.0074466466903687,
      "learning_rate": 1.5246e-05,
      "loss": 0.1422,
      "step": 5085
    },
    {
      "epoch": 9.384118190212373,
      "grad_norm": 0.9526377320289612,
      "learning_rate": 1.5249e-05,
      "loss": 0.1417,
      "step": 5086
    },
    {
      "epoch": 9.385964912280702,
      "grad_norm": 0.6263598203659058,
      "learning_rate": 1.5251999999999999e-05,
      "loss": 0.1326,
      "step": 5087
    },
    {
      "epoch": 9.387811634349031,
      "grad_norm": 0.7202224731445312,
      "learning_rate": 1.5254999999999999e-05,
      "loss": 0.152,
      "step": 5088
    },
    {
      "epoch": 9.389658356417359,
      "grad_norm": 0.7987571358680725,
      "learning_rate": 1.5258000000000002e-05,
      "loss": 0.1059,
      "step": 5089
    },
    {
      "epoch": 9.391505078485688,
      "grad_norm": 0.8620315790176392,
      "learning_rate": 1.5261000000000002e-05,
      "loss": 0.1033,
      "step": 5090
    },
    {
      "epoch": 9.393351800554017,
      "grad_norm": 0.6651105880737305,
      "learning_rate": 1.5264e-05,
      "loss": 0.1379,
      "step": 5091
    },
    {
      "epoch": 9.395198522622346,
      "grad_norm": 0.8910331130027771,
      "learning_rate": 1.5267e-05,
      "loss": 0.091,
      "step": 5092
    },
    {
      "epoch": 9.397045244690673,
      "grad_norm": 0.7889306545257568,
      "learning_rate": 1.527e-05,
      "loss": 0.0956,
      "step": 5093
    },
    {
      "epoch": 9.398891966759003,
      "grad_norm": 0.45242881774902344,
      "learning_rate": 1.5273e-05,
      "loss": 0.0403,
      "step": 5094
    },
    {
      "epoch": 9.400738688827332,
      "grad_norm": 0.44926634430885315,
      "learning_rate": 1.5276e-05,
      "loss": 0.0407,
      "step": 5095
    },
    {
      "epoch": 9.40258541089566,
      "grad_norm": 0.5183128118515015,
      "learning_rate": 1.5279e-05,
      "loss": 0.0396,
      "step": 5096
    },
    {
      "epoch": 9.404432132963988,
      "grad_norm": 0.7148613333702087,
      "learning_rate": 1.5282e-05,
      "loss": 0.0338,
      "step": 5097
    },
    {
      "epoch": 9.406278855032317,
      "grad_norm": 0.6550760269165039,
      "learning_rate": 1.5285e-05,
      "loss": 0.0405,
      "step": 5098
    },
    {
      "epoch": 9.408125577100646,
      "grad_norm": 0.7605519890785217,
      "learning_rate": 1.5288000000000003e-05,
      "loss": 0.0258,
      "step": 5099
    },
    {
      "epoch": 9.409972299168976,
      "grad_norm": 0.6842150092124939,
      "learning_rate": 1.5291000000000003e-05,
      "loss": 0.0257,
      "step": 5100
    },
    {
      "epoch": 9.411819021237303,
      "grad_norm": 0.653022050857544,
      "learning_rate": 1.5294000000000003e-05,
      "loss": 0.0464,
      "step": 5101
    },
    {
      "epoch": 9.413665743305632,
      "grad_norm": 0.610390841960907,
      "learning_rate": 1.5297e-05,
      "loss": 0.0324,
      "step": 5102
    },
    {
      "epoch": 9.415512465373961,
      "grad_norm": 0.4467082619667053,
      "learning_rate": 1.53e-05,
      "loss": 0.0303,
      "step": 5103
    },
    {
      "epoch": 9.41735918744229,
      "grad_norm": 0.7387998700141907,
      "learning_rate": 1.5303e-05,
      "loss": 0.0335,
      "step": 5104
    },
    {
      "epoch": 9.41920590951062,
      "grad_norm": 0.4913165271282196,
      "learning_rate": 1.5306e-05,
      "loss": 0.0213,
      "step": 5105
    },
    {
      "epoch": 9.421052631578947,
      "grad_norm": 0.8983688950538635,
      "learning_rate": 1.5309e-05,
      "loss": 0.0344,
      "step": 5106
    },
    {
      "epoch": 9.422899353647276,
      "grad_norm": 0.8142847418785095,
      "learning_rate": 1.5312e-05,
      "loss": 0.0212,
      "step": 5107
    },
    {
      "epoch": 9.424746075715605,
      "grad_norm": 0.7110851407051086,
      "learning_rate": 1.5314999999999998e-05,
      "loss": 0.0421,
      "step": 5108
    },
    {
      "epoch": 9.426592797783934,
      "grad_norm": 0.5607290267944336,
      "learning_rate": 1.5318e-05,
      "loss": 0.0224,
      "step": 5109
    },
    {
      "epoch": 9.428439519852262,
      "grad_norm": 0.4886728525161743,
      "learning_rate": 1.5321e-05,
      "loss": 0.0299,
      "step": 5110
    },
    {
      "epoch": 9.43028624192059,
      "grad_norm": 0.5855394005775452,
      "learning_rate": 1.5324e-05,
      "loss": 0.0333,
      "step": 5111
    },
    {
      "epoch": 9.43213296398892,
      "grad_norm": 0.7737128138542175,
      "learning_rate": 1.5327e-05,
      "loss": 0.0223,
      "step": 5112
    },
    {
      "epoch": 9.43397968605725,
      "grad_norm": 0.5125995874404907,
      "learning_rate": 1.533e-05,
      "loss": 0.0392,
      "step": 5113
    },
    {
      "epoch": 9.435826408125576,
      "grad_norm": 0.42056554555892944,
      "learning_rate": 1.5333e-05,
      "loss": 0.0208,
      "step": 5114
    },
    {
      "epoch": 9.437673130193906,
      "grad_norm": 0.7700082659721375,
      "learning_rate": 1.5336e-05,
      "loss": 0.0213,
      "step": 5115
    },
    {
      "epoch": 9.439519852262235,
      "grad_norm": 0.6500101685523987,
      "learning_rate": 1.5339e-05,
      "loss": 0.0304,
      "step": 5116
    },
    {
      "epoch": 9.441366574330564,
      "grad_norm": 0.7322611212730408,
      "learning_rate": 1.5342e-05,
      "loss": 0.043,
      "step": 5117
    },
    {
      "epoch": 9.443213296398891,
      "grad_norm": 1.0362110137939453,
      "learning_rate": 1.5345e-05,
      "loss": 0.034,
      "step": 5118
    },
    {
      "epoch": 9.44506001846722,
      "grad_norm": 0.7446939945220947,
      "learning_rate": 1.5348000000000003e-05,
      "loss": 0.0349,
      "step": 5119
    },
    {
      "epoch": 9.44690674053555,
      "grad_norm": 0.5886105298995972,
      "learning_rate": 1.5351000000000003e-05,
      "loss": 0.0264,
      "step": 5120
    },
    {
      "epoch": 9.448753462603879,
      "grad_norm": 0.48578187823295593,
      "learning_rate": 1.5354000000000002e-05,
      "loss": 0.0172,
      "step": 5121
    },
    {
      "epoch": 9.450600184672206,
      "grad_norm": 0.688308835029602,
      "learning_rate": 1.5357000000000002e-05,
      "loss": 0.0264,
      "step": 5122
    },
    {
      "epoch": 9.452446906740535,
      "grad_norm": 0.5412452816963196,
      "learning_rate": 1.5360000000000002e-05,
      "loss": 0.0138,
      "step": 5123
    },
    {
      "epoch": 9.454293628808864,
      "grad_norm": 0.7046781778335571,
      "learning_rate": 1.5363000000000002e-05,
      "loss": 0.0227,
      "step": 5124
    },
    {
      "epoch": 9.456140350877194,
      "grad_norm": 0.6773200631141663,
      "learning_rate": 1.5366e-05,
      "loss": 0.0427,
      "step": 5125
    },
    {
      "epoch": 9.45798707294552,
      "grad_norm": 1.3261581659317017,
      "learning_rate": 1.5368999999999998e-05,
      "loss": 0.0634,
      "step": 5126
    },
    {
      "epoch": 9.45983379501385,
      "grad_norm": 0.5345895886421204,
      "learning_rate": 1.5371999999999998e-05,
      "loss": 0.0376,
      "step": 5127
    },
    {
      "epoch": 9.46168051708218,
      "grad_norm": 0.697981595993042,
      "learning_rate": 1.5374999999999998e-05,
      "loss": 0.0487,
      "step": 5128
    },
    {
      "epoch": 9.463527239150508,
      "grad_norm": 0.8647690415382385,
      "learning_rate": 1.5377999999999997e-05,
      "loss": 0.2722,
      "step": 5129
    },
    {
      "epoch": 9.465373961218837,
      "grad_norm": 0.9083550572395325,
      "learning_rate": 1.5381e-05,
      "loss": 0.1937,
      "step": 5130
    },
    {
      "epoch": 9.467220683287165,
      "grad_norm": 0.7926074862480164,
      "learning_rate": 1.5384e-05,
      "loss": 0.2162,
      "step": 5131
    },
    {
      "epoch": 9.469067405355494,
      "grad_norm": 0.7760337591171265,
      "learning_rate": 1.5387e-05,
      "loss": 0.2132,
      "step": 5132
    },
    {
      "epoch": 9.470914127423823,
      "grad_norm": 0.8853809237480164,
      "learning_rate": 1.539e-05,
      "loss": 0.1809,
      "step": 5133
    },
    {
      "epoch": 9.472760849492152,
      "grad_norm": 1.001965045928955,
      "learning_rate": 1.5393e-05,
      "loss": 0.1805,
      "step": 5134
    },
    {
      "epoch": 9.47460757156048,
      "grad_norm": 1.8056459426879883,
      "learning_rate": 1.5396e-05,
      "loss": 0.1188,
      "step": 5135
    },
    {
      "epoch": 9.476454293628809,
      "grad_norm": 1.0790168046951294,
      "learning_rate": 1.5399e-05,
      "loss": 0.156,
      "step": 5136
    },
    {
      "epoch": 9.478301015697138,
      "grad_norm": 0.7434194087982178,
      "learning_rate": 1.5402e-05,
      "loss": 0.0947,
      "step": 5137
    },
    {
      "epoch": 9.480147737765467,
      "grad_norm": 0.6929402351379395,
      "learning_rate": 1.5405e-05,
      "loss": 0.0839,
      "step": 5138
    },
    {
      "epoch": 9.481994459833794,
      "grad_norm": 0.8843145966529846,
      "learning_rate": 1.5408e-05,
      "loss": 0.1342,
      "step": 5139
    },
    {
      "epoch": 9.483841181902124,
      "grad_norm": 0.37170612812042236,
      "learning_rate": 1.5411000000000002e-05,
      "loss": 0.0424,
      "step": 5140
    },
    {
      "epoch": 9.485687903970453,
      "grad_norm": 0.4109022617340088,
      "learning_rate": 1.5414000000000002e-05,
      "loss": 0.0548,
      "step": 5141
    },
    {
      "epoch": 9.487534626038782,
      "grad_norm": 0.4736909568309784,
      "learning_rate": 1.5417e-05,
      "loss": 0.0587,
      "step": 5142
    },
    {
      "epoch": 9.48938134810711,
      "grad_norm": 0.49091270565986633,
      "learning_rate": 1.542e-05,
      "loss": 0.0496,
      "step": 5143
    },
    {
      "epoch": 9.491228070175438,
      "grad_norm": 0.5085435509681702,
      "learning_rate": 1.5423e-05,
      "loss": 0.0316,
      "step": 5144
    },
    {
      "epoch": 9.493074792243767,
      "grad_norm": 0.350394606590271,
      "learning_rate": 1.5426e-05,
      "loss": 0.0277,
      "step": 5145
    },
    {
      "epoch": 9.494921514312097,
      "grad_norm": 0.6892826557159424,
      "learning_rate": 1.5429e-05,
      "loss": 0.0322,
      "step": 5146
    },
    {
      "epoch": 9.496768236380424,
      "grad_norm": 0.6446825861930847,
      "learning_rate": 1.5432e-05,
      "loss": 0.0246,
      "step": 5147
    },
    {
      "epoch": 9.498614958448753,
      "grad_norm": 0.79939204454422,
      "learning_rate": 1.5435e-05,
      "loss": 0.0732,
      "step": 5148
    },
    {
      "epoch": 9.500461680517082,
      "grad_norm": 0.6500365734100342,
      "learning_rate": 1.5438e-05,
      "loss": 0.0336,
      "step": 5149
    },
    {
      "epoch": 9.502308402585411,
      "grad_norm": 0.5178731679916382,
      "learning_rate": 1.5441000000000003e-05,
      "loss": 0.0348,
      "step": 5150
    },
    {
      "epoch": 9.504155124653739,
      "grad_norm": 0.3989321291446686,
      "learning_rate": 1.5444e-05,
      "loss": 0.018,
      "step": 5151
    },
    {
      "epoch": 9.506001846722068,
      "grad_norm": 0.9351484775543213,
      "learning_rate": 1.5447e-05,
      "loss": 0.0313,
      "step": 5152
    },
    {
      "epoch": 9.507848568790397,
      "grad_norm": 0.5877683162689209,
      "learning_rate": 1.545e-05,
      "loss": 0.0372,
      "step": 5153
    },
    {
      "epoch": 9.509695290858726,
      "grad_norm": 0.46480149030685425,
      "learning_rate": 1.5453e-05,
      "loss": 0.025,
      "step": 5154
    },
    {
      "epoch": 9.511542012927055,
      "grad_norm": 0.7051569223403931,
      "learning_rate": 1.5456e-05,
      "loss": 0.0375,
      "step": 5155
    },
    {
      "epoch": 9.513388734995383,
      "grad_norm": 0.5951721668243408,
      "learning_rate": 1.5459e-05,
      "loss": 0.0347,
      "step": 5156
    },
    {
      "epoch": 9.515235457063712,
      "grad_norm": 0.671683669090271,
      "learning_rate": 1.5462e-05,
      "loss": 0.0437,
      "step": 5157
    },
    {
      "epoch": 9.517082179132041,
      "grad_norm": 0.6410195231437683,
      "learning_rate": 1.5465e-05,
      "loss": 0.0548,
      "step": 5158
    },
    {
      "epoch": 9.51892890120037,
      "grad_norm": 0.5157730579376221,
      "learning_rate": 1.5467999999999998e-05,
      "loss": 0.0316,
      "step": 5159
    },
    {
      "epoch": 9.520775623268698,
      "grad_norm": 0.5009844303131104,
      "learning_rate": 1.5471e-05,
      "loss": 0.0218,
      "step": 5160
    },
    {
      "epoch": 9.522622345337027,
      "grad_norm": 0.39921513199806213,
      "learning_rate": 1.5474e-05,
      "loss": 0.0157,
      "step": 5161
    },
    {
      "epoch": 9.524469067405356,
      "grad_norm": 0.5840036273002625,
      "learning_rate": 1.5477e-05,
      "loss": 0.0308,
      "step": 5162
    },
    {
      "epoch": 9.526315789473685,
      "grad_norm": 0.5300602912902832,
      "learning_rate": 1.548e-05,
      "loss": 0.0255,
      "step": 5163
    },
    {
      "epoch": 9.528162511542012,
      "grad_norm": 0.5056509971618652,
      "learning_rate": 1.5483e-05,
      "loss": 0.0343,
      "step": 5164
    },
    {
      "epoch": 9.530009233610341,
      "grad_norm": 0.4027712941169739,
      "learning_rate": 1.5486e-05,
      "loss": 0.0291,
      "step": 5165
    },
    {
      "epoch": 9.53185595567867,
      "grad_norm": 0.46333715319633484,
      "learning_rate": 1.5489e-05,
      "loss": 0.0233,
      "step": 5166
    },
    {
      "epoch": 9.533702677747,
      "grad_norm": 0.651351273059845,
      "learning_rate": 1.5492e-05,
      "loss": 0.0306,
      "step": 5167
    },
    {
      "epoch": 9.535549399815327,
      "grad_norm": 0.5864402055740356,
      "learning_rate": 1.5495e-05,
      "loss": 0.023,
      "step": 5168
    },
    {
      "epoch": 9.537396121883656,
      "grad_norm": 0.5296105742454529,
      "learning_rate": 1.5498e-05,
      "loss": 0.0232,
      "step": 5169
    },
    {
      "epoch": 9.539242843951985,
      "grad_norm": 0.4870733916759491,
      "learning_rate": 1.5501000000000003e-05,
      "loss": 0.0235,
      "step": 5170
    },
    {
      "epoch": 9.541089566020315,
      "grad_norm": 0.5141538977622986,
      "learning_rate": 1.5504000000000003e-05,
      "loss": 0.0275,
      "step": 5171
    },
    {
      "epoch": 9.542936288088642,
      "grad_norm": 0.4707508683204651,
      "learning_rate": 1.5507000000000002e-05,
      "loss": 0.0225,
      "step": 5172
    },
    {
      "epoch": 9.544783010156971,
      "grad_norm": 0.8089979887008667,
      "learning_rate": 1.5510000000000002e-05,
      "loss": 0.0361,
      "step": 5173
    },
    {
      "epoch": 9.5466297322253,
      "grad_norm": 0.5475485920906067,
      "learning_rate": 1.5513000000000002e-05,
      "loss": 0.0279,
      "step": 5174
    },
    {
      "epoch": 9.54847645429363,
      "grad_norm": 2.153735399246216,
      "learning_rate": 1.5516000000000002e-05,
      "loss": 0.0351,
      "step": 5175
    },
    {
      "epoch": 9.550323176361957,
      "grad_norm": 1.0369545221328735,
      "learning_rate": 1.5518999999999998e-05,
      "loss": 0.0216,
      "step": 5176
    },
    {
      "epoch": 9.552169898430286,
      "grad_norm": 0.6523640751838684,
      "learning_rate": 1.5521999999999998e-05,
      "loss": 0.0243,
      "step": 5177
    },
    {
      "epoch": 9.554016620498615,
      "grad_norm": 0.887344479560852,
      "learning_rate": 1.5524999999999998e-05,
      "loss": 0.0567,
      "step": 5178
    },
    {
      "epoch": 9.555863342566944,
      "grad_norm": 1.1368447542190552,
      "learning_rate": 1.5527999999999998e-05,
      "loss": 0.2774,
      "step": 5179
    },
    {
      "epoch": 9.557710064635273,
      "grad_norm": 0.9110034704208374,
      "learning_rate": 1.5531e-05,
      "loss": 0.254,
      "step": 5180
    },
    {
      "epoch": 9.5595567867036,
      "grad_norm": 0.8574900031089783,
      "learning_rate": 1.5534e-05,
      "loss": 0.1996,
      "step": 5181
    },
    {
      "epoch": 9.56140350877193,
      "grad_norm": 0.7445184588432312,
      "learning_rate": 1.5537e-05,
      "loss": 0.1832,
      "step": 5182
    },
    {
      "epoch": 9.563250230840259,
      "grad_norm": 1.266960620880127,
      "learning_rate": 1.554e-05,
      "loss": 0.2373,
      "step": 5183
    },
    {
      "epoch": 9.565096952908588,
      "grad_norm": 0.8766116499900818,
      "learning_rate": 1.5543e-05,
      "loss": 0.1329,
      "step": 5184
    },
    {
      "epoch": 9.566943674976915,
      "grad_norm": 1.0444930791854858,
      "learning_rate": 1.5546e-05,
      "loss": 0.1283,
      "step": 5185
    },
    {
      "epoch": 9.568790397045245,
      "grad_norm": 0.7388608455657959,
      "learning_rate": 1.5549e-05,
      "loss": 0.1423,
      "step": 5186
    },
    {
      "epoch": 9.570637119113574,
      "grad_norm": 0.6338157057762146,
      "learning_rate": 1.5552e-05,
      "loss": 0.1518,
      "step": 5187
    },
    {
      "epoch": 9.572483841181903,
      "grad_norm": 0.5952925682067871,
      "learning_rate": 1.5555e-05,
      "loss": 0.1124,
      "step": 5188
    },
    {
      "epoch": 9.57433056325023,
      "grad_norm": 0.7095077037811279,
      "learning_rate": 1.5558e-05,
      "loss": 0.1288,
      "step": 5189
    },
    {
      "epoch": 9.57617728531856,
      "grad_norm": 0.5818418860435486,
      "learning_rate": 1.5561000000000002e-05,
      "loss": 0.0931,
      "step": 5190
    },
    {
      "epoch": 9.578024007386889,
      "grad_norm": 0.8665828704833984,
      "learning_rate": 1.5564000000000002e-05,
      "loss": 0.0822,
      "step": 5191
    },
    {
      "epoch": 9.579870729455218,
      "grad_norm": 0.9548938274383545,
      "learning_rate": 1.5567000000000002e-05,
      "loss": 0.0608,
      "step": 5192
    },
    {
      "epoch": 9.581717451523545,
      "grad_norm": 0.4729214310646057,
      "learning_rate": 1.5570000000000002e-05,
      "loss": 0.064,
      "step": 5193
    },
    {
      "epoch": 9.583564173591874,
      "grad_norm": 0.33552882075309753,
      "learning_rate": 1.5573e-05,
      "loss": 0.0273,
      "step": 5194
    },
    {
      "epoch": 9.585410895660203,
      "grad_norm": 0.8147560358047485,
      "learning_rate": 1.5576e-05,
      "loss": 0.0875,
      "step": 5195
    },
    {
      "epoch": 9.587257617728532,
      "grad_norm": 0.4922485947608948,
      "learning_rate": 1.5579e-05,
      "loss": 0.0491,
      "step": 5196
    },
    {
      "epoch": 9.58910433979686,
      "grad_norm": 3.5064287185668945,
      "learning_rate": 1.5582e-05,
      "loss": 0.0954,
      "step": 5197
    },
    {
      "epoch": 9.590951061865189,
      "grad_norm": 0.40575841069221497,
      "learning_rate": 1.5585e-05,
      "loss": 0.0368,
      "step": 5198
    },
    {
      "epoch": 9.592797783933518,
      "grad_norm": 0.425672322511673,
      "learning_rate": 1.5588e-05,
      "loss": 0.0279,
      "step": 5199
    },
    {
      "epoch": 9.594644506001847,
      "grad_norm": 0.4247608780860901,
      "learning_rate": 1.5591e-05,
      "loss": 0.0393,
      "step": 5200
    },
    {
      "epoch": 9.596491228070175,
      "grad_norm": 0.7301235795021057,
      "learning_rate": 1.5594e-05,
      "loss": 0.0619,
      "step": 5201
    },
    {
      "epoch": 9.598337950138504,
      "grad_norm": 0.4044654667377472,
      "learning_rate": 1.5597e-05,
      "loss": 0.0289,
      "step": 5202
    },
    {
      "epoch": 9.600184672206833,
      "grad_norm": 0.36088332533836365,
      "learning_rate": 1.56e-05,
      "loss": 0.0129,
      "step": 5203
    },
    {
      "epoch": 9.602031394275162,
      "grad_norm": 0.6777994632720947,
      "learning_rate": 1.5603e-05,
      "loss": 0.0395,
      "step": 5204
    },
    {
      "epoch": 9.603878116343491,
      "grad_norm": 0.2510547935962677,
      "learning_rate": 1.5606e-05,
      "loss": 0.0172,
      "step": 5205
    },
    {
      "epoch": 9.605724838411819,
      "grad_norm": 0.5264256596565247,
      "learning_rate": 1.5609e-05,
      "loss": 0.0282,
      "step": 5206
    },
    {
      "epoch": 9.607571560480148,
      "grad_norm": 0.5090929865837097,
      "learning_rate": 1.5612e-05,
      "loss": 0.0206,
      "step": 5207
    },
    {
      "epoch": 9.609418282548477,
      "grad_norm": 0.6527032852172852,
      "learning_rate": 1.5615e-05,
      "loss": 0.0314,
      "step": 5208
    },
    {
      "epoch": 9.611265004616806,
      "grad_norm": 0.5314533114433289,
      "learning_rate": 1.5618e-05,
      "loss": 0.0261,
      "step": 5209
    },
    {
      "epoch": 9.613111726685133,
      "grad_norm": 0.47839510440826416,
      "learning_rate": 1.5621000000000002e-05,
      "loss": 0.0234,
      "step": 5210
    },
    {
      "epoch": 9.614958448753463,
      "grad_norm": 0.5687962174415588,
      "learning_rate": 1.5624e-05,
      "loss": 0.0387,
      "step": 5211
    },
    {
      "epoch": 9.616805170821792,
      "grad_norm": 0.3908262848854065,
      "learning_rate": 1.5627e-05,
      "loss": 0.0175,
      "step": 5212
    },
    {
      "epoch": 9.61865189289012,
      "grad_norm": 0.4974033534526825,
      "learning_rate": 1.563e-05,
      "loss": 0.0267,
      "step": 5213
    },
    {
      "epoch": 9.620498614958448,
      "grad_norm": 0.7582276463508606,
      "learning_rate": 1.5633e-05,
      "loss": 0.0294,
      "step": 5214
    },
    {
      "epoch": 9.622345337026777,
      "grad_norm": 0.5293557047843933,
      "learning_rate": 1.5636e-05,
      "loss": 0.0246,
      "step": 5215
    },
    {
      "epoch": 9.624192059095106,
      "grad_norm": 0.5568930506706238,
      "learning_rate": 1.5639e-05,
      "loss": 0.0615,
      "step": 5216
    },
    {
      "epoch": 9.626038781163436,
      "grad_norm": 0.46155261993408203,
      "learning_rate": 1.5642e-05,
      "loss": 0.0252,
      "step": 5217
    },
    {
      "epoch": 9.627885503231763,
      "grad_norm": 0.7104062438011169,
      "learning_rate": 1.5645e-05,
      "loss": 0.0299,
      "step": 5218
    },
    {
      "epoch": 9.629732225300092,
      "grad_norm": 0.4467885196208954,
      "learning_rate": 1.5648e-05,
      "loss": 0.028,
      "step": 5219
    },
    {
      "epoch": 9.631578947368421,
      "grad_norm": 1.1483324766159058,
      "learning_rate": 1.5651000000000003e-05,
      "loss": 0.0344,
      "step": 5220
    },
    {
      "epoch": 9.63342566943675,
      "grad_norm": 0.6257169246673584,
      "learning_rate": 1.5654000000000003e-05,
      "loss": 0.0225,
      "step": 5221
    },
    {
      "epoch": 9.635272391505078,
      "grad_norm": 0.7972967028617859,
      "learning_rate": 1.5657000000000003e-05,
      "loss": 0.0343,
      "step": 5222
    },
    {
      "epoch": 9.637119113573407,
      "grad_norm": 0.5291518568992615,
      "learning_rate": 1.5660000000000003e-05,
      "loss": 0.0166,
      "step": 5223
    },
    {
      "epoch": 9.638965835641736,
      "grad_norm": 0.8691847324371338,
      "learning_rate": 1.5663000000000002e-05,
      "loss": 0.0359,
      "step": 5224
    },
    {
      "epoch": 9.640812557710065,
      "grad_norm": 0.5594950914382935,
      "learning_rate": 1.5666e-05,
      "loss": 0.0162,
      "step": 5225
    },
    {
      "epoch": 9.642659279778393,
      "grad_norm": 0.5064205527305603,
      "learning_rate": 1.5669e-05,
      "loss": 0.025,
      "step": 5226
    },
    {
      "epoch": 9.644506001846722,
      "grad_norm": 0.5738297700881958,
      "learning_rate": 1.5672e-05,
      "loss": 0.019,
      "step": 5227
    },
    {
      "epoch": 9.64635272391505,
      "grad_norm": 0.6175197958946228,
      "learning_rate": 1.5674999999999998e-05,
      "loss": 0.0538,
      "step": 5228
    },
    {
      "epoch": 9.64819944598338,
      "grad_norm": 2.0463693141937256,
      "learning_rate": 1.5677999999999998e-05,
      "loss": 0.3066,
      "step": 5229
    },
    {
      "epoch": 9.65004616805171,
      "grad_norm": 0.8259283304214478,
      "learning_rate": 1.5681e-05,
      "loss": 0.2584,
      "step": 5230
    },
    {
      "epoch": 9.651892890120036,
      "grad_norm": 0.6895362138748169,
      "learning_rate": 1.5684e-05,
      "loss": 0.191,
      "step": 5231
    },
    {
      "epoch": 9.653739612188366,
      "grad_norm": 0.6375547051429749,
      "learning_rate": 1.5687e-05,
      "loss": 0.1583,
      "step": 5232
    },
    {
      "epoch": 9.655586334256695,
      "grad_norm": 0.8149611353874207,
      "learning_rate": 1.569e-05,
      "loss": 0.1589,
      "step": 5233
    },
    {
      "epoch": 9.657433056325024,
      "grad_norm": 1.1168636083602905,
      "learning_rate": 1.5693e-05,
      "loss": 0.1955,
      "step": 5234
    },
    {
      "epoch": 9.659279778393351,
      "grad_norm": 1.1383134126663208,
      "learning_rate": 1.5696e-05,
      "loss": 0.182,
      "step": 5235
    },
    {
      "epoch": 9.66112650046168,
      "grad_norm": 0.8388357162475586,
      "learning_rate": 1.5699e-05,
      "loss": 0.138,
      "step": 5236
    },
    {
      "epoch": 9.66297322253001,
      "grad_norm": 0.8659719824790955,
      "learning_rate": 1.5702e-05,
      "loss": 0.1609,
      "step": 5237
    },
    {
      "epoch": 9.664819944598339,
      "grad_norm": 0.7714173793792725,
      "learning_rate": 1.5705e-05,
      "loss": 0.1067,
      "step": 5238
    },
    {
      "epoch": 9.666666666666666,
      "grad_norm": 0.5207611918449402,
      "learning_rate": 1.5708e-05,
      "loss": 0.0698,
      "step": 5239
    },
    {
      "epoch": 9.668513388734995,
      "grad_norm": 0.6840134859085083,
      "learning_rate": 1.5711000000000003e-05,
      "loss": 0.1231,
      "step": 5240
    },
    {
      "epoch": 9.670360110803324,
      "grad_norm": 1.1387262344360352,
      "learning_rate": 1.5714000000000002e-05,
      "loss": 0.1607,
      "step": 5241
    },
    {
      "epoch": 9.672206832871654,
      "grad_norm": 1.182800531387329,
      "learning_rate": 1.5717000000000002e-05,
      "loss": 0.093,
      "step": 5242
    },
    {
      "epoch": 9.67405355493998,
      "grad_norm": 0.5864893198013306,
      "learning_rate": 1.5720000000000002e-05,
      "loss": 0.0594,
      "step": 5243
    },
    {
      "epoch": 9.67590027700831,
      "grad_norm": 0.48163750767707825,
      "learning_rate": 1.5723000000000002e-05,
      "loss": 0.0634,
      "step": 5244
    },
    {
      "epoch": 9.67774699907664,
      "grad_norm": 0.5190600752830505,
      "learning_rate": 1.5726e-05,
      "loss": 0.0463,
      "step": 5245
    },
    {
      "epoch": 9.679593721144968,
      "grad_norm": 0.6937405467033386,
      "learning_rate": 1.5729e-05,
      "loss": 0.0225,
      "step": 5246
    },
    {
      "epoch": 9.681440443213296,
      "grad_norm": 0.33243173360824585,
      "learning_rate": 1.5732e-05,
      "loss": 0.0162,
      "step": 5247
    },
    {
      "epoch": 9.683287165281625,
      "grad_norm": 0.8670083284378052,
      "learning_rate": 1.5735e-05,
      "loss": 0.0254,
      "step": 5248
    },
    {
      "epoch": 9.685133887349954,
      "grad_norm": 0.6444321274757385,
      "learning_rate": 1.5737999999999997e-05,
      "loss": 0.027,
      "step": 5249
    },
    {
      "epoch": 9.686980609418283,
      "grad_norm": 0.6079081296920776,
      "learning_rate": 1.5741e-05,
      "loss": 0.048,
      "step": 5250
    },
    {
      "epoch": 9.68882733148661,
      "grad_norm": 0.6462955474853516,
      "learning_rate": 1.5744e-05,
      "loss": 0.0408,
      "step": 5251
    },
    {
      "epoch": 9.69067405355494,
      "grad_norm": 0.5033811330795288,
      "learning_rate": 1.5747e-05,
      "loss": 0.0309,
      "step": 5252
    },
    {
      "epoch": 9.692520775623269,
      "grad_norm": 0.43178778886795044,
      "learning_rate": 1.575e-05,
      "loss": 0.0221,
      "step": 5253
    },
    {
      "epoch": 9.694367497691598,
      "grad_norm": 0.7185611724853516,
      "learning_rate": 1.5753e-05,
      "loss": 0.0385,
      "step": 5254
    },
    {
      "epoch": 9.696214219759927,
      "grad_norm": 0.47433245182037354,
      "learning_rate": 1.5756e-05,
      "loss": 0.0174,
      "step": 5255
    },
    {
      "epoch": 9.698060941828254,
      "grad_norm": 0.3177076578140259,
      "learning_rate": 1.5759e-05,
      "loss": 0.0194,
      "step": 5256
    },
    {
      "epoch": 9.699907663896584,
      "grad_norm": 0.9755005836486816,
      "learning_rate": 1.5762e-05,
      "loss": 0.0293,
      "step": 5257
    },
    {
      "epoch": 9.701754385964913,
      "grad_norm": 0.418106347322464,
      "learning_rate": 1.5765e-05,
      "loss": 0.0269,
      "step": 5258
    },
    {
      "epoch": 9.703601108033242,
      "grad_norm": 0.6220974922180176,
      "learning_rate": 1.5768e-05,
      "loss": 0.0296,
      "step": 5259
    },
    {
      "epoch": 9.70544783010157,
      "grad_norm": 0.46836185455322266,
      "learning_rate": 1.5771e-05,
      "loss": 0.0562,
      "step": 5260
    },
    {
      "epoch": 9.707294552169898,
      "grad_norm": 0.5895121097564697,
      "learning_rate": 1.5774000000000002e-05,
      "loss": 0.0205,
      "step": 5261
    },
    {
      "epoch": 9.709141274238227,
      "grad_norm": 0.7849803566932678,
      "learning_rate": 1.5777e-05,
      "loss": 0.0263,
      "step": 5262
    },
    {
      "epoch": 9.710987996306557,
      "grad_norm": 0.47644442319869995,
      "learning_rate": 1.578e-05,
      "loss": 0.0294,
      "step": 5263
    },
    {
      "epoch": 9.712834718374884,
      "grad_norm": 0.5167056322097778,
      "learning_rate": 1.5783e-05,
      "loss": 0.0344,
      "step": 5264
    },
    {
      "epoch": 9.714681440443213,
      "grad_norm": 0.6035995483398438,
      "learning_rate": 1.5786e-05,
      "loss": 0.0229,
      "step": 5265
    },
    {
      "epoch": 9.716528162511542,
      "grad_norm": 0.35058900713920593,
      "learning_rate": 1.5789e-05,
      "loss": 0.0231,
      "step": 5266
    },
    {
      "epoch": 9.718374884579871,
      "grad_norm": 0.4649684429168701,
      "learning_rate": 1.5792e-05,
      "loss": 0.0298,
      "step": 5267
    },
    {
      "epoch": 9.720221606648199,
      "grad_norm": 0.44932129979133606,
      "learning_rate": 1.5795e-05,
      "loss": 0.0292,
      "step": 5268
    },
    {
      "epoch": 9.722068328716528,
      "grad_norm": 0.5442031025886536,
      "learning_rate": 1.5798e-05,
      "loss": 0.0238,
      "step": 5269
    },
    {
      "epoch": 9.723915050784857,
      "grad_norm": 0.40109357237815857,
      "learning_rate": 1.5801e-05,
      "loss": 0.0245,
      "step": 5270
    },
    {
      "epoch": 9.725761772853186,
      "grad_norm": 0.7246041297912598,
      "learning_rate": 1.5804000000000003e-05,
      "loss": 0.0334,
      "step": 5271
    },
    {
      "epoch": 9.727608494921514,
      "grad_norm": 0.42517438530921936,
      "learning_rate": 1.5807000000000003e-05,
      "loss": 0.0216,
      "step": 5272
    },
    {
      "epoch": 9.729455216989843,
      "grad_norm": 0.7946247458457947,
      "learning_rate": 1.5810000000000003e-05,
      "loss": 0.0425,
      "step": 5273
    },
    {
      "epoch": 9.731301939058172,
      "grad_norm": 0.7544342875480652,
      "learning_rate": 1.5813e-05,
      "loss": 0.0392,
      "step": 5274
    },
    {
      "epoch": 9.733148661126501,
      "grad_norm": 0.5644515752792358,
      "learning_rate": 1.5816e-05,
      "loss": 0.0273,
      "step": 5275
    },
    {
      "epoch": 9.734995383194828,
      "grad_norm": 0.614993155002594,
      "learning_rate": 1.5819e-05,
      "loss": 0.0538,
      "step": 5276
    },
    {
      "epoch": 9.736842105263158,
      "grad_norm": 1.1540549993515015,
      "learning_rate": 1.5822e-05,
      "loss": 0.0494,
      "step": 5277
    },
    {
      "epoch": 9.738688827331487,
      "grad_norm": 0.692342221736908,
      "learning_rate": 1.5825e-05,
      "loss": 0.0325,
      "step": 5278
    },
    {
      "epoch": 9.740535549399816,
      "grad_norm": 1.2595493793487549,
      "learning_rate": 1.5827999999999998e-05,
      "loss": 0.3012,
      "step": 5279
    },
    {
      "epoch": 9.742382271468145,
      "grad_norm": 0.8239694833755493,
      "learning_rate": 1.5830999999999998e-05,
      "loss": 0.239,
      "step": 5280
    },
    {
      "epoch": 9.744228993536472,
      "grad_norm": 0.7055354118347168,
      "learning_rate": 1.5834e-05,
      "loss": 0.2103,
      "step": 5281
    },
    {
      "epoch": 9.746075715604801,
      "grad_norm": 0.6928851008415222,
      "learning_rate": 1.5837e-05,
      "loss": 0.1638,
      "step": 5282
    },
    {
      "epoch": 9.74792243767313,
      "grad_norm": 1.6198242902755737,
      "learning_rate": 1.584e-05,
      "loss": 0.1653,
      "step": 5283
    },
    {
      "epoch": 9.749769159741458,
      "grad_norm": 1.1449764966964722,
      "learning_rate": 1.5843e-05,
      "loss": 0.2131,
      "step": 5284
    },
    {
      "epoch": 9.751615881809787,
      "grad_norm": 0.7790804505348206,
      "learning_rate": 1.5846e-05,
      "loss": 0.1631,
      "step": 5285
    },
    {
      "epoch": 9.753462603878116,
      "grad_norm": 0.7726112008094788,
      "learning_rate": 1.5849e-05,
      "loss": 0.1724,
      "step": 5286
    },
    {
      "epoch": 9.755309325946445,
      "grad_norm": 0.742899477481842,
      "learning_rate": 1.5852e-05,
      "loss": 0.1177,
      "step": 5287
    },
    {
      "epoch": 9.757156048014775,
      "grad_norm": 1.6110416650772095,
      "learning_rate": 1.5855e-05,
      "loss": 0.1097,
      "step": 5288
    },
    {
      "epoch": 9.759002770083102,
      "grad_norm": 0.6606073975563049,
      "learning_rate": 1.5858e-05,
      "loss": 0.0952,
      "step": 5289
    },
    {
      "epoch": 9.760849492151431,
      "grad_norm": 1.041293740272522,
      "learning_rate": 1.5861e-05,
      "loss": 0.0871,
      "step": 5290
    },
    {
      "epoch": 9.76269621421976,
      "grad_norm": 1.2442162036895752,
      "learning_rate": 1.5864000000000003e-05,
      "loss": 0.1605,
      "step": 5291
    },
    {
      "epoch": 9.76454293628809,
      "grad_norm": 0.8230966329574585,
      "learning_rate": 1.5867000000000002e-05,
      "loss": 0.101,
      "step": 5292
    },
    {
      "epoch": 9.766389658356417,
      "grad_norm": 0.45017313957214355,
      "learning_rate": 1.5870000000000002e-05,
      "loss": 0.0534,
      "step": 5293
    },
    {
      "epoch": 9.768236380424746,
      "grad_norm": 0.5587928295135498,
      "learning_rate": 1.5873000000000002e-05,
      "loss": 0.0422,
      "step": 5294
    },
    {
      "epoch": 9.770083102493075,
      "grad_norm": 0.41939765214920044,
      "learning_rate": 1.5876000000000002e-05,
      "loss": 0.0211,
      "step": 5295
    },
    {
      "epoch": 9.771929824561404,
      "grad_norm": 1.1486107110977173,
      "learning_rate": 1.5879e-05,
      "loss": 0.0782,
      "step": 5296
    },
    {
      "epoch": 9.773776546629731,
      "grad_norm": 0.740656852722168,
      "learning_rate": 1.5882e-05,
      "loss": 0.0474,
      "step": 5297
    },
    {
      "epoch": 9.77562326869806,
      "grad_norm": 0.46551042795181274,
      "learning_rate": 1.5884999999999998e-05,
      "loss": 0.0435,
      "step": 5298
    },
    {
      "epoch": 9.77746999076639,
      "grad_norm": 0.5009204745292664,
      "learning_rate": 1.5887999999999998e-05,
      "loss": 0.0341,
      "step": 5299
    },
    {
      "epoch": 9.779316712834719,
      "grad_norm": 0.34932154417037964,
      "learning_rate": 1.5890999999999997e-05,
      "loss": 0.0348,
      "step": 5300
    },
    {
      "epoch": 9.781163434903046,
      "grad_norm": 0.5163803696632385,
      "learning_rate": 1.5894e-05,
      "loss": 0.0389,
      "step": 5301
    },
    {
      "epoch": 9.783010156971375,
      "grad_norm": 0.5265400409698486,
      "learning_rate": 1.5897e-05,
      "loss": 0.0258,
      "step": 5302
    },
    {
      "epoch": 9.784856879039705,
      "grad_norm": 0.6005827784538269,
      "learning_rate": 1.59e-05,
      "loss": 0.0322,
      "step": 5303
    },
    {
      "epoch": 9.786703601108034,
      "grad_norm": 0.4704049825668335,
      "learning_rate": 1.5903e-05,
      "loss": 0.0343,
      "step": 5304
    },
    {
      "epoch": 9.788550323176363,
      "grad_norm": 0.4133896827697754,
      "learning_rate": 1.5906e-05,
      "loss": 0.0246,
      "step": 5305
    },
    {
      "epoch": 9.79039704524469,
      "grad_norm": 0.7306255102157593,
      "learning_rate": 1.5909e-05,
      "loss": 0.0306,
      "step": 5306
    },
    {
      "epoch": 9.79224376731302,
      "grad_norm": 1.2488560676574707,
      "learning_rate": 1.5912e-05,
      "loss": 0.0696,
      "step": 5307
    },
    {
      "epoch": 9.794090489381349,
      "grad_norm": 0.40144163370132446,
      "learning_rate": 1.5915e-05,
      "loss": 0.0272,
      "step": 5308
    },
    {
      "epoch": 9.795937211449676,
      "grad_norm": 0.4051397442817688,
      "learning_rate": 1.5918e-05,
      "loss": 0.0233,
      "step": 5309
    },
    {
      "epoch": 9.797783933518005,
      "grad_norm": 0.4358450472354889,
      "learning_rate": 1.5921e-05,
      "loss": 0.0322,
      "step": 5310
    },
    {
      "epoch": 9.799630655586334,
      "grad_norm": 0.5932910442352295,
      "learning_rate": 1.5924000000000002e-05,
      "loss": 0.0334,
      "step": 5311
    },
    {
      "epoch": 9.801477377654663,
      "grad_norm": 0.4162185490131378,
      "learning_rate": 1.5927000000000002e-05,
      "loss": 0.0206,
      "step": 5312
    },
    {
      "epoch": 9.803324099722992,
      "grad_norm": 0.5718238353729248,
      "learning_rate": 1.593e-05,
      "loss": 0.024,
      "step": 5313
    },
    {
      "epoch": 9.80517082179132,
      "grad_norm": 0.5045569539070129,
      "learning_rate": 1.5933e-05,
      "loss": 0.0296,
      "step": 5314
    },
    {
      "epoch": 9.807017543859649,
      "grad_norm": 0.4584674537181854,
      "learning_rate": 1.5936e-05,
      "loss": 0.0232,
      "step": 5315
    },
    {
      "epoch": 9.808864265927978,
      "grad_norm": 0.4227735698223114,
      "learning_rate": 1.5939e-05,
      "loss": 0.0343,
      "step": 5316
    },
    {
      "epoch": 9.810710987996307,
      "grad_norm": 0.8416896462440491,
      "learning_rate": 1.5942e-05,
      "loss": 0.0274,
      "step": 5317
    },
    {
      "epoch": 9.812557710064635,
      "grad_norm": 0.551142156124115,
      "learning_rate": 1.5945e-05,
      "loss": 0.0344,
      "step": 5318
    },
    {
      "epoch": 9.814404432132964,
      "grad_norm": 0.9386904239654541,
      "learning_rate": 1.5948e-05,
      "loss": 0.0517,
      "step": 5319
    },
    {
      "epoch": 9.816251154201293,
      "grad_norm": 0.8260375261306763,
      "learning_rate": 1.5951e-05,
      "loss": 0.0394,
      "step": 5320
    },
    {
      "epoch": 9.818097876269622,
      "grad_norm": 0.6223652362823486,
      "learning_rate": 1.5954000000000003e-05,
      "loss": 0.0328,
      "step": 5321
    },
    {
      "epoch": 9.81994459833795,
      "grad_norm": 1.2013148069381714,
      "learning_rate": 1.5957000000000003e-05,
      "loss": 0.0415,
      "step": 5322
    },
    {
      "epoch": 9.821791320406279,
      "grad_norm": 0.4488355815410614,
      "learning_rate": 1.596e-05,
      "loss": 0.0281,
      "step": 5323
    },
    {
      "epoch": 9.823638042474608,
      "grad_norm": 0.8077351450920105,
      "learning_rate": 1.5963e-05,
      "loss": 0.034,
      "step": 5324
    },
    {
      "epoch": 9.825484764542937,
      "grad_norm": 0.6670114398002625,
      "learning_rate": 1.5966e-05,
      "loss": 0.0314,
      "step": 5325
    },
    {
      "epoch": 9.827331486611264,
      "grad_norm": 0.5995292663574219,
      "learning_rate": 1.5969e-05,
      "loss": 0.0353,
      "step": 5326
    },
    {
      "epoch": 9.829178208679593,
      "grad_norm": 0.5224243402481079,
      "learning_rate": 1.5972e-05,
      "loss": 0.0346,
      "step": 5327
    },
    {
      "epoch": 9.831024930747922,
      "grad_norm": 1.5100773572921753,
      "learning_rate": 1.5975e-05,
      "loss": 0.0494,
      "step": 5328
    },
    {
      "epoch": 9.832871652816252,
      "grad_norm": 1.619434118270874,
      "learning_rate": 1.5978e-05,
      "loss": 0.2553,
      "step": 5329
    },
    {
      "epoch": 9.83471837488458,
      "grad_norm": 1.285167932510376,
      "learning_rate": 1.5980999999999998e-05,
      "loss": 0.2833,
      "step": 5330
    },
    {
      "epoch": 9.836565096952908,
      "grad_norm": 3.9550986289978027,
      "learning_rate": 1.5984e-05,
      "loss": 0.2325,
      "step": 5331
    },
    {
      "epoch": 9.838411819021237,
      "grad_norm": 0.7910865545272827,
      "learning_rate": 1.5987e-05,
      "loss": 0.2184,
      "step": 5332
    },
    {
      "epoch": 9.840258541089566,
      "grad_norm": 0.6185511350631714,
      "learning_rate": 1.599e-05,
      "loss": 0.1437,
      "step": 5333
    },
    {
      "epoch": 9.842105263157894,
      "grad_norm": 1.0045210123062134,
      "learning_rate": 1.5993e-05,
      "loss": 0.2016,
      "step": 5334
    },
    {
      "epoch": 9.843951985226223,
      "grad_norm": 0.8364124298095703,
      "learning_rate": 1.5996e-05,
      "loss": 0.1727,
      "step": 5335
    },
    {
      "epoch": 9.845798707294552,
      "grad_norm": 0.6798272728919983,
      "learning_rate": 1.5999e-05,
      "loss": 0.1071,
      "step": 5336
    },
    {
      "epoch": 9.847645429362881,
      "grad_norm": 1.0942049026489258,
      "learning_rate": 1.6002e-05,
      "loss": 0.1737,
      "step": 5337
    },
    {
      "epoch": 9.84949215143121,
      "grad_norm": 0.6949169635772705,
      "learning_rate": 1.6005e-05,
      "loss": 0.1016,
      "step": 5338
    },
    {
      "epoch": 9.851338873499538,
      "grad_norm": 1.0689623355865479,
      "learning_rate": 1.6008e-05,
      "loss": 0.1219,
      "step": 5339
    },
    {
      "epoch": 9.853185595567867,
      "grad_norm": 0.654719352722168,
      "learning_rate": 1.6011e-05,
      "loss": 0.0776,
      "step": 5340
    },
    {
      "epoch": 9.855032317636196,
      "grad_norm": 0.8142164945602417,
      "learning_rate": 1.6014000000000003e-05,
      "loss": 0.0958,
      "step": 5341
    },
    {
      "epoch": 9.856879039704525,
      "grad_norm": 0.4266529083251953,
      "learning_rate": 1.6017000000000003e-05,
      "loss": 0.042,
      "step": 5342
    },
    {
      "epoch": 9.858725761772853,
      "grad_norm": 0.3962545692920685,
      "learning_rate": 1.6020000000000002e-05,
      "loss": 0.0574,
      "step": 5343
    },
    {
      "epoch": 9.860572483841182,
      "grad_norm": 0.7206045985221863,
      "learning_rate": 1.6023000000000002e-05,
      "loss": 0.0626,
      "step": 5344
    },
    {
      "epoch": 9.86241920590951,
      "grad_norm": 0.48156070709228516,
      "learning_rate": 1.6026000000000002e-05,
      "loss": 0.0341,
      "step": 5345
    },
    {
      "epoch": 9.86426592797784,
      "grad_norm": 0.4663195312023163,
      "learning_rate": 1.6029000000000002e-05,
      "loss": 0.0355,
      "step": 5346
    },
    {
      "epoch": 9.866112650046167,
      "grad_norm": 0.30054986476898193,
      "learning_rate": 1.6032e-05,
      "loss": 0.026,
      "step": 5347
    },
    {
      "epoch": 9.867959372114496,
      "grad_norm": 0.8619183897972107,
      "learning_rate": 1.6034999999999998e-05,
      "loss": 0.1128,
      "step": 5348
    },
    {
      "epoch": 9.869806094182826,
      "grad_norm": 0.4234899580478668,
      "learning_rate": 1.6037999999999998e-05,
      "loss": 0.0335,
      "step": 5349
    },
    {
      "epoch": 9.871652816251155,
      "grad_norm": 0.6339716911315918,
      "learning_rate": 1.6040999999999998e-05,
      "loss": 0.0373,
      "step": 5350
    },
    {
      "epoch": 9.873499538319482,
      "grad_norm": 0.31937074661254883,
      "learning_rate": 1.6044e-05,
      "loss": 0.0212,
      "step": 5351
    },
    {
      "epoch": 9.875346260387811,
      "grad_norm": 0.38986673951148987,
      "learning_rate": 1.6047e-05,
      "loss": 0.022,
      "step": 5352
    },
    {
      "epoch": 9.87719298245614,
      "grad_norm": 0.32344773411750793,
      "learning_rate": 1.605e-05,
      "loss": 0.025,
      "step": 5353
    },
    {
      "epoch": 9.87903970452447,
      "grad_norm": 0.8871376514434814,
      "learning_rate": 1.6053e-05,
      "loss": 0.0318,
      "step": 5354
    },
    {
      "epoch": 9.880886426592799,
      "grad_norm": 0.5399925112724304,
      "learning_rate": 1.6056e-05,
      "loss": 0.0381,
      "step": 5355
    },
    {
      "epoch": 9.882733148661126,
      "grad_norm": 0.47385239601135254,
      "learning_rate": 1.6059e-05,
      "loss": 0.0299,
      "step": 5356
    },
    {
      "epoch": 9.884579870729455,
      "grad_norm": 0.38570722937583923,
      "learning_rate": 1.6062e-05,
      "loss": 0.0275,
      "step": 5357
    },
    {
      "epoch": 9.886426592797784,
      "grad_norm": 0.4128206670284271,
      "learning_rate": 1.6065e-05,
      "loss": 0.0209,
      "step": 5358
    },
    {
      "epoch": 9.888273314866112,
      "grad_norm": 0.5899372100830078,
      "learning_rate": 1.6068e-05,
      "loss": 0.0328,
      "step": 5359
    },
    {
      "epoch": 9.89012003693444,
      "grad_norm": 0.6390036940574646,
      "learning_rate": 1.6071e-05,
      "loss": 0.0197,
      "step": 5360
    },
    {
      "epoch": 9.89196675900277,
      "grad_norm": 0.531987726688385,
      "learning_rate": 1.6074000000000002e-05,
      "loss": 0.0251,
      "step": 5361
    },
    {
      "epoch": 9.8938134810711,
      "grad_norm": 0.4834712743759155,
      "learning_rate": 1.6077000000000002e-05,
      "loss": 0.0223,
      "step": 5362
    },
    {
      "epoch": 9.895660203139428,
      "grad_norm": 0.43583738803863525,
      "learning_rate": 1.6080000000000002e-05,
      "loss": 0.0277,
      "step": 5363
    },
    {
      "epoch": 9.897506925207756,
      "grad_norm": 0.49375590682029724,
      "learning_rate": 1.6083000000000002e-05,
      "loss": 0.0253,
      "step": 5364
    },
    {
      "epoch": 9.899353647276085,
      "grad_norm": 0.6179859042167664,
      "learning_rate": 1.6086e-05,
      "loss": 0.0212,
      "step": 5365
    },
    {
      "epoch": 9.901200369344414,
      "grad_norm": 0.5361495614051819,
      "learning_rate": 1.6089e-05,
      "loss": 0.0349,
      "step": 5366
    },
    {
      "epoch": 9.903047091412743,
      "grad_norm": 1.1141771078109741,
      "learning_rate": 1.6092e-05,
      "loss": 0.0301,
      "step": 5367
    },
    {
      "epoch": 9.90489381348107,
      "grad_norm": 0.7264317870140076,
      "learning_rate": 1.6095e-05,
      "loss": 0.0301,
      "step": 5368
    },
    {
      "epoch": 9.9067405355494,
      "grad_norm": 0.6218966245651245,
      "learning_rate": 1.6098e-05,
      "loss": 0.0379,
      "step": 5369
    },
    {
      "epoch": 9.908587257617729,
      "grad_norm": 0.8654443621635437,
      "learning_rate": 1.6101e-05,
      "loss": 0.0406,
      "step": 5370
    },
    {
      "epoch": 9.910433979686058,
      "grad_norm": 0.6884783506393433,
      "learning_rate": 1.6104000000000004e-05,
      "loss": 0.0286,
      "step": 5371
    },
    {
      "epoch": 9.912280701754385,
      "grad_norm": 0.5119009017944336,
      "learning_rate": 1.6107e-05,
      "loss": 0.0299,
      "step": 5372
    },
    {
      "epoch": 9.914127423822714,
      "grad_norm": 0.6698522567749023,
      "learning_rate": 1.611e-05,
      "loss": 0.0292,
      "step": 5373
    },
    {
      "epoch": 9.915974145891044,
      "grad_norm": 0.7594621181488037,
      "learning_rate": 1.6113e-05,
      "loss": 0.0333,
      "step": 5374
    },
    {
      "epoch": 9.917820867959373,
      "grad_norm": 0.4515618085861206,
      "learning_rate": 1.6116e-05,
      "loss": 0.0234,
      "step": 5375
    },
    {
      "epoch": 9.9196675900277,
      "grad_norm": 0.6473972797393799,
      "learning_rate": 1.6119e-05,
      "loss": 0.0345,
      "step": 5376
    },
    {
      "epoch": 9.92151431209603,
      "grad_norm": 0.7409936189651489,
      "learning_rate": 1.6122e-05,
      "loss": 0.0439,
      "step": 5377
    },
    {
      "epoch": 9.923361034164358,
      "grad_norm": 0.7425601482391357,
      "learning_rate": 1.6125e-05,
      "loss": 0.0607,
      "step": 5378
    },
    {
      "epoch": 9.925207756232687,
      "grad_norm": 0.9278039336204529,
      "learning_rate": 1.6128e-05,
      "loss": 0.2801,
      "step": 5379
    },
    {
      "epoch": 9.927054478301017,
      "grad_norm": 0.7289472818374634,
      "learning_rate": 1.6131e-05,
      "loss": 0.1843,
      "step": 5380
    },
    {
      "epoch": 9.928901200369344,
      "grad_norm": 1.1333955526351929,
      "learning_rate": 1.6134e-05,
      "loss": 0.1941,
      "step": 5381
    },
    {
      "epoch": 9.930747922437673,
      "grad_norm": 0.8633908629417419,
      "learning_rate": 1.6137e-05,
      "loss": 0.2418,
      "step": 5382
    },
    {
      "epoch": 9.932594644506002,
      "grad_norm": 0.829034686088562,
      "learning_rate": 1.614e-05,
      "loss": 0.1373,
      "step": 5383
    },
    {
      "epoch": 9.93444136657433,
      "grad_norm": 0.5944445729255676,
      "learning_rate": 1.6143e-05,
      "loss": 0.1451,
      "step": 5384
    },
    {
      "epoch": 9.936288088642659,
      "grad_norm": 0.7495405673980713,
      "learning_rate": 1.6146e-05,
      "loss": 0.1112,
      "step": 5385
    },
    {
      "epoch": 9.938134810710988,
      "grad_norm": 0.6639820337295532,
      "learning_rate": 1.6149e-05,
      "loss": 0.1407,
      "step": 5386
    },
    {
      "epoch": 9.939981532779317,
      "grad_norm": 0.7890506982803345,
      "learning_rate": 1.6152e-05,
      "loss": 0.1097,
      "step": 5387
    },
    {
      "epoch": 9.941828254847646,
      "grad_norm": 1.096908450126648,
      "learning_rate": 1.6155e-05,
      "loss": 0.0857,
      "step": 5388
    },
    {
      "epoch": 9.943674976915974,
      "grad_norm": 1.0947462320327759,
      "learning_rate": 1.6158e-05,
      "loss": 0.1497,
      "step": 5389
    },
    {
      "epoch": 9.945521698984303,
      "grad_norm": 0.8100894093513489,
      "learning_rate": 1.6161e-05,
      "loss": 0.0883,
      "step": 5390
    },
    {
      "epoch": 9.947368421052632,
      "grad_norm": 0.5456356406211853,
      "learning_rate": 1.6164e-05,
      "loss": 0.0527,
      "step": 5391
    },
    {
      "epoch": 9.949215143120961,
      "grad_norm": 0.5579842329025269,
      "learning_rate": 1.6167000000000003e-05,
      "loss": 0.0433,
      "step": 5392
    },
    {
      "epoch": 9.951061865189288,
      "grad_norm": 0.6612244248390198,
      "learning_rate": 1.6170000000000003e-05,
      "loss": 0.0377,
      "step": 5393
    },
    {
      "epoch": 9.952908587257618,
      "grad_norm": 0.4671032726764679,
      "learning_rate": 1.6173000000000003e-05,
      "loss": 0.0245,
      "step": 5394
    },
    {
      "epoch": 9.954755309325947,
      "grad_norm": 0.4639342725276947,
      "learning_rate": 1.6176000000000002e-05,
      "loss": 0.0338,
      "step": 5395
    },
    {
      "epoch": 9.956602031394276,
      "grad_norm": 0.38254842162132263,
      "learning_rate": 1.6179000000000002e-05,
      "loss": 0.0173,
      "step": 5396
    },
    {
      "epoch": 9.958448753462603,
      "grad_norm": 0.34761661291122437,
      "learning_rate": 1.6182e-05,
      "loss": 0.0267,
      "step": 5397
    },
    {
      "epoch": 9.960295475530932,
      "grad_norm": 0.5771934390068054,
      "learning_rate": 1.6185e-05,
      "loss": 0.0379,
      "step": 5398
    },
    {
      "epoch": 9.962142197599261,
      "grad_norm": 0.7256296873092651,
      "learning_rate": 1.6187999999999998e-05,
      "loss": 0.0503,
      "step": 5399
    },
    {
      "epoch": 9.96398891966759,
      "grad_norm": 0.4355110824108124,
      "learning_rate": 1.6190999999999998e-05,
      "loss": 0.0237,
      "step": 5400
    },
    {
      "epoch": 9.965835641735918,
      "grad_norm": 0.6564544439315796,
      "learning_rate": 1.6193999999999998e-05,
      "loss": 0.0701,
      "step": 5401
    },
    {
      "epoch": 9.967682363804247,
      "grad_norm": 0.4925614595413208,
      "learning_rate": 1.6197e-05,
      "loss": 0.0291,
      "step": 5402
    },
    {
      "epoch": 9.969529085872576,
      "grad_norm": 0.6966110467910767,
      "learning_rate": 1.62e-05,
      "loss": 0.0253,
      "step": 5403
    },
    {
      "epoch": 9.971375807940905,
      "grad_norm": 0.7741514444351196,
      "learning_rate": 1.6203e-05,
      "loss": 0.0212,
      "step": 5404
    },
    {
      "epoch": 9.973222530009235,
      "grad_norm": 0.5057776570320129,
      "learning_rate": 1.6206e-05,
      "loss": 0.0318,
      "step": 5405
    },
    {
      "epoch": 9.975069252077562,
      "grad_norm": 0.621398389339447,
      "learning_rate": 1.6209e-05,
      "loss": 0.0308,
      "step": 5406
    },
    {
      "epoch": 9.976915974145891,
      "grad_norm": 0.4574577510356903,
      "learning_rate": 1.6212e-05,
      "loss": 0.0226,
      "step": 5407
    },
    {
      "epoch": 9.97876269621422,
      "grad_norm": 0.5730082988739014,
      "learning_rate": 1.6215e-05,
      "loss": 0.0198,
      "step": 5408
    },
    {
      "epoch": 9.980609418282548,
      "grad_norm": 0.752869188785553,
      "learning_rate": 1.6218e-05,
      "loss": 0.0228,
      "step": 5409
    },
    {
      "epoch": 9.982456140350877,
      "grad_norm": 0.513892650604248,
      "learning_rate": 1.6221e-05,
      "loss": 0.0255,
      "step": 5410
    },
    {
      "epoch": 9.984302862419206,
      "grad_norm": 1.2083364725112915,
      "learning_rate": 1.6224e-05,
      "loss": 0.0294,
      "step": 5411
    },
    {
      "epoch": 9.986149584487535,
      "grad_norm": 0.5639918446540833,
      "learning_rate": 1.6227000000000002e-05,
      "loss": 0.0238,
      "step": 5412
    },
    {
      "epoch": 9.987996306555864,
      "grad_norm": 4.152475833892822,
      "learning_rate": 1.6230000000000002e-05,
      "loss": 0.0333,
      "step": 5413
    },
    {
      "epoch": 9.989843028624191,
      "grad_norm": 0.846754252910614,
      "learning_rate": 1.6233000000000002e-05,
      "loss": 0.0329,
      "step": 5414
    },
    {
      "epoch": 9.99168975069252,
      "grad_norm": 0.6781851053237915,
      "learning_rate": 1.6236000000000002e-05,
      "loss": 0.0388,
      "step": 5415
    },
    {
      "epoch": 9.99353647276085,
      "grad_norm": 0.8813036680221558,
      "learning_rate": 1.6239e-05,
      "loss": 0.0421,
      "step": 5416
    },
    {
      "epoch": 9.995383194829179,
      "grad_norm": 0.561560332775116,
      "learning_rate": 1.6242e-05,
      "loss": 0.0227,
      "step": 5417
    },
    {
      "epoch": 9.997229916897506,
      "grad_norm": 0.7628563642501831,
      "learning_rate": 1.6245e-05,
      "loss": 0.0295,
      "step": 5418
    },
    {
      "epoch": 9.999076638965835,
      "grad_norm": 0.5719330906867981,
      "learning_rate": 1.6248e-05,
      "loss": 0.0157,
      "step": 5419
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.5679962635040283,
      "learning_rate": 1.6251e-05,
      "loss": 0.0167,
      "step": 5420
    },
    {
      "epoch": 10.00184672206833,
      "grad_norm": 1.5406891107559204,
      "learning_rate": 1.6253999999999997e-05,
      "loss": 0.3427,
      "step": 5421
    },
    {
      "epoch": 10.003693444136658,
      "grad_norm": 0.8477674722671509,
      "learning_rate": 1.6257e-05,
      "loss": 0.2991,
      "step": 5422
    },
    {
      "epoch": 10.005540166204986,
      "grad_norm": 1.0089524984359741,
      "learning_rate": 1.626e-05,
      "loss": 0.2128,
      "step": 5423
    },
    {
      "epoch": 10.007386888273315,
      "grad_norm": 1.0328800678253174,
      "learning_rate": 1.6263e-05,
      "loss": 0.1977,
      "step": 5424
    },
    {
      "epoch": 10.009233610341644,
      "grad_norm": 0.9635773301124573,
      "learning_rate": 1.6266e-05,
      "loss": 0.2169,
      "step": 5425
    },
    {
      "epoch": 10.011080332409973,
      "grad_norm": 1.2987926006317139,
      "learning_rate": 1.6269e-05,
      "loss": 0.1697,
      "step": 5426
    },
    {
      "epoch": 10.0129270544783,
      "grad_norm": 0.6195570826530457,
      "learning_rate": 1.6272e-05,
      "loss": 0.1087,
      "step": 5427
    },
    {
      "epoch": 10.01477377654663,
      "grad_norm": 0.6779178977012634,
      "learning_rate": 1.6275e-05,
      "loss": 0.1107,
      "step": 5428
    },
    {
      "epoch": 10.016620498614959,
      "grad_norm": 0.6723268628120422,
      "learning_rate": 1.6278e-05,
      "loss": 0.1222,
      "step": 5429
    },
    {
      "epoch": 10.018467220683288,
      "grad_norm": 0.6540348529815674,
      "learning_rate": 1.6281e-05,
      "loss": 0.1055,
      "step": 5430
    },
    {
      "epoch": 10.020313942751615,
      "grad_norm": 0.764249324798584,
      "learning_rate": 1.6284e-05,
      "loss": 0.1055,
      "step": 5431
    },
    {
      "epoch": 10.022160664819944,
      "grad_norm": 0.5885652899742126,
      "learning_rate": 1.6287000000000002e-05,
      "loss": 0.0961,
      "step": 5432
    },
    {
      "epoch": 10.024007386888274,
      "grad_norm": 1.157502293586731,
      "learning_rate": 1.629e-05,
      "loss": 0.1551,
      "step": 5433
    },
    {
      "epoch": 10.025854108956603,
      "grad_norm": 0.5454378128051758,
      "learning_rate": 1.6293e-05,
      "loss": 0.0723,
      "step": 5434
    },
    {
      "epoch": 10.02770083102493,
      "grad_norm": 0.4833084046840668,
      "learning_rate": 1.6296e-05,
      "loss": 0.0472,
      "step": 5435
    },
    {
      "epoch": 10.02954755309326,
      "grad_norm": 0.7118651866912842,
      "learning_rate": 1.6299e-05,
      "loss": 0.0354,
      "step": 5436
    },
    {
      "epoch": 10.031394275161588,
      "grad_norm": 0.5361140370368958,
      "learning_rate": 1.6302e-05,
      "loss": 0.0412,
      "step": 5437
    },
    {
      "epoch": 10.033240997229917,
      "grad_norm": 0.43500009179115295,
      "learning_rate": 1.6305e-05,
      "loss": 0.0236,
      "step": 5438
    },
    {
      "epoch": 10.035087719298245,
      "grad_norm": 0.989443302154541,
      "learning_rate": 1.6308e-05,
      "loss": 0.0288,
      "step": 5439
    },
    {
      "epoch": 10.036934441366574,
      "grad_norm": 0.3764720857143402,
      "learning_rate": 1.6311e-05,
      "loss": 0.0343,
      "step": 5440
    },
    {
      "epoch": 10.038781163434903,
      "grad_norm": 0.3913370966911316,
      "learning_rate": 1.6314e-05,
      "loss": 0.019,
      "step": 5441
    },
    {
      "epoch": 10.040627885503232,
      "grad_norm": 1.057128667831421,
      "learning_rate": 1.6317000000000003e-05,
      "loss": 0.0333,
      "step": 5442
    },
    {
      "epoch": 10.04247460757156,
      "grad_norm": 0.6400501132011414,
      "learning_rate": 1.6320000000000003e-05,
      "loss": 0.0311,
      "step": 5443
    },
    {
      "epoch": 10.044321329639889,
      "grad_norm": 0.4104997515678406,
      "learning_rate": 1.6323000000000003e-05,
      "loss": 0.0223,
      "step": 5444
    },
    {
      "epoch": 10.046168051708218,
      "grad_norm": 0.33504095673561096,
      "learning_rate": 1.6326000000000003e-05,
      "loss": 0.0284,
      "step": 5445
    },
    {
      "epoch": 10.048014773776547,
      "grad_norm": 0.4359169602394104,
      "learning_rate": 1.6329e-05,
      "loss": 0.0251,
      "step": 5446
    },
    {
      "epoch": 10.049861495844876,
      "grad_norm": 0.46953126788139343,
      "learning_rate": 1.6332e-05,
      "loss": 0.0294,
      "step": 5447
    },
    {
      "epoch": 10.051708217913204,
      "grad_norm": 0.7887614369392395,
      "learning_rate": 1.6335e-05,
      "loss": 0.0274,
      "step": 5448
    },
    {
      "epoch": 10.053554939981533,
      "grad_norm": 0.3279338777065277,
      "learning_rate": 1.6338e-05,
      "loss": 0.022,
      "step": 5449
    },
    {
      "epoch": 10.055401662049862,
      "grad_norm": 0.5057511329650879,
      "learning_rate": 1.6340999999999998e-05,
      "loss": 0.0316,
      "step": 5450
    },
    {
      "epoch": 10.057248384118191,
      "grad_norm": 0.5215173363685608,
      "learning_rate": 1.6343999999999998e-05,
      "loss": 0.019,
      "step": 5451
    },
    {
      "epoch": 10.059095106186518,
      "grad_norm": 0.42165639996528625,
      "learning_rate": 1.6347e-05,
      "loss": 0.019,
      "step": 5452
    },
    {
      "epoch": 10.060941828254848,
      "grad_norm": 0.33034011721611023,
      "learning_rate": 1.635e-05,
      "loss": 0.0165,
      "step": 5453
    },
    {
      "epoch": 10.062788550323177,
      "grad_norm": 0.45684513449668884,
      "learning_rate": 1.6353e-05,
      "loss": 0.0194,
      "step": 5454
    },
    {
      "epoch": 10.064635272391506,
      "grad_norm": 0.5774962306022644,
      "learning_rate": 1.6356e-05,
      "loss": 0.0394,
      "step": 5455
    },
    {
      "epoch": 10.066481994459833,
      "grad_norm": 0.4379504323005676,
      "learning_rate": 1.6359e-05,
      "loss": 0.0187,
      "step": 5456
    },
    {
      "epoch": 10.068328716528162,
      "grad_norm": 0.7650175094604492,
      "learning_rate": 1.6362e-05,
      "loss": 0.0215,
      "step": 5457
    },
    {
      "epoch": 10.070175438596491,
      "grad_norm": 0.5422001481056213,
      "learning_rate": 1.6365e-05,
      "loss": 0.0536,
      "step": 5458
    },
    {
      "epoch": 10.07202216066482,
      "grad_norm": 0.333640456199646,
      "learning_rate": 1.6368e-05,
      "loss": 0.0154,
      "step": 5459
    },
    {
      "epoch": 10.073868882733148,
      "grad_norm": 1.1780778169631958,
      "learning_rate": 1.6371e-05,
      "loss": 0.0174,
      "step": 5460
    },
    {
      "epoch": 10.075715604801477,
      "grad_norm": 0.7993797063827515,
      "learning_rate": 1.6374e-05,
      "loss": 0.0287,
      "step": 5461
    },
    {
      "epoch": 10.077562326869806,
      "grad_norm": 0.7349343299865723,
      "learning_rate": 1.6377000000000003e-05,
      "loss": 0.0304,
      "step": 5462
    },
    {
      "epoch": 10.079409048938135,
      "grad_norm": 0.8012458086013794,
      "learning_rate": 1.6380000000000002e-05,
      "loss": 0.0229,
      "step": 5463
    },
    {
      "epoch": 10.081255771006463,
      "grad_norm": 0.5809488892555237,
      "learning_rate": 1.6383000000000002e-05,
      "loss": 0.026,
      "step": 5464
    },
    {
      "epoch": 10.083102493074792,
      "grad_norm": 0.6947927474975586,
      "learning_rate": 1.6386000000000002e-05,
      "loss": 0.0334,
      "step": 5465
    },
    {
      "epoch": 10.084949215143121,
      "grad_norm": 1.0784257650375366,
      "learning_rate": 1.6389000000000002e-05,
      "loss": 0.032,
      "step": 5466
    },
    {
      "epoch": 10.08679593721145,
      "grad_norm": 0.4056265950202942,
      "learning_rate": 1.6392e-05,
      "loss": 0.017,
      "step": 5467
    },
    {
      "epoch": 10.088642659279778,
      "grad_norm": 0.5124925374984741,
      "learning_rate": 1.6395e-05,
      "loss": 0.0279,
      "step": 5468
    },
    {
      "epoch": 10.090489381348107,
      "grad_norm": 1.012369990348816,
      "learning_rate": 1.6398e-05,
      "loss": 0.0267,
      "step": 5469
    },
    {
      "epoch": 10.092336103416436,
      "grad_norm": 0.6609354615211487,
      "learning_rate": 1.6400999999999998e-05,
      "loss": 0.0279,
      "step": 5470
    },
    {
      "epoch": 10.094182825484765,
      "grad_norm": 1.0703595876693726,
      "learning_rate": 1.6403999999999997e-05,
      "loss": 0.2938,
      "step": 5471
    },
    {
      "epoch": 10.096029547553094,
      "grad_norm": 0.7762547731399536,
      "learning_rate": 1.6407e-05,
      "loss": 0.1815,
      "step": 5472
    },
    {
      "epoch": 10.097876269621421,
      "grad_norm": 0.7497743368148804,
      "learning_rate": 1.641e-05,
      "loss": 0.1767,
      "step": 5473
    },
    {
      "epoch": 10.09972299168975,
      "grad_norm": 0.7290788888931274,
      "learning_rate": 1.6413e-05,
      "loss": 0.1451,
      "step": 5474
    },
    {
      "epoch": 10.10156971375808,
      "grad_norm": 0.6357746124267578,
      "learning_rate": 1.6416e-05,
      "loss": 0.141,
      "step": 5475
    },
    {
      "epoch": 10.103416435826409,
      "grad_norm": 0.7867652773857117,
      "learning_rate": 1.6419e-05,
      "loss": 0.1779,
      "step": 5476
    },
    {
      "epoch": 10.105263157894736,
      "grad_norm": 0.5439184308052063,
      "learning_rate": 1.6422e-05,
      "loss": 0.1147,
      "step": 5477
    },
    {
      "epoch": 10.107109879963065,
      "grad_norm": 0.6986681818962097,
      "learning_rate": 1.6425e-05,
      "loss": 0.174,
      "step": 5478
    },
    {
      "epoch": 10.108956602031395,
      "grad_norm": 0.5804990530014038,
      "learning_rate": 1.6428e-05,
      "loss": 0.0873,
      "step": 5479
    },
    {
      "epoch": 10.110803324099724,
      "grad_norm": 1.0001838207244873,
      "learning_rate": 1.6431e-05,
      "loss": 0.1138,
      "step": 5480
    },
    {
      "epoch": 10.112650046168051,
      "grad_norm": 0.825629711151123,
      "learning_rate": 1.6434e-05,
      "loss": 0.0916,
      "step": 5481
    },
    {
      "epoch": 10.11449676823638,
      "grad_norm": 0.7747324705123901,
      "learning_rate": 1.6437000000000002e-05,
      "loss": 0.0803,
      "step": 5482
    },
    {
      "epoch": 10.11634349030471,
      "grad_norm": 0.9804370999336243,
      "learning_rate": 1.6440000000000002e-05,
      "loss": 0.0415,
      "step": 5483
    },
    {
      "epoch": 10.118190212373039,
      "grad_norm": 0.7696126699447632,
      "learning_rate": 1.6443e-05,
      "loss": 0.0668,
      "step": 5484
    },
    {
      "epoch": 10.120036934441366,
      "grad_norm": 0.39905205368995667,
      "learning_rate": 1.6446e-05,
      "loss": 0.0409,
      "step": 5485
    },
    {
      "epoch": 10.121883656509695,
      "grad_norm": 0.40738794207572937,
      "learning_rate": 1.6449e-05,
      "loss": 0.0553,
      "step": 5486
    },
    {
      "epoch": 10.123730378578024,
      "grad_norm": 0.3792731761932373,
      "learning_rate": 1.6452e-05,
      "loss": 0.0366,
      "step": 5487
    },
    {
      "epoch": 10.125577100646353,
      "grad_norm": 0.538812518119812,
      "learning_rate": 1.6455e-05,
      "loss": 0.0338,
      "step": 5488
    },
    {
      "epoch": 10.12742382271468,
      "grad_norm": 0.6880744695663452,
      "learning_rate": 1.6458e-05,
      "loss": 0.0274,
      "step": 5489
    },
    {
      "epoch": 10.12927054478301,
      "grad_norm": 0.2780337333679199,
      "learning_rate": 1.6461e-05,
      "loss": 0.0161,
      "step": 5490
    },
    {
      "epoch": 10.131117266851339,
      "grad_norm": 0.5699805021286011,
      "learning_rate": 1.6464e-05,
      "loss": 0.02,
      "step": 5491
    },
    {
      "epoch": 10.132963988919668,
      "grad_norm": 0.7288239598274231,
      "learning_rate": 1.6467000000000003e-05,
      "loss": 0.0529,
      "step": 5492
    },
    {
      "epoch": 10.134810710987995,
      "grad_norm": 0.5717664361000061,
      "learning_rate": 1.6470000000000003e-05,
      "loss": 0.0315,
      "step": 5493
    },
    {
      "epoch": 10.136657433056325,
      "grad_norm": 0.36545732617378235,
      "learning_rate": 1.6473000000000003e-05,
      "loss": 0.0219,
      "step": 5494
    },
    {
      "epoch": 10.138504155124654,
      "grad_norm": 0.3427481949329376,
      "learning_rate": 1.6476e-05,
      "loss": 0.0238,
      "step": 5495
    },
    {
      "epoch": 10.140350877192983,
      "grad_norm": 0.461640328168869,
      "learning_rate": 1.6479e-05,
      "loss": 0.0238,
      "step": 5496
    },
    {
      "epoch": 10.142197599261312,
      "grad_norm": 0.5584130883216858,
      "learning_rate": 1.6482e-05,
      "loss": 0.0276,
      "step": 5497
    },
    {
      "epoch": 10.14404432132964,
      "grad_norm": 0.376208633184433,
      "learning_rate": 1.6485e-05,
      "loss": 0.0187,
      "step": 5498
    },
    {
      "epoch": 10.145891043397969,
      "grad_norm": 0.3295646607875824,
      "learning_rate": 1.6488e-05,
      "loss": 0.023,
      "step": 5499
    },
    {
      "epoch": 10.147737765466298,
      "grad_norm": 0.40510445833206177,
      "learning_rate": 1.6491e-05,
      "loss": 0.0235,
      "step": 5500
    },
    {
      "epoch": 10.149584487534627,
      "grad_norm": 0.594293475151062,
      "learning_rate": 1.6493999999999998e-05,
      "loss": 0.0188,
      "step": 5501
    },
    {
      "epoch": 10.151431209602954,
      "grad_norm": 1.6221593618392944,
      "learning_rate": 1.6497e-05,
      "loss": 0.0352,
      "step": 5502
    },
    {
      "epoch": 10.153277931671283,
      "grad_norm": 0.4105490446090698,
      "learning_rate": 1.65e-05,
      "loss": 0.0213,
      "step": 5503
    },
    {
      "epoch": 10.155124653739612,
      "grad_norm": 0.5947759747505188,
      "learning_rate": 1.6503e-05,
      "loss": 0.0268,
      "step": 5504
    },
    {
      "epoch": 10.156971375807942,
      "grad_norm": 0.3585164546966553,
      "learning_rate": 1.6506e-05,
      "loss": 0.0177,
      "step": 5505
    },
    {
      "epoch": 10.158818097876269,
      "grad_norm": 0.435248464345932,
      "learning_rate": 1.6509e-05,
      "loss": 0.0262,
      "step": 5506
    },
    {
      "epoch": 10.160664819944598,
      "grad_norm": 0.31813377141952515,
      "learning_rate": 1.6512e-05,
      "loss": 0.0187,
      "step": 5507
    },
    {
      "epoch": 10.162511542012927,
      "grad_norm": 0.569756031036377,
      "learning_rate": 1.6515e-05,
      "loss": 0.018,
      "step": 5508
    },
    {
      "epoch": 10.164358264081256,
      "grad_norm": 0.31594976782798767,
      "learning_rate": 1.6518e-05,
      "loss": 0.013,
      "step": 5509
    },
    {
      "epoch": 10.166204986149584,
      "grad_norm": 0.3902089297771454,
      "learning_rate": 1.6521e-05,
      "loss": 0.0197,
      "step": 5510
    },
    {
      "epoch": 10.168051708217913,
      "grad_norm": 0.5651479363441467,
      "learning_rate": 1.6524e-05,
      "loss": 0.0185,
      "step": 5511
    },
    {
      "epoch": 10.169898430286242,
      "grad_norm": 0.45485809445381165,
      "learning_rate": 1.6527e-05,
      "loss": 0.0256,
      "step": 5512
    },
    {
      "epoch": 10.171745152354571,
      "grad_norm": 1.6518807411193848,
      "learning_rate": 1.6530000000000003e-05,
      "loss": 0.0612,
      "step": 5513
    },
    {
      "epoch": 10.173591874422899,
      "grad_norm": 0.7069046497344971,
      "learning_rate": 1.6533000000000002e-05,
      "loss": 0.0348,
      "step": 5514
    },
    {
      "epoch": 10.175438596491228,
      "grad_norm": 0.6123530864715576,
      "learning_rate": 1.6536000000000002e-05,
      "loss": 0.0275,
      "step": 5515
    },
    {
      "epoch": 10.177285318559557,
      "grad_norm": 0.7783622145652771,
      "learning_rate": 1.6539000000000002e-05,
      "loss": 0.0379,
      "step": 5516
    },
    {
      "epoch": 10.179132040627886,
      "grad_norm": 0.5361220836639404,
      "learning_rate": 1.6542000000000002e-05,
      "loss": 0.0175,
      "step": 5517
    },
    {
      "epoch": 10.180978762696213,
      "grad_norm": 0.7196072340011597,
      "learning_rate": 1.6545e-05,
      "loss": 0.0519,
      "step": 5518
    },
    {
      "epoch": 10.182825484764543,
      "grad_norm": 0.8551514744758606,
      "learning_rate": 1.6548e-05,
      "loss": 0.0255,
      "step": 5519
    },
    {
      "epoch": 10.184672206832872,
      "grad_norm": 0.9311538338661194,
      "learning_rate": 1.6550999999999998e-05,
      "loss": 0.025,
      "step": 5520
    },
    {
      "epoch": 10.1865189289012,
      "grad_norm": 0.8415346145629883,
      "learning_rate": 1.6553999999999998e-05,
      "loss": 0.2651,
      "step": 5521
    },
    {
      "epoch": 10.18836565096953,
      "grad_norm": 0.9645931720733643,
      "learning_rate": 1.6556999999999998e-05,
      "loss": 0.2365,
      "step": 5522
    },
    {
      "epoch": 10.190212373037857,
      "grad_norm": 0.8362430930137634,
      "learning_rate": 1.656e-05,
      "loss": 0.2345,
      "step": 5523
    },
    {
      "epoch": 10.192059095106186,
      "grad_norm": 0.7902693152427673,
      "learning_rate": 1.6563e-05,
      "loss": 0.2174,
      "step": 5524
    },
    {
      "epoch": 10.193905817174516,
      "grad_norm": 0.5826321840286255,
      "learning_rate": 1.6566e-05,
      "loss": 0.1437,
      "step": 5525
    },
    {
      "epoch": 10.195752539242845,
      "grad_norm": 1.1524338722229004,
      "learning_rate": 1.6569e-05,
      "loss": 0.1946,
      "step": 5526
    },
    {
      "epoch": 10.197599261311172,
      "grad_norm": 0.7611707448959351,
      "learning_rate": 1.6572e-05,
      "loss": 0.1437,
      "step": 5527
    },
    {
      "epoch": 10.199445983379501,
      "grad_norm": 0.6333263516426086,
      "learning_rate": 1.6575e-05,
      "loss": 0.1312,
      "step": 5528
    },
    {
      "epoch": 10.20129270544783,
      "grad_norm": 0.619972288608551,
      "learning_rate": 1.6578e-05,
      "loss": 0.0986,
      "step": 5529
    },
    {
      "epoch": 10.20313942751616,
      "grad_norm": 0.6319988369941711,
      "learning_rate": 1.6581e-05,
      "loss": 0.0828,
      "step": 5530
    },
    {
      "epoch": 10.204986149584487,
      "grad_norm": 0.7655250430107117,
      "learning_rate": 1.6584e-05,
      "loss": 0.1116,
      "step": 5531
    },
    {
      "epoch": 10.206832871652816,
      "grad_norm": 0.5639950633049011,
      "learning_rate": 1.6587e-05,
      "loss": 0.0625,
      "step": 5532
    },
    {
      "epoch": 10.208679593721145,
      "grad_norm": 0.5196364521980286,
      "learning_rate": 1.6590000000000002e-05,
      "loss": 0.0564,
      "step": 5533
    },
    {
      "epoch": 10.210526315789474,
      "grad_norm": 0.5493685603141785,
      "learning_rate": 1.6593000000000002e-05,
      "loss": 0.048,
      "step": 5534
    },
    {
      "epoch": 10.212373037857802,
      "grad_norm": 0.6512869000434875,
      "learning_rate": 1.6596000000000002e-05,
      "loss": 0.065,
      "step": 5535
    },
    {
      "epoch": 10.21421975992613,
      "grad_norm": 0.5324305295944214,
      "learning_rate": 1.6599e-05,
      "loss": 0.0356,
      "step": 5536
    },
    {
      "epoch": 10.21606648199446,
      "grad_norm": 0.4373769164085388,
      "learning_rate": 1.6602e-05,
      "loss": 0.041,
      "step": 5537
    },
    {
      "epoch": 10.21791320406279,
      "grad_norm": 0.880097508430481,
      "learning_rate": 1.6605e-05,
      "loss": 0.0487,
      "step": 5538
    },
    {
      "epoch": 10.219759926131117,
      "grad_norm": 0.47192707657814026,
      "learning_rate": 1.6608e-05,
      "loss": 0.0328,
      "step": 5539
    },
    {
      "epoch": 10.221606648199446,
      "grad_norm": 0.4588438868522644,
      "learning_rate": 1.6611e-05,
      "loss": 0.0295,
      "step": 5540
    },
    {
      "epoch": 10.223453370267775,
      "grad_norm": 1.0013659000396729,
      "learning_rate": 1.6614e-05,
      "loss": 0.0305,
      "step": 5541
    },
    {
      "epoch": 10.225300092336104,
      "grad_norm": 0.5650994181632996,
      "learning_rate": 1.6617e-05,
      "loss": 0.0268,
      "step": 5542
    },
    {
      "epoch": 10.227146814404431,
      "grad_norm": 0.46585944294929504,
      "learning_rate": 1.6620000000000004e-05,
      "loss": 0.0288,
      "step": 5543
    },
    {
      "epoch": 10.22899353647276,
      "grad_norm": 0.528458833694458,
      "learning_rate": 1.6623e-05,
      "loss": 0.0294,
      "step": 5544
    },
    {
      "epoch": 10.23084025854109,
      "grad_norm": 0.4273636043071747,
      "learning_rate": 1.6626e-05,
      "loss": 0.0301,
      "step": 5545
    },
    {
      "epoch": 10.232686980609419,
      "grad_norm": 0.4764959514141083,
      "learning_rate": 1.6629e-05,
      "loss": 0.033,
      "step": 5546
    },
    {
      "epoch": 10.234533702677748,
      "grad_norm": 0.4240407943725586,
      "learning_rate": 1.6632e-05,
      "loss": 0.0642,
      "step": 5547
    },
    {
      "epoch": 10.236380424746075,
      "grad_norm": 0.4711458086967468,
      "learning_rate": 1.6635e-05,
      "loss": 0.0262,
      "step": 5548
    },
    {
      "epoch": 10.238227146814404,
      "grad_norm": 0.6241109371185303,
      "learning_rate": 1.6638e-05,
      "loss": 0.0384,
      "step": 5549
    },
    {
      "epoch": 10.240073868882734,
      "grad_norm": 0.37781938910484314,
      "learning_rate": 1.6641e-05,
      "loss": 0.0184,
      "step": 5550
    },
    {
      "epoch": 10.241920590951063,
      "grad_norm": 0.4634665548801422,
      "learning_rate": 1.6644e-05,
      "loss": 0.0156,
      "step": 5551
    },
    {
      "epoch": 10.24376731301939,
      "grad_norm": 0.5928099155426025,
      "learning_rate": 1.6647e-05,
      "loss": 0.0205,
      "step": 5552
    },
    {
      "epoch": 10.24561403508772,
      "grad_norm": 0.38055887818336487,
      "learning_rate": 1.665e-05,
      "loss": 0.022,
      "step": 5553
    },
    {
      "epoch": 10.247460757156048,
      "grad_norm": 0.8946309685707092,
      "learning_rate": 1.6653e-05,
      "loss": 0.0341,
      "step": 5554
    },
    {
      "epoch": 10.249307479224377,
      "grad_norm": 0.8753093481063843,
      "learning_rate": 1.6656e-05,
      "loss": 0.0254,
      "step": 5555
    },
    {
      "epoch": 10.251154201292705,
      "grad_norm": 0.44015374779701233,
      "learning_rate": 1.6659e-05,
      "loss": 0.0263,
      "step": 5556
    },
    {
      "epoch": 10.253000923361034,
      "grad_norm": 0.39704692363739014,
      "learning_rate": 1.6662e-05,
      "loss": 0.0269,
      "step": 5557
    },
    {
      "epoch": 10.254847645429363,
      "grad_norm": 0.4797463119029999,
      "learning_rate": 1.6665e-05,
      "loss": 0.0266,
      "step": 5558
    },
    {
      "epoch": 10.256694367497692,
      "grad_norm": 0.4630764126777649,
      "learning_rate": 1.6668e-05,
      "loss": 0.039,
      "step": 5559
    },
    {
      "epoch": 10.25854108956602,
      "grad_norm": 1.1624733209609985,
      "learning_rate": 1.6671e-05,
      "loss": 0.0357,
      "step": 5560
    },
    {
      "epoch": 10.260387811634349,
      "grad_norm": 0.6615108251571655,
      "learning_rate": 1.6674e-05,
      "loss": 0.0284,
      "step": 5561
    },
    {
      "epoch": 10.262234533702678,
      "grad_norm": 0.5912197828292847,
      "learning_rate": 1.6677e-05,
      "loss": 0.0233,
      "step": 5562
    },
    {
      "epoch": 10.264081255771007,
      "grad_norm": 0.5102552771568298,
      "learning_rate": 1.6680000000000003e-05,
      "loss": 0.0235,
      "step": 5563
    },
    {
      "epoch": 10.265927977839334,
      "grad_norm": 0.49518391489982605,
      "learning_rate": 1.6683000000000003e-05,
      "loss": 0.0322,
      "step": 5564
    },
    {
      "epoch": 10.267774699907664,
      "grad_norm": 0.9089075922966003,
      "learning_rate": 1.6686000000000003e-05,
      "loss": 0.024,
      "step": 5565
    },
    {
      "epoch": 10.269621421975993,
      "grad_norm": 0.49683696031570435,
      "learning_rate": 1.6689000000000002e-05,
      "loss": 0.0216,
      "step": 5566
    },
    {
      "epoch": 10.271468144044322,
      "grad_norm": 0.3209707736968994,
      "learning_rate": 1.6692000000000002e-05,
      "loss": 0.0191,
      "step": 5567
    },
    {
      "epoch": 10.27331486611265,
      "grad_norm": 0.567535400390625,
      "learning_rate": 1.6695000000000002e-05,
      "loss": 0.0253,
      "step": 5568
    },
    {
      "epoch": 10.275161588180978,
      "grad_norm": 0.8037060499191284,
      "learning_rate": 1.6698e-05,
      "loss": 0.0349,
      "step": 5569
    },
    {
      "epoch": 10.277008310249307,
      "grad_norm": 0.6364205479621887,
      "learning_rate": 1.6700999999999998e-05,
      "loss": 0.0263,
      "step": 5570
    },
    {
      "epoch": 10.278855032317637,
      "grad_norm": 1.1712576150894165,
      "learning_rate": 1.6703999999999998e-05,
      "loss": 0.2833,
      "step": 5571
    },
    {
      "epoch": 10.280701754385966,
      "grad_norm": 0.941981852054596,
      "learning_rate": 1.6706999999999998e-05,
      "loss": 0.247,
      "step": 5572
    },
    {
      "epoch": 10.282548476454293,
      "grad_norm": 0.8911486864089966,
      "learning_rate": 1.671e-05,
      "loss": 0.1956,
      "step": 5573
    },
    {
      "epoch": 10.284395198522622,
      "grad_norm": 0.7403378486633301,
      "learning_rate": 1.6713e-05,
      "loss": 0.1719,
      "step": 5574
    },
    {
      "epoch": 10.286241920590951,
      "grad_norm": 0.7996092438697815,
      "learning_rate": 1.6716e-05,
      "loss": 0.2207,
      "step": 5575
    },
    {
      "epoch": 10.28808864265928,
      "grad_norm": 0.6285210847854614,
      "learning_rate": 1.6719e-05,
      "loss": 0.1455,
      "step": 5576
    },
    {
      "epoch": 10.289935364727608,
      "grad_norm": 0.5551503896713257,
      "learning_rate": 1.6722e-05,
      "loss": 0.1133,
      "step": 5577
    },
    {
      "epoch": 10.291782086795937,
      "grad_norm": 0.8680622577667236,
      "learning_rate": 1.6725e-05,
      "loss": 0.1461,
      "step": 5578
    },
    {
      "epoch": 10.293628808864266,
      "grad_norm": 0.5695521235466003,
      "learning_rate": 1.6728e-05,
      "loss": 0.0919,
      "step": 5579
    },
    {
      "epoch": 10.295475530932595,
      "grad_norm": 0.557476818561554,
      "learning_rate": 1.6731e-05,
      "loss": 0.0961,
      "step": 5580
    },
    {
      "epoch": 10.297322253000923,
      "grad_norm": 0.6611366868019104,
      "learning_rate": 1.6734e-05,
      "loss": 0.083,
      "step": 5581
    },
    {
      "epoch": 10.299168975069252,
      "grad_norm": 0.5024586319923401,
      "learning_rate": 1.6737e-05,
      "loss": 0.0883,
      "step": 5582
    },
    {
      "epoch": 10.301015697137581,
      "grad_norm": 0.6985763311386108,
      "learning_rate": 1.6740000000000002e-05,
      "loss": 0.1743,
      "step": 5583
    },
    {
      "epoch": 10.30286241920591,
      "grad_norm": 1.2900344133377075,
      "learning_rate": 1.6743000000000002e-05,
      "loss": 0.0664,
      "step": 5584
    },
    {
      "epoch": 10.304709141274238,
      "grad_norm": 0.5589216947555542,
      "learning_rate": 1.6746000000000002e-05,
      "loss": 0.1042,
      "step": 5585
    },
    {
      "epoch": 10.306555863342567,
      "grad_norm": 0.47162994742393494,
      "learning_rate": 1.6749000000000002e-05,
      "loss": 0.0592,
      "step": 5586
    },
    {
      "epoch": 10.308402585410896,
      "grad_norm": 0.6191331148147583,
      "learning_rate": 1.6752e-05,
      "loss": 0.0937,
      "step": 5587
    },
    {
      "epoch": 10.310249307479225,
      "grad_norm": 0.3441529870033264,
      "learning_rate": 1.6755e-05,
      "loss": 0.03,
      "step": 5588
    },
    {
      "epoch": 10.312096029547552,
      "grad_norm": 0.2752731144428253,
      "learning_rate": 1.6758e-05,
      "loss": 0.0227,
      "step": 5589
    },
    {
      "epoch": 10.313942751615881,
      "grad_norm": 0.4862821698188782,
      "learning_rate": 1.6761e-05,
      "loss": 0.0195,
      "step": 5590
    },
    {
      "epoch": 10.31578947368421,
      "grad_norm": 0.40299078822135925,
      "learning_rate": 1.6764e-05,
      "loss": 0.0218,
      "step": 5591
    },
    {
      "epoch": 10.31763619575254,
      "grad_norm": 0.4200470745563507,
      "learning_rate": 1.6767e-05,
      "loss": 0.0201,
      "step": 5592
    },
    {
      "epoch": 10.319482917820867,
      "grad_norm": 0.7158517241477966,
      "learning_rate": 1.677e-05,
      "loss": 0.0411,
      "step": 5593
    },
    {
      "epoch": 10.321329639889196,
      "grad_norm": 0.8751426935195923,
      "learning_rate": 1.6773e-05,
      "loss": 0.025,
      "step": 5594
    },
    {
      "epoch": 10.323176361957525,
      "grad_norm": 0.5317754149436951,
      "learning_rate": 1.6776e-05,
      "loss": 0.0209,
      "step": 5595
    },
    {
      "epoch": 10.325023084025855,
      "grad_norm": 0.4970802664756775,
      "learning_rate": 1.6779e-05,
      "loss": 0.0179,
      "step": 5596
    },
    {
      "epoch": 10.326869806094184,
      "grad_norm": 0.6410852670669556,
      "learning_rate": 1.6782e-05,
      "loss": 0.0358,
      "step": 5597
    },
    {
      "epoch": 10.328716528162511,
      "grad_norm": 0.6019672155380249,
      "learning_rate": 1.6785e-05,
      "loss": 0.0274,
      "step": 5598
    },
    {
      "epoch": 10.33056325023084,
      "grad_norm": 0.484224796295166,
      "learning_rate": 1.6788e-05,
      "loss": 0.0514,
      "step": 5599
    },
    {
      "epoch": 10.33240997229917,
      "grad_norm": 0.7805466055870056,
      "learning_rate": 1.6791e-05,
      "loss": 0.019,
      "step": 5600
    },
    {
      "epoch": 10.334256694367498,
      "grad_norm": 0.4458668529987335,
      "learning_rate": 1.6794e-05,
      "loss": 0.0247,
      "step": 5601
    },
    {
      "epoch": 10.336103416435826,
      "grad_norm": 0.7102533578872681,
      "learning_rate": 1.6797e-05,
      "loss": 0.0279,
      "step": 5602
    },
    {
      "epoch": 10.337950138504155,
      "grad_norm": 0.5157403349876404,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.0196,
      "step": 5603
    },
    {
      "epoch": 10.339796860572484,
      "grad_norm": 0.41661450266838074,
      "learning_rate": 1.6803e-05,
      "loss": 0.0266,
      "step": 5604
    },
    {
      "epoch": 10.341643582640813,
      "grad_norm": 0.47312384843826294,
      "learning_rate": 1.6806e-05,
      "loss": 0.0152,
      "step": 5605
    },
    {
      "epoch": 10.34349030470914,
      "grad_norm": 0.9088107943534851,
      "learning_rate": 1.6809e-05,
      "loss": 0.0176,
      "step": 5606
    },
    {
      "epoch": 10.34533702677747,
      "grad_norm": 0.7505862712860107,
      "learning_rate": 1.6812e-05,
      "loss": 0.0283,
      "step": 5607
    },
    {
      "epoch": 10.347183748845799,
      "grad_norm": 0.49464118480682373,
      "learning_rate": 1.6815e-05,
      "loss": 0.0221,
      "step": 5608
    },
    {
      "epoch": 10.349030470914128,
      "grad_norm": 0.7184174656867981,
      "learning_rate": 1.6818e-05,
      "loss": 0.0328,
      "step": 5609
    },
    {
      "epoch": 10.350877192982455,
      "grad_norm": 0.5819430947303772,
      "learning_rate": 1.6821e-05,
      "loss": 0.0281,
      "step": 5610
    },
    {
      "epoch": 10.352723915050785,
      "grad_norm": 0.4279172718524933,
      "learning_rate": 1.6824e-05,
      "loss": 0.0223,
      "step": 5611
    },
    {
      "epoch": 10.354570637119114,
      "grad_norm": 0.764571487903595,
      "learning_rate": 1.6827e-05,
      "loss": 0.0269,
      "step": 5612
    },
    {
      "epoch": 10.356417359187443,
      "grad_norm": 0.6644816398620605,
      "learning_rate": 1.6830000000000003e-05,
      "loss": 0.0284,
      "step": 5613
    },
    {
      "epoch": 10.35826408125577,
      "grad_norm": 0.5781840085983276,
      "learning_rate": 1.6833000000000003e-05,
      "loss": 0.0405,
      "step": 5614
    },
    {
      "epoch": 10.3601108033241,
      "grad_norm": 0.5660284161567688,
      "learning_rate": 1.6836000000000003e-05,
      "loss": 0.0254,
      "step": 5615
    },
    {
      "epoch": 10.361957525392429,
      "grad_norm": 0.47080352902412415,
      "learning_rate": 1.6839000000000003e-05,
      "loss": 0.0246,
      "step": 5616
    },
    {
      "epoch": 10.363804247460758,
      "grad_norm": 0.5529908537864685,
      "learning_rate": 1.6842000000000002e-05,
      "loss": 0.0368,
      "step": 5617
    },
    {
      "epoch": 10.365650969529085,
      "grad_norm": 0.3346571624279022,
      "learning_rate": 1.6845e-05,
      "loss": 0.0129,
      "step": 5618
    },
    {
      "epoch": 10.367497691597414,
      "grad_norm": 0.9121596813201904,
      "learning_rate": 1.6848e-05,
      "loss": 0.0431,
      "step": 5619
    },
    {
      "epoch": 10.369344413665743,
      "grad_norm": 0.632863461971283,
      "learning_rate": 1.6851e-05,
      "loss": 0.0424,
      "step": 5620
    },
    {
      "epoch": 10.371191135734072,
      "grad_norm": 1.32807457447052,
      "learning_rate": 1.6853999999999998e-05,
      "loss": 0.2963,
      "step": 5621
    },
    {
      "epoch": 10.373037857802402,
      "grad_norm": 0.9964393973350525,
      "learning_rate": 1.6856999999999998e-05,
      "loss": 0.2261,
      "step": 5622
    },
    {
      "epoch": 10.374884579870729,
      "grad_norm": 1.0543564558029175,
      "learning_rate": 1.686e-05,
      "loss": 0.2349,
      "step": 5623
    },
    {
      "epoch": 10.376731301939058,
      "grad_norm": 0.904670000076294,
      "learning_rate": 1.6863e-05,
      "loss": 0.2302,
      "step": 5624
    },
    {
      "epoch": 10.378578024007387,
      "grad_norm": 2.453029155731201,
      "learning_rate": 1.6866e-05,
      "loss": 0.2143,
      "step": 5625
    },
    {
      "epoch": 10.380424746075716,
      "grad_norm": 0.6642909646034241,
      "learning_rate": 1.6869e-05,
      "loss": 0.1311,
      "step": 5626
    },
    {
      "epoch": 10.382271468144044,
      "grad_norm": 0.669899582862854,
      "learning_rate": 1.6872e-05,
      "loss": 0.1325,
      "step": 5627
    },
    {
      "epoch": 10.384118190212373,
      "grad_norm": 0.9939351677894592,
      "learning_rate": 1.6875e-05,
      "loss": 0.1419,
      "step": 5628
    },
    {
      "epoch": 10.385964912280702,
      "grad_norm": 0.9026162624359131,
      "learning_rate": 1.6878e-05,
      "loss": 0.0871,
      "step": 5629
    },
    {
      "epoch": 10.387811634349031,
      "grad_norm": 0.716627299785614,
      "learning_rate": 1.6881e-05,
      "loss": 0.0975,
      "step": 5630
    },
    {
      "epoch": 10.389658356417359,
      "grad_norm": 0.48917216062545776,
      "learning_rate": 1.6884e-05,
      "loss": 0.0757,
      "step": 5631
    },
    {
      "epoch": 10.391505078485688,
      "grad_norm": 0.6990576982498169,
      "learning_rate": 1.6887e-05,
      "loss": 0.1063,
      "step": 5632
    },
    {
      "epoch": 10.393351800554017,
      "grad_norm": 0.3678179681301117,
      "learning_rate": 1.689e-05,
      "loss": 0.0446,
      "step": 5633
    },
    {
      "epoch": 10.395198522622346,
      "grad_norm": 0.44908279180526733,
      "learning_rate": 1.6893000000000002e-05,
      "loss": 0.0445,
      "step": 5634
    },
    {
      "epoch": 10.397045244690673,
      "grad_norm": 0.5445517897605896,
      "learning_rate": 1.6896000000000002e-05,
      "loss": 0.0554,
      "step": 5635
    },
    {
      "epoch": 10.398891966759003,
      "grad_norm": 0.5841280817985535,
      "learning_rate": 1.6899000000000002e-05,
      "loss": 0.0216,
      "step": 5636
    },
    {
      "epoch": 10.400738688827332,
      "grad_norm": 0.38346558809280396,
      "learning_rate": 1.6902000000000002e-05,
      "loss": 0.0285,
      "step": 5637
    },
    {
      "epoch": 10.40258541089566,
      "grad_norm": 0.5853798389434814,
      "learning_rate": 1.6905e-05,
      "loss": 0.0246,
      "step": 5638
    },
    {
      "epoch": 10.404432132963988,
      "grad_norm": 0.5012631416320801,
      "learning_rate": 1.6908e-05,
      "loss": 0.0289,
      "step": 5639
    },
    {
      "epoch": 10.406278855032317,
      "grad_norm": 0.5409550666809082,
      "learning_rate": 1.6911e-05,
      "loss": 0.0317,
      "step": 5640
    },
    {
      "epoch": 10.408125577100646,
      "grad_norm": 0.390082448720932,
      "learning_rate": 1.6914e-05,
      "loss": 0.0208,
      "step": 5641
    },
    {
      "epoch": 10.409972299168976,
      "grad_norm": 0.6462604999542236,
      "learning_rate": 1.6916999999999997e-05,
      "loss": 0.0282,
      "step": 5642
    },
    {
      "epoch": 10.411819021237303,
      "grad_norm": 0.5268653631210327,
      "learning_rate": 1.6919999999999997e-05,
      "loss": 0.0282,
      "step": 5643
    },
    {
      "epoch": 10.413665743305632,
      "grad_norm": 0.41686591506004333,
      "learning_rate": 1.6923e-05,
      "loss": 0.0189,
      "step": 5644
    },
    {
      "epoch": 10.415512465373961,
      "grad_norm": 0.4445785582065582,
      "learning_rate": 1.6926e-05,
      "loss": 0.0317,
      "step": 5645
    },
    {
      "epoch": 10.41735918744229,
      "grad_norm": 1.0424392223358154,
      "learning_rate": 1.6929e-05,
      "loss": 0.0307,
      "step": 5646
    },
    {
      "epoch": 10.41920590951062,
      "grad_norm": 0.38552019000053406,
      "learning_rate": 1.6932e-05,
      "loss": 0.0336,
      "step": 5647
    },
    {
      "epoch": 10.421052631578947,
      "grad_norm": 0.5564932823181152,
      "learning_rate": 1.6935e-05,
      "loss": 0.0234,
      "step": 5648
    },
    {
      "epoch": 10.422899353647276,
      "grad_norm": 0.45230332016944885,
      "learning_rate": 1.6938e-05,
      "loss": 0.0286,
      "step": 5649
    },
    {
      "epoch": 10.424746075715605,
      "grad_norm": 0.5506929755210876,
      "learning_rate": 1.6941e-05,
      "loss": 0.035,
      "step": 5650
    },
    {
      "epoch": 10.426592797783934,
      "grad_norm": 0.39401277899742126,
      "learning_rate": 1.6944e-05,
      "loss": 0.0451,
      "step": 5651
    },
    {
      "epoch": 10.428439519852262,
      "grad_norm": 0.4682321846485138,
      "learning_rate": 1.6947e-05,
      "loss": 0.0223,
      "step": 5652
    },
    {
      "epoch": 10.43028624192059,
      "grad_norm": 0.4107759892940521,
      "learning_rate": 1.695e-05,
      "loss": 0.0194,
      "step": 5653
    },
    {
      "epoch": 10.43213296398892,
      "grad_norm": 0.38645249605178833,
      "learning_rate": 1.6953000000000002e-05,
      "loss": 0.0196,
      "step": 5654
    },
    {
      "epoch": 10.43397968605725,
      "grad_norm": 0.48322468996047974,
      "learning_rate": 1.6956e-05,
      "loss": 0.0163,
      "step": 5655
    },
    {
      "epoch": 10.435826408125576,
      "grad_norm": 0.5782933235168457,
      "learning_rate": 1.6959e-05,
      "loss": 0.0313,
      "step": 5656
    },
    {
      "epoch": 10.437673130193906,
      "grad_norm": 0.5378473997116089,
      "learning_rate": 1.6962e-05,
      "loss": 0.04,
      "step": 5657
    },
    {
      "epoch": 10.439519852262235,
      "grad_norm": 0.4453236162662506,
      "learning_rate": 1.6965e-05,
      "loss": 0.0222,
      "step": 5658
    },
    {
      "epoch": 10.441366574330564,
      "grad_norm": 0.6218374967575073,
      "learning_rate": 1.6968e-05,
      "loss": 0.0322,
      "step": 5659
    },
    {
      "epoch": 10.443213296398891,
      "grad_norm": 0.6248608231544495,
      "learning_rate": 1.6971e-05,
      "loss": 0.0274,
      "step": 5660
    },
    {
      "epoch": 10.44506001846722,
      "grad_norm": 0.4168800413608551,
      "learning_rate": 1.6974e-05,
      "loss": 0.0258,
      "step": 5661
    },
    {
      "epoch": 10.44690674053555,
      "grad_norm": 0.3333655297756195,
      "learning_rate": 1.6977e-05,
      "loss": 0.0237,
      "step": 5662
    },
    {
      "epoch": 10.448753462603879,
      "grad_norm": 0.28354814648628235,
      "learning_rate": 1.698e-05,
      "loss": 0.0163,
      "step": 5663
    },
    {
      "epoch": 10.450600184672206,
      "grad_norm": 0.5972402095794678,
      "learning_rate": 1.6983000000000003e-05,
      "loss": 0.033,
      "step": 5664
    },
    {
      "epoch": 10.452446906740535,
      "grad_norm": 0.3524409830570221,
      "learning_rate": 1.6986000000000003e-05,
      "loss": 0.0125,
      "step": 5665
    },
    {
      "epoch": 10.454293628808864,
      "grad_norm": 0.46399515867233276,
      "learning_rate": 1.6989000000000003e-05,
      "loss": 0.0194,
      "step": 5666
    },
    {
      "epoch": 10.456140350877194,
      "grad_norm": 0.5713216066360474,
      "learning_rate": 1.6992e-05,
      "loss": 0.0223,
      "step": 5667
    },
    {
      "epoch": 10.45798707294552,
      "grad_norm": 0.5732083916664124,
      "learning_rate": 1.6995e-05,
      "loss": 0.018,
      "step": 5668
    },
    {
      "epoch": 10.45983379501385,
      "grad_norm": 0.5675196647644043,
      "learning_rate": 1.6998e-05,
      "loss": 0.0243,
      "step": 5669
    },
    {
      "epoch": 10.46168051708218,
      "grad_norm": 0.6661143898963928,
      "learning_rate": 1.7001e-05,
      "loss": 0.0647,
      "step": 5670
    },
    {
      "epoch": 10.463527239150508,
      "grad_norm": 0.9836185574531555,
      "learning_rate": 1.7004e-05,
      "loss": 0.2418,
      "step": 5671
    },
    {
      "epoch": 10.465373961218837,
      "grad_norm": 0.9940314888954163,
      "learning_rate": 1.7006999999999998e-05,
      "loss": 0.2224,
      "step": 5672
    },
    {
      "epoch": 10.467220683287165,
      "grad_norm": 0.9100486636161804,
      "learning_rate": 1.7009999999999998e-05,
      "loss": 0.2137,
      "step": 5673
    },
    {
      "epoch": 10.469067405355494,
      "grad_norm": 0.7420975565910339,
      "learning_rate": 1.7013e-05,
      "loss": 0.1624,
      "step": 5674
    },
    {
      "epoch": 10.470914127423823,
      "grad_norm": 0.7102513909339905,
      "learning_rate": 1.7016e-05,
      "loss": 0.1962,
      "step": 5675
    },
    {
      "epoch": 10.472760849492152,
      "grad_norm": 0.7101590633392334,
      "learning_rate": 1.7019e-05,
      "loss": 0.1239,
      "step": 5676
    },
    {
      "epoch": 10.47460757156048,
      "grad_norm": 0.8599352240562439,
      "learning_rate": 1.7022e-05,
      "loss": 0.1271,
      "step": 5677
    },
    {
      "epoch": 10.476454293628809,
      "grad_norm": 0.9525429010391235,
      "learning_rate": 1.7025e-05,
      "loss": 0.1764,
      "step": 5678
    },
    {
      "epoch": 10.478301015697138,
      "grad_norm": 0.6291274428367615,
      "learning_rate": 1.7028e-05,
      "loss": 0.0883,
      "step": 5679
    },
    {
      "epoch": 10.480147737765467,
      "grad_norm": 1.0020326375961304,
      "learning_rate": 1.7031e-05,
      "loss": 0.1492,
      "step": 5680
    },
    {
      "epoch": 10.481994459833794,
      "grad_norm": 0.7298393249511719,
      "learning_rate": 1.7034e-05,
      "loss": 0.1428,
      "step": 5681
    },
    {
      "epoch": 10.483841181902124,
      "grad_norm": 0.6903399229049683,
      "learning_rate": 1.7037e-05,
      "loss": 0.1012,
      "step": 5682
    },
    {
      "epoch": 10.485687903970453,
      "grad_norm": 0.5499876141548157,
      "learning_rate": 1.704e-05,
      "loss": 0.0598,
      "step": 5683
    },
    {
      "epoch": 10.487534626038782,
      "grad_norm": 0.886182963848114,
      "learning_rate": 1.7043000000000003e-05,
      "loss": 0.1237,
      "step": 5684
    },
    {
      "epoch": 10.48938134810711,
      "grad_norm": 0.6885615587234497,
      "learning_rate": 1.7046000000000002e-05,
      "loss": 0.088,
      "step": 5685
    },
    {
      "epoch": 10.491228070175438,
      "grad_norm": 0.5300734639167786,
      "learning_rate": 1.7049000000000002e-05,
      "loss": 0.0481,
      "step": 5686
    },
    {
      "epoch": 10.493074792243767,
      "grad_norm": 0.6662781834602356,
      "learning_rate": 1.7052000000000002e-05,
      "loss": 0.0389,
      "step": 5687
    },
    {
      "epoch": 10.494921514312097,
      "grad_norm": 0.5047277808189392,
      "learning_rate": 1.7055000000000002e-05,
      "loss": 0.0163,
      "step": 5688
    },
    {
      "epoch": 10.496768236380424,
      "grad_norm": 0.5463379621505737,
      "learning_rate": 1.7058e-05,
      "loss": 0.04,
      "step": 5689
    },
    {
      "epoch": 10.498614958448753,
      "grad_norm": 0.5748884677886963,
      "learning_rate": 1.7061e-05,
      "loss": 0.0391,
      "step": 5690
    },
    {
      "epoch": 10.500461680517082,
      "grad_norm": 0.5658259987831116,
      "learning_rate": 1.7064e-05,
      "loss": 0.0291,
      "step": 5691
    },
    {
      "epoch": 10.502308402585411,
      "grad_norm": 0.36395278573036194,
      "learning_rate": 1.7066999999999998e-05,
      "loss": 0.0276,
      "step": 5692
    },
    {
      "epoch": 10.504155124653739,
      "grad_norm": 0.5200968980789185,
      "learning_rate": 1.7069999999999998e-05,
      "loss": 0.0344,
      "step": 5693
    },
    {
      "epoch": 10.506001846722068,
      "grad_norm": 0.41553211212158203,
      "learning_rate": 1.7073e-05,
      "loss": 0.0196,
      "step": 5694
    },
    {
      "epoch": 10.507848568790397,
      "grad_norm": 0.5262542366981506,
      "learning_rate": 1.7076e-05,
      "loss": 0.0177,
      "step": 5695
    },
    {
      "epoch": 10.509695290858726,
      "grad_norm": 0.4412112832069397,
      "learning_rate": 1.7079e-05,
      "loss": 0.025,
      "step": 5696
    },
    {
      "epoch": 10.511542012927055,
      "grad_norm": 0.37101516127586365,
      "learning_rate": 1.7082e-05,
      "loss": 0.0235,
      "step": 5697
    },
    {
      "epoch": 10.513388734995383,
      "grad_norm": 0.401090145111084,
      "learning_rate": 1.7085e-05,
      "loss": 0.023,
      "step": 5698
    },
    {
      "epoch": 10.515235457063712,
      "grad_norm": 0.46183472871780396,
      "learning_rate": 1.7088e-05,
      "loss": 0.0154,
      "step": 5699
    },
    {
      "epoch": 10.517082179132041,
      "grad_norm": 0.45999205112457275,
      "learning_rate": 1.7091e-05,
      "loss": 0.0305,
      "step": 5700
    },
    {
      "epoch": 10.51892890120037,
      "grad_norm": 0.444741815328598,
      "learning_rate": 1.7094e-05,
      "loss": 0.0275,
      "step": 5701
    },
    {
      "epoch": 10.520775623268698,
      "grad_norm": 0.4720512926578522,
      "learning_rate": 1.7097e-05,
      "loss": 0.0363,
      "step": 5702
    },
    {
      "epoch": 10.522622345337027,
      "grad_norm": 0.4139270782470703,
      "learning_rate": 1.71e-05,
      "loss": 0.0308,
      "step": 5703
    },
    {
      "epoch": 10.524469067405356,
      "grad_norm": 0.404547780752182,
      "learning_rate": 1.7103000000000002e-05,
      "loss": 0.0192,
      "step": 5704
    },
    {
      "epoch": 10.526315789473685,
      "grad_norm": 0.3558710813522339,
      "learning_rate": 1.7106000000000002e-05,
      "loss": 0.0198,
      "step": 5705
    },
    {
      "epoch": 10.528162511542012,
      "grad_norm": 0.8504428863525391,
      "learning_rate": 1.7109000000000002e-05,
      "loss": 0.0144,
      "step": 5706
    },
    {
      "epoch": 10.530009233610341,
      "grad_norm": 0.5197940468788147,
      "learning_rate": 1.7112e-05,
      "loss": 0.0244,
      "step": 5707
    },
    {
      "epoch": 10.53185595567867,
      "grad_norm": 0.5138944983482361,
      "learning_rate": 1.7115e-05,
      "loss": 0.0306,
      "step": 5708
    },
    {
      "epoch": 10.533702677747,
      "grad_norm": 0.530799925327301,
      "learning_rate": 1.7118e-05,
      "loss": 0.0253,
      "step": 5709
    },
    {
      "epoch": 10.535549399815327,
      "grad_norm": 0.751099705696106,
      "learning_rate": 1.7121e-05,
      "loss": 0.0442,
      "step": 5710
    },
    {
      "epoch": 10.537396121883656,
      "grad_norm": 0.4729883670806885,
      "learning_rate": 1.7124e-05,
      "loss": 0.0351,
      "step": 5711
    },
    {
      "epoch": 10.539242843951985,
      "grad_norm": 0.3551481068134308,
      "learning_rate": 1.7127e-05,
      "loss": 0.013,
      "step": 5712
    },
    {
      "epoch": 10.541089566020315,
      "grad_norm": 0.38660553097724915,
      "learning_rate": 1.713e-05,
      "loss": 0.0188,
      "step": 5713
    },
    {
      "epoch": 10.542936288088642,
      "grad_norm": 0.3219395577907562,
      "learning_rate": 1.7133000000000004e-05,
      "loss": 0.0166,
      "step": 5714
    },
    {
      "epoch": 10.544783010156971,
      "grad_norm": 0.4016413688659668,
      "learning_rate": 1.7136000000000003e-05,
      "loss": 0.0287,
      "step": 5715
    },
    {
      "epoch": 10.5466297322253,
      "grad_norm": 0.5673303008079529,
      "learning_rate": 1.7139e-05,
      "loss": 0.0269,
      "step": 5716
    },
    {
      "epoch": 10.54847645429363,
      "grad_norm": 0.6439066529273987,
      "learning_rate": 1.7142e-05,
      "loss": 0.0156,
      "step": 5717
    },
    {
      "epoch": 10.550323176361957,
      "grad_norm": 0.7035136818885803,
      "learning_rate": 1.7145e-05,
      "loss": 0.0264,
      "step": 5718
    },
    {
      "epoch": 10.552169898430286,
      "grad_norm": 0.5405717492103577,
      "learning_rate": 1.7148e-05,
      "loss": 0.0269,
      "step": 5719
    },
    {
      "epoch": 10.554016620498615,
      "grad_norm": 0.7340378761291504,
      "learning_rate": 1.7151e-05,
      "loss": 0.0369,
      "step": 5720
    },
    {
      "epoch": 10.555863342566944,
      "grad_norm": 0.8675962686538696,
      "learning_rate": 1.7154e-05,
      "loss": 0.2132,
      "step": 5721
    },
    {
      "epoch": 10.557710064635273,
      "grad_norm": 0.6791530847549438,
      "learning_rate": 1.7157e-05,
      "loss": 0.2311,
      "step": 5722
    },
    {
      "epoch": 10.5595567867036,
      "grad_norm": 0.6052650213241577,
      "learning_rate": 1.716e-05,
      "loss": 0.1708,
      "step": 5723
    },
    {
      "epoch": 10.56140350877193,
      "grad_norm": 0.9645447134971619,
      "learning_rate": 1.7163e-05,
      "loss": 0.1836,
      "step": 5724
    },
    {
      "epoch": 10.563250230840259,
      "grad_norm": 1.0505342483520508,
      "learning_rate": 1.7166e-05,
      "loss": 0.1674,
      "step": 5725
    },
    {
      "epoch": 10.565096952908588,
      "grad_norm": 0.829861581325531,
      "learning_rate": 1.7169e-05,
      "loss": 0.1579,
      "step": 5726
    },
    {
      "epoch": 10.566943674976915,
      "grad_norm": 0.6052542328834534,
      "learning_rate": 1.7172e-05,
      "loss": 0.1572,
      "step": 5727
    },
    {
      "epoch": 10.568790397045245,
      "grad_norm": 0.6950606107711792,
      "learning_rate": 1.7175e-05,
      "loss": 0.1262,
      "step": 5728
    },
    {
      "epoch": 10.570637119113574,
      "grad_norm": 1.024465799331665,
      "learning_rate": 1.7178e-05,
      "loss": 0.1287,
      "step": 5729
    },
    {
      "epoch": 10.572483841181903,
      "grad_norm": 0.4541202485561371,
      "learning_rate": 1.7181e-05,
      "loss": 0.0765,
      "step": 5730
    },
    {
      "epoch": 10.57433056325023,
      "grad_norm": 0.7202557325363159,
      "learning_rate": 1.7184e-05,
      "loss": 0.0876,
      "step": 5731
    },
    {
      "epoch": 10.57617728531856,
      "grad_norm": 0.8201732635498047,
      "learning_rate": 1.7187e-05,
      "loss": 0.1163,
      "step": 5732
    },
    {
      "epoch": 10.578024007386889,
      "grad_norm": 1.2063844203948975,
      "learning_rate": 1.719e-05,
      "loss": 0.0757,
      "step": 5733
    },
    {
      "epoch": 10.579870729455218,
      "grad_norm": 0.510230541229248,
      "learning_rate": 1.7193000000000003e-05,
      "loss": 0.0441,
      "step": 5734
    },
    {
      "epoch": 10.581717451523545,
      "grad_norm": 0.6762166619300842,
      "learning_rate": 1.7196000000000003e-05,
      "loss": 0.1082,
      "step": 5735
    },
    {
      "epoch": 10.583564173591874,
      "grad_norm": 0.4172094762325287,
      "learning_rate": 1.7199000000000003e-05,
      "loss": 0.0368,
      "step": 5736
    },
    {
      "epoch": 10.585410895660203,
      "grad_norm": 0.3209916949272156,
      "learning_rate": 1.7202000000000002e-05,
      "loss": 0.0304,
      "step": 5737
    },
    {
      "epoch": 10.587257617728532,
      "grad_norm": 0.4255470335483551,
      "learning_rate": 1.7205000000000002e-05,
      "loss": 0.0551,
      "step": 5738
    },
    {
      "epoch": 10.58910433979686,
      "grad_norm": 0.3540935814380646,
      "learning_rate": 1.7208000000000002e-05,
      "loss": 0.0178,
      "step": 5739
    },
    {
      "epoch": 10.590951061865189,
      "grad_norm": 0.44914117455482483,
      "learning_rate": 1.7211000000000002e-05,
      "loss": 0.0627,
      "step": 5740
    },
    {
      "epoch": 10.592797783933518,
      "grad_norm": 0.4884379208087921,
      "learning_rate": 1.7213999999999998e-05,
      "loss": 0.0196,
      "step": 5741
    },
    {
      "epoch": 10.594644506001847,
      "grad_norm": 0.2642185091972351,
      "learning_rate": 1.7216999999999998e-05,
      "loss": 0.0188,
      "step": 5742
    },
    {
      "epoch": 10.596491228070175,
      "grad_norm": 0.42987197637557983,
      "learning_rate": 1.7219999999999998e-05,
      "loss": 0.0245,
      "step": 5743
    },
    {
      "epoch": 10.598337950138504,
      "grad_norm": 0.3514634966850281,
      "learning_rate": 1.7223e-05,
      "loss": 0.0217,
      "step": 5744
    },
    {
      "epoch": 10.600184672206833,
      "grad_norm": 0.24382612109184265,
      "learning_rate": 1.7226e-05,
      "loss": 0.01,
      "step": 5745
    },
    {
      "epoch": 10.602031394275162,
      "grad_norm": 0.6637547016143799,
      "learning_rate": 1.7229e-05,
      "loss": 0.0287,
      "step": 5746
    },
    {
      "epoch": 10.603878116343491,
      "grad_norm": 0.6304849982261658,
      "learning_rate": 1.7232e-05,
      "loss": 0.0218,
      "step": 5747
    },
    {
      "epoch": 10.605724838411819,
      "grad_norm": 0.5097060203552246,
      "learning_rate": 1.7235e-05,
      "loss": 0.0198,
      "step": 5748
    },
    {
      "epoch": 10.607571560480148,
      "grad_norm": 0.3933012783527374,
      "learning_rate": 1.7238e-05,
      "loss": 0.0127,
      "step": 5749
    },
    {
      "epoch": 10.609418282548477,
      "grad_norm": 0.3572508692741394,
      "learning_rate": 1.7241e-05,
      "loss": 0.0583,
      "step": 5750
    },
    {
      "epoch": 10.611265004616806,
      "grad_norm": 0.5863717198371887,
      "learning_rate": 1.7244e-05,
      "loss": 0.0528,
      "step": 5751
    },
    {
      "epoch": 10.613111726685133,
      "grad_norm": 0.44540056586265564,
      "learning_rate": 1.7247e-05,
      "loss": 0.0233,
      "step": 5752
    },
    {
      "epoch": 10.614958448753463,
      "grad_norm": 0.4500695765018463,
      "learning_rate": 1.725e-05,
      "loss": 0.0226,
      "step": 5753
    },
    {
      "epoch": 10.616805170821792,
      "grad_norm": 0.9237388968467712,
      "learning_rate": 1.7253e-05,
      "loss": 0.0319,
      "step": 5754
    },
    {
      "epoch": 10.61865189289012,
      "grad_norm": 0.5861974954605103,
      "learning_rate": 1.7256000000000002e-05,
      "loss": 0.0345,
      "step": 5755
    },
    {
      "epoch": 10.620498614958448,
      "grad_norm": 0.4686151444911957,
      "learning_rate": 1.7259000000000002e-05,
      "loss": 0.0279,
      "step": 5756
    },
    {
      "epoch": 10.622345337026777,
      "grad_norm": 0.40479278564453125,
      "learning_rate": 1.7262000000000002e-05,
      "loss": 0.0149,
      "step": 5757
    },
    {
      "epoch": 10.624192059095106,
      "grad_norm": 0.5644612312316895,
      "learning_rate": 1.7265e-05,
      "loss": 0.0185,
      "step": 5758
    },
    {
      "epoch": 10.626038781163436,
      "grad_norm": 0.7194874882698059,
      "learning_rate": 1.7268e-05,
      "loss": 0.029,
      "step": 5759
    },
    {
      "epoch": 10.627885503231763,
      "grad_norm": 0.32438820600509644,
      "learning_rate": 1.7271e-05,
      "loss": 0.0162,
      "step": 5760
    },
    {
      "epoch": 10.629732225300092,
      "grad_norm": 0.47052034735679626,
      "learning_rate": 1.7274e-05,
      "loss": 0.0193,
      "step": 5761
    },
    {
      "epoch": 10.631578947368421,
      "grad_norm": 1.1280709505081177,
      "learning_rate": 1.7277e-05,
      "loss": 0.035,
      "step": 5762
    },
    {
      "epoch": 10.63342566943675,
      "grad_norm": 0.469835489988327,
      "learning_rate": 1.728e-05,
      "loss": 0.0168,
      "step": 5763
    },
    {
      "epoch": 10.635272391505078,
      "grad_norm": 0.5156711339950562,
      "learning_rate": 1.7283e-05,
      "loss": 0.0178,
      "step": 5764
    },
    {
      "epoch": 10.637119113573407,
      "grad_norm": 0.5281715989112854,
      "learning_rate": 1.7286e-05,
      "loss": 0.0275,
      "step": 5765
    },
    {
      "epoch": 10.638965835641736,
      "grad_norm": 0.45756691694259644,
      "learning_rate": 1.7289e-05,
      "loss": 0.025,
      "step": 5766
    },
    {
      "epoch": 10.640812557710065,
      "grad_norm": 0.4730209708213806,
      "learning_rate": 1.7292e-05,
      "loss": 0.026,
      "step": 5767
    },
    {
      "epoch": 10.642659279778393,
      "grad_norm": 1.0519452095031738,
      "learning_rate": 1.7295e-05,
      "loss": 0.0384,
      "step": 5768
    },
    {
      "epoch": 10.644506001846722,
      "grad_norm": 0.6030709743499756,
      "learning_rate": 1.7298e-05,
      "loss": 0.0318,
      "step": 5769
    },
    {
      "epoch": 10.64635272391505,
      "grad_norm": 1.0561301708221436,
      "learning_rate": 1.7301e-05,
      "loss": 0.0237,
      "step": 5770
    },
    {
      "epoch": 10.64819944598338,
      "grad_norm": 0.9371124505996704,
      "learning_rate": 1.7304e-05,
      "loss": 0.2677,
      "step": 5771
    },
    {
      "epoch": 10.65004616805171,
      "grad_norm": 0.6510640382766724,
      "learning_rate": 1.7307e-05,
      "loss": 0.1784,
      "step": 5772
    },
    {
      "epoch": 10.651892890120036,
      "grad_norm": 0.9221227765083313,
      "learning_rate": 1.731e-05,
      "loss": 0.197,
      "step": 5773
    },
    {
      "epoch": 10.653739612188366,
      "grad_norm": 0.7787302136421204,
      "learning_rate": 1.7313e-05,
      "loss": 0.1623,
      "step": 5774
    },
    {
      "epoch": 10.655586334256695,
      "grad_norm": 0.6928793787956238,
      "learning_rate": 1.7316e-05,
      "loss": 0.1452,
      "step": 5775
    },
    {
      "epoch": 10.657433056325024,
      "grad_norm": 0.580687403678894,
      "learning_rate": 1.7319e-05,
      "loss": 0.1221,
      "step": 5776
    },
    {
      "epoch": 10.659279778393351,
      "grad_norm": 0.8013032674789429,
      "learning_rate": 1.7322e-05,
      "loss": 0.1252,
      "step": 5777
    },
    {
      "epoch": 10.66112650046168,
      "grad_norm": 0.755961000919342,
      "learning_rate": 1.7325e-05,
      "loss": 0.1432,
      "step": 5778
    },
    {
      "epoch": 10.66297322253001,
      "grad_norm": 1.1269781589508057,
      "learning_rate": 1.7328e-05,
      "loss": 0.142,
      "step": 5779
    },
    {
      "epoch": 10.664819944598339,
      "grad_norm": 0.8808785676956177,
      "learning_rate": 1.7331e-05,
      "loss": 0.1661,
      "step": 5780
    },
    {
      "epoch": 10.666666666666666,
      "grad_norm": 0.7470980286598206,
      "learning_rate": 1.7334e-05,
      "loss": 0.0968,
      "step": 5781
    },
    {
      "epoch": 10.668513388734995,
      "grad_norm": 0.519422173500061,
      "learning_rate": 1.7337e-05,
      "loss": 0.0657,
      "step": 5782
    },
    {
      "epoch": 10.670360110803324,
      "grad_norm": 0.6593102812767029,
      "learning_rate": 1.734e-05,
      "loss": 0.1161,
      "step": 5783
    },
    {
      "epoch": 10.672206832871654,
      "grad_norm": 0.48402902483940125,
      "learning_rate": 1.7343e-05,
      "loss": 0.0666,
      "step": 5784
    },
    {
      "epoch": 10.67405355493998,
      "grad_norm": 0.3188153803348541,
      "learning_rate": 1.7346000000000003e-05,
      "loss": 0.0356,
      "step": 5785
    },
    {
      "epoch": 10.67590027700831,
      "grad_norm": 0.5182211995124817,
      "learning_rate": 1.7349000000000003e-05,
      "loss": 0.0434,
      "step": 5786
    },
    {
      "epoch": 10.67774699907664,
      "grad_norm": 0.5140071511268616,
      "learning_rate": 1.7352000000000003e-05,
      "loss": 0.0403,
      "step": 5787
    },
    {
      "epoch": 10.679593721144968,
      "grad_norm": 0.4483383297920227,
      "learning_rate": 1.7355000000000002e-05,
      "loss": 0.0437,
      "step": 5788
    },
    {
      "epoch": 10.681440443213296,
      "grad_norm": 1.0645132064819336,
      "learning_rate": 1.7358000000000002e-05,
      "loss": 0.0577,
      "step": 5789
    },
    {
      "epoch": 10.683287165281625,
      "grad_norm": 0.7384029626846313,
      "learning_rate": 1.7361e-05,
      "loss": 0.0997,
      "step": 5790
    },
    {
      "epoch": 10.685133887349954,
      "grad_norm": 0.4034633934497833,
      "learning_rate": 1.7364e-05,
      "loss": 0.0212,
      "step": 5791
    },
    {
      "epoch": 10.686980609418283,
      "grad_norm": 0.2344970703125,
      "learning_rate": 1.7366999999999998e-05,
      "loss": 0.0175,
      "step": 5792
    },
    {
      "epoch": 10.68882733148661,
      "grad_norm": 0.49585863947868347,
      "learning_rate": 1.7369999999999998e-05,
      "loss": 0.0317,
      "step": 5793
    },
    {
      "epoch": 10.69067405355494,
      "grad_norm": 0.5580799579620361,
      "learning_rate": 1.7372999999999998e-05,
      "loss": 0.0279,
      "step": 5794
    },
    {
      "epoch": 10.692520775623269,
      "grad_norm": 0.3624635934829712,
      "learning_rate": 1.7376e-05,
      "loss": 0.0176,
      "step": 5795
    },
    {
      "epoch": 10.694367497691598,
      "grad_norm": 0.5163046717643738,
      "learning_rate": 1.7379e-05,
      "loss": 0.0277,
      "step": 5796
    },
    {
      "epoch": 10.696214219759927,
      "grad_norm": 0.467898964881897,
      "learning_rate": 1.7382e-05,
      "loss": 0.0254,
      "step": 5797
    },
    {
      "epoch": 10.698060941828254,
      "grad_norm": 0.3116570711135864,
      "learning_rate": 1.7385e-05,
      "loss": 0.0142,
      "step": 5798
    },
    {
      "epoch": 10.699907663896584,
      "grad_norm": 0.5890852808952332,
      "learning_rate": 1.7388e-05,
      "loss": 0.0153,
      "step": 5799
    },
    {
      "epoch": 10.701754385964913,
      "grad_norm": 0.37547069787979126,
      "learning_rate": 1.7391e-05,
      "loss": 0.0155,
      "step": 5800
    },
    {
      "epoch": 10.703601108033242,
      "grad_norm": 0.5420143008232117,
      "learning_rate": 1.7394e-05,
      "loss": 0.0314,
      "step": 5801
    },
    {
      "epoch": 10.70544783010157,
      "grad_norm": 0.5886558294296265,
      "learning_rate": 1.7397e-05,
      "loss": 0.0313,
      "step": 5802
    },
    {
      "epoch": 10.707294552169898,
      "grad_norm": 0.6461412310600281,
      "learning_rate": 1.74e-05,
      "loss": 0.0356,
      "step": 5803
    },
    {
      "epoch": 10.709141274238227,
      "grad_norm": 0.4910597801208496,
      "learning_rate": 1.7403e-05,
      "loss": 0.0198,
      "step": 5804
    },
    {
      "epoch": 10.710987996306557,
      "grad_norm": 0.2973109483718872,
      "learning_rate": 1.7406000000000002e-05,
      "loss": 0.0152,
      "step": 5805
    },
    {
      "epoch": 10.712834718374884,
      "grad_norm": 0.40196624398231506,
      "learning_rate": 1.7409000000000002e-05,
      "loss": 0.0179,
      "step": 5806
    },
    {
      "epoch": 10.714681440443213,
      "grad_norm": 0.4059551954269409,
      "learning_rate": 1.7412000000000002e-05,
      "loss": 0.0191,
      "step": 5807
    },
    {
      "epoch": 10.716528162511542,
      "grad_norm": 0.8953748941421509,
      "learning_rate": 1.7415000000000002e-05,
      "loss": 0.0298,
      "step": 5808
    },
    {
      "epoch": 10.718374884579871,
      "grad_norm": 0.9182073473930359,
      "learning_rate": 1.7418e-05,
      "loss": 0.0306,
      "step": 5809
    },
    {
      "epoch": 10.720221606648199,
      "grad_norm": 0.5914534330368042,
      "learning_rate": 1.7421e-05,
      "loss": 0.0303,
      "step": 5810
    },
    {
      "epoch": 10.722068328716528,
      "grad_norm": 0.5903384685516357,
      "learning_rate": 1.7424e-05,
      "loss": 0.0245,
      "step": 5811
    },
    {
      "epoch": 10.723915050784857,
      "grad_norm": 0.49383530020713806,
      "learning_rate": 1.7427e-05,
      "loss": 0.0261,
      "step": 5812
    },
    {
      "epoch": 10.725761772853186,
      "grad_norm": 0.5673595666885376,
      "learning_rate": 1.743e-05,
      "loss": 0.0287,
      "step": 5813
    },
    {
      "epoch": 10.727608494921514,
      "grad_norm": 0.6683375835418701,
      "learning_rate": 1.7432999999999997e-05,
      "loss": 0.0309,
      "step": 5814
    },
    {
      "epoch": 10.729455216989843,
      "grad_norm": 0.34646183252334595,
      "learning_rate": 1.7436e-05,
      "loss": 0.014,
      "step": 5815
    },
    {
      "epoch": 10.731301939058172,
      "grad_norm": 0.599746584892273,
      "learning_rate": 1.7439e-05,
      "loss": 0.0249,
      "step": 5816
    },
    {
      "epoch": 10.733148661126501,
      "grad_norm": 0.7076974511146545,
      "learning_rate": 1.7442e-05,
      "loss": 0.037,
      "step": 5817
    },
    {
      "epoch": 10.734995383194828,
      "grad_norm": 0.9104524850845337,
      "learning_rate": 1.7445e-05,
      "loss": 0.0311,
      "step": 5818
    },
    {
      "epoch": 10.736842105263158,
      "grad_norm": 0.5769959092140198,
      "learning_rate": 1.7448e-05,
      "loss": 0.0355,
      "step": 5819
    },
    {
      "epoch": 10.738688827331487,
      "grad_norm": 0.5935896039009094,
      "learning_rate": 1.7451e-05,
      "loss": 0.021,
      "step": 5820
    },
    {
      "epoch": 10.740535549399816,
      "grad_norm": 1.0296519994735718,
      "learning_rate": 1.7454e-05,
      "loss": 0.2168,
      "step": 5821
    },
    {
      "epoch": 10.742382271468145,
      "grad_norm": 0.8563361167907715,
      "learning_rate": 1.7457e-05,
      "loss": 0.2389,
      "step": 5822
    },
    {
      "epoch": 10.744228993536472,
      "grad_norm": 0.8937773704528809,
      "learning_rate": 1.746e-05,
      "loss": 0.2175,
      "step": 5823
    },
    {
      "epoch": 10.746075715604801,
      "grad_norm": 0.9618704915046692,
      "learning_rate": 1.7463e-05,
      "loss": 0.148,
      "step": 5824
    },
    {
      "epoch": 10.74792243767313,
      "grad_norm": 0.9022472500801086,
      "learning_rate": 1.7466000000000002e-05,
      "loss": 0.2013,
      "step": 5825
    },
    {
      "epoch": 10.749769159741458,
      "grad_norm": 0.6519712209701538,
      "learning_rate": 1.7469e-05,
      "loss": 0.1351,
      "step": 5826
    },
    {
      "epoch": 10.751615881809787,
      "grad_norm": 0.5046953558921814,
      "learning_rate": 1.7472e-05,
      "loss": 0.1077,
      "step": 5827
    },
    {
      "epoch": 10.753462603878116,
      "grad_norm": 1.0729660987854004,
      "learning_rate": 1.7475e-05,
      "loss": 0.1374,
      "step": 5828
    },
    {
      "epoch": 10.755309325946445,
      "grad_norm": 0.735659658908844,
      "learning_rate": 1.7478e-05,
      "loss": 0.1447,
      "step": 5829
    },
    {
      "epoch": 10.757156048014775,
      "grad_norm": 0.8504224419593811,
      "learning_rate": 1.7481e-05,
      "loss": 0.1169,
      "step": 5830
    },
    {
      "epoch": 10.759002770083102,
      "grad_norm": 0.7671719789505005,
      "learning_rate": 1.7484e-05,
      "loss": 0.0831,
      "step": 5831
    },
    {
      "epoch": 10.760849492151431,
      "grad_norm": 0.3581356108188629,
      "learning_rate": 1.7487e-05,
      "loss": 0.043,
      "step": 5832
    },
    {
      "epoch": 10.76269621421976,
      "grad_norm": 0.5683116316795349,
      "learning_rate": 1.749e-05,
      "loss": 0.1233,
      "step": 5833
    },
    {
      "epoch": 10.76454293628809,
      "grad_norm": 0.4286747872829437,
      "learning_rate": 1.7493e-05,
      "loss": 0.0561,
      "step": 5834
    },
    {
      "epoch": 10.766389658356417,
      "grad_norm": 0.7286791801452637,
      "learning_rate": 1.7496000000000003e-05,
      "loss": 0.0465,
      "step": 5835
    },
    {
      "epoch": 10.768236380424746,
      "grad_norm": 0.6198620796203613,
      "learning_rate": 1.7499000000000003e-05,
      "loss": 0.0647,
      "step": 5836
    },
    {
      "epoch": 10.770083102493075,
      "grad_norm": 0.39466071128845215,
      "learning_rate": 1.7502000000000003e-05,
      "loss": 0.0338,
      "step": 5837
    },
    {
      "epoch": 10.771929824561404,
      "grad_norm": 0.5323166251182556,
      "learning_rate": 1.7505000000000003e-05,
      "loss": 0.0309,
      "step": 5838
    },
    {
      "epoch": 10.773776546629731,
      "grad_norm": 0.6481252312660217,
      "learning_rate": 1.7508e-05,
      "loss": 0.0314,
      "step": 5839
    },
    {
      "epoch": 10.77562326869806,
      "grad_norm": 0.623769223690033,
      "learning_rate": 1.7511e-05,
      "loss": 0.0316,
      "step": 5840
    },
    {
      "epoch": 10.77746999076639,
      "grad_norm": 0.6676685810089111,
      "learning_rate": 1.7514e-05,
      "loss": 0.0421,
      "step": 5841
    },
    {
      "epoch": 10.779316712834719,
      "grad_norm": 0.5272042155265808,
      "learning_rate": 1.7517e-05,
      "loss": 0.0426,
      "step": 5842
    },
    {
      "epoch": 10.781163434903046,
      "grad_norm": 0.6411356925964355,
      "learning_rate": 1.7519999999999998e-05,
      "loss": 0.0367,
      "step": 5843
    },
    {
      "epoch": 10.783010156971375,
      "grad_norm": 0.7479405999183655,
      "learning_rate": 1.7522999999999998e-05,
      "loss": 0.02,
      "step": 5844
    },
    {
      "epoch": 10.784856879039705,
      "grad_norm": 0.37660306692123413,
      "learning_rate": 1.7526e-05,
      "loss": 0.0288,
      "step": 5845
    },
    {
      "epoch": 10.786703601108034,
      "grad_norm": 0.9735251665115356,
      "learning_rate": 1.7529e-05,
      "loss": 0.0539,
      "step": 5846
    },
    {
      "epoch": 10.788550323176363,
      "grad_norm": 0.6767516136169434,
      "learning_rate": 1.7532e-05,
      "loss": 0.0234,
      "step": 5847
    },
    {
      "epoch": 10.79039704524469,
      "grad_norm": 0.42775028944015503,
      "learning_rate": 1.7535e-05,
      "loss": 0.0186,
      "step": 5848
    },
    {
      "epoch": 10.79224376731302,
      "grad_norm": 0.3126576542854309,
      "learning_rate": 1.7538e-05,
      "loss": 0.0109,
      "step": 5849
    },
    {
      "epoch": 10.794090489381349,
      "grad_norm": 0.41386958956718445,
      "learning_rate": 1.7541e-05,
      "loss": 0.029,
      "step": 5850
    },
    {
      "epoch": 10.795937211449676,
      "grad_norm": 0.33370286226272583,
      "learning_rate": 1.7544e-05,
      "loss": 0.0178,
      "step": 5851
    },
    {
      "epoch": 10.797783933518005,
      "grad_norm": 0.6642718315124512,
      "learning_rate": 1.7547e-05,
      "loss": 0.0294,
      "step": 5852
    },
    {
      "epoch": 10.799630655586334,
      "grad_norm": 0.6200350522994995,
      "learning_rate": 1.755e-05,
      "loss": 0.0254,
      "step": 5853
    },
    {
      "epoch": 10.801477377654663,
      "grad_norm": 1.2748810052871704,
      "learning_rate": 1.7553e-05,
      "loss": 0.0452,
      "step": 5854
    },
    {
      "epoch": 10.803324099722992,
      "grad_norm": 0.811354398727417,
      "learning_rate": 1.7556000000000003e-05,
      "loss": 0.0254,
      "step": 5855
    },
    {
      "epoch": 10.80517082179132,
      "grad_norm": 0.5516785979270935,
      "learning_rate": 1.7559000000000002e-05,
      "loss": 0.0271,
      "step": 5856
    },
    {
      "epoch": 10.807017543859649,
      "grad_norm": 0.5956955552101135,
      "learning_rate": 1.7562000000000002e-05,
      "loss": 0.0197,
      "step": 5857
    },
    {
      "epoch": 10.808864265927978,
      "grad_norm": 0.531806468963623,
      "learning_rate": 1.7565000000000002e-05,
      "loss": 0.0289,
      "step": 5858
    },
    {
      "epoch": 10.810710987996307,
      "grad_norm": 0.5544119477272034,
      "learning_rate": 1.7568000000000002e-05,
      "loss": 0.0229,
      "step": 5859
    },
    {
      "epoch": 10.812557710064635,
      "grad_norm": 0.528141975402832,
      "learning_rate": 1.7571e-05,
      "loss": 0.0207,
      "step": 5860
    },
    {
      "epoch": 10.814404432132964,
      "grad_norm": 0.5564522743225098,
      "learning_rate": 1.7574e-05,
      "loss": 0.0293,
      "step": 5861
    },
    {
      "epoch": 10.816251154201293,
      "grad_norm": 0.3546724319458008,
      "learning_rate": 1.7577e-05,
      "loss": 0.0113,
      "step": 5862
    },
    {
      "epoch": 10.818097876269622,
      "grad_norm": 0.46226778626441956,
      "learning_rate": 1.758e-05,
      "loss": 0.0212,
      "step": 5863
    },
    {
      "epoch": 10.81994459833795,
      "grad_norm": 0.38217854499816895,
      "learning_rate": 1.7582999999999998e-05,
      "loss": 0.0236,
      "step": 5864
    },
    {
      "epoch": 10.821791320406279,
      "grad_norm": 0.6104041934013367,
      "learning_rate": 1.7586e-05,
      "loss": 0.0207,
      "step": 5865
    },
    {
      "epoch": 10.823638042474608,
      "grad_norm": 0.4689196050167084,
      "learning_rate": 1.7589e-05,
      "loss": 0.0197,
      "step": 5866
    },
    {
      "epoch": 10.825484764542937,
      "grad_norm": 0.4690294563770294,
      "learning_rate": 1.7592e-05,
      "loss": 0.0218,
      "step": 5867
    },
    {
      "epoch": 10.827331486611264,
      "grad_norm": 0.8422415852546692,
      "learning_rate": 1.7595e-05,
      "loss": 0.0335,
      "step": 5868
    },
    {
      "epoch": 10.829178208679593,
      "grad_norm": 0.9045748114585876,
      "learning_rate": 1.7598e-05,
      "loss": 0.0321,
      "step": 5869
    },
    {
      "epoch": 10.831024930747922,
      "grad_norm": 1.0264004468917847,
      "learning_rate": 1.7601e-05,
      "loss": 0.0389,
      "step": 5870
    },
    {
      "epoch": 10.832871652816252,
      "grad_norm": 1.2848174571990967,
      "learning_rate": 1.7604e-05,
      "loss": 0.2561,
      "step": 5871
    },
    {
      "epoch": 10.83471837488458,
      "grad_norm": 1.1149464845657349,
      "learning_rate": 1.7607e-05,
      "loss": 0.2667,
      "step": 5872
    },
    {
      "epoch": 10.836565096952908,
      "grad_norm": 1.0092140436172485,
      "learning_rate": 1.761e-05,
      "loss": 0.2504,
      "step": 5873
    },
    {
      "epoch": 10.838411819021237,
      "grad_norm": 0.6515405774116516,
      "learning_rate": 1.7613e-05,
      "loss": 0.153,
      "step": 5874
    },
    {
      "epoch": 10.840258541089566,
      "grad_norm": 0.7586976885795593,
      "learning_rate": 1.7616000000000002e-05,
      "loss": 0.2003,
      "step": 5875
    },
    {
      "epoch": 10.842105263157894,
      "grad_norm": 0.5531761646270752,
      "learning_rate": 1.7619000000000002e-05,
      "loss": 0.1495,
      "step": 5876
    },
    {
      "epoch": 10.843951985226223,
      "grad_norm": 0.696130633354187,
      "learning_rate": 1.7622000000000002e-05,
      "loss": 0.127,
      "step": 5877
    },
    {
      "epoch": 10.845798707294552,
      "grad_norm": 0.8173731565475464,
      "learning_rate": 1.7625e-05,
      "loss": 0.1449,
      "step": 5878
    },
    {
      "epoch": 10.847645429362881,
      "grad_norm": 1.2938200235366821,
      "learning_rate": 1.7628e-05,
      "loss": 0.1448,
      "step": 5879
    },
    {
      "epoch": 10.84949215143121,
      "grad_norm": 0.8170304894447327,
      "learning_rate": 1.7631e-05,
      "loss": 0.0966,
      "step": 5880
    },
    {
      "epoch": 10.851338873499538,
      "grad_norm": 0.8002492189407349,
      "learning_rate": 1.7634e-05,
      "loss": 0.0673,
      "step": 5881
    },
    {
      "epoch": 10.853185595567867,
      "grad_norm": 0.7990866899490356,
      "learning_rate": 1.7637e-05,
      "loss": 0.0818,
      "step": 5882
    },
    {
      "epoch": 10.855032317636196,
      "grad_norm": 0.5102277994155884,
      "learning_rate": 1.764e-05,
      "loss": 0.0492,
      "step": 5883
    },
    {
      "epoch": 10.856879039704525,
      "grad_norm": 1.3296751976013184,
      "learning_rate": 1.7643e-05,
      "loss": 0.0644,
      "step": 5884
    },
    {
      "epoch": 10.858725761772853,
      "grad_norm": 0.59538733959198,
      "learning_rate": 1.7646e-05,
      "loss": 0.05,
      "step": 5885
    },
    {
      "epoch": 10.860572483841182,
      "grad_norm": 0.4024272561073303,
      "learning_rate": 1.7649000000000003e-05,
      "loss": 0.0357,
      "step": 5886
    },
    {
      "epoch": 10.86241920590951,
      "grad_norm": 0.5509514212608337,
      "learning_rate": 1.7652000000000003e-05,
      "loss": 0.0313,
      "step": 5887
    },
    {
      "epoch": 10.86426592797784,
      "grad_norm": 0.6461734771728516,
      "learning_rate": 1.7655e-05,
      "loss": 0.049,
      "step": 5888
    },
    {
      "epoch": 10.866112650046167,
      "grad_norm": 0.7171344757080078,
      "learning_rate": 1.7658e-05,
      "loss": 0.032,
      "step": 5889
    },
    {
      "epoch": 10.867959372114496,
      "grad_norm": 0.5926627516746521,
      "learning_rate": 1.7661e-05,
      "loss": 0.0275,
      "step": 5890
    },
    {
      "epoch": 10.869806094182826,
      "grad_norm": 0.38975149393081665,
      "learning_rate": 1.7664e-05,
      "loss": 0.0204,
      "step": 5891
    },
    {
      "epoch": 10.871652816251155,
      "grad_norm": 0.32457953691482544,
      "learning_rate": 1.7667e-05,
      "loss": 0.0259,
      "step": 5892
    },
    {
      "epoch": 10.873499538319482,
      "grad_norm": 0.29443931579589844,
      "learning_rate": 1.767e-05,
      "loss": 0.0175,
      "step": 5893
    },
    {
      "epoch": 10.875346260387811,
      "grad_norm": 0.36705857515335083,
      "learning_rate": 1.7673e-05,
      "loss": 0.0233,
      "step": 5894
    },
    {
      "epoch": 10.87719298245614,
      "grad_norm": 0.4953431487083435,
      "learning_rate": 1.7675999999999998e-05,
      "loss": 0.0372,
      "step": 5895
    },
    {
      "epoch": 10.87903970452447,
      "grad_norm": 0.5245679616928101,
      "learning_rate": 1.7679e-05,
      "loss": 0.0296,
      "step": 5896
    },
    {
      "epoch": 10.880886426592799,
      "grad_norm": 0.3879951536655426,
      "learning_rate": 1.7682e-05,
      "loss": 0.0238,
      "step": 5897
    },
    {
      "epoch": 10.882733148661126,
      "grad_norm": 0.8937625885009766,
      "learning_rate": 1.7685e-05,
      "loss": 0.0178,
      "step": 5898
    },
    {
      "epoch": 10.884579870729455,
      "grad_norm": 0.47701773047447205,
      "learning_rate": 1.7688e-05,
      "loss": 0.0272,
      "step": 5899
    },
    {
      "epoch": 10.886426592797784,
      "grad_norm": 0.32832854986190796,
      "learning_rate": 1.7691e-05,
      "loss": 0.0151,
      "step": 5900
    },
    {
      "epoch": 10.888273314866112,
      "grad_norm": 0.4349646270275116,
      "learning_rate": 1.7694e-05,
      "loss": 0.032,
      "step": 5901
    },
    {
      "epoch": 10.89012003693444,
      "grad_norm": 0.47184181213378906,
      "learning_rate": 1.7697e-05,
      "loss": 0.0194,
      "step": 5902
    },
    {
      "epoch": 10.89196675900277,
      "grad_norm": 0.29045549035072327,
      "learning_rate": 1.77e-05,
      "loss": 0.0117,
      "step": 5903
    },
    {
      "epoch": 10.8938134810711,
      "grad_norm": 0.3673584759235382,
      "learning_rate": 1.7703e-05,
      "loss": 0.0145,
      "step": 5904
    },
    {
      "epoch": 10.895660203139428,
      "grad_norm": 0.41271263360977173,
      "learning_rate": 1.7706e-05,
      "loss": 0.0216,
      "step": 5905
    },
    {
      "epoch": 10.897506925207756,
      "grad_norm": 0.38137903809547424,
      "learning_rate": 1.7709000000000003e-05,
      "loss": 0.0228,
      "step": 5906
    },
    {
      "epoch": 10.899353647276085,
      "grad_norm": 0.46189069747924805,
      "learning_rate": 1.7712000000000003e-05,
      "loss": 0.0269,
      "step": 5907
    },
    {
      "epoch": 10.901200369344414,
      "grad_norm": 0.9427136182785034,
      "learning_rate": 1.7715000000000002e-05,
      "loss": 0.0381,
      "step": 5908
    },
    {
      "epoch": 10.903047091412743,
      "grad_norm": 0.7765113115310669,
      "learning_rate": 1.7718000000000002e-05,
      "loss": 0.0273,
      "step": 5909
    },
    {
      "epoch": 10.90489381348107,
      "grad_norm": 0.5198538899421692,
      "learning_rate": 1.7721000000000002e-05,
      "loss": 0.0262,
      "step": 5910
    },
    {
      "epoch": 10.9067405355494,
      "grad_norm": 0.4149373471736908,
      "learning_rate": 1.7724000000000002e-05,
      "loss": 0.0241,
      "step": 5911
    },
    {
      "epoch": 10.908587257617729,
      "grad_norm": 0.39832204580307007,
      "learning_rate": 1.7727e-05,
      "loss": 0.0174,
      "step": 5912
    },
    {
      "epoch": 10.910433979686058,
      "grad_norm": 0.46863245964050293,
      "learning_rate": 1.7729999999999998e-05,
      "loss": 0.0162,
      "step": 5913
    },
    {
      "epoch": 10.912280701754385,
      "grad_norm": 0.5959219932556152,
      "learning_rate": 1.7732999999999998e-05,
      "loss": 0.0229,
      "step": 5914
    },
    {
      "epoch": 10.914127423822714,
      "grad_norm": 0.626492977142334,
      "learning_rate": 1.7735999999999998e-05,
      "loss": 0.0334,
      "step": 5915
    },
    {
      "epoch": 10.915974145891044,
      "grad_norm": 0.35012662410736084,
      "learning_rate": 1.7739e-05,
      "loss": 0.0142,
      "step": 5916
    },
    {
      "epoch": 10.917820867959373,
      "grad_norm": 0.5902463793754578,
      "learning_rate": 1.7742e-05,
      "loss": 0.0281,
      "step": 5917
    },
    {
      "epoch": 10.9196675900277,
      "grad_norm": 0.5554072856903076,
      "learning_rate": 1.7745e-05,
      "loss": 0.017,
      "step": 5918
    },
    {
      "epoch": 10.92151431209603,
      "grad_norm": 0.5523650646209717,
      "learning_rate": 1.7748e-05,
      "loss": 0.0333,
      "step": 5919
    },
    {
      "epoch": 10.923361034164358,
      "grad_norm": 0.5115478038787842,
      "learning_rate": 1.7751e-05,
      "loss": 0.0231,
      "step": 5920
    },
    {
      "epoch": 10.925207756232687,
      "grad_norm": 0.8140024542808533,
      "learning_rate": 1.7754e-05,
      "loss": 0.2674,
      "step": 5921
    },
    {
      "epoch": 10.927054478301017,
      "grad_norm": 1.0150563716888428,
      "learning_rate": 1.7757e-05,
      "loss": 0.1972,
      "step": 5922
    },
    {
      "epoch": 10.928901200369344,
      "grad_norm": 0.8180074095726013,
      "learning_rate": 1.776e-05,
      "loss": 0.1416,
      "step": 5923
    },
    {
      "epoch": 10.930747922437673,
      "grad_norm": 0.6263332962989807,
      "learning_rate": 1.7763e-05,
      "loss": 0.154,
      "step": 5924
    },
    {
      "epoch": 10.932594644506002,
      "grad_norm": 0.7633740901947021,
      "learning_rate": 1.7766e-05,
      "loss": 0.1867,
      "step": 5925
    },
    {
      "epoch": 10.93444136657433,
      "grad_norm": 0.7075972557067871,
      "learning_rate": 1.7769000000000002e-05,
      "loss": 0.1033,
      "step": 5926
    },
    {
      "epoch": 10.936288088642659,
      "grad_norm": 0.9612365961074829,
      "learning_rate": 1.7772000000000002e-05,
      "loss": 0.1497,
      "step": 5927
    },
    {
      "epoch": 10.938134810710988,
      "grad_norm": 0.5302220582962036,
      "learning_rate": 1.7775000000000002e-05,
      "loss": 0.0802,
      "step": 5928
    },
    {
      "epoch": 10.939981532779317,
      "grad_norm": 0.7335618734359741,
      "learning_rate": 1.7778e-05,
      "loss": 0.1157,
      "step": 5929
    },
    {
      "epoch": 10.941828254847646,
      "grad_norm": 0.5661278963088989,
      "learning_rate": 1.7781e-05,
      "loss": 0.0775,
      "step": 5930
    },
    {
      "epoch": 10.943674976915974,
      "grad_norm": 0.580007016658783,
      "learning_rate": 1.7784e-05,
      "loss": 0.0852,
      "step": 5931
    },
    {
      "epoch": 10.945521698984303,
      "grad_norm": 0.7283288836479187,
      "learning_rate": 1.7787e-05,
      "loss": 0.0515,
      "step": 5932
    },
    {
      "epoch": 10.947368421052632,
      "grad_norm": 0.47408390045166016,
      "learning_rate": 1.779e-05,
      "loss": 0.0598,
      "step": 5933
    },
    {
      "epoch": 10.949215143120961,
      "grad_norm": 0.4382071793079376,
      "learning_rate": 1.7793e-05,
      "loss": 0.0297,
      "step": 5934
    },
    {
      "epoch": 10.951061865189288,
      "grad_norm": 0.4971558153629303,
      "learning_rate": 1.7796e-05,
      "loss": 0.0495,
      "step": 5935
    },
    {
      "epoch": 10.952908587257618,
      "grad_norm": 1.1129035949707031,
      "learning_rate": 1.7799000000000004e-05,
      "loss": 0.043,
      "step": 5936
    },
    {
      "epoch": 10.954755309325947,
      "grad_norm": 0.4687868654727936,
      "learning_rate": 1.7802e-05,
      "loss": 0.0481,
      "step": 5937
    },
    {
      "epoch": 10.956602031394276,
      "grad_norm": 0.611916184425354,
      "learning_rate": 1.7805e-05,
      "loss": 0.0697,
      "step": 5938
    },
    {
      "epoch": 10.958448753462603,
      "grad_norm": 0.5590572357177734,
      "learning_rate": 1.7808e-05,
      "loss": 0.0174,
      "step": 5939
    },
    {
      "epoch": 10.960295475530932,
      "grad_norm": 0.5645524263381958,
      "learning_rate": 1.7811e-05,
      "loss": 0.0227,
      "step": 5940
    },
    {
      "epoch": 10.962142197599261,
      "grad_norm": 0.5157508254051208,
      "learning_rate": 1.7814e-05,
      "loss": 0.0246,
      "step": 5941
    },
    {
      "epoch": 10.96398891966759,
      "grad_norm": 0.4864080250263214,
      "learning_rate": 1.7817e-05,
      "loss": 0.0357,
      "step": 5942
    },
    {
      "epoch": 10.965835641735918,
      "grad_norm": 0.3735795021057129,
      "learning_rate": 1.782e-05,
      "loss": 0.0199,
      "step": 5943
    },
    {
      "epoch": 10.967682363804247,
      "grad_norm": 0.5223242044448853,
      "learning_rate": 1.7823e-05,
      "loss": 0.0257,
      "step": 5944
    },
    {
      "epoch": 10.969529085872576,
      "grad_norm": 0.3293151259422302,
      "learning_rate": 1.7826e-05,
      "loss": 0.0186,
      "step": 5945
    },
    {
      "epoch": 10.971375807940905,
      "grad_norm": 0.6055980920791626,
      "learning_rate": 1.7829e-05,
      "loss": 0.0302,
      "step": 5946
    },
    {
      "epoch": 10.973222530009235,
      "grad_norm": 0.3997729420661926,
      "learning_rate": 1.7832e-05,
      "loss": 0.0225,
      "step": 5947
    },
    {
      "epoch": 10.975069252077562,
      "grad_norm": 0.49927785992622375,
      "learning_rate": 1.7835e-05,
      "loss": 0.025,
      "step": 5948
    },
    {
      "epoch": 10.976915974145891,
      "grad_norm": 0.8552719950675964,
      "learning_rate": 1.7838e-05,
      "loss": 0.037,
      "step": 5949
    },
    {
      "epoch": 10.97876269621422,
      "grad_norm": 0.7387699484825134,
      "learning_rate": 1.7841e-05,
      "loss": 0.0343,
      "step": 5950
    },
    {
      "epoch": 10.980609418282548,
      "grad_norm": 0.6182773113250732,
      "learning_rate": 1.7844e-05,
      "loss": 0.0288,
      "step": 5951
    },
    {
      "epoch": 10.982456140350877,
      "grad_norm": 0.6504404544830322,
      "learning_rate": 1.7847e-05,
      "loss": 0.0281,
      "step": 5952
    },
    {
      "epoch": 10.984302862419206,
      "grad_norm": 0.4910828173160553,
      "learning_rate": 1.785e-05,
      "loss": 0.0247,
      "step": 5953
    },
    {
      "epoch": 10.986149584487535,
      "grad_norm": 0.4561193585395813,
      "learning_rate": 1.7853e-05,
      "loss": 0.0296,
      "step": 5954
    },
    {
      "epoch": 10.987996306555864,
      "grad_norm": 0.7066456079483032,
      "learning_rate": 1.7856e-05,
      "loss": 0.0344,
      "step": 5955
    },
    {
      "epoch": 10.989843028624191,
      "grad_norm": 0.5314071178436279,
      "learning_rate": 1.7859000000000003e-05,
      "loss": 0.0296,
      "step": 5956
    },
    {
      "epoch": 10.99168975069252,
      "grad_norm": 0.5812799334526062,
      "learning_rate": 1.7862000000000003e-05,
      "loss": 0.0211,
      "step": 5957
    },
    {
      "epoch": 10.99353647276085,
      "grad_norm": 0.2950487732887268,
      "learning_rate": 1.7865000000000003e-05,
      "loss": 0.0136,
      "step": 5958
    },
    {
      "epoch": 10.995383194829179,
      "grad_norm": 1.0167781114578247,
      "learning_rate": 1.7868000000000002e-05,
      "loss": 0.041,
      "step": 5959
    },
    {
      "epoch": 10.997229916897506,
      "grad_norm": 0.41852226853370667,
      "learning_rate": 1.7871000000000002e-05,
      "loss": 0.0189,
      "step": 5960
    },
    {
      "epoch": 10.999076638965835,
      "grad_norm": 0.6753515005111694,
      "learning_rate": 1.7874000000000002e-05,
      "loss": 0.0239,
      "step": 5961
    },
    {
      "epoch": 11.0,
      "grad_norm": 0.3744104504585266,
      "learning_rate": 1.7877e-05,
      "loss": 0.0189,
      "step": 5962
    },
    {
      "epoch": 11.00184672206833,
      "grad_norm": 0.7426442503929138,
      "learning_rate": 1.7879999999999998e-05,
      "loss": 0.2933,
      "step": 5963
    },
    {
      "epoch": 11.003693444136658,
      "grad_norm": 0.6798139810562134,
      "learning_rate": 1.7882999999999998e-05,
      "loss": 0.1877,
      "step": 5964
    },
    {
      "epoch": 11.005540166204986,
      "grad_norm": 0.5603246688842773,
      "learning_rate": 1.7885999999999998e-05,
      "loss": 0.2133,
      "step": 5965
    },
    {
      "epoch": 11.007386888273315,
      "grad_norm": 0.8107830286026001,
      "learning_rate": 1.7889e-05,
      "loss": 0.1866,
      "step": 5966
    },
    {
      "epoch": 11.009233610341644,
      "grad_norm": 0.7005266547203064,
      "learning_rate": 1.7892e-05,
      "loss": 0.1359,
      "step": 5967
    },
    {
      "epoch": 11.011080332409973,
      "grad_norm": 0.8084649443626404,
      "learning_rate": 1.7895e-05,
      "loss": 0.1772,
      "step": 5968
    },
    {
      "epoch": 11.0129270544783,
      "grad_norm": 0.8491911292076111,
      "learning_rate": 1.7898e-05,
      "loss": 0.1294,
      "step": 5969
    },
    {
      "epoch": 11.01477377654663,
      "grad_norm": 0.7015768885612488,
      "learning_rate": 1.7901e-05,
      "loss": 0.1238,
      "step": 5970
    },
    {
      "epoch": 11.016620498614959,
      "grad_norm": 0.7217311263084412,
      "learning_rate": 1.7904e-05,
      "loss": 0.088,
      "step": 5971
    },
    {
      "epoch": 11.018467220683288,
      "grad_norm": 0.6328301429748535,
      "learning_rate": 1.7907e-05,
      "loss": 0.0878,
      "step": 5972
    },
    {
      "epoch": 11.020313942751615,
      "grad_norm": 0.6157311201095581,
      "learning_rate": 1.791e-05,
      "loss": 0.0828,
      "step": 5973
    },
    {
      "epoch": 11.022160664819944,
      "grad_norm": 0.5200340747833252,
      "learning_rate": 1.7913e-05,
      "loss": 0.0616,
      "step": 5974
    },
    {
      "epoch": 11.024007386888274,
      "grad_norm": 1.4208979606628418,
      "learning_rate": 1.7916e-05,
      "loss": 0.188,
      "step": 5975
    },
    {
      "epoch": 11.025854108956603,
      "grad_norm": 0.47305992245674133,
      "learning_rate": 1.7919000000000002e-05,
      "loss": 0.0323,
      "step": 5976
    },
    {
      "epoch": 11.02770083102493,
      "grad_norm": 0.46868613362312317,
      "learning_rate": 1.7922000000000002e-05,
      "loss": 0.044,
      "step": 5977
    },
    {
      "epoch": 11.02954755309326,
      "grad_norm": 0.8581869006156921,
      "learning_rate": 1.7925000000000002e-05,
      "loss": 0.0702,
      "step": 5978
    },
    {
      "epoch": 11.031394275161588,
      "grad_norm": 0.5283058285713196,
      "learning_rate": 1.7928000000000002e-05,
      "loss": 0.0424,
      "step": 5979
    },
    {
      "epoch": 11.033240997229917,
      "grad_norm": 0.5160033106803894,
      "learning_rate": 1.7931e-05,
      "loss": 0.0484,
      "step": 5980
    },
    {
      "epoch": 11.035087719298245,
      "grad_norm": 0.8869902491569519,
      "learning_rate": 1.7934e-05,
      "loss": 0.0201,
      "step": 5981
    },
    {
      "epoch": 11.036934441366574,
      "grad_norm": 0.5846513509750366,
      "learning_rate": 1.7937e-05,
      "loss": 0.0361,
      "step": 5982
    },
    {
      "epoch": 11.038781163434903,
      "grad_norm": 0.3070891797542572,
      "learning_rate": 1.794e-05,
      "loss": 0.018,
      "step": 5983
    },
    {
      "epoch": 11.040627885503232,
      "grad_norm": 0.6062940359115601,
      "learning_rate": 1.7943e-05,
      "loss": 0.0414,
      "step": 5984
    },
    {
      "epoch": 11.04247460757156,
      "grad_norm": 0.3984304964542389,
      "learning_rate": 1.7946e-05,
      "loss": 0.0183,
      "step": 5985
    },
    {
      "epoch": 11.044321329639889,
      "grad_norm": 0.4931993782520294,
      "learning_rate": 1.7949e-05,
      "loss": 0.0151,
      "step": 5986
    },
    {
      "epoch": 11.046168051708218,
      "grad_norm": 0.6660001873970032,
      "learning_rate": 1.7952e-05,
      "loss": 0.0291,
      "step": 5987
    },
    {
      "epoch": 11.048014773776547,
      "grad_norm": 0.6292256116867065,
      "learning_rate": 1.7955e-05,
      "loss": 0.0191,
      "step": 5988
    },
    {
      "epoch": 11.049861495844876,
      "grad_norm": 0.4108259975910187,
      "learning_rate": 1.7958e-05,
      "loss": 0.0171,
      "step": 5989
    },
    {
      "epoch": 11.051708217913204,
      "grad_norm": 0.4854716658592224,
      "learning_rate": 1.7961e-05,
      "loss": 0.0188,
      "step": 5990
    },
    {
      "epoch": 11.053554939981533,
      "grad_norm": 0.9647286534309387,
      "learning_rate": 1.7964e-05,
      "loss": 0.0191,
      "step": 5991
    },
    {
      "epoch": 11.055401662049862,
      "grad_norm": 0.733239471912384,
      "learning_rate": 1.7967e-05,
      "loss": 0.0599,
      "step": 5992
    },
    {
      "epoch": 11.057248384118191,
      "grad_norm": 0.3580063283443451,
      "learning_rate": 1.797e-05,
      "loss": 0.0164,
      "step": 5993
    },
    {
      "epoch": 11.059095106186518,
      "grad_norm": 0.40893301367759705,
      "learning_rate": 1.7973e-05,
      "loss": 0.0144,
      "step": 5994
    },
    {
      "epoch": 11.060941828254848,
      "grad_norm": 0.3047679364681244,
      "learning_rate": 1.7976e-05,
      "loss": 0.0136,
      "step": 5995
    },
    {
      "epoch": 11.062788550323177,
      "grad_norm": 0.45307955145835876,
      "learning_rate": 1.7979000000000002e-05,
      "loss": 0.0204,
      "step": 5996
    },
    {
      "epoch": 11.064635272391506,
      "grad_norm": 0.8502277135848999,
      "learning_rate": 1.7982e-05,
      "loss": 0.0208,
      "step": 5997
    },
    {
      "epoch": 11.066481994459833,
      "grad_norm": 0.6238176822662354,
      "learning_rate": 1.7985e-05,
      "loss": 0.0282,
      "step": 5998
    },
    {
      "epoch": 11.068328716528162,
      "grad_norm": 0.42590227723121643,
      "learning_rate": 1.7988e-05,
      "loss": 0.0182,
      "step": 5999
    },
    {
      "epoch": 11.070175438596491,
      "grad_norm": 0.554565966129303,
      "learning_rate": 1.7991e-05,
      "loss": 0.0299,
      "step": 6000
    },
    {
      "epoch": 11.070175438596491,
      "eval_cer": 0.11734971615211136,
      "eval_loss": 0.328770250082016,
      "eval_runtime": 16.1595,
      "eval_samples_per_second": 18.812,
      "eval_steps_per_second": 0.619,
      "eval_wer": 0.41596316193399846,
      "step": 6000
    },
    {
      "epoch": 11.07202216066482,
      "grad_norm": 0.7954382300376892,
      "learning_rate": 1.7994e-05,
      "loss": 0.0324,
      "step": 6001
    },
    {
      "epoch": 11.073868882733148,
      "grad_norm": 0.3789304494857788,
      "learning_rate": 1.7997e-05,
      "loss": 0.0172,
      "step": 6002
    },
    {
      "epoch": 11.075715604801477,
      "grad_norm": 0.36132848262786865,
      "learning_rate": 1.8e-05,
      "loss": 0.0156,
      "step": 6003
    },
    {
      "epoch": 11.077562326869806,
      "grad_norm": 0.7410117387771606,
      "learning_rate": 1.8003e-05,
      "loss": 0.0261,
      "step": 6004
    },
    {
      "epoch": 11.079409048938135,
      "grad_norm": 0.46199312806129456,
      "learning_rate": 1.8006e-05,
      "loss": 0.0183,
      "step": 6005
    },
    {
      "epoch": 11.081255771006463,
      "grad_norm": 0.5138466358184814,
      "learning_rate": 1.8009e-05,
      "loss": 0.0241,
      "step": 6006
    },
    {
      "epoch": 11.083102493074792,
      "grad_norm": 0.7309773564338684,
      "learning_rate": 1.8012000000000003e-05,
      "loss": 0.0251,
      "step": 6007
    },
    {
      "epoch": 11.084949215143121,
      "grad_norm": 0.40339091420173645,
      "learning_rate": 1.8015000000000003e-05,
      "loss": 0.014,
      "step": 6008
    },
    {
      "epoch": 11.08679593721145,
      "grad_norm": 0.7696073055267334,
      "learning_rate": 1.8018000000000003e-05,
      "loss": 0.0278,
      "step": 6009
    },
    {
      "epoch": 11.088642659279778,
      "grad_norm": 0.9116730690002441,
      "learning_rate": 1.8021000000000002e-05,
      "loss": 0.0336,
      "step": 6010
    },
    {
      "epoch": 11.090489381348107,
      "grad_norm": 1.1889814138412476,
      "learning_rate": 1.8024e-05,
      "loss": 0.0413,
      "step": 6011
    },
    {
      "epoch": 11.092336103416436,
      "grad_norm": 1.3130958080291748,
      "learning_rate": 1.8027e-05,
      "loss": 0.0312,
      "step": 6012
    },
    {
      "epoch": 11.094182825484765,
      "grad_norm": 0.7441951036453247,
      "learning_rate": 1.803e-05,
      "loss": 0.2037,
      "step": 6013
    },
    {
      "epoch": 11.096029547553094,
      "grad_norm": 0.6344949007034302,
      "learning_rate": 1.8032999999999998e-05,
      "loss": 0.1776,
      "step": 6014
    },
    {
      "epoch": 11.097876269621421,
      "grad_norm": 0.8395406007766724,
      "learning_rate": 1.8035999999999998e-05,
      "loss": 0.236,
      "step": 6015
    },
    {
      "epoch": 11.09972299168975,
      "grad_norm": 0.7685767412185669,
      "learning_rate": 1.8038999999999998e-05,
      "loss": 0.1972,
      "step": 6016
    },
    {
      "epoch": 11.10156971375808,
      "grad_norm": 0.8736318945884705,
      "learning_rate": 1.8042e-05,
      "loss": 0.2188,
      "step": 6017
    },
    {
      "epoch": 11.103416435826409,
      "grad_norm": 0.5640191435813904,
      "learning_rate": 1.8045e-05,
      "loss": 0.0933,
      "step": 6018
    },
    {
      "epoch": 11.105263157894736,
      "grad_norm": 0.521339476108551,
      "learning_rate": 1.8048e-05,
      "loss": 0.0808,
      "step": 6019
    },
    {
      "epoch": 11.107109879963065,
      "grad_norm": 0.6846213936805725,
      "learning_rate": 1.8051e-05,
      "loss": 0.1305,
      "step": 6020
    },
    {
      "epoch": 11.108956602031395,
      "grad_norm": 0.7579799890518188,
      "learning_rate": 1.8054e-05,
      "loss": 0.0898,
      "step": 6021
    },
    {
      "epoch": 11.110803324099724,
      "grad_norm": 0.5198628306388855,
      "learning_rate": 1.8057e-05,
      "loss": 0.0999,
      "step": 6022
    },
    {
      "epoch": 11.112650046168051,
      "grad_norm": 0.5691999793052673,
      "learning_rate": 1.806e-05,
      "loss": 0.0805,
      "step": 6023
    },
    {
      "epoch": 11.11449676823638,
      "grad_norm": 0.5658872723579407,
      "learning_rate": 1.8063e-05,
      "loss": 0.0517,
      "step": 6024
    },
    {
      "epoch": 11.11634349030471,
      "grad_norm": 0.47111114859580994,
      "learning_rate": 1.8066e-05,
      "loss": 0.0653,
      "step": 6025
    },
    {
      "epoch": 11.118190212373039,
      "grad_norm": 0.5160207748413086,
      "learning_rate": 1.8069e-05,
      "loss": 0.0678,
      "step": 6026
    },
    {
      "epoch": 11.120036934441366,
      "grad_norm": 0.5169432163238525,
      "learning_rate": 1.8072000000000002e-05,
      "loss": 0.0425,
      "step": 6027
    },
    {
      "epoch": 11.121883656509695,
      "grad_norm": 0.3853415548801422,
      "learning_rate": 1.8075000000000002e-05,
      "loss": 0.0303,
      "step": 6028
    },
    {
      "epoch": 11.123730378578024,
      "grad_norm": 0.4425393044948578,
      "learning_rate": 1.8078000000000002e-05,
      "loss": 0.0147,
      "step": 6029
    },
    {
      "epoch": 11.125577100646353,
      "grad_norm": 0.46627745032310486,
      "learning_rate": 1.8081000000000002e-05,
      "loss": 0.0306,
      "step": 6030
    },
    {
      "epoch": 11.12742382271468,
      "grad_norm": 0.4932944178581238,
      "learning_rate": 1.8084e-05,
      "loss": 0.0487,
      "step": 6031
    },
    {
      "epoch": 11.12927054478301,
      "grad_norm": 0.43220511078834534,
      "learning_rate": 1.8087e-05,
      "loss": 0.01,
      "step": 6032
    },
    {
      "epoch": 11.131117266851339,
      "grad_norm": 1.317031741142273,
      "learning_rate": 1.809e-05,
      "loss": 0.0253,
      "step": 6033
    },
    {
      "epoch": 11.132963988919668,
      "grad_norm": 0.2857363820075989,
      "learning_rate": 1.8093e-05,
      "loss": 0.0114,
      "step": 6034
    },
    {
      "epoch": 11.134810710987995,
      "grad_norm": 1.058005690574646,
      "learning_rate": 1.8096e-05,
      "loss": 0.0329,
      "step": 6035
    },
    {
      "epoch": 11.136657433056325,
      "grad_norm": 0.6060735583305359,
      "learning_rate": 1.8098999999999997e-05,
      "loss": 0.0236,
      "step": 6036
    },
    {
      "epoch": 11.138504155124654,
      "grad_norm": 0.4277653694152832,
      "learning_rate": 1.8102e-05,
      "loss": 0.0267,
      "step": 6037
    },
    {
      "epoch": 11.140350877192983,
      "grad_norm": 0.42562127113342285,
      "learning_rate": 1.8105e-05,
      "loss": 0.0375,
      "step": 6038
    },
    {
      "epoch": 11.142197599261312,
      "grad_norm": 0.4103083312511444,
      "learning_rate": 1.8108e-05,
      "loss": 0.0179,
      "step": 6039
    },
    {
      "epoch": 11.14404432132964,
      "grad_norm": 0.5249040722846985,
      "learning_rate": 1.8111e-05,
      "loss": 0.0222,
      "step": 6040
    },
    {
      "epoch": 11.145891043397969,
      "grad_norm": 0.3144191801548004,
      "learning_rate": 1.8114e-05,
      "loss": 0.0161,
      "step": 6041
    },
    {
      "epoch": 11.147737765466298,
      "grad_norm": 0.9872508645057678,
      "learning_rate": 1.8117e-05,
      "loss": 0.0518,
      "step": 6042
    },
    {
      "epoch": 11.149584487534627,
      "grad_norm": 0.44013455510139465,
      "learning_rate": 1.812e-05,
      "loss": 0.0265,
      "step": 6043
    },
    {
      "epoch": 11.151431209602954,
      "grad_norm": 0.4432675540447235,
      "learning_rate": 1.8123e-05,
      "loss": 0.0194,
      "step": 6044
    },
    {
      "epoch": 11.153277931671283,
      "grad_norm": 0.3006099462509155,
      "learning_rate": 1.8126e-05,
      "loss": 0.0187,
      "step": 6045
    },
    {
      "epoch": 11.155124653739612,
      "grad_norm": 0.3285147547721863,
      "learning_rate": 1.8129e-05,
      "loss": 0.0195,
      "step": 6046
    },
    {
      "epoch": 11.156971375807942,
      "grad_norm": 0.29914382100105286,
      "learning_rate": 1.8132000000000002e-05,
      "loss": 0.0152,
      "step": 6047
    },
    {
      "epoch": 11.158818097876269,
      "grad_norm": 1.1159223318099976,
      "learning_rate": 1.8135000000000002e-05,
      "loss": 0.0245,
      "step": 6048
    },
    {
      "epoch": 11.160664819944598,
      "grad_norm": 0.45235589146614075,
      "learning_rate": 1.8138e-05,
      "loss": 0.0177,
      "step": 6049
    },
    {
      "epoch": 11.162511542012927,
      "grad_norm": 0.5673660039901733,
      "learning_rate": 1.8141e-05,
      "loss": 0.027,
      "step": 6050
    },
    {
      "epoch": 11.164358264081256,
      "grad_norm": 0.4398684799671173,
      "learning_rate": 1.8144e-05,
      "loss": 0.021,
      "step": 6051
    },
    {
      "epoch": 11.166204986149584,
      "grad_norm": 0.9551856517791748,
      "learning_rate": 1.8147e-05,
      "loss": 0.0465,
      "step": 6052
    },
    {
      "epoch": 11.168051708217913,
      "grad_norm": 0.7103539705276489,
      "learning_rate": 1.815e-05,
      "loss": 0.0261,
      "step": 6053
    },
    {
      "epoch": 11.169898430286242,
      "grad_norm": 1.8120602369308472,
      "learning_rate": 1.8153e-05,
      "loss": 0.0183,
      "step": 6054
    },
    {
      "epoch": 11.171745152354571,
      "grad_norm": 0.4405721127986908,
      "learning_rate": 1.8156e-05,
      "loss": 0.0201,
      "step": 6055
    },
    {
      "epoch": 11.173591874422899,
      "grad_norm": 0.4213368892669678,
      "learning_rate": 1.8159e-05,
      "loss": 0.0172,
      "step": 6056
    },
    {
      "epoch": 11.175438596491228,
      "grad_norm": 0.4763179123401642,
      "learning_rate": 1.8162000000000003e-05,
      "loss": 0.0106,
      "step": 6057
    },
    {
      "epoch": 11.177285318559557,
      "grad_norm": 0.33773693442344666,
      "learning_rate": 1.8165000000000003e-05,
      "loss": 0.0176,
      "step": 6058
    },
    {
      "epoch": 11.179132040627886,
      "grad_norm": 0.7333492636680603,
      "learning_rate": 1.8168000000000003e-05,
      "loss": 0.0281,
      "step": 6059
    },
    {
      "epoch": 11.180978762696213,
      "grad_norm": 0.3161907494068146,
      "learning_rate": 1.8171e-05,
      "loss": 0.0119,
      "step": 6060
    },
    {
      "epoch": 11.182825484764543,
      "grad_norm": 0.7705785632133484,
      "learning_rate": 1.8174e-05,
      "loss": 0.024,
      "step": 6061
    },
    {
      "epoch": 11.184672206832872,
      "grad_norm": 0.6240018606185913,
      "learning_rate": 1.8177e-05,
      "loss": 0.0143,
      "step": 6062
    },
    {
      "epoch": 11.1865189289012,
      "grad_norm": 1.3212826251983643,
      "learning_rate": 1.818e-05,
      "loss": 0.3074,
      "step": 6063
    },
    {
      "epoch": 11.18836565096953,
      "grad_norm": 0.6692222356796265,
      "learning_rate": 1.8183e-05,
      "loss": 0.1793,
      "step": 6064
    },
    {
      "epoch": 11.190212373037857,
      "grad_norm": 0.6634356379508972,
      "learning_rate": 1.8186e-05,
      "loss": 0.14,
      "step": 6065
    },
    {
      "epoch": 11.192059095106186,
      "grad_norm": 0.6485190987586975,
      "learning_rate": 1.8188999999999998e-05,
      "loss": 0.1288,
      "step": 6066
    },
    {
      "epoch": 11.193905817174516,
      "grad_norm": 0.7990904450416565,
      "learning_rate": 1.8192e-05,
      "loss": 0.1893,
      "step": 6067
    },
    {
      "epoch": 11.195752539242845,
      "grad_norm": 0.7302971482276917,
      "learning_rate": 1.8195e-05,
      "loss": 0.1558,
      "step": 6068
    },
    {
      "epoch": 11.197599261311172,
      "grad_norm": 0.6865899562835693,
      "learning_rate": 1.8198e-05,
      "loss": 0.1101,
      "step": 6069
    },
    {
      "epoch": 11.199445983379501,
      "grad_norm": 0.6729971766471863,
      "learning_rate": 1.8201e-05,
      "loss": 0.1269,
      "step": 6070
    },
    {
      "epoch": 11.20129270544783,
      "grad_norm": 0.725504457950592,
      "learning_rate": 1.8204e-05,
      "loss": 0.13,
      "step": 6071
    },
    {
      "epoch": 11.20313942751616,
      "grad_norm": 0.5052322149276733,
      "learning_rate": 1.8207e-05,
      "loss": 0.0995,
      "step": 6072
    },
    {
      "epoch": 11.204986149584487,
      "grad_norm": 0.8587621450424194,
      "learning_rate": 1.821e-05,
      "loss": 0.1422,
      "step": 6073
    },
    {
      "epoch": 11.206832871652816,
      "grad_norm": 0.4450884461402893,
      "learning_rate": 1.8213e-05,
      "loss": 0.0913,
      "step": 6074
    },
    {
      "epoch": 11.208679593721145,
      "grad_norm": 0.6525385975837708,
      "learning_rate": 1.8216e-05,
      "loss": 0.0399,
      "step": 6075
    },
    {
      "epoch": 11.210526315789474,
      "grad_norm": 0.4793529510498047,
      "learning_rate": 1.8219e-05,
      "loss": 0.0398,
      "step": 6076
    },
    {
      "epoch": 11.212373037857802,
      "grad_norm": 0.6193368434906006,
      "learning_rate": 1.8222000000000003e-05,
      "loss": 0.0483,
      "step": 6077
    },
    {
      "epoch": 11.21421975992613,
      "grad_norm": 0.2656480669975281,
      "learning_rate": 1.8225000000000003e-05,
      "loss": 0.0243,
      "step": 6078
    },
    {
      "epoch": 11.21606648199446,
      "grad_norm": 0.3633423447608948,
      "learning_rate": 1.8228000000000002e-05,
      "loss": 0.0286,
      "step": 6079
    },
    {
      "epoch": 11.21791320406279,
      "grad_norm": 0.4554925858974457,
      "learning_rate": 1.8231000000000002e-05,
      "loss": 0.0353,
      "step": 6080
    },
    {
      "epoch": 11.219759926131117,
      "grad_norm": 0.3139352798461914,
      "learning_rate": 1.8234000000000002e-05,
      "loss": 0.0155,
      "step": 6081
    },
    {
      "epoch": 11.221606648199446,
      "grad_norm": 0.5080939531326294,
      "learning_rate": 1.8237000000000002e-05,
      "loss": 0.0253,
      "step": 6082
    },
    {
      "epoch": 11.223453370267775,
      "grad_norm": 0.4644097089767456,
      "learning_rate": 1.824e-05,
      "loss": 0.0212,
      "step": 6083
    },
    {
      "epoch": 11.225300092336104,
      "grad_norm": 0.6743317246437073,
      "learning_rate": 1.8243e-05,
      "loss": 0.0294,
      "step": 6084
    },
    {
      "epoch": 11.227146814404431,
      "grad_norm": 0.3275095224380493,
      "learning_rate": 1.8245999999999998e-05,
      "loss": 0.027,
      "step": 6085
    },
    {
      "epoch": 11.22899353647276,
      "grad_norm": 0.5828282833099365,
      "learning_rate": 1.8248999999999998e-05,
      "loss": 0.027,
      "step": 6086
    },
    {
      "epoch": 11.23084025854109,
      "grad_norm": 0.4786163866519928,
      "learning_rate": 1.8252e-05,
      "loss": 0.0278,
      "step": 6087
    },
    {
      "epoch": 11.232686980609419,
      "grad_norm": 0.35602590441703796,
      "learning_rate": 1.8255e-05,
      "loss": 0.0196,
      "step": 6088
    },
    {
      "epoch": 11.234533702677748,
      "grad_norm": 0.9777085185050964,
      "learning_rate": 1.8258e-05,
      "loss": 0.0278,
      "step": 6089
    },
    {
      "epoch": 11.236380424746075,
      "grad_norm": 0.6795394420623779,
      "learning_rate": 1.8261e-05,
      "loss": 0.028,
      "step": 6090
    },
    {
      "epoch": 11.238227146814404,
      "grad_norm": 0.5188271403312683,
      "learning_rate": 1.8264e-05,
      "loss": 0.0392,
      "step": 6091
    },
    {
      "epoch": 11.240073868882734,
      "grad_norm": 0.4623538851737976,
      "learning_rate": 1.8267e-05,
      "loss": 0.0272,
      "step": 6092
    },
    {
      "epoch": 11.241920590951063,
      "grad_norm": 0.7097045183181763,
      "learning_rate": 1.827e-05,
      "loss": 0.0191,
      "step": 6093
    },
    {
      "epoch": 11.24376731301939,
      "grad_norm": 0.4057754874229431,
      "learning_rate": 1.8273e-05,
      "loss": 0.0201,
      "step": 6094
    },
    {
      "epoch": 11.24561403508772,
      "grad_norm": 0.5271565914154053,
      "learning_rate": 1.8276e-05,
      "loss": 0.019,
      "step": 6095
    },
    {
      "epoch": 11.247460757156048,
      "grad_norm": 0.84198397397995,
      "learning_rate": 1.8279e-05,
      "loss": 0.0287,
      "step": 6096
    },
    {
      "epoch": 11.249307479224377,
      "grad_norm": 0.3388877213001251,
      "learning_rate": 1.8282000000000002e-05,
      "loss": 0.0182,
      "step": 6097
    },
    {
      "epoch": 11.251154201292705,
      "grad_norm": 0.4032248854637146,
      "learning_rate": 1.8285000000000002e-05,
      "loss": 0.0219,
      "step": 6098
    },
    {
      "epoch": 11.253000923361034,
      "grad_norm": 0.6903819441795349,
      "learning_rate": 1.8288000000000002e-05,
      "loss": 0.0176,
      "step": 6099
    },
    {
      "epoch": 11.254847645429363,
      "grad_norm": 0.46638402342796326,
      "learning_rate": 1.8291e-05,
      "loss": 0.0293,
      "step": 6100
    },
    {
      "epoch": 11.256694367497692,
      "grad_norm": 0.4857083559036255,
      "learning_rate": 1.8294e-05,
      "loss": 0.0208,
      "step": 6101
    },
    {
      "epoch": 11.25854108956602,
      "grad_norm": 0.5676124095916748,
      "learning_rate": 1.8297e-05,
      "loss": 0.0231,
      "step": 6102
    },
    {
      "epoch": 11.260387811634349,
      "grad_norm": 0.3779642581939697,
      "learning_rate": 1.83e-05,
      "loss": 0.0264,
      "step": 6103
    },
    {
      "epoch": 11.262234533702678,
      "grad_norm": 0.30721354484558105,
      "learning_rate": 1.8303e-05,
      "loss": 0.0181,
      "step": 6104
    },
    {
      "epoch": 11.264081255771007,
      "grad_norm": 0.6499035358428955,
      "learning_rate": 1.8306e-05,
      "loss": 0.0265,
      "step": 6105
    },
    {
      "epoch": 11.265927977839334,
      "grad_norm": 0.5521036386489868,
      "learning_rate": 1.8309e-05,
      "loss": 0.0199,
      "step": 6106
    },
    {
      "epoch": 11.267774699907664,
      "grad_norm": 0.7346552610397339,
      "learning_rate": 1.8312000000000004e-05,
      "loss": 0.0283,
      "step": 6107
    },
    {
      "epoch": 11.269621421975993,
      "grad_norm": 0.5562924742698669,
      "learning_rate": 1.8315000000000003e-05,
      "loss": 0.0279,
      "step": 6108
    },
    {
      "epoch": 11.271468144044322,
      "grad_norm": 0.6505104899406433,
      "learning_rate": 1.8318e-05,
      "loss": 0.0257,
      "step": 6109
    },
    {
      "epoch": 11.27331486611265,
      "grad_norm": 0.8869682550430298,
      "learning_rate": 1.8321e-05,
      "loss": 0.0219,
      "step": 6110
    },
    {
      "epoch": 11.275161588180978,
      "grad_norm": 0.5087664723396301,
      "learning_rate": 1.8324e-05,
      "loss": 0.0255,
      "step": 6111
    },
    {
      "epoch": 11.277008310249307,
      "grad_norm": 0.5345026254653931,
      "learning_rate": 1.8327e-05,
      "loss": 0.0456,
      "step": 6112
    },
    {
      "epoch": 11.278855032317637,
      "grad_norm": 0.959027886390686,
      "learning_rate": 1.833e-05,
      "loss": 0.2465,
      "step": 6113
    },
    {
      "epoch": 11.280701754385966,
      "grad_norm": 1.0771859884262085,
      "learning_rate": 1.8333e-05,
      "loss": 0.2627,
      "step": 6114
    },
    {
      "epoch": 11.282548476454293,
      "grad_norm": 0.6319298148155212,
      "learning_rate": 1.8336e-05,
      "loss": 0.1879,
      "step": 6115
    },
    {
      "epoch": 11.284395198522622,
      "grad_norm": 0.6334912180900574,
      "learning_rate": 1.8339e-05,
      "loss": 0.1726,
      "step": 6116
    },
    {
      "epoch": 11.286241920590951,
      "grad_norm": 0.6496388912200928,
      "learning_rate": 1.8342e-05,
      "loss": 0.0909,
      "step": 6117
    },
    {
      "epoch": 11.28808864265928,
      "grad_norm": 1.0599894523620605,
      "learning_rate": 1.8345e-05,
      "loss": 0.1608,
      "step": 6118
    },
    {
      "epoch": 11.289935364727608,
      "grad_norm": 0.7013213634490967,
      "learning_rate": 1.8348e-05,
      "loss": 0.1448,
      "step": 6119
    },
    {
      "epoch": 11.291782086795937,
      "grad_norm": 0.5151294469833374,
      "learning_rate": 1.8351e-05,
      "loss": 0.1267,
      "step": 6120
    },
    {
      "epoch": 11.293628808864266,
      "grad_norm": 0.6530570983886719,
      "learning_rate": 1.8354e-05,
      "loss": 0.108,
      "step": 6121
    },
    {
      "epoch": 11.295475530932595,
      "grad_norm": 0.672336757183075,
      "learning_rate": 1.8357e-05,
      "loss": 0.1433,
      "step": 6122
    },
    {
      "epoch": 11.297322253000923,
      "grad_norm": 0.5117024779319763,
      "learning_rate": 1.836e-05,
      "loss": 0.0575,
      "step": 6123
    },
    {
      "epoch": 11.299168975069252,
      "grad_norm": 0.4188872277736664,
      "learning_rate": 1.8363e-05,
      "loss": 0.0657,
      "step": 6124
    },
    {
      "epoch": 11.301015697137581,
      "grad_norm": 0.5835525393486023,
      "learning_rate": 1.8366e-05,
      "loss": 0.0622,
      "step": 6125
    },
    {
      "epoch": 11.30286241920591,
      "grad_norm": 0.9023607969284058,
      "learning_rate": 1.8369e-05,
      "loss": 0.0952,
      "step": 6126
    },
    {
      "epoch": 11.304709141274238,
      "grad_norm": 0.35532012581825256,
      "learning_rate": 1.8372000000000003e-05,
      "loss": 0.025,
      "step": 6127
    },
    {
      "epoch": 11.306555863342567,
      "grad_norm": 0.37712958455085754,
      "learning_rate": 1.8375000000000003e-05,
      "loss": 0.025,
      "step": 6128
    },
    {
      "epoch": 11.308402585410896,
      "grad_norm": 0.4042429029941559,
      "learning_rate": 1.8378000000000003e-05,
      "loss": 0.026,
      "step": 6129
    },
    {
      "epoch": 11.310249307479225,
      "grad_norm": 0.461452454328537,
      "learning_rate": 1.8381000000000002e-05,
      "loss": 0.029,
      "step": 6130
    },
    {
      "epoch": 11.312096029547552,
      "grad_norm": 0.32719576358795166,
      "learning_rate": 1.8384000000000002e-05,
      "loss": 0.0254,
      "step": 6131
    },
    {
      "epoch": 11.313942751615881,
      "grad_norm": 0.559114396572113,
      "learning_rate": 1.8387000000000002e-05,
      "loss": 0.0451,
      "step": 6132
    },
    {
      "epoch": 11.31578947368421,
      "grad_norm": 0.38609135150909424,
      "learning_rate": 1.8390000000000002e-05,
      "loss": 0.0148,
      "step": 6133
    },
    {
      "epoch": 11.31763619575254,
      "grad_norm": 0.3838483393192291,
      "learning_rate": 1.8392999999999998e-05,
      "loss": 0.0324,
      "step": 6134
    },
    {
      "epoch": 11.319482917820867,
      "grad_norm": 0.6396909952163696,
      "learning_rate": 1.8395999999999998e-05,
      "loss": 0.0264,
      "step": 6135
    },
    {
      "epoch": 11.321329639889196,
      "grad_norm": 0.7127343416213989,
      "learning_rate": 1.8398999999999998e-05,
      "loss": 0.0166,
      "step": 6136
    },
    {
      "epoch": 11.323176361957525,
      "grad_norm": 0.8143025636672974,
      "learning_rate": 1.8401999999999998e-05,
      "loss": 0.0261,
      "step": 6137
    },
    {
      "epoch": 11.325023084025855,
      "grad_norm": 1.8625601530075073,
      "learning_rate": 1.8405e-05,
      "loss": 0.0181,
      "step": 6138
    },
    {
      "epoch": 11.326869806094184,
      "grad_norm": 0.5424591302871704,
      "learning_rate": 1.8408e-05,
      "loss": 0.0152,
      "step": 6139
    },
    {
      "epoch": 11.328716528162511,
      "grad_norm": 0.31574827432632446,
      "learning_rate": 1.8411e-05,
      "loss": 0.01,
      "step": 6140
    },
    {
      "epoch": 11.33056325023084,
      "grad_norm": 0.47474777698516846,
      "learning_rate": 1.8414e-05,
      "loss": 0.0234,
      "step": 6141
    },
    {
      "epoch": 11.33240997229917,
      "grad_norm": 0.48354342579841614,
      "learning_rate": 1.8417e-05,
      "loss": 0.0156,
      "step": 6142
    },
    {
      "epoch": 11.334256694367498,
      "grad_norm": 0.9422298669815063,
      "learning_rate": 1.842e-05,
      "loss": 0.0511,
      "step": 6143
    },
    {
      "epoch": 11.336103416435826,
      "grad_norm": 0.9375277161598206,
      "learning_rate": 1.8423e-05,
      "loss": 0.0296,
      "step": 6144
    },
    {
      "epoch": 11.337950138504155,
      "grad_norm": 0.85575270652771,
      "learning_rate": 1.8426e-05,
      "loss": 0.0257,
      "step": 6145
    },
    {
      "epoch": 11.339796860572484,
      "grad_norm": 0.4999704957008362,
      "learning_rate": 1.8429e-05,
      "loss": 0.0224,
      "step": 6146
    },
    {
      "epoch": 11.341643582640813,
      "grad_norm": 0.7743487358093262,
      "learning_rate": 1.8432e-05,
      "loss": 0.0349,
      "step": 6147
    },
    {
      "epoch": 11.34349030470914,
      "grad_norm": 0.547167956829071,
      "learning_rate": 1.8435000000000002e-05,
      "loss": 0.0262,
      "step": 6148
    },
    {
      "epoch": 11.34533702677747,
      "grad_norm": 0.46982502937316895,
      "learning_rate": 1.8438000000000002e-05,
      "loss": 0.0505,
      "step": 6149
    },
    {
      "epoch": 11.347183748845799,
      "grad_norm": 0.4561302959918976,
      "learning_rate": 1.8441000000000002e-05,
      "loss": 0.019,
      "step": 6150
    },
    {
      "epoch": 11.349030470914128,
      "grad_norm": 0.556189239025116,
      "learning_rate": 1.8444e-05,
      "loss": 0.0222,
      "step": 6151
    },
    {
      "epoch": 11.350877192982455,
      "grad_norm": 2.062488555908203,
      "learning_rate": 1.8447e-05,
      "loss": 0.0387,
      "step": 6152
    },
    {
      "epoch": 11.352723915050785,
      "grad_norm": 0.48797592520713806,
      "learning_rate": 1.845e-05,
      "loss": 0.0207,
      "step": 6153
    },
    {
      "epoch": 11.354570637119114,
      "grad_norm": 0.6511245965957642,
      "learning_rate": 1.8453e-05,
      "loss": 0.0292,
      "step": 6154
    },
    {
      "epoch": 11.356417359187443,
      "grad_norm": 0.4582718014717102,
      "learning_rate": 1.8456e-05,
      "loss": 0.0186,
      "step": 6155
    },
    {
      "epoch": 11.35826408125577,
      "grad_norm": 0.567861020565033,
      "learning_rate": 1.8459e-05,
      "loss": 0.0283,
      "step": 6156
    },
    {
      "epoch": 11.3601108033241,
      "grad_norm": 0.6611009240150452,
      "learning_rate": 1.8462e-05,
      "loss": 0.0327,
      "step": 6157
    },
    {
      "epoch": 11.361957525392429,
      "grad_norm": 0.4739968776702881,
      "learning_rate": 1.8465e-05,
      "loss": 0.024,
      "step": 6158
    },
    {
      "epoch": 11.363804247460758,
      "grad_norm": 0.5592001676559448,
      "learning_rate": 1.8468e-05,
      "loss": 0.0183,
      "step": 6159
    },
    {
      "epoch": 11.365650969529085,
      "grad_norm": 0.6979249119758606,
      "learning_rate": 1.8471e-05,
      "loss": 0.0497,
      "step": 6160
    },
    {
      "epoch": 11.367497691597414,
      "grad_norm": 0.5059146881103516,
      "learning_rate": 1.8474e-05,
      "loss": 0.0225,
      "step": 6161
    },
    {
      "epoch": 11.369344413665743,
      "grad_norm": 0.6794499754905701,
      "learning_rate": 1.8477e-05,
      "loss": 0.0312,
      "step": 6162
    },
    {
      "epoch": 11.371191135734072,
      "grad_norm": 1.0225704908370972,
      "learning_rate": 1.848e-05,
      "loss": 0.2726,
      "step": 6163
    },
    {
      "epoch": 11.373037857802402,
      "grad_norm": 0.9803306460380554,
      "learning_rate": 1.8483e-05,
      "loss": 0.2974,
      "step": 6164
    },
    {
      "epoch": 11.374884579870729,
      "grad_norm": 0.6645657420158386,
      "learning_rate": 1.8486e-05,
      "loss": 0.1743,
      "step": 6165
    },
    {
      "epoch": 11.376731301939058,
      "grad_norm": 0.9934237003326416,
      "learning_rate": 1.8489e-05,
      "loss": 0.1558,
      "step": 6166
    },
    {
      "epoch": 11.378578024007387,
      "grad_norm": 0.7072994709014893,
      "learning_rate": 1.8492e-05,
      "loss": 0.1928,
      "step": 6167
    },
    {
      "epoch": 11.380424746075716,
      "grad_norm": 0.7351928353309631,
      "learning_rate": 1.8495e-05,
      "loss": 0.1044,
      "step": 6168
    },
    {
      "epoch": 11.382271468144044,
      "grad_norm": 0.6384725570678711,
      "learning_rate": 1.8498e-05,
      "loss": 0.1161,
      "step": 6169
    },
    {
      "epoch": 11.384118190212373,
      "grad_norm": 0.5761701464653015,
      "learning_rate": 1.8501e-05,
      "loss": 0.0978,
      "step": 6170
    },
    {
      "epoch": 11.385964912280702,
      "grad_norm": 0.7269904613494873,
      "learning_rate": 1.8504e-05,
      "loss": 0.1354,
      "step": 6171
    },
    {
      "epoch": 11.387811634349031,
      "grad_norm": 0.6286356449127197,
      "learning_rate": 1.8507e-05,
      "loss": 0.0969,
      "step": 6172
    },
    {
      "epoch": 11.389658356417359,
      "grad_norm": 0.6741904616355896,
      "learning_rate": 1.851e-05,
      "loss": 0.1379,
      "step": 6173
    },
    {
      "epoch": 11.391505078485688,
      "grad_norm": 0.5379699468612671,
      "learning_rate": 1.8513e-05,
      "loss": 0.0718,
      "step": 6174
    },
    {
      "epoch": 11.393351800554017,
      "grad_norm": 0.6479658484458923,
      "learning_rate": 1.8516e-05,
      "loss": 0.1091,
      "step": 6175
    },
    {
      "epoch": 11.395198522622346,
      "grad_norm": 1.7053743600845337,
      "learning_rate": 1.8519e-05,
      "loss": 0.1225,
      "step": 6176
    },
    {
      "epoch": 11.397045244690673,
      "grad_norm": 0.46991604566574097,
      "learning_rate": 1.8522e-05,
      "loss": 0.0673,
      "step": 6177
    },
    {
      "epoch": 11.398891966759003,
      "grad_norm": 0.8959211707115173,
      "learning_rate": 1.8525000000000003e-05,
      "loss": 0.0555,
      "step": 6178
    },
    {
      "epoch": 11.400738688827332,
      "grad_norm": 0.554658830165863,
      "learning_rate": 1.8528000000000003e-05,
      "loss": 0.0563,
      "step": 6179
    },
    {
      "epoch": 11.40258541089566,
      "grad_norm": 0.3450296223163605,
      "learning_rate": 1.8531000000000003e-05,
      "loss": 0.0186,
      "step": 6180
    },
    {
      "epoch": 11.404432132963988,
      "grad_norm": 0.7503721714019775,
      "learning_rate": 1.8534000000000002e-05,
      "loss": 0.0257,
      "step": 6181
    },
    {
      "epoch": 11.406278855032317,
      "grad_norm": 0.41149789094924927,
      "learning_rate": 1.8537000000000002e-05,
      "loss": 0.0299,
      "step": 6182
    },
    {
      "epoch": 11.408125577100646,
      "grad_norm": 0.8231745362281799,
      "learning_rate": 1.854e-05,
      "loss": 0.0378,
      "step": 6183
    },
    {
      "epoch": 11.409972299168976,
      "grad_norm": 0.61480313539505,
      "learning_rate": 1.8543e-05,
      "loss": 0.0604,
      "step": 6184
    },
    {
      "epoch": 11.411819021237303,
      "grad_norm": 0.42844507098197937,
      "learning_rate": 1.8545999999999998e-05,
      "loss": 0.0346,
      "step": 6185
    },
    {
      "epoch": 11.413665743305632,
      "grad_norm": 0.4954689145088196,
      "learning_rate": 1.8548999999999998e-05,
      "loss": 0.0253,
      "step": 6186
    },
    {
      "epoch": 11.415512465373961,
      "grad_norm": 0.35354283452033997,
      "learning_rate": 1.8551999999999998e-05,
      "loss": 0.0176,
      "step": 6187
    },
    {
      "epoch": 11.41735918744229,
      "grad_norm": 0.3825988173484802,
      "learning_rate": 1.8555e-05,
      "loss": 0.0297,
      "step": 6188
    },
    {
      "epoch": 11.41920590951062,
      "grad_norm": 0.43082669377326965,
      "learning_rate": 1.8558e-05,
      "loss": 0.0344,
      "step": 6189
    },
    {
      "epoch": 11.421052631578947,
      "grad_norm": 0.5307301878929138,
      "learning_rate": 1.8561e-05,
      "loss": 0.0179,
      "step": 6190
    },
    {
      "epoch": 11.422899353647276,
      "grad_norm": 0.6526667475700378,
      "learning_rate": 1.8564e-05,
      "loss": 0.0481,
      "step": 6191
    },
    {
      "epoch": 11.424746075715605,
      "grad_norm": 0.42508047819137573,
      "learning_rate": 1.8567e-05,
      "loss": 0.0164,
      "step": 6192
    },
    {
      "epoch": 11.426592797783934,
      "grad_norm": 0.4299699664115906,
      "learning_rate": 1.857e-05,
      "loss": 0.023,
      "step": 6193
    },
    {
      "epoch": 11.428439519852262,
      "grad_norm": 0.43284791707992554,
      "learning_rate": 1.8573e-05,
      "loss": 0.0188,
      "step": 6194
    },
    {
      "epoch": 11.43028624192059,
      "grad_norm": 0.6751835942268372,
      "learning_rate": 1.8576e-05,
      "loss": 0.0214,
      "step": 6195
    },
    {
      "epoch": 11.43213296398892,
      "grad_norm": 0.4137125015258789,
      "learning_rate": 1.8579e-05,
      "loss": 0.0177,
      "step": 6196
    },
    {
      "epoch": 11.43397968605725,
      "grad_norm": 0.48743513226509094,
      "learning_rate": 1.8582e-05,
      "loss": 0.0232,
      "step": 6197
    },
    {
      "epoch": 11.435826408125576,
      "grad_norm": 0.4469239413738251,
      "learning_rate": 1.8585000000000002e-05,
      "loss": 0.0229,
      "step": 6198
    },
    {
      "epoch": 11.437673130193906,
      "grad_norm": 0.30634063482284546,
      "learning_rate": 1.8588000000000002e-05,
      "loss": 0.0131,
      "step": 6199
    },
    {
      "epoch": 11.439519852262235,
      "grad_norm": 0.774604082107544,
      "learning_rate": 1.8591000000000002e-05,
      "loss": 0.0274,
      "step": 6200
    },
    {
      "epoch": 11.441366574330564,
      "grad_norm": 0.377434641122818,
      "learning_rate": 1.8594000000000002e-05,
      "loss": 0.0169,
      "step": 6201
    },
    {
      "epoch": 11.443213296398891,
      "grad_norm": 0.41127023100852966,
      "learning_rate": 1.8597e-05,
      "loss": 0.0165,
      "step": 6202
    },
    {
      "epoch": 11.44506001846722,
      "grad_norm": 0.40584078431129456,
      "learning_rate": 1.86e-05,
      "loss": 0.0197,
      "step": 6203
    },
    {
      "epoch": 11.44690674053555,
      "grad_norm": 0.5642501711845398,
      "learning_rate": 1.8603e-05,
      "loss": 0.0229,
      "step": 6204
    },
    {
      "epoch": 11.448753462603879,
      "grad_norm": 0.6988409757614136,
      "learning_rate": 1.8606e-05,
      "loss": 0.0124,
      "step": 6205
    },
    {
      "epoch": 11.450600184672206,
      "grad_norm": 0.958296000957489,
      "learning_rate": 1.8609e-05,
      "loss": 0.0309,
      "step": 6206
    },
    {
      "epoch": 11.452446906740535,
      "grad_norm": 0.7275720238685608,
      "learning_rate": 1.8612e-05,
      "loss": 0.019,
      "step": 6207
    },
    {
      "epoch": 11.454293628808864,
      "grad_norm": 0.789011538028717,
      "learning_rate": 1.8615e-05,
      "loss": 0.0253,
      "step": 6208
    },
    {
      "epoch": 11.456140350877194,
      "grad_norm": 0.9203380346298218,
      "learning_rate": 1.8618e-05,
      "loss": 0.0239,
      "step": 6209
    },
    {
      "epoch": 11.45798707294552,
      "grad_norm": 0.726800799369812,
      "learning_rate": 1.8621e-05,
      "loss": 0.0213,
      "step": 6210
    },
    {
      "epoch": 11.45983379501385,
      "grad_norm": 0.308773398399353,
      "learning_rate": 1.8624e-05,
      "loss": 0.0125,
      "step": 6211
    },
    {
      "epoch": 11.46168051708218,
      "grad_norm": 0.7262300848960876,
      "learning_rate": 1.8627e-05,
      "loss": 0.0334,
      "step": 6212
    },
    {
      "epoch": 11.463527239150508,
      "grad_norm": 2.23403000831604,
      "learning_rate": 1.863e-05,
      "loss": 0.3048,
      "step": 6213
    },
    {
      "epoch": 11.465373961218837,
      "grad_norm": 1.738344430923462,
      "learning_rate": 1.8633e-05,
      "loss": 0.3017,
      "step": 6214
    },
    {
      "epoch": 11.467220683287165,
      "grad_norm": 0.6851328015327454,
      "learning_rate": 1.8636e-05,
      "loss": 0.2065,
      "step": 6215
    },
    {
      "epoch": 11.469067405355494,
      "grad_norm": 0.9628781080245972,
      "learning_rate": 1.8639e-05,
      "loss": 0.2222,
      "step": 6216
    },
    {
      "epoch": 11.470914127423823,
      "grad_norm": 0.6785116195678711,
      "learning_rate": 1.8642e-05,
      "loss": 0.1355,
      "step": 6217
    },
    {
      "epoch": 11.472760849492152,
      "grad_norm": 0.8106775879859924,
      "learning_rate": 1.8645000000000002e-05,
      "loss": 0.1877,
      "step": 6218
    },
    {
      "epoch": 11.47460757156048,
      "grad_norm": 0.8037397861480713,
      "learning_rate": 1.8648000000000002e-05,
      "loss": 0.1401,
      "step": 6219
    },
    {
      "epoch": 11.476454293628809,
      "grad_norm": 0.49321749806404114,
      "learning_rate": 1.8651e-05,
      "loss": 0.1098,
      "step": 6220
    },
    {
      "epoch": 11.478301015697138,
      "grad_norm": 0.5776592493057251,
      "learning_rate": 1.8654e-05,
      "loss": 0.0997,
      "step": 6221
    },
    {
      "epoch": 11.480147737765467,
      "grad_norm": 0.9313098192214966,
      "learning_rate": 1.8657e-05,
      "loss": 0.1167,
      "step": 6222
    },
    {
      "epoch": 11.481994459833794,
      "grad_norm": 0.700075626373291,
      "learning_rate": 1.866e-05,
      "loss": 0.086,
      "step": 6223
    },
    {
      "epoch": 11.483841181902124,
      "grad_norm": 0.33606475591659546,
      "learning_rate": 1.8663e-05,
      "loss": 0.0487,
      "step": 6224
    },
    {
      "epoch": 11.485687903970453,
      "grad_norm": 0.6985853910446167,
      "learning_rate": 1.8666e-05,
      "loss": 0.0404,
      "step": 6225
    },
    {
      "epoch": 11.487534626038782,
      "grad_norm": 0.6496925354003906,
      "learning_rate": 1.8669e-05,
      "loss": 0.0782,
      "step": 6226
    },
    {
      "epoch": 11.48938134810711,
      "grad_norm": 1.1872271299362183,
      "learning_rate": 1.8672e-05,
      "loss": 0.0995,
      "step": 6227
    },
    {
      "epoch": 11.491228070175438,
      "grad_norm": 0.6918002367019653,
      "learning_rate": 1.8675000000000003e-05,
      "loss": 0.0593,
      "step": 6228
    },
    {
      "epoch": 11.493074792243767,
      "grad_norm": 0.3916986584663391,
      "learning_rate": 1.8678000000000003e-05,
      "loss": 0.0308,
      "step": 6229
    },
    {
      "epoch": 11.494921514312097,
      "grad_norm": 0.46228450536727905,
      "learning_rate": 1.8681000000000003e-05,
      "loss": 0.0223,
      "step": 6230
    },
    {
      "epoch": 11.496768236380424,
      "grad_norm": 0.48111462593078613,
      "learning_rate": 1.8684000000000003e-05,
      "loss": 0.0312,
      "step": 6231
    },
    {
      "epoch": 11.498614958448753,
      "grad_norm": 0.38682618737220764,
      "learning_rate": 1.8687e-05,
      "loss": 0.0189,
      "step": 6232
    },
    {
      "epoch": 11.500461680517082,
      "grad_norm": 0.3328859210014343,
      "learning_rate": 1.869e-05,
      "loss": 0.0166,
      "step": 6233
    },
    {
      "epoch": 11.502308402585411,
      "grad_norm": 0.512199878692627,
      "learning_rate": 1.8693e-05,
      "loss": 0.0185,
      "step": 6234
    },
    {
      "epoch": 11.504155124653739,
      "grad_norm": 0.524346113204956,
      "learning_rate": 1.8696e-05,
      "loss": 0.0273,
      "step": 6235
    },
    {
      "epoch": 11.506001846722068,
      "grad_norm": 0.5263787508010864,
      "learning_rate": 1.8699e-05,
      "loss": 0.0285,
      "step": 6236
    },
    {
      "epoch": 11.507848568790397,
      "grad_norm": 0.6654219031333923,
      "learning_rate": 1.8701999999999998e-05,
      "loss": 0.0287,
      "step": 6237
    },
    {
      "epoch": 11.509695290858726,
      "grad_norm": 0.4475488066673279,
      "learning_rate": 1.8705e-05,
      "loss": 0.043,
      "step": 6238
    },
    {
      "epoch": 11.511542012927055,
      "grad_norm": 0.4326750934123993,
      "learning_rate": 1.8708e-05,
      "loss": 0.03,
      "step": 6239
    },
    {
      "epoch": 11.513388734995383,
      "grad_norm": 0.4968683123588562,
      "learning_rate": 1.8711e-05,
      "loss": 0.0169,
      "step": 6240
    },
    {
      "epoch": 11.515235457063712,
      "grad_norm": 0.9535714983940125,
      "learning_rate": 1.8714e-05,
      "loss": 0.0247,
      "step": 6241
    },
    {
      "epoch": 11.517082179132041,
      "grad_norm": 0.7456192970275879,
      "learning_rate": 1.8717e-05,
      "loss": 0.0248,
      "step": 6242
    },
    {
      "epoch": 11.51892890120037,
      "grad_norm": 0.23908020555973053,
      "learning_rate": 1.872e-05,
      "loss": 0.0109,
      "step": 6243
    },
    {
      "epoch": 11.520775623268698,
      "grad_norm": 0.42601415514945984,
      "learning_rate": 1.8723e-05,
      "loss": 0.0504,
      "step": 6244
    },
    {
      "epoch": 11.522622345337027,
      "grad_norm": 0.7600030899047852,
      "learning_rate": 1.8726e-05,
      "loss": 0.0212,
      "step": 6245
    },
    {
      "epoch": 11.524469067405356,
      "grad_norm": 0.8499584197998047,
      "learning_rate": 1.8729e-05,
      "loss": 0.0238,
      "step": 6246
    },
    {
      "epoch": 11.526315789473685,
      "grad_norm": 0.5940492749214172,
      "learning_rate": 1.8732e-05,
      "loss": 0.0173,
      "step": 6247
    },
    {
      "epoch": 11.528162511542012,
      "grad_norm": 0.3725167512893677,
      "learning_rate": 1.8735000000000003e-05,
      "loss": 0.02,
      "step": 6248
    },
    {
      "epoch": 11.530009233610341,
      "grad_norm": 0.6236770749092102,
      "learning_rate": 1.8738000000000003e-05,
      "loss": 0.0195,
      "step": 6249
    },
    {
      "epoch": 11.53185595567867,
      "grad_norm": 0.4146452248096466,
      "learning_rate": 1.8741000000000002e-05,
      "loss": 0.0167,
      "step": 6250
    },
    {
      "epoch": 11.533702677747,
      "grad_norm": 0.3959144353866577,
      "learning_rate": 1.8744000000000002e-05,
      "loss": 0.0193,
      "step": 6251
    },
    {
      "epoch": 11.535549399815327,
      "grad_norm": 0.6457730531692505,
      "learning_rate": 1.8747000000000002e-05,
      "loss": 0.0273,
      "step": 6252
    },
    {
      "epoch": 11.537396121883656,
      "grad_norm": 0.6635228991508484,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.0299,
      "step": 6253
    },
    {
      "epoch": 11.539242843951985,
      "grad_norm": 0.41056013107299805,
      "learning_rate": 1.8753e-05,
      "loss": 0.0349,
      "step": 6254
    },
    {
      "epoch": 11.541089566020315,
      "grad_norm": 0.3606818616390228,
      "learning_rate": 1.8756e-05,
      "loss": 0.0117,
      "step": 6255
    },
    {
      "epoch": 11.542936288088642,
      "grad_norm": 1.001387357711792,
      "learning_rate": 1.8759e-05,
      "loss": 0.0178,
      "step": 6256
    },
    {
      "epoch": 11.544783010156971,
      "grad_norm": 0.4944041967391968,
      "learning_rate": 1.8761999999999998e-05,
      "loss": 0.0241,
      "step": 6257
    },
    {
      "epoch": 11.5466297322253,
      "grad_norm": 0.4109015166759491,
      "learning_rate": 1.8764999999999997e-05,
      "loss": 0.0159,
      "step": 6258
    },
    {
      "epoch": 11.54847645429363,
      "grad_norm": 0.4980219006538391,
      "learning_rate": 1.8768e-05,
      "loss": 0.0221,
      "step": 6259
    },
    {
      "epoch": 11.550323176361957,
      "grad_norm": 0.5004519820213318,
      "learning_rate": 1.8771e-05,
      "loss": 0.025,
      "step": 6260
    },
    {
      "epoch": 11.552169898430286,
      "grad_norm": 0.7837686538696289,
      "learning_rate": 1.8774e-05,
      "loss": 0.0397,
      "step": 6261
    },
    {
      "epoch": 11.554016620498615,
      "grad_norm": 1.5615583658218384,
      "learning_rate": 1.8777e-05,
      "loss": 0.0869,
      "step": 6262
    },
    {
      "epoch": 11.555863342566944,
      "grad_norm": 0.7100450992584229,
      "learning_rate": 1.878e-05,
      "loss": 0.2417,
      "step": 6263
    },
    {
      "epoch": 11.557710064635273,
      "grad_norm": 0.5554599761962891,
      "learning_rate": 1.8783e-05,
      "loss": 0.1802,
      "step": 6264
    },
    {
      "epoch": 11.5595567867036,
      "grad_norm": 0.6480225324630737,
      "learning_rate": 1.8786e-05,
      "loss": 0.1449,
      "step": 6265
    },
    {
      "epoch": 11.56140350877193,
      "grad_norm": 0.6713318228721619,
      "learning_rate": 1.8789e-05,
      "loss": 0.1696,
      "step": 6266
    },
    {
      "epoch": 11.563250230840259,
      "grad_norm": 0.7170022130012512,
      "learning_rate": 1.8792e-05,
      "loss": 0.1916,
      "step": 6267
    },
    {
      "epoch": 11.565096952908588,
      "grad_norm": 0.5778694748878479,
      "learning_rate": 1.8795e-05,
      "loss": 0.126,
      "step": 6268
    },
    {
      "epoch": 11.566943674976915,
      "grad_norm": 0.5431328415870667,
      "learning_rate": 1.8798000000000002e-05,
      "loss": 0.108,
      "step": 6269
    },
    {
      "epoch": 11.568790397045245,
      "grad_norm": 1.2238951921463013,
      "learning_rate": 1.8801000000000002e-05,
      "loss": 0.1114,
      "step": 6270
    },
    {
      "epoch": 11.570637119113574,
      "grad_norm": 0.8291918039321899,
      "learning_rate": 1.8804e-05,
      "loss": 0.0839,
      "step": 6271
    },
    {
      "epoch": 11.572483841181903,
      "grad_norm": 0.6827449202537537,
      "learning_rate": 1.8807e-05,
      "loss": 0.1024,
      "step": 6272
    },
    {
      "epoch": 11.57433056325023,
      "grad_norm": 0.4937664568424225,
      "learning_rate": 1.881e-05,
      "loss": 0.0618,
      "step": 6273
    },
    {
      "epoch": 11.57617728531856,
      "grad_norm": 0.4901401400566101,
      "learning_rate": 1.8813e-05,
      "loss": 0.0651,
      "step": 6274
    },
    {
      "epoch": 11.578024007386889,
      "grad_norm": 0.8407455086708069,
      "learning_rate": 1.8816e-05,
      "loss": 0.1219,
      "step": 6275
    },
    {
      "epoch": 11.579870729455218,
      "grad_norm": 0.5447347164154053,
      "learning_rate": 1.8819e-05,
      "loss": 0.0682,
      "step": 6276
    },
    {
      "epoch": 11.581717451523545,
      "grad_norm": 0.3770175278186798,
      "learning_rate": 1.8822e-05,
      "loss": 0.0362,
      "step": 6277
    },
    {
      "epoch": 11.583564173591874,
      "grad_norm": 0.48820844292640686,
      "learning_rate": 1.8825e-05,
      "loss": 0.0351,
      "step": 6278
    },
    {
      "epoch": 11.585410895660203,
      "grad_norm": 0.5128350257873535,
      "learning_rate": 1.8828000000000003e-05,
      "loss": 0.0315,
      "step": 6279
    },
    {
      "epoch": 11.587257617728532,
      "grad_norm": 0.6119294762611389,
      "learning_rate": 1.8831000000000003e-05,
      "loss": 0.0252,
      "step": 6280
    },
    {
      "epoch": 11.58910433979686,
      "grad_norm": 0.5184559226036072,
      "learning_rate": 1.8834e-05,
      "loss": 0.0221,
      "step": 6281
    },
    {
      "epoch": 11.590951061865189,
      "grad_norm": 0.6163107752799988,
      "learning_rate": 1.8837e-05,
      "loss": 0.0258,
      "step": 6282
    },
    {
      "epoch": 11.592797783933518,
      "grad_norm": 0.5317324995994568,
      "learning_rate": 1.884e-05,
      "loss": 0.033,
      "step": 6283
    },
    {
      "epoch": 11.594644506001847,
      "grad_norm": 0.4561191201210022,
      "learning_rate": 1.8843e-05,
      "loss": 0.0296,
      "step": 6284
    },
    {
      "epoch": 11.596491228070175,
      "grad_norm": 0.3529854714870453,
      "learning_rate": 1.8846e-05,
      "loss": 0.0212,
      "step": 6285
    },
    {
      "epoch": 11.598337950138504,
      "grad_norm": 0.39690741896629333,
      "learning_rate": 1.8849e-05,
      "loss": 0.016,
      "step": 6286
    },
    {
      "epoch": 11.600184672206833,
      "grad_norm": 0.6497731804847717,
      "learning_rate": 1.8852e-05,
      "loss": 0.0233,
      "step": 6287
    },
    {
      "epoch": 11.602031394275162,
      "grad_norm": 0.5435981750488281,
      "learning_rate": 1.8854999999999998e-05,
      "loss": 0.0331,
      "step": 6288
    },
    {
      "epoch": 11.603878116343491,
      "grad_norm": 0.724307119846344,
      "learning_rate": 1.8858e-05,
      "loss": 0.029,
      "step": 6289
    },
    {
      "epoch": 11.605724838411819,
      "grad_norm": 0.4700302183628082,
      "learning_rate": 1.8861e-05,
      "loss": 0.0328,
      "step": 6290
    },
    {
      "epoch": 11.607571560480148,
      "grad_norm": 0.39719411730766296,
      "learning_rate": 1.8864e-05,
      "loss": 0.0263,
      "step": 6291
    },
    {
      "epoch": 11.609418282548477,
      "grad_norm": 0.38685211539268494,
      "learning_rate": 1.8867e-05,
      "loss": 0.0182,
      "step": 6292
    },
    {
      "epoch": 11.611265004616806,
      "grad_norm": 0.3751578629016876,
      "learning_rate": 1.887e-05,
      "loss": 0.0185,
      "step": 6293
    },
    {
      "epoch": 11.613111726685133,
      "grad_norm": 0.38899144530296326,
      "learning_rate": 1.8873e-05,
      "loss": 0.0106,
      "step": 6294
    },
    {
      "epoch": 11.614958448753463,
      "grad_norm": 0.6312898993492126,
      "learning_rate": 1.8876e-05,
      "loss": 0.029,
      "step": 6295
    },
    {
      "epoch": 11.616805170821792,
      "grad_norm": 1.0704835653305054,
      "learning_rate": 1.8879e-05,
      "loss": 0.0254,
      "step": 6296
    },
    {
      "epoch": 11.61865189289012,
      "grad_norm": 0.9887194633483887,
      "learning_rate": 1.8882e-05,
      "loss": 0.0248,
      "step": 6297
    },
    {
      "epoch": 11.620498614958448,
      "grad_norm": 0.5434224009513855,
      "learning_rate": 1.8885e-05,
      "loss": 0.0297,
      "step": 6298
    },
    {
      "epoch": 11.622345337026777,
      "grad_norm": 0.506160318851471,
      "learning_rate": 1.8888000000000003e-05,
      "loss": 0.0176,
      "step": 6299
    },
    {
      "epoch": 11.624192059095106,
      "grad_norm": 1.1333355903625488,
      "learning_rate": 1.8891000000000003e-05,
      "loss": 0.045,
      "step": 6300
    },
    {
      "epoch": 11.626038781163436,
      "grad_norm": 0.3217025399208069,
      "learning_rate": 1.8894000000000002e-05,
      "loss": 0.0174,
      "step": 6301
    },
    {
      "epoch": 11.627885503231763,
      "grad_norm": 0.4421088397502899,
      "learning_rate": 1.8897000000000002e-05,
      "loss": 0.0144,
      "step": 6302
    },
    {
      "epoch": 11.629732225300092,
      "grad_norm": 0.5572022199630737,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 0.0194,
      "step": 6303
    },
    {
      "epoch": 11.631578947368421,
      "grad_norm": 0.292294979095459,
      "learning_rate": 1.8903000000000002e-05,
      "loss": 0.0143,
      "step": 6304
    },
    {
      "epoch": 11.63342566943675,
      "grad_norm": 1.2303966283798218,
      "learning_rate": 1.8906e-05,
      "loss": 0.0317,
      "step": 6305
    },
    {
      "epoch": 11.635272391505078,
      "grad_norm": 0.5086663365364075,
      "learning_rate": 1.8908999999999998e-05,
      "loss": 0.022,
      "step": 6306
    },
    {
      "epoch": 11.637119113573407,
      "grad_norm": 0.9124029874801636,
      "learning_rate": 1.8911999999999998e-05,
      "loss": 0.0346,
      "step": 6307
    },
    {
      "epoch": 11.638965835641736,
      "grad_norm": 0.47102823853492737,
      "learning_rate": 1.8914999999999998e-05,
      "loss": 0.0181,
      "step": 6308
    },
    {
      "epoch": 11.640812557710065,
      "grad_norm": 0.624832808971405,
      "learning_rate": 1.8918e-05,
      "loss": 0.0241,
      "step": 6309
    },
    {
      "epoch": 11.642659279778393,
      "grad_norm": 0.7668011784553528,
      "learning_rate": 1.8921e-05,
      "loss": 0.0271,
      "step": 6310
    },
    {
      "epoch": 11.644506001846722,
      "grad_norm": 0.4850684404373169,
      "learning_rate": 1.8924e-05,
      "loss": 0.0169,
      "step": 6311
    },
    {
      "epoch": 11.64635272391505,
      "grad_norm": 0.40155303478240967,
      "learning_rate": 1.8927e-05,
      "loss": 0.0182,
      "step": 6312
    },
    {
      "epoch": 11.64819944598338,
      "grad_norm": 0.7572173476219177,
      "learning_rate": 1.893e-05,
      "loss": 0.2137,
      "step": 6313
    },
    {
      "epoch": 11.65004616805171,
      "grad_norm": 0.5553044080734253,
      "learning_rate": 1.8933e-05,
      "loss": 0.1823,
      "step": 6314
    },
    {
      "epoch": 11.651892890120036,
      "grad_norm": 0.7447746396064758,
      "learning_rate": 1.8936e-05,
      "loss": 0.1886,
      "step": 6315
    },
    {
      "epoch": 11.653739612188366,
      "grad_norm": 0.6007423400878906,
      "learning_rate": 1.8939e-05,
      "loss": 0.1405,
      "step": 6316
    },
    {
      "epoch": 11.655586334256695,
      "grad_norm": 0.6235414147377014,
      "learning_rate": 1.8942e-05,
      "loss": 0.1075,
      "step": 6317
    },
    {
      "epoch": 11.657433056325024,
      "grad_norm": 0.6585131287574768,
      "learning_rate": 1.8945e-05,
      "loss": 0.1024,
      "step": 6318
    },
    {
      "epoch": 11.659279778393351,
      "grad_norm": 0.7461692690849304,
      "learning_rate": 1.8948000000000002e-05,
      "loss": 0.1454,
      "step": 6319
    },
    {
      "epoch": 11.66112650046168,
      "grad_norm": 0.5178083181381226,
      "learning_rate": 1.8951000000000002e-05,
      "loss": 0.0975,
      "step": 6320
    },
    {
      "epoch": 11.66297322253001,
      "grad_norm": 0.7293455004692078,
      "learning_rate": 1.8954000000000002e-05,
      "loss": 0.1067,
      "step": 6321
    },
    {
      "epoch": 11.664819944598339,
      "grad_norm": 0.44974884390830994,
      "learning_rate": 1.8957e-05,
      "loss": 0.0661,
      "step": 6322
    },
    {
      "epoch": 11.666666666666666,
      "grad_norm": 0.45195817947387695,
      "learning_rate": 1.896e-05,
      "loss": 0.0724,
      "step": 6323
    },
    {
      "epoch": 11.668513388734995,
      "grad_norm": 0.5777408480644226,
      "learning_rate": 1.8963e-05,
      "loss": 0.0848,
      "step": 6324
    },
    {
      "epoch": 11.670360110803324,
      "grad_norm": 0.6494334936141968,
      "learning_rate": 1.8966e-05,
      "loss": 0.0469,
      "step": 6325
    },
    {
      "epoch": 11.672206832871654,
      "grad_norm": 0.5034980773925781,
      "learning_rate": 1.8969e-05,
      "loss": 0.0652,
      "step": 6326
    },
    {
      "epoch": 11.67405355493998,
      "grad_norm": 0.37976697087287903,
      "learning_rate": 1.8972e-05,
      "loss": 0.0352,
      "step": 6327
    },
    {
      "epoch": 11.67590027700831,
      "grad_norm": 0.932072639465332,
      "learning_rate": 1.8975e-05,
      "loss": 0.0583,
      "step": 6328
    },
    {
      "epoch": 11.67774699907664,
      "grad_norm": 0.4678153395652771,
      "learning_rate": 1.8978000000000004e-05,
      "loss": 0.0266,
      "step": 6329
    },
    {
      "epoch": 11.679593721144968,
      "grad_norm": 0.45993855595588684,
      "learning_rate": 1.8981e-05,
      "loss": 0.0398,
      "step": 6330
    },
    {
      "epoch": 11.681440443213296,
      "grad_norm": 0.587837815284729,
      "learning_rate": 1.8984e-05,
      "loss": 0.053,
      "step": 6331
    },
    {
      "epoch": 11.683287165281625,
      "grad_norm": 0.5369887948036194,
      "learning_rate": 1.8987e-05,
      "loss": 0.0946,
      "step": 6332
    },
    {
      "epoch": 11.685133887349954,
      "grad_norm": 0.803807258605957,
      "learning_rate": 1.899e-05,
      "loss": 0.0356,
      "step": 6333
    },
    {
      "epoch": 11.686980609418283,
      "grad_norm": 0.4701552391052246,
      "learning_rate": 1.8993e-05,
      "loss": 0.0325,
      "step": 6334
    },
    {
      "epoch": 11.68882733148661,
      "grad_norm": 0.6736271381378174,
      "learning_rate": 1.8996e-05,
      "loss": 0.0449,
      "step": 6335
    },
    {
      "epoch": 11.69067405355494,
      "grad_norm": 0.35552453994750977,
      "learning_rate": 1.8999e-05,
      "loss": 0.0203,
      "step": 6336
    },
    {
      "epoch": 11.692520775623269,
      "grad_norm": 0.3159559965133667,
      "learning_rate": 1.9002e-05,
      "loss": 0.0209,
      "step": 6337
    },
    {
      "epoch": 11.694367497691598,
      "grad_norm": 0.38443759083747864,
      "learning_rate": 1.9005e-05,
      "loss": 0.0223,
      "step": 6338
    },
    {
      "epoch": 11.696214219759927,
      "grad_norm": 0.20609219372272491,
      "learning_rate": 1.9008e-05,
      "loss": 0.0133,
      "step": 6339
    },
    {
      "epoch": 11.698060941828254,
      "grad_norm": 0.5753526091575623,
      "learning_rate": 1.9011e-05,
      "loss": 0.0636,
      "step": 6340
    },
    {
      "epoch": 11.699907663896584,
      "grad_norm": 0.38154879212379456,
      "learning_rate": 1.9014e-05,
      "loss": 0.0193,
      "step": 6341
    },
    {
      "epoch": 11.701754385964913,
      "grad_norm": 0.485611230134964,
      "learning_rate": 1.9017e-05,
      "loss": 0.0177,
      "step": 6342
    },
    {
      "epoch": 11.703601108033242,
      "grad_norm": 0.3644583523273468,
      "learning_rate": 1.902e-05,
      "loss": 0.0158,
      "step": 6343
    },
    {
      "epoch": 11.70544783010157,
      "grad_norm": 0.33003613352775574,
      "learning_rate": 1.9023e-05,
      "loss": 0.0141,
      "step": 6344
    },
    {
      "epoch": 11.707294552169898,
      "grad_norm": 0.33550500869750977,
      "learning_rate": 1.9026e-05,
      "loss": 0.0136,
      "step": 6345
    },
    {
      "epoch": 11.709141274238227,
      "grad_norm": 0.3046441376209259,
      "learning_rate": 1.9029e-05,
      "loss": 0.0159,
      "step": 6346
    },
    {
      "epoch": 11.710987996306557,
      "grad_norm": 0.6863639950752258,
      "learning_rate": 1.9032e-05,
      "loss": 0.0201,
      "step": 6347
    },
    {
      "epoch": 11.712834718374884,
      "grad_norm": 0.7514137029647827,
      "learning_rate": 1.9035e-05,
      "loss": 0.0305,
      "step": 6348
    },
    {
      "epoch": 11.714681440443213,
      "grad_norm": 0.5391896963119507,
      "learning_rate": 1.9038000000000003e-05,
      "loss": 0.0214,
      "step": 6349
    },
    {
      "epoch": 11.716528162511542,
      "grad_norm": 0.5416777729988098,
      "learning_rate": 1.9041000000000003e-05,
      "loss": 0.0147,
      "step": 6350
    },
    {
      "epoch": 11.718374884579871,
      "grad_norm": 0.7514315843582153,
      "learning_rate": 1.9044000000000003e-05,
      "loss": 0.019,
      "step": 6351
    },
    {
      "epoch": 11.720221606648199,
      "grad_norm": 0.8173295259475708,
      "learning_rate": 1.9047000000000002e-05,
      "loss": 0.0279,
      "step": 6352
    },
    {
      "epoch": 11.722068328716528,
      "grad_norm": 0.7046154141426086,
      "learning_rate": 1.9050000000000002e-05,
      "loss": 0.0148,
      "step": 6353
    },
    {
      "epoch": 11.723915050784857,
      "grad_norm": 0.33618953824043274,
      "learning_rate": 1.9053000000000002e-05,
      "loss": 0.0141,
      "step": 6354
    },
    {
      "epoch": 11.725761772853186,
      "grad_norm": 0.4373946189880371,
      "learning_rate": 1.9056e-05,
      "loss": 0.0264,
      "step": 6355
    },
    {
      "epoch": 11.727608494921514,
      "grad_norm": 0.7646742463111877,
      "learning_rate": 1.9058999999999998e-05,
      "loss": 0.0249,
      "step": 6356
    },
    {
      "epoch": 11.729455216989843,
      "grad_norm": 0.9046465158462524,
      "learning_rate": 1.9061999999999998e-05,
      "loss": 0.0266,
      "step": 6357
    },
    {
      "epoch": 11.731301939058172,
      "grad_norm": 0.4174250662326813,
      "learning_rate": 1.9064999999999998e-05,
      "loss": 0.0228,
      "step": 6358
    },
    {
      "epoch": 11.733148661126501,
      "grad_norm": 0.756666898727417,
      "learning_rate": 1.9068e-05,
      "loss": 0.0473,
      "step": 6359
    },
    {
      "epoch": 11.734995383194828,
      "grad_norm": 0.5958936214447021,
      "learning_rate": 1.9071e-05,
      "loss": 0.0234,
      "step": 6360
    },
    {
      "epoch": 11.736842105263158,
      "grad_norm": 0.7871098518371582,
      "learning_rate": 1.9074e-05,
      "loss": 0.0198,
      "step": 6361
    },
    {
      "epoch": 11.738688827331487,
      "grad_norm": 1.7256369590759277,
      "learning_rate": 1.9077e-05,
      "loss": 0.0317,
      "step": 6362
    },
    {
      "epoch": 11.740535549399816,
      "grad_norm": 0.892645537853241,
      "learning_rate": 1.908e-05,
      "loss": 0.2726,
      "step": 6363
    },
    {
      "epoch": 11.742382271468145,
      "grad_norm": 0.7114629745483398,
      "learning_rate": 1.9083e-05,
      "loss": 0.2411,
      "step": 6364
    },
    {
      "epoch": 11.744228993536472,
      "grad_norm": 0.6901366710662842,
      "learning_rate": 1.9086e-05,
      "loss": 0.1997,
      "step": 6365
    },
    {
      "epoch": 11.746075715604801,
      "grad_norm": 0.7185481786727905,
      "learning_rate": 1.9089e-05,
      "loss": 0.164,
      "step": 6366
    },
    {
      "epoch": 11.74792243767313,
      "grad_norm": 0.676231324672699,
      "learning_rate": 1.9092e-05,
      "loss": 0.1472,
      "step": 6367
    },
    {
      "epoch": 11.749769159741458,
      "grad_norm": 0.5787920355796814,
      "learning_rate": 1.9095e-05,
      "loss": 0.1133,
      "step": 6368
    },
    {
      "epoch": 11.751615881809787,
      "grad_norm": 0.5355517268180847,
      "learning_rate": 1.9098000000000002e-05,
      "loss": 0.117,
      "step": 6369
    },
    {
      "epoch": 11.753462603878116,
      "grad_norm": 0.7404063940048218,
      "learning_rate": 1.9101000000000002e-05,
      "loss": 0.1369,
      "step": 6370
    },
    {
      "epoch": 11.755309325946445,
      "grad_norm": 0.5565574765205383,
      "learning_rate": 1.9104000000000002e-05,
      "loss": 0.0798,
      "step": 6371
    },
    {
      "epoch": 11.757156048014775,
      "grad_norm": 0.5662729144096375,
      "learning_rate": 1.9107000000000002e-05,
      "loss": 0.086,
      "step": 6372
    },
    {
      "epoch": 11.759002770083102,
      "grad_norm": 0.5577512383460999,
      "learning_rate": 1.911e-05,
      "loss": 0.0923,
      "step": 6373
    },
    {
      "epoch": 11.760849492151431,
      "grad_norm": 0.5840328931808472,
      "learning_rate": 1.9113e-05,
      "loss": 0.0757,
      "step": 6374
    },
    {
      "epoch": 11.76269621421976,
      "grad_norm": 0.564666748046875,
      "learning_rate": 1.9116e-05,
      "loss": 0.0569,
      "step": 6375
    },
    {
      "epoch": 11.76454293628809,
      "grad_norm": 0.8744934797286987,
      "learning_rate": 1.9119e-05,
      "loss": 0.1026,
      "step": 6376
    },
    {
      "epoch": 11.766389658356417,
      "grad_norm": 0.41327589750289917,
      "learning_rate": 1.9122e-05,
      "loss": 0.0332,
      "step": 6377
    },
    {
      "epoch": 11.768236380424746,
      "grad_norm": 0.43064767122268677,
      "learning_rate": 1.9125e-05,
      "loss": 0.0283,
      "step": 6378
    },
    {
      "epoch": 11.770083102493075,
      "grad_norm": 0.46121013164520264,
      "learning_rate": 1.9128e-05,
      "loss": 0.0629,
      "step": 6379
    },
    {
      "epoch": 11.771929824561404,
      "grad_norm": 0.3944796621799469,
      "learning_rate": 1.9131e-05,
      "loss": 0.0398,
      "step": 6380
    },
    {
      "epoch": 11.773776546629731,
      "grad_norm": 0.3431522250175476,
      "learning_rate": 1.9134e-05,
      "loss": 0.0369,
      "step": 6381
    },
    {
      "epoch": 11.77562326869806,
      "grad_norm": 0.3767950236797333,
      "learning_rate": 1.9137e-05,
      "loss": 0.0288,
      "step": 6382
    },
    {
      "epoch": 11.77746999076639,
      "grad_norm": 0.3588629364967346,
      "learning_rate": 1.914e-05,
      "loss": 0.0206,
      "step": 6383
    },
    {
      "epoch": 11.779316712834719,
      "grad_norm": 0.5769614577293396,
      "learning_rate": 1.9143e-05,
      "loss": 0.0299,
      "step": 6384
    },
    {
      "epoch": 11.781163434903046,
      "grad_norm": 0.49329817295074463,
      "learning_rate": 1.9146e-05,
      "loss": 0.033,
      "step": 6385
    },
    {
      "epoch": 11.783010156971375,
      "grad_norm": 0.38085702061653137,
      "learning_rate": 1.9149e-05,
      "loss": 0.0218,
      "step": 6386
    },
    {
      "epoch": 11.784856879039705,
      "grad_norm": 0.970015287399292,
      "learning_rate": 1.9152e-05,
      "loss": 0.0306,
      "step": 6387
    },
    {
      "epoch": 11.786703601108034,
      "grad_norm": 0.4538666903972626,
      "learning_rate": 1.9155e-05,
      "loss": 0.0162,
      "step": 6388
    },
    {
      "epoch": 11.788550323176363,
      "grad_norm": 0.45507508516311646,
      "learning_rate": 1.9158e-05,
      "loss": 0.0243,
      "step": 6389
    },
    {
      "epoch": 11.79039704524469,
      "grad_norm": 0.4752728343009949,
      "learning_rate": 1.9161000000000002e-05,
      "loss": 0.0184,
      "step": 6390
    },
    {
      "epoch": 11.79224376731302,
      "grad_norm": 0.3834904432296753,
      "learning_rate": 1.9164e-05,
      "loss": 0.0195,
      "step": 6391
    },
    {
      "epoch": 11.794090489381349,
      "grad_norm": 0.42033740878105164,
      "learning_rate": 1.9167e-05,
      "loss": 0.0208,
      "step": 6392
    },
    {
      "epoch": 11.795937211449676,
      "grad_norm": 0.40084317326545715,
      "learning_rate": 1.917e-05,
      "loss": 0.0186,
      "step": 6393
    },
    {
      "epoch": 11.797783933518005,
      "grad_norm": 0.5597448348999023,
      "learning_rate": 1.9173e-05,
      "loss": 0.0245,
      "step": 6394
    },
    {
      "epoch": 11.799630655586334,
      "grad_norm": 0.3703473210334778,
      "learning_rate": 1.9176e-05,
      "loss": 0.0216,
      "step": 6395
    },
    {
      "epoch": 11.801477377654663,
      "grad_norm": 0.39047250151634216,
      "learning_rate": 1.9179e-05,
      "loss": 0.0139,
      "step": 6396
    },
    {
      "epoch": 11.803324099722992,
      "grad_norm": 0.5017368197441101,
      "learning_rate": 1.9182e-05,
      "loss": 0.0182,
      "step": 6397
    },
    {
      "epoch": 11.80517082179132,
      "grad_norm": 0.7702073454856873,
      "learning_rate": 1.9185e-05,
      "loss": 0.0157,
      "step": 6398
    },
    {
      "epoch": 11.807017543859649,
      "grad_norm": 0.5346407890319824,
      "learning_rate": 1.9188e-05,
      "loss": 0.0236,
      "step": 6399
    },
    {
      "epoch": 11.808864265927978,
      "grad_norm": 0.37275490164756775,
      "learning_rate": 1.9191000000000003e-05,
      "loss": 0.0215,
      "step": 6400
    },
    {
      "epoch": 11.810710987996307,
      "grad_norm": 0.6201837062835693,
      "learning_rate": 1.9194000000000003e-05,
      "loss": 0.0226,
      "step": 6401
    },
    {
      "epoch": 11.812557710064635,
      "grad_norm": 1.1140520572662354,
      "learning_rate": 1.9197000000000003e-05,
      "loss": 0.0433,
      "step": 6402
    },
    {
      "epoch": 11.814404432132964,
      "grad_norm": 0.8839095234870911,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.0261,
      "step": 6403
    },
    {
      "epoch": 11.816251154201293,
      "grad_norm": 0.5836665034294128,
      "learning_rate": 1.9203e-05,
      "loss": 0.0299,
      "step": 6404
    },
    {
      "epoch": 11.818097876269622,
      "grad_norm": 0.7881624102592468,
      "learning_rate": 1.9206e-05,
      "loss": 0.0281,
      "step": 6405
    },
    {
      "epoch": 11.81994459833795,
      "grad_norm": 0.7433090806007385,
      "learning_rate": 1.9209e-05,
      "loss": 0.0277,
      "step": 6406
    },
    {
      "epoch": 11.821791320406279,
      "grad_norm": 0.5852258205413818,
      "learning_rate": 1.9212e-05,
      "loss": 0.0233,
      "step": 6407
    },
    {
      "epoch": 11.823638042474608,
      "grad_norm": 0.374634712934494,
      "learning_rate": 1.9214999999999998e-05,
      "loss": 0.0174,
      "step": 6408
    },
    {
      "epoch": 11.825484764542937,
      "grad_norm": 0.4605027735233307,
      "learning_rate": 1.9217999999999998e-05,
      "loss": 0.0194,
      "step": 6409
    },
    {
      "epoch": 11.827331486611264,
      "grad_norm": 0.6278563737869263,
      "learning_rate": 1.9221e-05,
      "loss": 0.0233,
      "step": 6410
    },
    {
      "epoch": 11.829178208679593,
      "grad_norm": 0.7279020547866821,
      "learning_rate": 1.9224e-05,
      "loss": 0.0452,
      "step": 6411
    },
    {
      "epoch": 11.831024930747922,
      "grad_norm": 0.5646888613700867,
      "learning_rate": 1.9227e-05,
      "loss": 0.0137,
      "step": 6412
    },
    {
      "epoch": 11.832871652816252,
      "grad_norm": 0.8535590171813965,
      "learning_rate": 1.923e-05,
      "loss": 0.2184,
      "step": 6413
    },
    {
      "epoch": 11.83471837488458,
      "grad_norm": 0.6590369939804077,
      "learning_rate": 1.9233e-05,
      "loss": 0.2622,
      "step": 6414
    },
    {
      "epoch": 11.836565096952908,
      "grad_norm": 0.7224426865577698,
      "learning_rate": 1.9236e-05,
      "loss": 0.2415,
      "step": 6415
    },
    {
      "epoch": 11.838411819021237,
      "grad_norm": 0.6342539191246033,
      "learning_rate": 1.9239e-05,
      "loss": 0.1965,
      "step": 6416
    },
    {
      "epoch": 11.840258541089566,
      "grad_norm": 0.6734256148338318,
      "learning_rate": 1.9242e-05,
      "loss": 0.1593,
      "step": 6417
    },
    {
      "epoch": 11.842105263157894,
      "grad_norm": 0.6946101188659668,
      "learning_rate": 1.9245e-05,
      "loss": 0.1726,
      "step": 6418
    },
    {
      "epoch": 11.843951985226223,
      "grad_norm": 0.6345155835151672,
      "learning_rate": 1.9248e-05,
      "loss": 0.1261,
      "step": 6419
    },
    {
      "epoch": 11.845798707294552,
      "grad_norm": 0.8070105910301208,
      "learning_rate": 1.9251000000000003e-05,
      "loss": 0.0942,
      "step": 6420
    },
    {
      "epoch": 11.847645429362881,
      "grad_norm": 0.6948537230491638,
      "learning_rate": 1.9254000000000002e-05,
      "loss": 0.1272,
      "step": 6421
    },
    {
      "epoch": 11.84949215143121,
      "grad_norm": 1.096514105796814,
      "learning_rate": 1.9257000000000002e-05,
      "loss": 0.135,
      "step": 6422
    },
    {
      "epoch": 11.851338873499538,
      "grad_norm": 0.5865067839622498,
      "learning_rate": 1.9260000000000002e-05,
      "loss": 0.1008,
      "step": 6423
    },
    {
      "epoch": 11.853185595567867,
      "grad_norm": 0.5337828397750854,
      "learning_rate": 1.9263000000000002e-05,
      "loss": 0.0691,
      "step": 6424
    },
    {
      "epoch": 11.855032317636196,
      "grad_norm": 0.5951547026634216,
      "learning_rate": 1.9266e-05,
      "loss": 0.0889,
      "step": 6425
    },
    {
      "epoch": 11.856879039704525,
      "grad_norm": 0.45236048102378845,
      "learning_rate": 1.9269e-05,
      "loss": 0.0799,
      "step": 6426
    },
    {
      "epoch": 11.858725761772853,
      "grad_norm": 0.4453403353691101,
      "learning_rate": 1.9272e-05,
      "loss": 0.0491,
      "step": 6427
    },
    {
      "epoch": 11.860572483841182,
      "grad_norm": 0.6715244650840759,
      "learning_rate": 1.9275e-05,
      "loss": 0.0551,
      "step": 6428
    },
    {
      "epoch": 11.86241920590951,
      "grad_norm": 0.465982049703598,
      "learning_rate": 1.9277999999999997e-05,
      "loss": 0.0275,
      "step": 6429
    },
    {
      "epoch": 11.86426592797784,
      "grad_norm": 0.4710586667060852,
      "learning_rate": 1.9281e-05,
      "loss": 0.0405,
      "step": 6430
    },
    {
      "epoch": 11.866112650046167,
      "grad_norm": 0.37974944710731506,
      "learning_rate": 1.9284e-05,
      "loss": 0.0258,
      "step": 6431
    },
    {
      "epoch": 11.867959372114496,
      "grad_norm": 0.44384387135505676,
      "learning_rate": 1.9287e-05,
      "loss": 0.0256,
      "step": 6432
    },
    {
      "epoch": 11.869806094182826,
      "grad_norm": 0.536176323890686,
      "learning_rate": 1.929e-05,
      "loss": 0.0261,
      "step": 6433
    },
    {
      "epoch": 11.871652816251155,
      "grad_norm": 0.44399988651275635,
      "learning_rate": 1.9293e-05,
      "loss": 0.0306,
      "step": 6434
    },
    {
      "epoch": 11.873499538319482,
      "grad_norm": 0.47737666964530945,
      "learning_rate": 1.9296e-05,
      "loss": 0.0285,
      "step": 6435
    },
    {
      "epoch": 11.875346260387811,
      "grad_norm": 0.35225576162338257,
      "learning_rate": 1.9299e-05,
      "loss": 0.0196,
      "step": 6436
    },
    {
      "epoch": 11.87719298245614,
      "grad_norm": 0.5185000896453857,
      "learning_rate": 1.9302e-05,
      "loss": 0.025,
      "step": 6437
    },
    {
      "epoch": 11.87903970452447,
      "grad_norm": 0.33857572078704834,
      "learning_rate": 1.9305e-05,
      "loss": 0.0132,
      "step": 6438
    },
    {
      "epoch": 11.880886426592799,
      "grad_norm": 0.40006375312805176,
      "learning_rate": 1.9308e-05,
      "loss": 0.0192,
      "step": 6439
    },
    {
      "epoch": 11.882733148661126,
      "grad_norm": 0.3900461494922638,
      "learning_rate": 1.9311000000000002e-05,
      "loss": 0.0238,
      "step": 6440
    },
    {
      "epoch": 11.884579870729455,
      "grad_norm": 0.22454053163528442,
      "learning_rate": 1.9314000000000002e-05,
      "loss": 0.0128,
      "step": 6441
    },
    {
      "epoch": 11.886426592797784,
      "grad_norm": 0.6170540452003479,
      "learning_rate": 1.9317e-05,
      "loss": 0.0245,
      "step": 6442
    },
    {
      "epoch": 11.888273314866112,
      "grad_norm": 0.4073467254638672,
      "learning_rate": 1.932e-05,
      "loss": 0.0151,
      "step": 6443
    },
    {
      "epoch": 11.89012003693444,
      "grad_norm": 0.4349503517150879,
      "learning_rate": 1.9323e-05,
      "loss": 0.0193,
      "step": 6444
    },
    {
      "epoch": 11.89196675900277,
      "grad_norm": 0.6105451583862305,
      "learning_rate": 1.9326e-05,
      "loss": 0.0266,
      "step": 6445
    },
    {
      "epoch": 11.8938134810711,
      "grad_norm": 0.44563114643096924,
      "learning_rate": 1.9329e-05,
      "loss": 0.0201,
      "step": 6446
    },
    {
      "epoch": 11.895660203139428,
      "grad_norm": 0.37493470311164856,
      "learning_rate": 1.9332e-05,
      "loss": 0.0161,
      "step": 6447
    },
    {
      "epoch": 11.897506925207756,
      "grad_norm": 0.5489336252212524,
      "learning_rate": 1.9335e-05,
      "loss": 0.026,
      "step": 6448
    },
    {
      "epoch": 11.899353647276085,
      "grad_norm": 0.4425601661205292,
      "learning_rate": 1.9338e-05,
      "loss": 0.0341,
      "step": 6449
    },
    {
      "epoch": 11.901200369344414,
      "grad_norm": 1.0750174522399902,
      "learning_rate": 1.9341000000000003e-05,
      "loss": 0.0274,
      "step": 6450
    },
    {
      "epoch": 11.903047091412743,
      "grad_norm": 0.5695386528968811,
      "learning_rate": 1.9344000000000003e-05,
      "loss": 0.0247,
      "step": 6451
    },
    {
      "epoch": 11.90489381348107,
      "grad_norm": 0.953221321105957,
      "learning_rate": 1.9347000000000003e-05,
      "loss": 0.0357,
      "step": 6452
    },
    {
      "epoch": 11.9067405355494,
      "grad_norm": 0.5750436782836914,
      "learning_rate": 1.935e-05,
      "loss": 0.0235,
      "step": 6453
    },
    {
      "epoch": 11.908587257617729,
      "grad_norm": 0.4252550005912781,
      "learning_rate": 1.9353e-05,
      "loss": 0.0249,
      "step": 6454
    },
    {
      "epoch": 11.910433979686058,
      "grad_norm": 1.0517467260360718,
      "learning_rate": 1.9356e-05,
      "loss": 0.0216,
      "step": 6455
    },
    {
      "epoch": 11.912280701754385,
      "grad_norm": 0.45456618070602417,
      "learning_rate": 1.9359e-05,
      "loss": 0.0185,
      "step": 6456
    },
    {
      "epoch": 11.914127423822714,
      "grad_norm": 0.9214330315589905,
      "learning_rate": 1.9362e-05,
      "loss": 0.0264,
      "step": 6457
    },
    {
      "epoch": 11.915974145891044,
      "grad_norm": 0.6516225337982178,
      "learning_rate": 1.9365e-05,
      "loss": 0.0231,
      "step": 6458
    },
    {
      "epoch": 11.917820867959373,
      "grad_norm": 0.8239295482635498,
      "learning_rate": 1.9367999999999998e-05,
      "loss": 0.0239,
      "step": 6459
    },
    {
      "epoch": 11.9196675900277,
      "grad_norm": 0.7495092153549194,
      "learning_rate": 1.9371e-05,
      "loss": 0.0236,
      "step": 6460
    },
    {
      "epoch": 11.92151431209603,
      "grad_norm": 1.0203250646591187,
      "learning_rate": 1.9374e-05,
      "loss": 0.0328,
      "step": 6461
    },
    {
      "epoch": 11.923361034164358,
      "grad_norm": 1.0962752103805542,
      "learning_rate": 1.9377e-05,
      "loss": 0.0556,
      "step": 6462
    },
    {
      "epoch": 11.925207756232687,
      "grad_norm": 0.7170693278312683,
      "learning_rate": 1.938e-05,
      "loss": 0.2026,
      "step": 6463
    },
    {
      "epoch": 11.927054478301017,
      "grad_norm": 0.6785096526145935,
      "learning_rate": 1.9383e-05,
      "loss": 0.163,
      "step": 6464
    },
    {
      "epoch": 11.928901200369344,
      "grad_norm": 0.4773499369621277,
      "learning_rate": 1.9386e-05,
      "loss": 0.1217,
      "step": 6465
    },
    {
      "epoch": 11.930747922437673,
      "grad_norm": 0.9946964383125305,
      "learning_rate": 1.9389e-05,
      "loss": 0.1997,
      "step": 6466
    },
    {
      "epoch": 11.932594644506002,
      "grad_norm": 0.8184168934822083,
      "learning_rate": 1.9392e-05,
      "loss": 0.1777,
      "step": 6467
    },
    {
      "epoch": 11.93444136657433,
      "grad_norm": 0.7503487467765808,
      "learning_rate": 1.9395e-05,
      "loss": 0.1585,
      "step": 6468
    },
    {
      "epoch": 11.936288088642659,
      "grad_norm": 0.4815266728401184,
      "learning_rate": 1.9398e-05,
      "loss": 0.1039,
      "step": 6469
    },
    {
      "epoch": 11.938134810710988,
      "grad_norm": 0.8742775917053223,
      "learning_rate": 1.9401000000000003e-05,
      "loss": 0.1137,
      "step": 6470
    },
    {
      "epoch": 11.939981532779317,
      "grad_norm": 0.5850298404693604,
      "learning_rate": 1.9404000000000003e-05,
      "loss": 0.0984,
      "step": 6471
    },
    {
      "epoch": 11.941828254847646,
      "grad_norm": 0.6854799389839172,
      "learning_rate": 1.9407000000000002e-05,
      "loss": 0.0732,
      "step": 6472
    },
    {
      "epoch": 11.943674976915974,
      "grad_norm": 0.5733482837677002,
      "learning_rate": 1.9410000000000002e-05,
      "loss": 0.0735,
      "step": 6473
    },
    {
      "epoch": 11.945521698984303,
      "grad_norm": 0.5619057416915894,
      "learning_rate": 1.9413000000000002e-05,
      "loss": 0.059,
      "step": 6474
    },
    {
      "epoch": 11.947368421052632,
      "grad_norm": 0.5128675699234009,
      "learning_rate": 1.9416000000000002e-05,
      "loss": 0.0609,
      "step": 6475
    },
    {
      "epoch": 11.949215143120961,
      "grad_norm": 0.5612236261367798,
      "learning_rate": 1.9419e-05,
      "loss": 0.0606,
      "step": 6476
    },
    {
      "epoch": 11.951061865189288,
      "grad_norm": 0.5712298154830933,
      "learning_rate": 1.9422e-05,
      "loss": 0.037,
      "step": 6477
    },
    {
      "epoch": 11.952908587257618,
      "grad_norm": 0.8684074282646179,
      "learning_rate": 1.9424999999999998e-05,
      "loss": 0.0637,
      "step": 6478
    },
    {
      "epoch": 11.954755309325947,
      "grad_norm": 0.3099026083946228,
      "learning_rate": 1.9427999999999998e-05,
      "loss": 0.0245,
      "step": 6479
    },
    {
      "epoch": 11.956602031394276,
      "grad_norm": 0.2949822247028351,
      "learning_rate": 1.9431e-05,
      "loss": 0.0138,
      "step": 6480
    },
    {
      "epoch": 11.958448753462603,
      "grad_norm": 0.3665012717247009,
      "learning_rate": 1.9434e-05,
      "loss": 0.0161,
      "step": 6481
    },
    {
      "epoch": 11.960295475530932,
      "grad_norm": 0.3798087239265442,
      "learning_rate": 1.9437e-05,
      "loss": 0.0244,
      "step": 6482
    },
    {
      "epoch": 11.962142197599261,
      "grad_norm": 0.525526225566864,
      "learning_rate": 1.944e-05,
      "loss": 0.0402,
      "step": 6483
    },
    {
      "epoch": 11.96398891966759,
      "grad_norm": 0.45501410961151123,
      "learning_rate": 1.9443e-05,
      "loss": 0.0238,
      "step": 6484
    },
    {
      "epoch": 11.965835641735918,
      "grad_norm": 0.26545315980911255,
      "learning_rate": 1.9446e-05,
      "loss": 0.0135,
      "step": 6485
    },
    {
      "epoch": 11.967682363804247,
      "grad_norm": 0.6194910407066345,
      "learning_rate": 1.9449e-05,
      "loss": 0.0347,
      "step": 6486
    },
    {
      "epoch": 11.969529085872576,
      "grad_norm": 0.5148172378540039,
      "learning_rate": 1.9452e-05,
      "loss": 0.0699,
      "step": 6487
    },
    {
      "epoch": 11.971375807940905,
      "grad_norm": 0.6581459045410156,
      "learning_rate": 1.9455e-05,
      "loss": 0.0313,
      "step": 6488
    },
    {
      "epoch": 11.973222530009235,
      "grad_norm": 0.6907075047492981,
      "learning_rate": 1.9458e-05,
      "loss": 0.0248,
      "step": 6489
    },
    {
      "epoch": 11.975069252077562,
      "grad_norm": 0.4608518183231354,
      "learning_rate": 1.9461000000000002e-05,
      "loss": 0.0242,
      "step": 6490
    },
    {
      "epoch": 11.976915974145891,
      "grad_norm": 0.48929011821746826,
      "learning_rate": 1.9464000000000002e-05,
      "loss": 0.0171,
      "step": 6491
    },
    {
      "epoch": 11.97876269621422,
      "grad_norm": 0.5067629218101501,
      "learning_rate": 1.9467000000000002e-05,
      "loss": 0.0204,
      "step": 6492
    },
    {
      "epoch": 11.980609418282548,
      "grad_norm": 0.5316839814186096,
      "learning_rate": 1.947e-05,
      "loss": 0.0255,
      "step": 6493
    },
    {
      "epoch": 11.982456140350877,
      "grad_norm": 0.7034479975700378,
      "learning_rate": 1.9473e-05,
      "loss": 0.0407,
      "step": 6494
    },
    {
      "epoch": 11.984302862419206,
      "grad_norm": 0.469872385263443,
      "learning_rate": 1.9476e-05,
      "loss": 0.0268,
      "step": 6495
    },
    {
      "epoch": 11.986149584487535,
      "grad_norm": 0.49888551235198975,
      "learning_rate": 1.9479e-05,
      "loss": 0.0164,
      "step": 6496
    },
    {
      "epoch": 11.987996306555864,
      "grad_norm": 0.5648143887519836,
      "learning_rate": 1.9482e-05,
      "loss": 0.0296,
      "step": 6497
    },
    {
      "epoch": 11.989843028624191,
      "grad_norm": 0.43154245615005493,
      "learning_rate": 1.9485e-05,
      "loss": 0.023,
      "step": 6498
    },
    {
      "epoch": 11.99168975069252,
      "grad_norm": 0.7657442092895508,
      "learning_rate": 1.9488e-05,
      "loss": 0.0136,
      "step": 6499
    },
    {
      "epoch": 11.99353647276085,
      "grad_norm": 0.6309219002723694,
      "learning_rate": 1.9491000000000004e-05,
      "loss": 0.0256,
      "step": 6500
    },
    {
      "epoch": 11.995383194829179,
      "grad_norm": 0.5489864945411682,
      "learning_rate": 1.9494000000000003e-05,
      "loss": 0.0134,
      "step": 6501
    },
    {
      "epoch": 11.997229916897506,
      "grad_norm": 0.5915862321853638,
      "learning_rate": 1.9497e-05,
      "loss": 0.0278,
      "step": 6502
    },
    {
      "epoch": 11.999076638965835,
      "grad_norm": 1.2053042650222778,
      "learning_rate": 1.95e-05,
      "loss": 0.045,
      "step": 6503
    },
    {
      "epoch": 12.0,
      "grad_norm": 0.3713594377040863,
      "learning_rate": 1.9503e-05,
      "loss": 0.0098,
      "step": 6504
    },
    {
      "epoch": 12.00184672206833,
      "grad_norm": 2.1398563385009766,
      "learning_rate": 1.9506e-05,
      "loss": 0.2887,
      "step": 6505
    },
    {
      "epoch": 12.003693444136658,
      "grad_norm": 0.7182486653327942,
      "learning_rate": 1.9509e-05,
      "loss": 0.2153,
      "step": 6506
    },
    {
      "epoch": 12.005540166204986,
      "grad_norm": 0.6150814294815063,
      "learning_rate": 1.9512e-05,
      "loss": 0.1447,
      "step": 6507
    },
    {
      "epoch": 12.007386888273315,
      "grad_norm": 0.772539496421814,
      "learning_rate": 1.9515e-05,
      "loss": 0.1322,
      "step": 6508
    },
    {
      "epoch": 12.009233610341644,
      "grad_norm": 1.2308114767074585,
      "learning_rate": 1.9518e-05,
      "loss": 0.1985,
      "step": 6509
    },
    {
      "epoch": 12.011080332409973,
      "grad_norm": 1.0095338821411133,
      "learning_rate": 1.9520999999999998e-05,
      "loss": 0.1299,
      "step": 6510
    },
    {
      "epoch": 12.0129270544783,
      "grad_norm": 0.43991515040397644,
      "learning_rate": 1.9524e-05,
      "loss": 0.0782,
      "step": 6511
    },
    {
      "epoch": 12.01477377654663,
      "grad_norm": 0.7752163410186768,
      "learning_rate": 1.9527e-05,
      "loss": 0.128,
      "step": 6512
    },
    {
      "epoch": 12.016620498614959,
      "grad_norm": 0.5800342559814453,
      "learning_rate": 1.953e-05,
      "loss": 0.0866,
      "step": 6513
    },
    {
      "epoch": 12.018467220683288,
      "grad_norm": 0.5003458261489868,
      "learning_rate": 1.9533e-05,
      "loss": 0.0812,
      "step": 6514
    },
    {
      "epoch": 12.020313942751615,
      "grad_norm": 0.6201184988021851,
      "learning_rate": 1.9536e-05,
      "loss": 0.0759,
      "step": 6515
    },
    {
      "epoch": 12.022160664819944,
      "grad_norm": 0.4879046678543091,
      "learning_rate": 1.9539e-05,
      "loss": 0.075,
      "step": 6516
    },
    {
      "epoch": 12.024007386888274,
      "grad_norm": 0.5425206422805786,
      "learning_rate": 1.9542e-05,
      "loss": 0.0834,
      "step": 6517
    },
    {
      "epoch": 12.025854108956603,
      "grad_norm": 0.3138076961040497,
      "learning_rate": 1.9545e-05,
      "loss": 0.0351,
      "step": 6518
    },
    {
      "epoch": 12.02770083102493,
      "grad_norm": 0.32039162516593933,
      "learning_rate": 1.9548e-05,
      "loss": 0.0247,
      "step": 6519
    },
    {
      "epoch": 12.02954755309326,
      "grad_norm": 0.28697481751441956,
      "learning_rate": 1.9551e-05,
      "loss": 0.023,
      "step": 6520
    },
    {
      "epoch": 12.031394275161588,
      "grad_norm": 0.4406968355178833,
      "learning_rate": 1.9554000000000003e-05,
      "loss": 0.0343,
      "step": 6521
    },
    {
      "epoch": 12.033240997229917,
      "grad_norm": 0.437920480966568,
      "learning_rate": 1.9557000000000003e-05,
      "loss": 0.0232,
      "step": 6522
    },
    {
      "epoch": 12.035087719298245,
      "grad_norm": 0.24330918490886688,
      "learning_rate": 1.9560000000000002e-05,
      "loss": 0.0188,
      "step": 6523
    },
    {
      "epoch": 12.036934441366574,
      "grad_norm": 0.4772188663482666,
      "learning_rate": 1.9563000000000002e-05,
      "loss": 0.0483,
      "step": 6524
    },
    {
      "epoch": 12.038781163434903,
      "grad_norm": 0.4449160397052765,
      "learning_rate": 1.9566000000000002e-05,
      "loss": 0.0279,
      "step": 6525
    },
    {
      "epoch": 12.040627885503232,
      "grad_norm": 0.481763631105423,
      "learning_rate": 1.9569000000000002e-05,
      "loss": 0.0334,
      "step": 6526
    },
    {
      "epoch": 12.04247460757156,
      "grad_norm": 0.786405086517334,
      "learning_rate": 1.9571999999999998e-05,
      "loss": 0.0255,
      "step": 6527
    },
    {
      "epoch": 12.044321329639889,
      "grad_norm": 0.2664114832878113,
      "learning_rate": 1.9574999999999998e-05,
      "loss": 0.0123,
      "step": 6528
    },
    {
      "epoch": 12.046168051708218,
      "grad_norm": 0.2534761428833008,
      "learning_rate": 1.9577999999999998e-05,
      "loss": 0.0123,
      "step": 6529
    },
    {
      "epoch": 12.048014773776547,
      "grad_norm": 0.4776705205440521,
      "learning_rate": 1.9580999999999998e-05,
      "loss": 0.0211,
      "step": 6530
    },
    {
      "epoch": 12.049861495844876,
      "grad_norm": 0.8299447298049927,
      "learning_rate": 1.9584e-05,
      "loss": 0.0158,
      "step": 6531
    },
    {
      "epoch": 12.051708217913204,
      "grad_norm": 0.2538312077522278,
      "learning_rate": 1.9587e-05,
      "loss": 0.011,
      "step": 6532
    },
    {
      "epoch": 12.053554939981533,
      "grad_norm": 1.139421820640564,
      "learning_rate": 1.959e-05,
      "loss": 0.021,
      "step": 6533
    },
    {
      "epoch": 12.055401662049862,
      "grad_norm": 0.7884008288383484,
      "learning_rate": 1.9593e-05,
      "loss": 0.1168,
      "step": 6534
    },
    {
      "epoch": 12.057248384118191,
      "grad_norm": 0.7113057374954224,
      "learning_rate": 1.9596e-05,
      "loss": 0.0277,
      "step": 6535
    },
    {
      "epoch": 12.059095106186518,
      "grad_norm": 0.9404953718185425,
      "learning_rate": 1.9599e-05,
      "loss": 0.0284,
      "step": 6536
    },
    {
      "epoch": 12.060941828254848,
      "grad_norm": 0.6621551513671875,
      "learning_rate": 1.9602e-05,
      "loss": 0.0208,
      "step": 6537
    },
    {
      "epoch": 12.062788550323177,
      "grad_norm": 0.6471493244171143,
      "learning_rate": 1.9605e-05,
      "loss": 0.0229,
      "step": 6538
    },
    {
      "epoch": 12.064635272391506,
      "grad_norm": 0.6462122797966003,
      "learning_rate": 1.9608e-05,
      "loss": 0.0257,
      "step": 6539
    },
    {
      "epoch": 12.066481994459833,
      "grad_norm": 0.6691087484359741,
      "learning_rate": 1.9611e-05,
      "loss": 0.0228,
      "step": 6540
    },
    {
      "epoch": 12.068328716528162,
      "grad_norm": 0.5461102724075317,
      "learning_rate": 1.9614000000000002e-05,
      "loss": 0.046,
      "step": 6541
    },
    {
      "epoch": 12.070175438596491,
      "grad_norm": 0.46700987219810486,
      "learning_rate": 1.9617000000000002e-05,
      "loss": 0.0109,
      "step": 6542
    },
    {
      "epoch": 12.07202216066482,
      "grad_norm": 0.4373733401298523,
      "learning_rate": 1.9620000000000002e-05,
      "loss": 0.0279,
      "step": 6543
    },
    {
      "epoch": 12.073868882733148,
      "grad_norm": 0.3557780981063843,
      "learning_rate": 1.9623e-05,
      "loss": 0.0134,
      "step": 6544
    },
    {
      "epoch": 12.075715604801477,
      "grad_norm": 0.5521689653396606,
      "learning_rate": 1.9626e-05,
      "loss": 0.0166,
      "step": 6545
    },
    {
      "epoch": 12.077562326869806,
      "grad_norm": 0.43459439277648926,
      "learning_rate": 1.9629e-05,
      "loss": 0.016,
      "step": 6546
    },
    {
      "epoch": 12.079409048938135,
      "grad_norm": 1.5827715396881104,
      "learning_rate": 1.9632e-05,
      "loss": 0.0308,
      "step": 6547
    },
    {
      "epoch": 12.081255771006463,
      "grad_norm": 0.5385817885398865,
      "learning_rate": 1.9635e-05,
      "loss": 0.0255,
      "step": 6548
    },
    {
      "epoch": 12.083102493074792,
      "grad_norm": 0.44712623953819275,
      "learning_rate": 1.9638e-05,
      "loss": 0.0099,
      "step": 6549
    },
    {
      "epoch": 12.084949215143121,
      "grad_norm": 0.621837317943573,
      "learning_rate": 1.9641e-05,
      "loss": 0.0273,
      "step": 6550
    },
    {
      "epoch": 12.08679593721145,
      "grad_norm": 0.321278840303421,
      "learning_rate": 1.9644e-05,
      "loss": 0.0159,
      "step": 6551
    },
    {
      "epoch": 12.088642659279778,
      "grad_norm": 0.6715978980064392,
      "learning_rate": 1.9647e-05,
      "loss": 0.0202,
      "step": 6552
    },
    {
      "epoch": 12.090489381348107,
      "grad_norm": 0.5194762349128723,
      "learning_rate": 1.965e-05,
      "loss": 0.0208,
      "step": 6553
    },
    {
      "epoch": 12.092336103416436,
      "grad_norm": 0.6484201550483704,
      "learning_rate": 1.9653e-05,
      "loss": 0.0149,
      "step": 6554
    },
    {
      "epoch": 12.094182825484765,
      "grad_norm": 1.2438277006149292,
      "learning_rate": 1.9656e-05,
      "loss": 0.2754,
      "step": 6555
    },
    {
      "epoch": 12.096029547553094,
      "grad_norm": 0.9930234551429749,
      "learning_rate": 1.9659e-05,
      "loss": 0.239,
      "step": 6556
    },
    {
      "epoch": 12.097876269621421,
      "grad_norm": 0.6995410323143005,
      "learning_rate": 1.9662e-05,
      "loss": 0.2009,
      "step": 6557
    },
    {
      "epoch": 12.09972299168975,
      "grad_norm": 0.8049606084823608,
      "learning_rate": 1.9665e-05,
      "loss": 0.1279,
      "step": 6558
    },
    {
      "epoch": 12.10156971375808,
      "grad_norm": 0.649700939655304,
      "learning_rate": 1.9668e-05,
      "loss": 0.1453,
      "step": 6559
    },
    {
      "epoch": 12.103416435826409,
      "grad_norm": 0.6611593961715698,
      "learning_rate": 1.9671e-05,
      "loss": 0.1059,
      "step": 6560
    },
    {
      "epoch": 12.105263157894736,
      "grad_norm": 0.6356304883956909,
      "learning_rate": 1.9674000000000002e-05,
      "loss": 0.1879,
      "step": 6561
    },
    {
      "epoch": 12.107109879963065,
      "grad_norm": 2.028852939605713,
      "learning_rate": 1.9677e-05,
      "loss": 0.1257,
      "step": 6562
    },
    {
      "epoch": 12.108956602031395,
      "grad_norm": 0.5349612236022949,
      "learning_rate": 1.968e-05,
      "loss": 0.1681,
      "step": 6563
    },
    {
      "epoch": 12.110803324099724,
      "grad_norm": 1.0658071041107178,
      "learning_rate": 1.9683e-05,
      "loss": 0.086,
      "step": 6564
    },
    {
      "epoch": 12.112650046168051,
      "grad_norm": 0.8170069456100464,
      "learning_rate": 1.9686e-05,
      "loss": 0.064,
      "step": 6565
    },
    {
      "epoch": 12.11449676823638,
      "grad_norm": 0.5723968744277954,
      "learning_rate": 1.9689e-05,
      "loss": 0.0567,
      "step": 6566
    },
    {
      "epoch": 12.11634349030471,
      "grad_norm": 0.8841872811317444,
      "learning_rate": 1.9692e-05,
      "loss": 0.1255,
      "step": 6567
    },
    {
      "epoch": 12.118190212373039,
      "grad_norm": 0.3957323431968689,
      "learning_rate": 1.9695e-05,
      "loss": 0.0482,
      "step": 6568
    },
    {
      "epoch": 12.120036934441366,
      "grad_norm": 0.4203268885612488,
      "learning_rate": 1.9698e-05,
      "loss": 0.0498,
      "step": 6569
    },
    {
      "epoch": 12.121883656509695,
      "grad_norm": 0.48907235264778137,
      "learning_rate": 1.9701e-05,
      "loss": 0.0271,
      "step": 6570
    },
    {
      "epoch": 12.123730378578024,
      "grad_norm": 0.6788593530654907,
      "learning_rate": 1.9704000000000003e-05,
      "loss": 0.0692,
      "step": 6571
    },
    {
      "epoch": 12.125577100646353,
      "grad_norm": 0.3560887575149536,
      "learning_rate": 1.9707000000000003e-05,
      "loss": 0.0281,
      "step": 6572
    },
    {
      "epoch": 12.12742382271468,
      "grad_norm": 0.5257298946380615,
      "learning_rate": 1.9710000000000003e-05,
      "loss": 0.0747,
      "step": 6573
    },
    {
      "epoch": 12.12927054478301,
      "grad_norm": 0.41419556736946106,
      "learning_rate": 1.9713000000000003e-05,
      "loss": 0.0192,
      "step": 6574
    },
    {
      "epoch": 12.131117266851339,
      "grad_norm": 0.40501436591148376,
      "learning_rate": 1.9716000000000002e-05,
      "loss": 0.0148,
      "step": 6575
    },
    {
      "epoch": 12.132963988919668,
      "grad_norm": 0.5695074796676636,
      "learning_rate": 1.9719e-05,
      "loss": 0.0535,
      "step": 6576
    },
    {
      "epoch": 12.134810710987995,
      "grad_norm": 0.6081413626670837,
      "learning_rate": 1.9722e-05,
      "loss": 0.0284,
      "step": 6577
    },
    {
      "epoch": 12.136657433056325,
      "grad_norm": 0.30088096857070923,
      "learning_rate": 1.9725e-05,
      "loss": 0.0154,
      "step": 6578
    },
    {
      "epoch": 12.138504155124654,
      "grad_norm": 0.37103888392448425,
      "learning_rate": 1.9727999999999998e-05,
      "loss": 0.0171,
      "step": 6579
    },
    {
      "epoch": 12.140350877192983,
      "grad_norm": 0.4621252417564392,
      "learning_rate": 1.9730999999999998e-05,
      "loss": 0.0283,
      "step": 6580
    },
    {
      "epoch": 12.142197599261312,
      "grad_norm": 0.26825007796287537,
      "learning_rate": 1.9734e-05,
      "loss": 0.0201,
      "step": 6581
    },
    {
      "epoch": 12.14404432132964,
      "grad_norm": 0.4614250659942627,
      "learning_rate": 1.9737e-05,
      "loss": 0.0158,
      "step": 6582
    },
    {
      "epoch": 12.145891043397969,
      "grad_norm": 0.3444186747074127,
      "learning_rate": 1.974e-05,
      "loss": 0.0149,
      "step": 6583
    },
    {
      "epoch": 12.147737765466298,
      "grad_norm": 0.34118175506591797,
      "learning_rate": 1.9743e-05,
      "loss": 0.0165,
      "step": 6584
    },
    {
      "epoch": 12.149584487534627,
      "grad_norm": 0.669164776802063,
      "learning_rate": 1.9746e-05,
      "loss": 0.0141,
      "step": 6585
    },
    {
      "epoch": 12.151431209602954,
      "grad_norm": 0.532076358795166,
      "learning_rate": 1.9749e-05,
      "loss": 0.0254,
      "step": 6586
    },
    {
      "epoch": 12.153277931671283,
      "grad_norm": 0.40641987323760986,
      "learning_rate": 1.9752e-05,
      "loss": 0.0146,
      "step": 6587
    },
    {
      "epoch": 12.155124653739612,
      "grad_norm": 0.3384653627872467,
      "learning_rate": 1.9755e-05,
      "loss": 0.014,
      "step": 6588
    },
    {
      "epoch": 12.156971375807942,
      "grad_norm": 0.7094037532806396,
      "learning_rate": 1.9758e-05,
      "loss": 0.0129,
      "step": 6589
    },
    {
      "epoch": 12.158818097876269,
      "grad_norm": 0.683089017868042,
      "learning_rate": 1.9761e-05,
      "loss": 0.0229,
      "step": 6590
    },
    {
      "epoch": 12.160664819944598,
      "grad_norm": 0.5520561337471008,
      "learning_rate": 1.9764000000000003e-05,
      "loss": 0.0253,
      "step": 6591
    },
    {
      "epoch": 12.162511542012927,
      "grad_norm": 0.6962130665779114,
      "learning_rate": 1.9767000000000002e-05,
      "loss": 0.0172,
      "step": 6592
    },
    {
      "epoch": 12.164358264081256,
      "grad_norm": 0.5265188813209534,
      "learning_rate": 1.9770000000000002e-05,
      "loss": 0.0173,
      "step": 6593
    },
    {
      "epoch": 12.166204986149584,
      "grad_norm": 0.9148104190826416,
      "learning_rate": 1.9773000000000002e-05,
      "loss": 0.0247,
      "step": 6594
    },
    {
      "epoch": 12.168051708217913,
      "grad_norm": 0.5359881520271301,
      "learning_rate": 1.9776000000000002e-05,
      "loss": 0.0196,
      "step": 6595
    },
    {
      "epoch": 12.169898430286242,
      "grad_norm": 0.41873955726623535,
      "learning_rate": 1.9779e-05,
      "loss": 0.0125,
      "step": 6596
    },
    {
      "epoch": 12.171745152354571,
      "grad_norm": 0.39210107922554016,
      "learning_rate": 1.9782e-05,
      "loss": 0.0191,
      "step": 6597
    },
    {
      "epoch": 12.173591874422899,
      "grad_norm": 0.6035261750221252,
      "learning_rate": 1.9785e-05,
      "loss": 0.0198,
      "step": 6598
    },
    {
      "epoch": 12.175438596491228,
      "grad_norm": 0.3195289373397827,
      "learning_rate": 1.9788e-05,
      "loss": 0.0159,
      "step": 6599
    },
    {
      "epoch": 12.177285318559557,
      "grad_norm": 0.5464214086532593,
      "learning_rate": 1.9791e-05,
      "loss": 0.0224,
      "step": 6600
    },
    {
      "epoch": 12.179132040627886,
      "grad_norm": 0.7400136590003967,
      "learning_rate": 1.9794e-05,
      "loss": 0.026,
      "step": 6601
    },
    {
      "epoch": 12.180978762696213,
      "grad_norm": 0.496610552072525,
      "learning_rate": 1.9797e-05,
      "loss": 0.0165,
      "step": 6602
    },
    {
      "epoch": 12.182825484764543,
      "grad_norm": 0.47159847617149353,
      "learning_rate": 1.98e-05,
      "loss": 0.0208,
      "step": 6603
    },
    {
      "epoch": 12.184672206832872,
      "grad_norm": 0.48338207602500916,
      "learning_rate": 1.9803e-05,
      "loss": 0.0202,
      "step": 6604
    },
    {
      "epoch": 12.1865189289012,
      "grad_norm": 1.570766568183899,
      "learning_rate": 1.9806e-05,
      "loss": 0.2364,
      "step": 6605
    },
    {
      "epoch": 12.18836565096953,
      "grad_norm": 0.74299556016922,
      "learning_rate": 1.9809e-05,
      "loss": 0.1835,
      "step": 6606
    },
    {
      "epoch": 12.190212373037857,
      "grad_norm": 0.6581249833106995,
      "learning_rate": 1.9812e-05,
      "loss": 0.1466,
      "step": 6607
    },
    {
      "epoch": 12.192059095106186,
      "grad_norm": 0.6629277467727661,
      "learning_rate": 1.9815e-05,
      "loss": 0.1645,
      "step": 6608
    },
    {
      "epoch": 12.193905817174516,
      "grad_norm": 0.5991044640541077,
      "learning_rate": 1.9818e-05,
      "loss": 0.1617,
      "step": 6609
    },
    {
      "epoch": 12.195752539242845,
      "grad_norm": 1.113797664642334,
      "learning_rate": 1.9821e-05,
      "loss": 0.1707,
      "step": 6610
    },
    {
      "epoch": 12.197599261311172,
      "grad_norm": 0.6141255497932434,
      "learning_rate": 1.9824000000000002e-05,
      "loss": 0.0937,
      "step": 6611
    },
    {
      "epoch": 12.199445983379501,
      "grad_norm": 0.6133359670639038,
      "learning_rate": 1.9827000000000002e-05,
      "loss": 0.0931,
      "step": 6612
    },
    {
      "epoch": 12.20129270544783,
      "grad_norm": 0.44140446186065674,
      "learning_rate": 1.983e-05,
      "loss": 0.0858,
      "step": 6613
    },
    {
      "epoch": 12.20313942751616,
      "grad_norm": 0.5552379488945007,
      "learning_rate": 1.9833e-05,
      "loss": 0.0709,
      "step": 6614
    },
    {
      "epoch": 12.204986149584487,
      "grad_norm": 0.42327743768692017,
      "learning_rate": 1.9836e-05,
      "loss": 0.0575,
      "step": 6615
    },
    {
      "epoch": 12.206832871652816,
      "grad_norm": 0.8752537369728088,
      "learning_rate": 1.9839e-05,
      "loss": 0.0878,
      "step": 6616
    },
    {
      "epoch": 12.208679593721145,
      "grad_norm": 0.7336311936378479,
      "learning_rate": 1.9842e-05,
      "loss": 0.1506,
      "step": 6617
    },
    {
      "epoch": 12.210526315789474,
      "grad_norm": 0.6331095695495605,
      "learning_rate": 1.9845e-05,
      "loss": 0.0754,
      "step": 6618
    },
    {
      "epoch": 12.212373037857802,
      "grad_norm": 0.6478675007820129,
      "learning_rate": 1.9848e-05,
      "loss": 0.046,
      "step": 6619
    },
    {
      "epoch": 12.21421975992613,
      "grad_norm": 0.46887344121932983,
      "learning_rate": 1.9851e-05,
      "loss": 0.0586,
      "step": 6620
    },
    {
      "epoch": 12.21606648199446,
      "grad_norm": 0.4673554599285126,
      "learning_rate": 1.9854000000000003e-05,
      "loss": 0.0213,
      "step": 6621
    },
    {
      "epoch": 12.21791320406279,
      "grad_norm": 0.3019261658191681,
      "learning_rate": 1.9857000000000003e-05,
      "loss": 0.0208,
      "step": 6622
    },
    {
      "epoch": 12.219759926131117,
      "grad_norm": 0.55424565076828,
      "learning_rate": 1.9860000000000003e-05,
      "loss": 0.0172,
      "step": 6623
    },
    {
      "epoch": 12.221606648199446,
      "grad_norm": 0.5971589088439941,
      "learning_rate": 1.9863000000000003e-05,
      "loss": 0.0189,
      "step": 6624
    },
    {
      "epoch": 12.223453370267775,
      "grad_norm": 2.0746729373931885,
      "learning_rate": 1.9866e-05,
      "loss": 0.0486,
      "step": 6625
    },
    {
      "epoch": 12.225300092336104,
      "grad_norm": 0.390958309173584,
      "learning_rate": 1.9869e-05,
      "loss": 0.0159,
      "step": 6626
    },
    {
      "epoch": 12.227146814404431,
      "grad_norm": 0.7522118091583252,
      "learning_rate": 1.9872e-05,
      "loss": 0.0211,
      "step": 6627
    },
    {
      "epoch": 12.22899353647276,
      "grad_norm": 0.38090553879737854,
      "learning_rate": 1.9875e-05,
      "loss": 0.0176,
      "step": 6628
    },
    {
      "epoch": 12.23084025854109,
      "grad_norm": 0.8301336169242859,
      "learning_rate": 1.9878e-05,
      "loss": 0.0378,
      "step": 6629
    },
    {
      "epoch": 12.232686980609419,
      "grad_norm": 0.6119270324707031,
      "learning_rate": 1.9880999999999998e-05,
      "loss": 0.0143,
      "step": 6630
    },
    {
      "epoch": 12.234533702677748,
      "grad_norm": 0.5565687417984009,
      "learning_rate": 1.9883999999999998e-05,
      "loss": 0.0337,
      "step": 6631
    },
    {
      "epoch": 12.236380424746075,
      "grad_norm": 0.3857928216457367,
      "learning_rate": 1.9887e-05,
      "loss": 0.0198,
      "step": 6632
    },
    {
      "epoch": 12.238227146814404,
      "grad_norm": 0.39611518383026123,
      "learning_rate": 1.989e-05,
      "loss": 0.0233,
      "step": 6633
    },
    {
      "epoch": 12.240073868882734,
      "grad_norm": 0.4528099596500397,
      "learning_rate": 1.9893e-05,
      "loss": 0.0185,
      "step": 6634
    },
    {
      "epoch": 12.241920590951063,
      "grad_norm": 0.666111946105957,
      "learning_rate": 1.9896e-05,
      "loss": 0.0155,
      "step": 6635
    },
    {
      "epoch": 12.24376731301939,
      "grad_norm": 1.013671636581421,
      "learning_rate": 1.9899e-05,
      "loss": 0.0189,
      "step": 6636
    },
    {
      "epoch": 12.24561403508772,
      "grad_norm": 0.293468177318573,
      "learning_rate": 1.9902e-05,
      "loss": 0.0166,
      "step": 6637
    },
    {
      "epoch": 12.247460757156048,
      "grad_norm": 0.6467782855033875,
      "learning_rate": 1.9905e-05,
      "loss": 0.0114,
      "step": 6638
    },
    {
      "epoch": 12.249307479224377,
      "grad_norm": 0.5225874185562134,
      "learning_rate": 1.9908e-05,
      "loss": 0.0245,
      "step": 6639
    },
    {
      "epoch": 12.251154201292705,
      "grad_norm": 0.6721553206443787,
      "learning_rate": 1.9911e-05,
      "loss": 0.0244,
      "step": 6640
    },
    {
      "epoch": 12.253000923361034,
      "grad_norm": 0.7446433305740356,
      "learning_rate": 1.9914e-05,
      "loss": 0.0272,
      "step": 6641
    },
    {
      "epoch": 12.254847645429363,
      "grad_norm": 0.3481638431549072,
      "learning_rate": 1.9917000000000003e-05,
      "loss": 0.021,
      "step": 6642
    },
    {
      "epoch": 12.256694367497692,
      "grad_norm": 0.5886558890342712,
      "learning_rate": 1.9920000000000002e-05,
      "loss": 0.017,
      "step": 6643
    },
    {
      "epoch": 12.25854108956602,
      "grad_norm": 0.3663330078125,
      "learning_rate": 1.9923000000000002e-05,
      "loss": 0.0147,
      "step": 6644
    },
    {
      "epoch": 12.260387811634349,
      "grad_norm": 0.4481843113899231,
      "learning_rate": 1.9926000000000002e-05,
      "loss": 0.0211,
      "step": 6645
    },
    {
      "epoch": 12.262234533702678,
      "grad_norm": 0.6207283735275269,
      "learning_rate": 1.9929000000000002e-05,
      "loss": 0.0242,
      "step": 6646
    },
    {
      "epoch": 12.264081255771007,
      "grad_norm": 0.5860519409179688,
      "learning_rate": 1.9932e-05,
      "loss": 0.0208,
      "step": 6647
    },
    {
      "epoch": 12.265927977839334,
      "grad_norm": 0.34904736280441284,
      "learning_rate": 1.9935e-05,
      "loss": 0.0233,
      "step": 6648
    },
    {
      "epoch": 12.267774699907664,
      "grad_norm": 0.446857213973999,
      "learning_rate": 1.9938e-05,
      "loss": 0.0171,
      "step": 6649
    },
    {
      "epoch": 12.269621421975993,
      "grad_norm": 0.3671776354312897,
      "learning_rate": 1.9940999999999998e-05,
      "loss": 0.0146,
      "step": 6650
    },
    {
      "epoch": 12.271468144044322,
      "grad_norm": 0.499612033367157,
      "learning_rate": 1.9943999999999997e-05,
      "loss": 0.0178,
      "step": 6651
    },
    {
      "epoch": 12.27331486611265,
      "grad_norm": 0.40178707242012024,
      "learning_rate": 1.9947e-05,
      "loss": 0.0225,
      "step": 6652
    },
    {
      "epoch": 12.275161588180978,
      "grad_norm": 1.6798856258392334,
      "learning_rate": 1.995e-05,
      "loss": 0.0262,
      "step": 6653
    },
    {
      "epoch": 12.277008310249307,
      "grad_norm": 0.5195391774177551,
      "learning_rate": 1.9953e-05,
      "loss": 0.02,
      "step": 6654
    },
    {
      "epoch": 12.278855032317637,
      "grad_norm": 1.4945707321166992,
      "learning_rate": 1.9956e-05,
      "loss": 0.2249,
      "step": 6655
    },
    {
      "epoch": 12.280701754385966,
      "grad_norm": 1.618351936340332,
      "learning_rate": 1.9959e-05,
      "loss": 0.2945,
      "step": 6656
    },
    {
      "epoch": 12.282548476454293,
      "grad_norm": 1.1166245937347412,
      "learning_rate": 1.9962e-05,
      "loss": 0.1938,
      "step": 6657
    },
    {
      "epoch": 12.284395198522622,
      "grad_norm": 1.0007214546203613,
      "learning_rate": 1.9965e-05,
      "loss": 0.1808,
      "step": 6658
    },
    {
      "epoch": 12.286241920590951,
      "grad_norm": 0.8794845938682556,
      "learning_rate": 1.9968e-05,
      "loss": 0.1573,
      "step": 6659
    },
    {
      "epoch": 12.28808864265928,
      "grad_norm": 0.7759224772453308,
      "learning_rate": 1.9971e-05,
      "loss": 0.1279,
      "step": 6660
    },
    {
      "epoch": 12.289935364727608,
      "grad_norm": 0.6087407469749451,
      "learning_rate": 1.9974e-05,
      "loss": 0.1017,
      "step": 6661
    },
    {
      "epoch": 12.291782086795937,
      "grad_norm": 0.4634784162044525,
      "learning_rate": 1.9977000000000002e-05,
      "loss": 0.0812,
      "step": 6662
    },
    {
      "epoch": 12.293628808864266,
      "grad_norm": 0.5533508062362671,
      "learning_rate": 1.9980000000000002e-05,
      "loss": 0.1524,
      "step": 6663
    },
    {
      "epoch": 12.295475530932595,
      "grad_norm": 0.7511826753616333,
      "learning_rate": 1.9983e-05,
      "loss": 0.09,
      "step": 6664
    },
    {
      "epoch": 12.297322253000923,
      "grad_norm": 0.5132535696029663,
      "learning_rate": 1.9986e-05,
      "loss": 0.0795,
      "step": 6665
    },
    {
      "epoch": 12.299168975069252,
      "grad_norm": 0.4026097357273102,
      "learning_rate": 1.9989e-05,
      "loss": 0.0406,
      "step": 6666
    },
    {
      "epoch": 12.301015697137581,
      "grad_norm": 0.5257237553596497,
      "learning_rate": 1.9992e-05,
      "loss": 0.0618,
      "step": 6667
    },
    {
      "epoch": 12.30286241920591,
      "grad_norm": 0.329614520072937,
      "learning_rate": 1.9995e-05,
      "loss": 0.0445,
      "step": 6668
    },
    {
      "epoch": 12.304709141274238,
      "grad_norm": 0.41097500920295715,
      "learning_rate": 1.9998e-05,
      "loss": 0.0325,
      "step": 6669
    },
    {
      "epoch": 12.306555863342567,
      "grad_norm": 1.6208419799804688,
      "learning_rate": 2.0001e-05,
      "loss": 0.0479,
      "step": 6670
    },
    {
      "epoch": 12.308402585410896,
      "grad_norm": 0.6512314677238464,
      "learning_rate": 2.0004e-05,
      "loss": 0.029,
      "step": 6671
    },
    {
      "epoch": 12.310249307479225,
      "grad_norm": 0.41237813234329224,
      "learning_rate": 2.0007000000000003e-05,
      "loss": 0.042,
      "step": 6672
    },
    {
      "epoch": 12.312096029547552,
      "grad_norm": 0.4640461206436157,
      "learning_rate": 2.0010000000000003e-05,
      "loss": 0.024,
      "step": 6673
    },
    {
      "epoch": 12.313942751615881,
      "grad_norm": 0.32338428497314453,
      "learning_rate": 2.0013e-05,
      "loss": 0.0277,
      "step": 6674
    },
    {
      "epoch": 12.31578947368421,
      "grad_norm": 0.561813235282898,
      "learning_rate": 2.0016e-05,
      "loss": 0.0236,
      "step": 6675
    },
    {
      "epoch": 12.31763619575254,
      "grad_norm": 0.48433899879455566,
      "learning_rate": 2.0019e-05,
      "loss": 0.043,
      "step": 6676
    },
    {
      "epoch": 12.319482917820867,
      "grad_norm": 0.35797983407974243,
      "learning_rate": 2.0022e-05,
      "loss": 0.0152,
      "step": 6677
    },
    {
      "epoch": 12.321329639889196,
      "grad_norm": 0.24485398828983307,
      "learning_rate": 2.0025e-05,
      "loss": 0.019,
      "step": 6678
    },
    {
      "epoch": 12.323176361957525,
      "grad_norm": 0.5601616501808167,
      "learning_rate": 2.0028e-05,
      "loss": 0.0288,
      "step": 6679
    },
    {
      "epoch": 12.325023084025855,
      "grad_norm": 0.3700091242790222,
      "learning_rate": 2.0031e-05,
      "loss": 0.0105,
      "step": 6680
    },
    {
      "epoch": 12.326869806094184,
      "grad_norm": 0.9464983344078064,
      "learning_rate": 2.0033999999999998e-05,
      "loss": 0.0188,
      "step": 6681
    },
    {
      "epoch": 12.328716528162511,
      "grad_norm": 1.1551624536514282,
      "learning_rate": 2.0037e-05,
      "loss": 0.0764,
      "step": 6682
    },
    {
      "epoch": 12.33056325023084,
      "grad_norm": 0.6184530258178711,
      "learning_rate": 2.004e-05,
      "loss": 0.0151,
      "step": 6683
    },
    {
      "epoch": 12.33240997229917,
      "grad_norm": 0.33395180106163025,
      "learning_rate": 2.0043e-05,
      "loss": 0.0082,
      "step": 6684
    },
    {
      "epoch": 12.334256694367498,
      "grad_norm": 0.39083316922187805,
      "learning_rate": 2.0046e-05,
      "loss": 0.0223,
      "step": 6685
    },
    {
      "epoch": 12.336103416435826,
      "grad_norm": 0.45690304040908813,
      "learning_rate": 2.0049e-05,
      "loss": 0.026,
      "step": 6686
    },
    {
      "epoch": 12.337950138504155,
      "grad_norm": 0.3044404685497284,
      "learning_rate": 2.0052e-05,
      "loss": 0.0211,
      "step": 6687
    },
    {
      "epoch": 12.339796860572484,
      "grad_norm": 0.7949528694152832,
      "learning_rate": 2.0055e-05,
      "loss": 0.0256,
      "step": 6688
    },
    {
      "epoch": 12.341643582640813,
      "grad_norm": 0.6159335374832153,
      "learning_rate": 2.0058e-05,
      "loss": 0.0214,
      "step": 6689
    },
    {
      "epoch": 12.34349030470914,
      "grad_norm": 0.48883113265037537,
      "learning_rate": 2.0061e-05,
      "loss": 0.0376,
      "step": 6690
    },
    {
      "epoch": 12.34533702677747,
      "grad_norm": 0.36205312609672546,
      "learning_rate": 2.0064e-05,
      "loss": 0.0169,
      "step": 6691
    },
    {
      "epoch": 12.347183748845799,
      "grad_norm": 0.4154253900051117,
      "learning_rate": 2.0067000000000003e-05,
      "loss": 0.0132,
      "step": 6692
    },
    {
      "epoch": 12.349030470914128,
      "grad_norm": 0.32109948992729187,
      "learning_rate": 2.0070000000000003e-05,
      "loss": 0.0204,
      "step": 6693
    },
    {
      "epoch": 12.350877192982455,
      "grad_norm": 0.6490615010261536,
      "learning_rate": 2.0073000000000002e-05,
      "loss": 0.028,
      "step": 6694
    },
    {
      "epoch": 12.352723915050785,
      "grad_norm": 0.5937842130661011,
      "learning_rate": 2.0076000000000002e-05,
      "loss": 0.0215,
      "step": 6695
    },
    {
      "epoch": 12.354570637119114,
      "grad_norm": 0.41471704840660095,
      "learning_rate": 2.0079000000000002e-05,
      "loss": 0.016,
      "step": 6696
    },
    {
      "epoch": 12.356417359187443,
      "grad_norm": 0.4338700473308563,
      "learning_rate": 2.0082000000000002e-05,
      "loss": 0.0115,
      "step": 6697
    },
    {
      "epoch": 12.35826408125577,
      "grad_norm": 0.6470513939857483,
      "learning_rate": 2.0085e-05,
      "loss": 0.0306,
      "step": 6698
    },
    {
      "epoch": 12.3601108033241,
      "grad_norm": 0.6698824763298035,
      "learning_rate": 2.0087999999999998e-05,
      "loss": 0.0135,
      "step": 6699
    },
    {
      "epoch": 12.361957525392429,
      "grad_norm": 0.37007206678390503,
      "learning_rate": 2.0090999999999998e-05,
      "loss": 0.0146,
      "step": 6700
    },
    {
      "epoch": 12.363804247460758,
      "grad_norm": 0.4813641905784607,
      "learning_rate": 2.0093999999999998e-05,
      "loss": 0.0202,
      "step": 6701
    },
    {
      "epoch": 12.365650969529085,
      "grad_norm": 1.0598350763320923,
      "learning_rate": 2.0097e-05,
      "loss": 0.0451,
      "step": 6702
    },
    {
      "epoch": 12.367497691597414,
      "grad_norm": 0.682828426361084,
      "learning_rate": 2.01e-05,
      "loss": 0.0215,
      "step": 6703
    },
    {
      "epoch": 12.369344413665743,
      "grad_norm": 0.82156902551651,
      "learning_rate": 2.0103e-05,
      "loss": 0.0499,
      "step": 6704
    },
    {
      "epoch": 12.371191135734072,
      "grad_norm": 0.9906166195869446,
      "learning_rate": 2.0106e-05,
      "loss": 0.213,
      "step": 6705
    },
    {
      "epoch": 12.373037857802402,
      "grad_norm": 0.7346987128257751,
      "learning_rate": 2.0109e-05,
      "loss": 0.2196,
      "step": 6706
    },
    {
      "epoch": 12.374884579870729,
      "grad_norm": 0.6778160929679871,
      "learning_rate": 2.0112e-05,
      "loss": 0.1236,
      "step": 6707
    },
    {
      "epoch": 12.376731301939058,
      "grad_norm": 0.6001963019371033,
      "learning_rate": 2.0115e-05,
      "loss": 0.1193,
      "step": 6708
    },
    {
      "epoch": 12.378578024007387,
      "grad_norm": 0.7126210331916809,
      "learning_rate": 2.0118e-05,
      "loss": 0.1331,
      "step": 6709
    },
    {
      "epoch": 12.380424746075716,
      "grad_norm": 0.580418586730957,
      "learning_rate": 2.0121e-05,
      "loss": 0.1309,
      "step": 6710
    },
    {
      "epoch": 12.382271468144044,
      "grad_norm": 0.837908923625946,
      "learning_rate": 2.0124e-05,
      "loss": 0.1743,
      "step": 6711
    },
    {
      "epoch": 12.384118190212373,
      "grad_norm": 0.47805464267730713,
      "learning_rate": 2.0127000000000002e-05,
      "loss": 0.0906,
      "step": 6712
    },
    {
      "epoch": 12.385964912280702,
      "grad_norm": 0.5674648880958557,
      "learning_rate": 2.0130000000000002e-05,
      "loss": 0.0921,
      "step": 6713
    },
    {
      "epoch": 12.387811634349031,
      "grad_norm": 0.4743923246860504,
      "learning_rate": 2.0133000000000002e-05,
      "loss": 0.0776,
      "step": 6714
    },
    {
      "epoch": 12.389658356417359,
      "grad_norm": 0.807295024394989,
      "learning_rate": 2.0136e-05,
      "loss": 0.0961,
      "step": 6715
    },
    {
      "epoch": 12.391505078485688,
      "grad_norm": 0.6291029453277588,
      "learning_rate": 2.0139e-05,
      "loss": 0.1024,
      "step": 6716
    },
    {
      "epoch": 12.393351800554017,
      "grad_norm": 0.6043853759765625,
      "learning_rate": 2.0142e-05,
      "loss": 0.0667,
      "step": 6717
    },
    {
      "epoch": 12.395198522622346,
      "grad_norm": 0.8836755752563477,
      "learning_rate": 2.0145e-05,
      "loss": 0.0419,
      "step": 6718
    },
    {
      "epoch": 12.397045244690673,
      "grad_norm": 0.7132802605628967,
      "learning_rate": 2.0148e-05,
      "loss": 0.0285,
      "step": 6719
    },
    {
      "epoch": 12.398891966759003,
      "grad_norm": 0.3178471028804779,
      "learning_rate": 2.0151e-05,
      "loss": 0.0358,
      "step": 6720
    },
    {
      "epoch": 12.400738688827332,
      "grad_norm": 0.4992874264717102,
      "learning_rate": 2.0154e-05,
      "loss": 0.036,
      "step": 6721
    },
    {
      "epoch": 12.40258541089566,
      "grad_norm": 0.38336145877838135,
      "learning_rate": 2.0157000000000004e-05,
      "loss": 0.0297,
      "step": 6722
    },
    {
      "epoch": 12.404432132963988,
      "grad_norm": 0.37750759720802307,
      "learning_rate": 2.016e-05,
      "loss": 0.0243,
      "step": 6723
    },
    {
      "epoch": 12.406278855032317,
      "grad_norm": 0.3396998941898346,
      "learning_rate": 2.0163e-05,
      "loss": 0.02,
      "step": 6724
    },
    {
      "epoch": 12.408125577100646,
      "grad_norm": 0.5006681680679321,
      "learning_rate": 2.0166e-05,
      "loss": 0.0113,
      "step": 6725
    },
    {
      "epoch": 12.409972299168976,
      "grad_norm": 0.29943355917930603,
      "learning_rate": 2.0169e-05,
      "loss": 0.013,
      "step": 6726
    },
    {
      "epoch": 12.411819021237303,
      "grad_norm": 0.2731862962245941,
      "learning_rate": 2.0172e-05,
      "loss": 0.0179,
      "step": 6727
    },
    {
      "epoch": 12.413665743305632,
      "grad_norm": 0.4055343270301819,
      "learning_rate": 2.0175e-05,
      "loss": 0.0269,
      "step": 6728
    },
    {
      "epoch": 12.415512465373961,
      "grad_norm": 0.4607265889644623,
      "learning_rate": 2.0178e-05,
      "loss": 0.0172,
      "step": 6729
    },
    {
      "epoch": 12.41735918744229,
      "grad_norm": 0.17374105751514435,
      "learning_rate": 2.0181e-05,
      "loss": 0.0122,
      "step": 6730
    },
    {
      "epoch": 12.41920590951062,
      "grad_norm": 0.30624425411224365,
      "learning_rate": 2.0184e-05,
      "loss": 0.0162,
      "step": 6731
    },
    {
      "epoch": 12.421052631578947,
      "grad_norm": 0.30509719252586365,
      "learning_rate": 2.0187000000000002e-05,
      "loss": 0.0176,
      "step": 6732
    },
    {
      "epoch": 12.422899353647276,
      "grad_norm": 0.6953483819961548,
      "learning_rate": 2.019e-05,
      "loss": 0.0228,
      "step": 6733
    },
    {
      "epoch": 12.424746075715605,
      "grad_norm": 0.5839991569519043,
      "learning_rate": 2.0193e-05,
      "loss": 0.0239,
      "step": 6734
    },
    {
      "epoch": 12.426592797783934,
      "grad_norm": 0.2980464994907379,
      "learning_rate": 2.0196e-05,
      "loss": 0.0231,
      "step": 6735
    },
    {
      "epoch": 12.428439519852262,
      "grad_norm": 0.4845972955226898,
      "learning_rate": 2.0199e-05,
      "loss": 0.0149,
      "step": 6736
    },
    {
      "epoch": 12.43028624192059,
      "grad_norm": 0.5114170908927917,
      "learning_rate": 2.0202e-05,
      "loss": 0.0225,
      "step": 6737
    },
    {
      "epoch": 12.43213296398892,
      "grad_norm": 0.4214576780796051,
      "learning_rate": 2.0205e-05,
      "loss": 0.0167,
      "step": 6738
    },
    {
      "epoch": 12.43397968605725,
      "grad_norm": 0.5541107058525085,
      "learning_rate": 2.0208e-05,
      "loss": 0.0245,
      "step": 6739
    },
    {
      "epoch": 12.435826408125576,
      "grad_norm": 0.40229126811027527,
      "learning_rate": 2.0211e-05,
      "loss": 0.0147,
      "step": 6740
    },
    {
      "epoch": 12.437673130193906,
      "grad_norm": 0.7117041349411011,
      "learning_rate": 2.0214e-05,
      "loss": 0.0153,
      "step": 6741
    },
    {
      "epoch": 12.439519852262235,
      "grad_norm": 0.4517172574996948,
      "learning_rate": 2.0217000000000003e-05,
      "loss": 0.0233,
      "step": 6742
    },
    {
      "epoch": 12.441366574330564,
      "grad_norm": 0.6834757924079895,
      "learning_rate": 2.0220000000000003e-05,
      "loss": 0.0289,
      "step": 6743
    },
    {
      "epoch": 12.443213296398891,
      "grad_norm": 0.5634806156158447,
      "learning_rate": 2.0223000000000003e-05,
      "loss": 0.0321,
      "step": 6744
    },
    {
      "epoch": 12.44506001846722,
      "grad_norm": 0.42304977774620056,
      "learning_rate": 2.0226000000000003e-05,
      "loss": 0.0151,
      "step": 6745
    },
    {
      "epoch": 12.44690674053555,
      "grad_norm": 0.3649885654449463,
      "learning_rate": 2.0229000000000002e-05,
      "loss": 0.0168,
      "step": 6746
    },
    {
      "epoch": 12.448753462603879,
      "grad_norm": 0.44206562638282776,
      "learning_rate": 2.0232000000000002e-05,
      "loss": 0.0143,
      "step": 6747
    },
    {
      "epoch": 12.450600184672206,
      "grad_norm": 0.7201863527297974,
      "learning_rate": 2.0235e-05,
      "loss": 0.032,
      "step": 6748
    },
    {
      "epoch": 12.452446906740535,
      "grad_norm": 0.5601069927215576,
      "learning_rate": 2.0238e-05,
      "loss": 0.0189,
      "step": 6749
    },
    {
      "epoch": 12.454293628808864,
      "grad_norm": 0.4897482395172119,
      "learning_rate": 2.0240999999999998e-05,
      "loss": 0.0178,
      "step": 6750
    },
    {
      "epoch": 12.456140350877194,
      "grad_norm": 0.6514797210693359,
      "learning_rate": 2.0243999999999998e-05,
      "loss": 0.026,
      "step": 6751
    },
    {
      "epoch": 12.45798707294552,
      "grad_norm": 0.6403403282165527,
      "learning_rate": 2.0247e-05,
      "loss": 0.0298,
      "step": 6752
    },
    {
      "epoch": 12.45983379501385,
      "grad_norm": 0.4038439691066742,
      "learning_rate": 2.025e-05,
      "loss": 0.0197,
      "step": 6753
    },
    {
      "epoch": 12.46168051708218,
      "grad_norm": 0.5625746846199036,
      "learning_rate": 2.0253e-05,
      "loss": 0.0279,
      "step": 6754
    },
    {
      "epoch": 12.463527239150508,
      "grad_norm": 1.6312752962112427,
      "learning_rate": 2.0256e-05,
      "loss": 0.3294,
      "step": 6755
    },
    {
      "epoch": 12.465373961218837,
      "grad_norm": 0.7647871375083923,
      "learning_rate": 2.0259e-05,
      "loss": 0.1652,
      "step": 6756
    },
    {
      "epoch": 12.467220683287165,
      "grad_norm": 0.8227677941322327,
      "learning_rate": 2.0262e-05,
      "loss": 0.1735,
      "step": 6757
    },
    {
      "epoch": 12.469067405355494,
      "grad_norm": 0.6325469017028809,
      "learning_rate": 2.0265e-05,
      "loss": 0.1444,
      "step": 6758
    },
    {
      "epoch": 12.470914127423823,
      "grad_norm": 0.7547577619552612,
      "learning_rate": 2.0268e-05,
      "loss": 0.1389,
      "step": 6759
    },
    {
      "epoch": 12.472760849492152,
      "grad_norm": 0.5040287971496582,
      "learning_rate": 2.0271e-05,
      "loss": 0.084,
      "step": 6760
    },
    {
      "epoch": 12.47460757156048,
      "grad_norm": 0.5050820708274841,
      "learning_rate": 2.0274e-05,
      "loss": 0.1005,
      "step": 6761
    },
    {
      "epoch": 12.476454293628809,
      "grad_norm": 0.8370978832244873,
      "learning_rate": 2.0277e-05,
      "loss": 0.101,
      "step": 6762
    },
    {
      "epoch": 12.478301015697138,
      "grad_norm": 0.5290765762329102,
      "learning_rate": 2.0280000000000002e-05,
      "loss": 0.0826,
      "step": 6763
    },
    {
      "epoch": 12.480147737765467,
      "grad_norm": 0.529080331325531,
      "learning_rate": 2.0283000000000002e-05,
      "loss": 0.0676,
      "step": 6764
    },
    {
      "epoch": 12.481994459833794,
      "grad_norm": 0.6170281171798706,
      "learning_rate": 2.0286000000000002e-05,
      "loss": 0.1273,
      "step": 6765
    },
    {
      "epoch": 12.483841181902124,
      "grad_norm": 0.574787974357605,
      "learning_rate": 2.0289000000000002e-05,
      "loss": 0.0825,
      "step": 6766
    },
    {
      "epoch": 12.485687903970453,
      "grad_norm": 0.6429061889648438,
      "learning_rate": 2.0292e-05,
      "loss": 0.0628,
      "step": 6767
    },
    {
      "epoch": 12.487534626038782,
      "grad_norm": 0.4455440938472748,
      "learning_rate": 2.0295e-05,
      "loss": 0.0438,
      "step": 6768
    },
    {
      "epoch": 12.48938134810711,
      "grad_norm": 0.45541685819625854,
      "learning_rate": 2.0298e-05,
      "loss": 0.0672,
      "step": 6769
    },
    {
      "epoch": 12.491228070175438,
      "grad_norm": 0.4998338222503662,
      "learning_rate": 2.0301e-05,
      "loss": 0.0338,
      "step": 6770
    },
    {
      "epoch": 12.493074792243767,
      "grad_norm": 0.4116666913032532,
      "learning_rate": 2.0304e-05,
      "loss": 0.0408,
      "step": 6771
    },
    {
      "epoch": 12.494921514312097,
      "grad_norm": 0.3885330557823181,
      "learning_rate": 2.0307e-05,
      "loss": 0.0224,
      "step": 6772
    },
    {
      "epoch": 12.496768236380424,
      "grad_norm": 0.3494364619255066,
      "learning_rate": 2.031e-05,
      "loss": 0.0545,
      "step": 6773
    },
    {
      "epoch": 12.498614958448753,
      "grad_norm": 0.2879297733306885,
      "learning_rate": 2.0313e-05,
      "loss": 0.0503,
      "step": 6774
    },
    {
      "epoch": 12.500461680517082,
      "grad_norm": 0.3762493133544922,
      "learning_rate": 2.0316e-05,
      "loss": 0.0205,
      "step": 6775
    },
    {
      "epoch": 12.502308402585411,
      "grad_norm": 0.25101399421691895,
      "learning_rate": 2.0319e-05,
      "loss": 0.0166,
      "step": 6776
    },
    {
      "epoch": 12.504155124653739,
      "grad_norm": 0.49378249049186707,
      "learning_rate": 2.0322e-05,
      "loss": 0.0296,
      "step": 6777
    },
    {
      "epoch": 12.506001846722068,
      "grad_norm": 0.6454524993896484,
      "learning_rate": 2.0325e-05,
      "loss": 0.0301,
      "step": 6778
    },
    {
      "epoch": 12.507848568790397,
      "grad_norm": 0.372597336769104,
      "learning_rate": 2.0328e-05,
      "loss": 0.023,
      "step": 6779
    },
    {
      "epoch": 12.509695290858726,
      "grad_norm": 0.47888875007629395,
      "learning_rate": 2.0331e-05,
      "loss": 0.0267,
      "step": 6780
    },
    {
      "epoch": 12.511542012927055,
      "grad_norm": 0.2966492772102356,
      "learning_rate": 2.0334e-05,
      "loss": 0.0163,
      "step": 6781
    },
    {
      "epoch": 12.513388734995383,
      "grad_norm": 0.7563128471374512,
      "learning_rate": 2.0337e-05,
      "loss": 0.0213,
      "step": 6782
    },
    {
      "epoch": 12.515235457063712,
      "grad_norm": 0.3697977662086487,
      "learning_rate": 2.0340000000000002e-05,
      "loss": 0.0184,
      "step": 6783
    },
    {
      "epoch": 12.517082179132041,
      "grad_norm": 0.5888333916664124,
      "learning_rate": 2.0343e-05,
      "loss": 0.0147,
      "step": 6784
    },
    {
      "epoch": 12.51892890120037,
      "grad_norm": 0.4508650004863739,
      "learning_rate": 2.0346e-05,
      "loss": 0.0167,
      "step": 6785
    },
    {
      "epoch": 12.520775623268698,
      "grad_norm": 0.3996213972568512,
      "learning_rate": 2.0349e-05,
      "loss": 0.0117,
      "step": 6786
    },
    {
      "epoch": 12.522622345337027,
      "grad_norm": 0.3682596981525421,
      "learning_rate": 2.0352e-05,
      "loss": 0.0102,
      "step": 6787
    },
    {
      "epoch": 12.524469067405356,
      "grad_norm": 0.3281971216201782,
      "learning_rate": 2.0355e-05,
      "loss": 0.0099,
      "step": 6788
    },
    {
      "epoch": 12.526315789473685,
      "grad_norm": 0.6341251730918884,
      "learning_rate": 2.0358e-05,
      "loss": 0.0175,
      "step": 6789
    },
    {
      "epoch": 12.528162511542012,
      "grad_norm": 0.8982787728309631,
      "learning_rate": 2.0361e-05,
      "loss": 0.0379,
      "step": 6790
    },
    {
      "epoch": 12.530009233610341,
      "grad_norm": 0.629838228225708,
      "learning_rate": 2.0364e-05,
      "loss": 0.0259,
      "step": 6791
    },
    {
      "epoch": 12.53185595567867,
      "grad_norm": 0.4660007655620575,
      "learning_rate": 2.0367e-05,
      "loss": 0.0206,
      "step": 6792
    },
    {
      "epoch": 12.533702677747,
      "grad_norm": 0.7335576415061951,
      "learning_rate": 2.0370000000000003e-05,
      "loss": 0.0227,
      "step": 6793
    },
    {
      "epoch": 12.535549399815327,
      "grad_norm": 0.5546920299530029,
      "learning_rate": 2.0373000000000003e-05,
      "loss": 0.0224,
      "step": 6794
    },
    {
      "epoch": 12.537396121883656,
      "grad_norm": 0.5457835793495178,
      "learning_rate": 2.0376000000000003e-05,
      "loss": 0.0274,
      "step": 6795
    },
    {
      "epoch": 12.539242843951985,
      "grad_norm": 1.0218298435211182,
      "learning_rate": 2.0379000000000003e-05,
      "loss": 0.0188,
      "step": 6796
    },
    {
      "epoch": 12.541089566020315,
      "grad_norm": 0.6698922514915466,
      "learning_rate": 2.0382e-05,
      "loss": 0.0175,
      "step": 6797
    },
    {
      "epoch": 12.542936288088642,
      "grad_norm": 0.38196083903312683,
      "learning_rate": 2.0385e-05,
      "loss": 0.0291,
      "step": 6798
    },
    {
      "epoch": 12.544783010156971,
      "grad_norm": 0.3500259816646576,
      "learning_rate": 2.0388e-05,
      "loss": 0.0162,
      "step": 6799
    },
    {
      "epoch": 12.5466297322253,
      "grad_norm": 0.7685902714729309,
      "learning_rate": 2.0391e-05,
      "loss": 0.0223,
      "step": 6800
    },
    {
      "epoch": 12.54847645429363,
      "grad_norm": 0.901985764503479,
      "learning_rate": 2.0393999999999998e-05,
      "loss": 0.0397,
      "step": 6801
    },
    {
      "epoch": 12.550323176361957,
      "grad_norm": 0.3865233063697815,
      "learning_rate": 2.0396999999999998e-05,
      "loss": 0.0209,
      "step": 6802
    },
    {
      "epoch": 12.552169898430286,
      "grad_norm": 0.5133811831474304,
      "learning_rate": 2.04e-05,
      "loss": 0.0193,
      "step": 6803
    },
    {
      "epoch": 12.554016620498615,
      "grad_norm": 0.9574827551841736,
      "learning_rate": 2.0403e-05,
      "loss": 0.0598,
      "step": 6804
    },
    {
      "epoch": 12.555863342566944,
      "grad_norm": 1.2635304927825928,
      "learning_rate": 2.0406e-05,
      "loss": 0.2437,
      "step": 6805
    },
    {
      "epoch": 12.557710064635273,
      "grad_norm": 0.9194162487983704,
      "learning_rate": 2.0409e-05,
      "loss": 0.235,
      "step": 6806
    },
    {
      "epoch": 12.5595567867036,
      "grad_norm": 0.9686319231987,
      "learning_rate": 2.0412e-05,
      "loss": 0.149,
      "step": 6807
    },
    {
      "epoch": 12.56140350877193,
      "grad_norm": 0.5710583925247192,
      "learning_rate": 2.0415e-05,
      "loss": 0.1096,
      "step": 6808
    },
    {
      "epoch": 12.563250230840259,
      "grad_norm": 0.924613893032074,
      "learning_rate": 2.0418e-05,
      "loss": 0.1968,
      "step": 6809
    },
    {
      "epoch": 12.565096952908588,
      "grad_norm": 0.6345779299736023,
      "learning_rate": 2.0421e-05,
      "loss": 0.1546,
      "step": 6810
    },
    {
      "epoch": 12.566943674976915,
      "grad_norm": 0.5243431329727173,
      "learning_rate": 2.0424e-05,
      "loss": 0.0736,
      "step": 6811
    },
    {
      "epoch": 12.568790397045245,
      "grad_norm": 1.152990698814392,
      "learning_rate": 2.0427e-05,
      "loss": 0.1234,
      "step": 6812
    },
    {
      "epoch": 12.570637119113574,
      "grad_norm": 0.544137179851532,
      "learning_rate": 2.0430000000000003e-05,
      "loss": 0.0915,
      "step": 6813
    },
    {
      "epoch": 12.572483841181903,
      "grad_norm": 0.5432739853858948,
      "learning_rate": 2.0433000000000002e-05,
      "loss": 0.0802,
      "step": 6814
    },
    {
      "epoch": 12.57433056325023,
      "grad_norm": 0.4892103970050812,
      "learning_rate": 2.0436000000000002e-05,
      "loss": 0.1198,
      "step": 6815
    },
    {
      "epoch": 12.57617728531856,
      "grad_norm": 0.639560878276825,
      "learning_rate": 2.0439000000000002e-05,
      "loss": 0.0567,
      "step": 6816
    },
    {
      "epoch": 12.578024007386889,
      "grad_norm": 0.40163639187812805,
      "learning_rate": 2.0442000000000002e-05,
      "loss": 0.0621,
      "step": 6817
    },
    {
      "epoch": 12.579870729455218,
      "grad_norm": 0.42498865723609924,
      "learning_rate": 2.0445e-05,
      "loss": 0.0504,
      "step": 6818
    },
    {
      "epoch": 12.581717451523545,
      "grad_norm": 0.5056062936782837,
      "learning_rate": 2.0448e-05,
      "loss": 0.0545,
      "step": 6819
    },
    {
      "epoch": 12.583564173591874,
      "grad_norm": 0.46628761291503906,
      "learning_rate": 2.0451e-05,
      "loss": 0.0268,
      "step": 6820
    },
    {
      "epoch": 12.585410895660203,
      "grad_norm": 0.3696845769882202,
      "learning_rate": 2.0454e-05,
      "loss": 0.0186,
      "step": 6821
    },
    {
      "epoch": 12.587257617728532,
      "grad_norm": 0.3414132297039032,
      "learning_rate": 2.0456999999999997e-05,
      "loss": 0.0394,
      "step": 6822
    },
    {
      "epoch": 12.58910433979686,
      "grad_norm": 0.35303443670272827,
      "learning_rate": 2.046e-05,
      "loss": 0.0295,
      "step": 6823
    },
    {
      "epoch": 12.590951061865189,
      "grad_norm": 0.5817875266075134,
      "learning_rate": 2.0463e-05,
      "loss": 0.0226,
      "step": 6824
    },
    {
      "epoch": 12.592797783933518,
      "grad_norm": 0.33950066566467285,
      "learning_rate": 2.0466e-05,
      "loss": 0.0222,
      "step": 6825
    },
    {
      "epoch": 12.594644506001847,
      "grad_norm": 0.43999528884887695,
      "learning_rate": 2.0469e-05,
      "loss": 0.0229,
      "step": 6826
    },
    {
      "epoch": 12.596491228070175,
      "grad_norm": 0.3061814308166504,
      "learning_rate": 2.0472e-05,
      "loss": 0.0181,
      "step": 6827
    },
    {
      "epoch": 12.598337950138504,
      "grad_norm": 0.351764053106308,
      "learning_rate": 2.0475e-05,
      "loss": 0.0159,
      "step": 6828
    },
    {
      "epoch": 12.600184672206833,
      "grad_norm": 0.2815607488155365,
      "learning_rate": 2.0478e-05,
      "loss": 0.0184,
      "step": 6829
    },
    {
      "epoch": 12.602031394275162,
      "grad_norm": 0.5808914303779602,
      "learning_rate": 2.0481e-05,
      "loss": 0.036,
      "step": 6830
    },
    {
      "epoch": 12.603878116343491,
      "grad_norm": 0.41375821828842163,
      "learning_rate": 2.0484e-05,
      "loss": 0.0205,
      "step": 6831
    },
    {
      "epoch": 12.605724838411819,
      "grad_norm": 0.6034269332885742,
      "learning_rate": 2.0487e-05,
      "loss": 0.0296,
      "step": 6832
    },
    {
      "epoch": 12.607571560480148,
      "grad_norm": 0.5789977312088013,
      "learning_rate": 2.0490000000000002e-05,
      "loss": 0.0231,
      "step": 6833
    },
    {
      "epoch": 12.609418282548477,
      "grad_norm": 0.4856737554073334,
      "learning_rate": 2.0493000000000002e-05,
      "loss": 0.0156,
      "step": 6834
    },
    {
      "epoch": 12.611265004616806,
      "grad_norm": 0.5737854242324829,
      "learning_rate": 2.0496e-05,
      "loss": 0.0147,
      "step": 6835
    },
    {
      "epoch": 12.613111726685133,
      "grad_norm": 0.44736528396606445,
      "learning_rate": 2.0499e-05,
      "loss": 0.013,
      "step": 6836
    },
    {
      "epoch": 12.614958448753463,
      "grad_norm": 0.6260068416595459,
      "learning_rate": 2.0502e-05,
      "loss": 0.0262,
      "step": 6837
    },
    {
      "epoch": 12.616805170821792,
      "grad_norm": 0.31467726826667786,
      "learning_rate": 2.0505e-05,
      "loss": 0.0112,
      "step": 6838
    },
    {
      "epoch": 12.61865189289012,
      "grad_norm": 0.3031906485557556,
      "learning_rate": 2.0508e-05,
      "loss": 0.0175,
      "step": 6839
    },
    {
      "epoch": 12.620498614958448,
      "grad_norm": 0.5624115467071533,
      "learning_rate": 2.0511e-05,
      "loss": 0.0116,
      "step": 6840
    },
    {
      "epoch": 12.622345337026777,
      "grad_norm": 0.6537050008773804,
      "learning_rate": 2.0514e-05,
      "loss": 0.015,
      "step": 6841
    },
    {
      "epoch": 12.624192059095106,
      "grad_norm": 0.8279240727424622,
      "learning_rate": 2.0517e-05,
      "loss": 0.0222,
      "step": 6842
    },
    {
      "epoch": 12.626038781163436,
      "grad_norm": 0.6681758761405945,
      "learning_rate": 2.0520000000000003e-05,
      "loss": 0.0225,
      "step": 6843
    },
    {
      "epoch": 12.627885503231763,
      "grad_norm": 0.6742988228797913,
      "learning_rate": 2.0523000000000003e-05,
      "loss": 0.0287,
      "step": 6844
    },
    {
      "epoch": 12.629732225300092,
      "grad_norm": 0.3357849717140198,
      "learning_rate": 2.0526000000000003e-05,
      "loss": 0.0173,
      "step": 6845
    },
    {
      "epoch": 12.631578947368421,
      "grad_norm": 0.5499638915061951,
      "learning_rate": 2.0529e-05,
      "loss": 0.0165,
      "step": 6846
    },
    {
      "epoch": 12.63342566943675,
      "grad_norm": 0.4188133776187897,
      "learning_rate": 2.0532e-05,
      "loss": 0.0148,
      "step": 6847
    },
    {
      "epoch": 12.635272391505078,
      "grad_norm": 0.7594406008720398,
      "learning_rate": 2.0535e-05,
      "loss": 0.0307,
      "step": 6848
    },
    {
      "epoch": 12.637119113573407,
      "grad_norm": 0.7319937348365784,
      "learning_rate": 2.0538e-05,
      "loss": 0.0342,
      "step": 6849
    },
    {
      "epoch": 12.638965835641736,
      "grad_norm": 0.6182621717453003,
      "learning_rate": 2.0541e-05,
      "loss": 0.0176,
      "step": 6850
    },
    {
      "epoch": 12.640812557710065,
      "grad_norm": 0.4317701756954193,
      "learning_rate": 2.0544e-05,
      "loss": 0.0197,
      "step": 6851
    },
    {
      "epoch": 12.642659279778393,
      "grad_norm": 0.9269976615905762,
      "learning_rate": 2.0546999999999998e-05,
      "loss": 0.0304,
      "step": 6852
    },
    {
      "epoch": 12.644506001846722,
      "grad_norm": 0.6142908930778503,
      "learning_rate": 2.055e-05,
      "loss": 0.0223,
      "step": 6853
    },
    {
      "epoch": 12.64635272391505,
      "grad_norm": 1.2634549140930176,
      "learning_rate": 2.0553e-05,
      "loss": 0.0243,
      "step": 6854
    },
    {
      "epoch": 12.64819944598338,
      "grad_norm": 0.7594727873802185,
      "learning_rate": 2.0556e-05,
      "loss": 0.2272,
      "step": 6855
    },
    {
      "epoch": 12.65004616805171,
      "grad_norm": 0.7709804177284241,
      "learning_rate": 2.0559e-05,
      "loss": 0.2132,
      "step": 6856
    },
    {
      "epoch": 12.651892890120036,
      "grad_norm": 0.6140466928482056,
      "learning_rate": 2.0562e-05,
      "loss": 0.1399,
      "step": 6857
    },
    {
      "epoch": 12.653739612188366,
      "grad_norm": 0.6246731877326965,
      "learning_rate": 2.0565e-05,
      "loss": 0.1812,
      "step": 6858
    },
    {
      "epoch": 12.655586334256695,
      "grad_norm": 1.0530415773391724,
      "learning_rate": 2.0568e-05,
      "loss": 0.1882,
      "step": 6859
    },
    {
      "epoch": 12.657433056325024,
      "grad_norm": 0.48276087641716003,
      "learning_rate": 2.0571e-05,
      "loss": 0.0938,
      "step": 6860
    },
    {
      "epoch": 12.659279778393351,
      "grad_norm": 0.5793479681015015,
      "learning_rate": 2.0574e-05,
      "loss": 0.1123,
      "step": 6861
    },
    {
      "epoch": 12.66112650046168,
      "grad_norm": 0.8910770416259766,
      "learning_rate": 2.0577e-05,
      "loss": 0.1138,
      "step": 6862
    },
    {
      "epoch": 12.66297322253001,
      "grad_norm": 0.6055312752723694,
      "learning_rate": 2.0580000000000003e-05,
      "loss": 0.0824,
      "step": 6863
    },
    {
      "epoch": 12.664819944598339,
      "grad_norm": 0.6477535963058472,
      "learning_rate": 2.0583000000000003e-05,
      "loss": 0.1224,
      "step": 6864
    },
    {
      "epoch": 12.666666666666666,
      "grad_norm": 0.44331997632980347,
      "learning_rate": 2.0586000000000002e-05,
      "loss": 0.0543,
      "step": 6865
    },
    {
      "epoch": 12.668513388734995,
      "grad_norm": 0.6174275875091553,
      "learning_rate": 2.0589000000000002e-05,
      "loss": 0.0554,
      "step": 6866
    },
    {
      "epoch": 12.670360110803324,
      "grad_norm": 0.5855109691619873,
      "learning_rate": 2.0592000000000002e-05,
      "loss": 0.0414,
      "step": 6867
    },
    {
      "epoch": 12.672206832871654,
      "grad_norm": 0.666130542755127,
      "learning_rate": 2.0595000000000002e-05,
      "loss": 0.0647,
      "step": 6868
    },
    {
      "epoch": 12.67405355493998,
      "grad_norm": 0.5880280137062073,
      "learning_rate": 2.0598e-05,
      "loss": 0.0468,
      "step": 6869
    },
    {
      "epoch": 12.67590027700831,
      "grad_norm": 0.3871926963329315,
      "learning_rate": 2.0601e-05,
      "loss": 0.0756,
      "step": 6870
    },
    {
      "epoch": 12.67774699907664,
      "grad_norm": 0.45748671889305115,
      "learning_rate": 2.0603999999999998e-05,
      "loss": 0.0349,
      "step": 6871
    },
    {
      "epoch": 12.679593721144968,
      "grad_norm": 0.37007564306259155,
      "learning_rate": 2.0606999999999998e-05,
      "loss": 0.0257,
      "step": 6872
    },
    {
      "epoch": 12.681440443213296,
      "grad_norm": 0.5523969531059265,
      "learning_rate": 2.061e-05,
      "loss": 0.0196,
      "step": 6873
    },
    {
      "epoch": 12.683287165281625,
      "grad_norm": 0.5408273339271545,
      "learning_rate": 2.0613e-05,
      "loss": 0.0339,
      "step": 6874
    },
    {
      "epoch": 12.685133887349954,
      "grad_norm": 0.42680907249450684,
      "learning_rate": 2.0616e-05,
      "loss": 0.0147,
      "step": 6875
    },
    {
      "epoch": 12.686980609418283,
      "grad_norm": 0.4050646424293518,
      "learning_rate": 2.0619e-05,
      "loss": 0.0205,
      "step": 6876
    },
    {
      "epoch": 12.68882733148661,
      "grad_norm": 0.3192436695098877,
      "learning_rate": 2.0622e-05,
      "loss": 0.0152,
      "step": 6877
    },
    {
      "epoch": 12.69067405355494,
      "grad_norm": 0.49936342239379883,
      "learning_rate": 2.0625e-05,
      "loss": 0.0176,
      "step": 6878
    },
    {
      "epoch": 12.692520775623269,
      "grad_norm": 0.3989526331424713,
      "learning_rate": 2.0628e-05,
      "loss": 0.0192,
      "step": 6879
    },
    {
      "epoch": 12.694367497691598,
      "grad_norm": 0.46741893887519836,
      "learning_rate": 2.0631e-05,
      "loss": 0.0197,
      "step": 6880
    },
    {
      "epoch": 12.696214219759927,
      "grad_norm": 0.3675607740879059,
      "learning_rate": 2.0634e-05,
      "loss": 0.0128,
      "step": 6881
    },
    {
      "epoch": 12.698060941828254,
      "grad_norm": 0.48503541946411133,
      "learning_rate": 2.0637e-05,
      "loss": 0.0245,
      "step": 6882
    },
    {
      "epoch": 12.699907663896584,
      "grad_norm": 0.6410048007965088,
      "learning_rate": 2.064e-05,
      "loss": 0.0165,
      "step": 6883
    },
    {
      "epoch": 12.701754385964913,
      "grad_norm": 0.6283408999443054,
      "learning_rate": 2.0643000000000002e-05,
      "loss": 0.0239,
      "step": 6884
    },
    {
      "epoch": 12.703601108033242,
      "grad_norm": 0.49065521359443665,
      "learning_rate": 2.0646000000000002e-05,
      "loss": 0.028,
      "step": 6885
    },
    {
      "epoch": 12.70544783010157,
      "grad_norm": 0.47825199365615845,
      "learning_rate": 2.0649e-05,
      "loss": 0.0219,
      "step": 6886
    },
    {
      "epoch": 12.707294552169898,
      "grad_norm": 2.441694736480713,
      "learning_rate": 2.0652e-05,
      "loss": 0.0195,
      "step": 6887
    },
    {
      "epoch": 12.709141274238227,
      "grad_norm": 0.7367351651191711,
      "learning_rate": 2.0655e-05,
      "loss": 0.0298,
      "step": 6888
    },
    {
      "epoch": 12.710987996306557,
      "grad_norm": 0.46378958225250244,
      "learning_rate": 2.0658e-05,
      "loss": 0.018,
      "step": 6889
    },
    {
      "epoch": 12.712834718374884,
      "grad_norm": 0.43975552916526794,
      "learning_rate": 2.0661e-05,
      "loss": 0.02,
      "step": 6890
    },
    {
      "epoch": 12.714681440443213,
      "grad_norm": 0.5574910640716553,
      "learning_rate": 2.0664e-05,
      "loss": 0.0221,
      "step": 6891
    },
    {
      "epoch": 12.716528162511542,
      "grad_norm": 0.459857314825058,
      "learning_rate": 2.0667e-05,
      "loss": 0.0153,
      "step": 6892
    },
    {
      "epoch": 12.718374884579871,
      "grad_norm": 0.4340030550956726,
      "learning_rate": 2.067e-05,
      "loss": 0.0166,
      "step": 6893
    },
    {
      "epoch": 12.720221606648199,
      "grad_norm": 0.8469353914260864,
      "learning_rate": 2.0673000000000003e-05,
      "loss": 0.0214,
      "step": 6894
    },
    {
      "epoch": 12.722068328716528,
      "grad_norm": 0.48159655928611755,
      "learning_rate": 2.0676e-05,
      "loss": 0.022,
      "step": 6895
    },
    {
      "epoch": 12.723915050784857,
      "grad_norm": 1.0634124279022217,
      "learning_rate": 2.0679e-05,
      "loss": 0.0298,
      "step": 6896
    },
    {
      "epoch": 12.725761772853186,
      "grad_norm": 0.3850836157798767,
      "learning_rate": 2.0682e-05,
      "loss": 0.0136,
      "step": 6897
    },
    {
      "epoch": 12.727608494921514,
      "grad_norm": 0.5385677814483643,
      "learning_rate": 2.0685e-05,
      "loss": 0.0265,
      "step": 6898
    },
    {
      "epoch": 12.729455216989843,
      "grad_norm": 0.7085365653038025,
      "learning_rate": 2.0688e-05,
      "loss": 0.0128,
      "step": 6899
    },
    {
      "epoch": 12.731301939058172,
      "grad_norm": 0.5235032439231873,
      "learning_rate": 2.0691e-05,
      "loss": 0.0233,
      "step": 6900
    },
    {
      "epoch": 12.733148661126501,
      "grad_norm": 0.6124913096427917,
      "learning_rate": 2.0694e-05,
      "loss": 0.0154,
      "step": 6901
    },
    {
      "epoch": 12.734995383194828,
      "grad_norm": 0.4355025291442871,
      "learning_rate": 2.0697e-05,
      "loss": 0.0184,
      "step": 6902
    },
    {
      "epoch": 12.736842105263158,
      "grad_norm": 0.6916985511779785,
      "learning_rate": 2.07e-05,
      "loss": 0.0301,
      "step": 6903
    },
    {
      "epoch": 12.738688827331487,
      "grad_norm": 0.7459660768508911,
      "learning_rate": 2.0703e-05,
      "loss": 0.0351,
      "step": 6904
    },
    {
      "epoch": 12.740535549399816,
      "grad_norm": 0.7205683588981628,
      "learning_rate": 2.0706e-05,
      "loss": 0.206,
      "step": 6905
    },
    {
      "epoch": 12.742382271468145,
      "grad_norm": 1.1066662073135376,
      "learning_rate": 2.0709e-05,
      "loss": 0.2021,
      "step": 6906
    },
    {
      "epoch": 12.744228993536472,
      "grad_norm": 0.7279828190803528,
      "learning_rate": 2.0712e-05,
      "loss": 0.2561,
      "step": 6907
    },
    {
      "epoch": 12.746075715604801,
      "grad_norm": 0.4925101101398468,
      "learning_rate": 2.0715e-05,
      "loss": 0.121,
      "step": 6908
    },
    {
      "epoch": 12.74792243767313,
      "grad_norm": 0.7707668542861938,
      "learning_rate": 2.0718e-05,
      "loss": 0.1928,
      "step": 6909
    },
    {
      "epoch": 12.749769159741458,
      "grad_norm": 0.7203795313835144,
      "learning_rate": 2.0721e-05,
      "loss": 0.1554,
      "step": 6910
    },
    {
      "epoch": 12.751615881809787,
      "grad_norm": 0.5161910653114319,
      "learning_rate": 2.0724e-05,
      "loss": 0.1218,
      "step": 6911
    },
    {
      "epoch": 12.753462603878116,
      "grad_norm": 0.6372627019882202,
      "learning_rate": 2.0727e-05,
      "loss": 0.0966,
      "step": 6912
    },
    {
      "epoch": 12.755309325946445,
      "grad_norm": 0.4801217019557953,
      "learning_rate": 2.073e-05,
      "loss": 0.0698,
      "step": 6913
    },
    {
      "epoch": 12.757156048014775,
      "grad_norm": 0.4829772710800171,
      "learning_rate": 2.0733000000000003e-05,
      "loss": 0.0823,
      "step": 6914
    },
    {
      "epoch": 12.759002770083102,
      "grad_norm": 0.41738224029541016,
      "learning_rate": 2.0736000000000003e-05,
      "loss": 0.0433,
      "step": 6915
    },
    {
      "epoch": 12.760849492151431,
      "grad_norm": 0.7026256322860718,
      "learning_rate": 2.0739000000000003e-05,
      "loss": 0.0618,
      "step": 6916
    },
    {
      "epoch": 12.76269621421976,
      "grad_norm": 0.8881515860557556,
      "learning_rate": 2.0742000000000002e-05,
      "loss": 0.1348,
      "step": 6917
    },
    {
      "epoch": 12.76454293628809,
      "grad_norm": 0.6656627655029297,
      "learning_rate": 2.0745000000000002e-05,
      "loss": 0.0435,
      "step": 6918
    },
    {
      "epoch": 12.766389658356417,
      "grad_norm": 0.6259805560112,
      "learning_rate": 2.0748000000000002e-05,
      "loss": 0.0728,
      "step": 6919
    },
    {
      "epoch": 12.768236380424746,
      "grad_norm": 0.7157806754112244,
      "learning_rate": 2.0751e-05,
      "loss": 0.0545,
      "step": 6920
    },
    {
      "epoch": 12.770083102493075,
      "grad_norm": 0.39127853512763977,
      "learning_rate": 2.0753999999999998e-05,
      "loss": 0.0329,
      "step": 6921
    },
    {
      "epoch": 12.771929824561404,
      "grad_norm": 1.0186903476715088,
      "learning_rate": 2.0756999999999998e-05,
      "loss": 0.0205,
      "step": 6922
    },
    {
      "epoch": 12.773776546629731,
      "grad_norm": 0.7008572816848755,
      "learning_rate": 2.0759999999999998e-05,
      "loss": 0.0323,
      "step": 6923
    },
    {
      "epoch": 12.77562326869806,
      "grad_norm": 0.5930014848709106,
      "learning_rate": 2.0763e-05,
      "loss": 0.0316,
      "step": 6924
    },
    {
      "epoch": 12.77746999076639,
      "grad_norm": 0.3785812258720398,
      "learning_rate": 2.0766e-05,
      "loss": 0.0181,
      "step": 6925
    },
    {
      "epoch": 12.779316712834719,
      "grad_norm": 0.3636646568775177,
      "learning_rate": 2.0769e-05,
      "loss": 0.0156,
      "step": 6926
    },
    {
      "epoch": 12.781163434903046,
      "grad_norm": 0.42580145597457886,
      "learning_rate": 2.0772e-05,
      "loss": 0.0301,
      "step": 6927
    },
    {
      "epoch": 12.783010156971375,
      "grad_norm": 0.23977164924144745,
      "learning_rate": 2.0775e-05,
      "loss": 0.0156,
      "step": 6928
    },
    {
      "epoch": 12.784856879039705,
      "grad_norm": 0.4149671494960785,
      "learning_rate": 2.0778e-05,
      "loss": 0.021,
      "step": 6929
    },
    {
      "epoch": 12.786703601108034,
      "grad_norm": 0.6454666256904602,
      "learning_rate": 2.0781e-05,
      "loss": 0.029,
      "step": 6930
    },
    {
      "epoch": 12.788550323176363,
      "grad_norm": 1.481869101524353,
      "learning_rate": 2.0784e-05,
      "loss": 0.0257,
      "step": 6931
    },
    {
      "epoch": 12.79039704524469,
      "grad_norm": 0.35012781620025635,
      "learning_rate": 2.0787e-05,
      "loss": 0.0194,
      "step": 6932
    },
    {
      "epoch": 12.79224376731302,
      "grad_norm": 0.32981836795806885,
      "learning_rate": 2.079e-05,
      "loss": 0.0232,
      "step": 6933
    },
    {
      "epoch": 12.794090489381349,
      "grad_norm": 0.683250904083252,
      "learning_rate": 2.0793000000000002e-05,
      "loss": 0.0243,
      "step": 6934
    },
    {
      "epoch": 12.795937211449676,
      "grad_norm": 0.5012224912643433,
      "learning_rate": 2.0796000000000002e-05,
      "loss": 0.0142,
      "step": 6935
    },
    {
      "epoch": 12.797783933518005,
      "grad_norm": 1.3063740730285645,
      "learning_rate": 2.0799000000000002e-05,
      "loss": 0.0624,
      "step": 6936
    },
    {
      "epoch": 12.799630655586334,
      "grad_norm": 0.4880787134170532,
      "learning_rate": 2.0802000000000002e-05,
      "loss": 0.027,
      "step": 6937
    },
    {
      "epoch": 12.801477377654663,
      "grad_norm": 0.23685327172279358,
      "learning_rate": 2.0805e-05,
      "loss": 0.0076,
      "step": 6938
    },
    {
      "epoch": 12.803324099722992,
      "grad_norm": 0.5517817139625549,
      "learning_rate": 2.0808e-05,
      "loss": 0.0151,
      "step": 6939
    },
    {
      "epoch": 12.80517082179132,
      "grad_norm": 0.28913408517837524,
      "learning_rate": 2.0811e-05,
      "loss": 0.0151,
      "step": 6940
    },
    {
      "epoch": 12.807017543859649,
      "grad_norm": 0.5635241866111755,
      "learning_rate": 2.0814e-05,
      "loss": 0.0377,
      "step": 6941
    },
    {
      "epoch": 12.808864265927978,
      "grad_norm": 0.5010838508605957,
      "learning_rate": 2.0817e-05,
      "loss": 0.0202,
      "step": 6942
    },
    {
      "epoch": 12.810710987996307,
      "grad_norm": 0.3855663537979126,
      "learning_rate": 2.082e-05,
      "loss": 0.0138,
      "step": 6943
    },
    {
      "epoch": 12.812557710064635,
      "grad_norm": 0.8619406223297119,
      "learning_rate": 2.0823e-05,
      "loss": 0.0257,
      "step": 6944
    },
    {
      "epoch": 12.814404432132964,
      "grad_norm": 0.30185648798942566,
      "learning_rate": 2.0826e-05,
      "loss": 0.0166,
      "step": 6945
    },
    {
      "epoch": 12.816251154201293,
      "grad_norm": 0.6880849003791809,
      "learning_rate": 2.0829e-05,
      "loss": 0.0246,
      "step": 6946
    },
    {
      "epoch": 12.818097876269622,
      "grad_norm": 0.5926246643066406,
      "learning_rate": 2.0832e-05,
      "loss": 0.0255,
      "step": 6947
    },
    {
      "epoch": 12.81994459833795,
      "grad_norm": 0.4034042954444885,
      "learning_rate": 2.0835e-05,
      "loss": 0.0137,
      "step": 6948
    },
    {
      "epoch": 12.821791320406279,
      "grad_norm": 0.7444289326667786,
      "learning_rate": 2.0838e-05,
      "loss": 0.0219,
      "step": 6949
    },
    {
      "epoch": 12.823638042474608,
      "grad_norm": 0.2735901474952698,
      "learning_rate": 2.0841e-05,
      "loss": 0.0143,
      "step": 6950
    },
    {
      "epoch": 12.825484764542937,
      "grad_norm": 0.4763954281806946,
      "learning_rate": 2.0844e-05,
      "loss": 0.0129,
      "step": 6951
    },
    {
      "epoch": 12.827331486611264,
      "grad_norm": 0.7796971797943115,
      "learning_rate": 2.0847e-05,
      "loss": 0.0207,
      "step": 6952
    },
    {
      "epoch": 12.829178208679593,
      "grad_norm": 0.6892587542533875,
      "learning_rate": 2.085e-05,
      "loss": 0.034,
      "step": 6953
    },
    {
      "epoch": 12.831024930747922,
      "grad_norm": 2.045210361480713,
      "learning_rate": 2.0853000000000002e-05,
      "loss": 0.0358,
      "step": 6954
    },
    {
      "epoch": 12.832871652816252,
      "grad_norm": 0.6591195464134216,
      "learning_rate": 2.0856e-05,
      "loss": 0.2097,
      "step": 6955
    },
    {
      "epoch": 12.83471837488458,
      "grad_norm": 0.7082017660140991,
      "learning_rate": 2.0859e-05,
      "loss": 0.2276,
      "step": 6956
    },
    {
      "epoch": 12.836565096952908,
      "grad_norm": 0.5350101590156555,
      "learning_rate": 2.0862e-05,
      "loss": 0.1638,
      "step": 6957
    },
    {
      "epoch": 12.838411819021237,
      "grad_norm": 0.6802604794502258,
      "learning_rate": 2.0865e-05,
      "loss": 0.2066,
      "step": 6958
    },
    {
      "epoch": 12.840258541089566,
      "grad_norm": 0.9725978970527649,
      "learning_rate": 2.0868e-05,
      "loss": 0.1875,
      "step": 6959
    },
    {
      "epoch": 12.842105263157894,
      "grad_norm": 0.6738961338996887,
      "learning_rate": 2.0871e-05,
      "loss": 0.1481,
      "step": 6960
    },
    {
      "epoch": 12.843951985226223,
      "grad_norm": 0.5680983662605286,
      "learning_rate": 2.0874e-05,
      "loss": 0.1173,
      "step": 6961
    },
    {
      "epoch": 12.845798707294552,
      "grad_norm": 0.45616841316223145,
      "learning_rate": 2.0877e-05,
      "loss": 0.0983,
      "step": 6962
    },
    {
      "epoch": 12.847645429362881,
      "grad_norm": 0.598066508769989,
      "learning_rate": 2.088e-05,
      "loss": 0.1052,
      "step": 6963
    },
    {
      "epoch": 12.84949215143121,
      "grad_norm": 1.4628897905349731,
      "learning_rate": 2.0883000000000003e-05,
      "loss": 0.1242,
      "step": 6964
    },
    {
      "epoch": 12.851338873499538,
      "grad_norm": 0.8236523866653442,
      "learning_rate": 2.0886000000000003e-05,
      "loss": 0.0854,
      "step": 6965
    },
    {
      "epoch": 12.853185595567867,
      "grad_norm": 0.4810824990272522,
      "learning_rate": 2.0889000000000003e-05,
      "loss": 0.0581,
      "step": 6966
    },
    {
      "epoch": 12.855032317636196,
      "grad_norm": 0.3711375296115875,
      "learning_rate": 2.0892000000000003e-05,
      "loss": 0.0426,
      "step": 6967
    },
    {
      "epoch": 12.856879039704525,
      "grad_norm": 0.347176730632782,
      "learning_rate": 2.0895000000000002e-05,
      "loss": 0.051,
      "step": 6968
    },
    {
      "epoch": 12.858725761772853,
      "grad_norm": 0.33575108647346497,
      "learning_rate": 2.0898e-05,
      "loss": 0.0452,
      "step": 6969
    },
    {
      "epoch": 12.860572483841182,
      "grad_norm": 0.35422179102897644,
      "learning_rate": 2.0901e-05,
      "loss": 0.0251,
      "step": 6970
    },
    {
      "epoch": 12.86241920590951,
      "grad_norm": 0.23786962032318115,
      "learning_rate": 2.0904e-05,
      "loss": 0.0274,
      "step": 6971
    },
    {
      "epoch": 12.86426592797784,
      "grad_norm": 0.4184572696685791,
      "learning_rate": 2.0906999999999998e-05,
      "loss": 0.0333,
      "step": 6972
    },
    {
      "epoch": 12.866112650046167,
      "grad_norm": 0.3609471023082733,
      "learning_rate": 2.0909999999999998e-05,
      "loss": 0.0286,
      "step": 6973
    },
    {
      "epoch": 12.867959372114496,
      "grad_norm": 0.7237544655799866,
      "learning_rate": 2.0913e-05,
      "loss": 0.0644,
      "step": 6974
    },
    {
      "epoch": 12.869806094182826,
      "grad_norm": 0.49987494945526123,
      "learning_rate": 2.0916e-05,
      "loss": 0.0315,
      "step": 6975
    },
    {
      "epoch": 12.871652816251155,
      "grad_norm": 0.291019082069397,
      "learning_rate": 2.0919e-05,
      "loss": 0.0161,
      "step": 6976
    },
    {
      "epoch": 12.873499538319482,
      "grad_norm": 0.3615555465221405,
      "learning_rate": 2.0922e-05,
      "loss": 0.0203,
      "step": 6977
    },
    {
      "epoch": 12.875346260387811,
      "grad_norm": 0.2607676088809967,
      "learning_rate": 2.0925e-05,
      "loss": 0.0171,
      "step": 6978
    },
    {
      "epoch": 12.87719298245614,
      "grad_norm": 0.30133187770843506,
      "learning_rate": 2.0928e-05,
      "loss": 0.0241,
      "step": 6979
    },
    {
      "epoch": 12.87903970452447,
      "grad_norm": 0.30178168416023254,
      "learning_rate": 2.0931e-05,
      "loss": 0.0131,
      "step": 6980
    },
    {
      "epoch": 12.880886426592799,
      "grad_norm": 0.31973060965538025,
      "learning_rate": 2.0934e-05,
      "loss": 0.0171,
      "step": 6981
    },
    {
      "epoch": 12.882733148661126,
      "grad_norm": 0.4676547348499298,
      "learning_rate": 2.0937e-05,
      "loss": 0.0241,
      "step": 6982
    },
    {
      "epoch": 12.884579870729455,
      "grad_norm": 0.49283716082572937,
      "learning_rate": 2.094e-05,
      "loss": 0.0262,
      "step": 6983
    },
    {
      "epoch": 12.886426592797784,
      "grad_norm": 0.651775062084198,
      "learning_rate": 2.0943000000000003e-05,
      "loss": 0.0193,
      "step": 6984
    },
    {
      "epoch": 12.888273314866112,
      "grad_norm": 0.41634607315063477,
      "learning_rate": 2.0946000000000002e-05,
      "loss": 0.0168,
      "step": 6985
    },
    {
      "epoch": 12.89012003693444,
      "grad_norm": 0.4039091169834137,
      "learning_rate": 2.0949000000000002e-05,
      "loss": 0.019,
      "step": 6986
    },
    {
      "epoch": 12.89196675900277,
      "grad_norm": 0.8984862565994263,
      "learning_rate": 2.0952000000000002e-05,
      "loss": 0.0301,
      "step": 6987
    },
    {
      "epoch": 12.8938134810711,
      "grad_norm": 0.3584153950214386,
      "learning_rate": 2.0955000000000002e-05,
      "loss": 0.0124,
      "step": 6988
    },
    {
      "epoch": 12.895660203139428,
      "grad_norm": 0.27817732095718384,
      "learning_rate": 2.0958e-05,
      "loss": 0.0133,
      "step": 6989
    },
    {
      "epoch": 12.897506925207756,
      "grad_norm": 0.5320377945899963,
      "learning_rate": 2.0961e-05,
      "loss": 0.0254,
      "step": 6990
    },
    {
      "epoch": 12.899353647276085,
      "grad_norm": 0.3643032908439636,
      "learning_rate": 2.0964e-05,
      "loss": 0.0121,
      "step": 6991
    },
    {
      "epoch": 12.901200369344414,
      "grad_norm": 0.5568037033081055,
      "learning_rate": 2.0967e-05,
      "loss": 0.035,
      "step": 6992
    },
    {
      "epoch": 12.903047091412743,
      "grad_norm": 0.30891767144203186,
      "learning_rate": 2.097e-05,
      "loss": 0.0108,
      "step": 6993
    },
    {
      "epoch": 12.90489381348107,
      "grad_norm": 0.7795283794403076,
      "learning_rate": 2.0973e-05,
      "loss": 0.0205,
      "step": 6994
    },
    {
      "epoch": 12.9067405355494,
      "grad_norm": 0.4496384561061859,
      "learning_rate": 2.0976e-05,
      "loss": 0.0161,
      "step": 6995
    },
    {
      "epoch": 12.908587257617729,
      "grad_norm": 0.37556585669517517,
      "learning_rate": 2.0979e-05,
      "loss": 0.0177,
      "step": 6996
    },
    {
      "epoch": 12.910433979686058,
      "grad_norm": 0.6202804446220398,
      "learning_rate": 2.0982e-05,
      "loss": 0.028,
      "step": 6997
    },
    {
      "epoch": 12.912280701754385,
      "grad_norm": 0.3959607183933258,
      "learning_rate": 2.0985e-05,
      "loss": 0.0165,
      "step": 6998
    },
    {
      "epoch": 12.914127423822714,
      "grad_norm": 0.8058693408966064,
      "learning_rate": 2.0988e-05,
      "loss": 0.0333,
      "step": 6999
    },
    {
      "epoch": 12.915974145891044,
      "grad_norm": 0.4133374094963074,
      "learning_rate": 2.0991e-05,
      "loss": 0.0162,
      "step": 7000
    },
    {
      "epoch": 12.915974145891044,
      "eval_cer": 0.11390206599787438,
      "eval_loss": 0.3320147395133972,
      "eval_runtime": 15.7794,
      "eval_samples_per_second": 19.266,
      "eval_steps_per_second": 0.634,
      "eval_wer": 0.3983115886415963,
      "step": 7000
    },
    {
      "epoch": 12.917820867959373,
      "grad_norm": 0.383863627910614,
      "learning_rate": 2.0994e-05,
      "loss": 0.0181,
      "step": 7001
    },
    {
      "epoch": 12.9196675900277,
      "grad_norm": 0.5668813586235046,
      "learning_rate": 2.0997e-05,
      "loss": 0.0153,
      "step": 7002
    },
    {
      "epoch": 12.92151431209603,
      "grad_norm": 0.4858815371990204,
      "learning_rate": 2.1e-05,
      "loss": 0.0214,
      "step": 7003
    },
    {
      "epoch": 12.923361034164358,
      "grad_norm": 0.3549613356590271,
      "learning_rate": 2.1003e-05,
      "loss": 0.0165,
      "step": 7004
    },
    {
      "epoch": 12.925207756232687,
      "grad_norm": 0.8246420621871948,
      "learning_rate": 2.1006000000000002e-05,
      "loss": 0.2249,
      "step": 7005
    },
    {
      "epoch": 12.927054478301017,
      "grad_norm": 0.8446950316429138,
      "learning_rate": 2.1009e-05,
      "loss": 0.224,
      "step": 7006
    },
    {
      "epoch": 12.928901200369344,
      "grad_norm": 0.9744293689727783,
      "learning_rate": 2.1012e-05,
      "loss": 0.1771,
      "step": 7007
    },
    {
      "epoch": 12.930747922437673,
      "grad_norm": 0.7374587655067444,
      "learning_rate": 2.1015e-05,
      "loss": 0.1285,
      "step": 7008
    },
    {
      "epoch": 12.932594644506002,
      "grad_norm": 0.9851799607276917,
      "learning_rate": 2.1018e-05,
      "loss": 0.1067,
      "step": 7009
    },
    {
      "epoch": 12.93444136657433,
      "grad_norm": 0.5872266888618469,
      "learning_rate": 2.1021e-05,
      "loss": 0.1105,
      "step": 7010
    },
    {
      "epoch": 12.936288088642659,
      "grad_norm": 0.7037659287452698,
      "learning_rate": 2.1024e-05,
      "loss": 0.1388,
      "step": 7011
    },
    {
      "epoch": 12.938134810710988,
      "grad_norm": 0.7483109831809998,
      "learning_rate": 2.1027e-05,
      "loss": 0.0796,
      "step": 7012
    },
    {
      "epoch": 12.939981532779317,
      "grad_norm": 0.5599822998046875,
      "learning_rate": 2.103e-05,
      "loss": 0.0798,
      "step": 7013
    },
    {
      "epoch": 12.941828254847646,
      "grad_norm": 0.5699829459190369,
      "learning_rate": 2.1033e-05,
      "loss": 0.1105,
      "step": 7014
    },
    {
      "epoch": 12.943674976915974,
      "grad_norm": 0.6335070133209229,
      "learning_rate": 2.1036000000000003e-05,
      "loss": 0.0509,
      "step": 7015
    },
    {
      "epoch": 12.945521698984303,
      "grad_norm": 0.4124175012111664,
      "learning_rate": 2.1039000000000003e-05,
      "loss": 0.0774,
      "step": 7016
    },
    {
      "epoch": 12.947368421052632,
      "grad_norm": 0.5848485827445984,
      "learning_rate": 2.1042000000000003e-05,
      "loss": 0.035,
      "step": 7017
    },
    {
      "epoch": 12.949215143120961,
      "grad_norm": 0.35119831562042236,
      "learning_rate": 2.1045e-05,
      "loss": 0.0339,
      "step": 7018
    },
    {
      "epoch": 12.951061865189288,
      "grad_norm": 0.38088950514793396,
      "learning_rate": 2.1048e-05,
      "loss": 0.0242,
      "step": 7019
    },
    {
      "epoch": 12.952908587257618,
      "grad_norm": 0.2906529903411865,
      "learning_rate": 2.1051e-05,
      "loss": 0.0179,
      "step": 7020
    },
    {
      "epoch": 12.954755309325947,
      "grad_norm": 0.25520816445350647,
      "learning_rate": 2.1054e-05,
      "loss": 0.0201,
      "step": 7021
    },
    {
      "epoch": 12.956602031394276,
      "grad_norm": 0.8681545257568359,
      "learning_rate": 2.1057e-05,
      "loss": 0.0407,
      "step": 7022
    },
    {
      "epoch": 12.958448753462603,
      "grad_norm": 0.6451759934425354,
      "learning_rate": 2.1059999999999998e-05,
      "loss": 0.0216,
      "step": 7023
    },
    {
      "epoch": 12.960295475530932,
      "grad_norm": 0.24663574993610382,
      "learning_rate": 2.1062999999999998e-05,
      "loss": 0.0137,
      "step": 7024
    },
    {
      "epoch": 12.962142197599261,
      "grad_norm": 0.44320282340049744,
      "learning_rate": 2.1066e-05,
      "loss": 0.0271,
      "step": 7025
    },
    {
      "epoch": 12.96398891966759,
      "grad_norm": 0.673911988735199,
      "learning_rate": 2.1069e-05,
      "loss": 0.0374,
      "step": 7026
    },
    {
      "epoch": 12.965835641735918,
      "grad_norm": 0.3014264702796936,
      "learning_rate": 2.1072e-05,
      "loss": 0.0093,
      "step": 7027
    },
    {
      "epoch": 12.967682363804247,
      "grad_norm": 0.35082635283470154,
      "learning_rate": 2.1075e-05,
      "loss": 0.0301,
      "step": 7028
    },
    {
      "epoch": 12.969529085872576,
      "grad_norm": 0.41949841380119324,
      "learning_rate": 2.1078e-05,
      "loss": 0.0239,
      "step": 7029
    },
    {
      "epoch": 12.971375807940905,
      "grad_norm": 0.42678096890449524,
      "learning_rate": 2.1081e-05,
      "loss": 0.0199,
      "step": 7030
    },
    {
      "epoch": 12.973222530009235,
      "grad_norm": 0.36272698640823364,
      "learning_rate": 2.1084e-05,
      "loss": 0.0106,
      "step": 7031
    },
    {
      "epoch": 12.975069252077562,
      "grad_norm": 0.9446842670440674,
      "learning_rate": 2.1087e-05,
      "loss": 0.052,
      "step": 7032
    },
    {
      "epoch": 12.976915974145891,
      "grad_norm": 0.46493256092071533,
      "learning_rate": 2.109e-05,
      "loss": 0.0224,
      "step": 7033
    },
    {
      "epoch": 12.97876269621422,
      "grad_norm": 0.5906376838684082,
      "learning_rate": 2.1093e-05,
      "loss": 0.0228,
      "step": 7034
    },
    {
      "epoch": 12.980609418282548,
      "grad_norm": 0.463929682970047,
      "learning_rate": 2.1096000000000003e-05,
      "loss": 0.0156,
      "step": 7035
    },
    {
      "epoch": 12.982456140350877,
      "grad_norm": 0.4736950099468231,
      "learning_rate": 2.1099000000000002e-05,
      "loss": 0.0205,
      "step": 7036
    },
    {
      "epoch": 12.984302862419206,
      "grad_norm": 0.48260924220085144,
      "learning_rate": 2.1102000000000002e-05,
      "loss": 0.0207,
      "step": 7037
    },
    {
      "epoch": 12.986149584487535,
      "grad_norm": 0.6692067384719849,
      "learning_rate": 2.1105000000000002e-05,
      "loss": 0.0157,
      "step": 7038
    },
    {
      "epoch": 12.987996306555864,
      "grad_norm": 0.6906869411468506,
      "learning_rate": 2.1108000000000002e-05,
      "loss": 0.0335,
      "step": 7039
    },
    {
      "epoch": 12.989843028624191,
      "grad_norm": 0.5933719873428345,
      "learning_rate": 2.1111e-05,
      "loss": 0.0265,
      "step": 7040
    },
    {
      "epoch": 12.99168975069252,
      "grad_norm": 0.36224380135536194,
      "learning_rate": 2.1114e-05,
      "loss": 0.0172,
      "step": 7041
    },
    {
      "epoch": 12.99353647276085,
      "grad_norm": 0.5262947082519531,
      "learning_rate": 2.1117e-05,
      "loss": 0.0283,
      "step": 7042
    },
    {
      "epoch": 12.995383194829179,
      "grad_norm": 0.4658443331718445,
      "learning_rate": 2.1119999999999998e-05,
      "loss": 0.0214,
      "step": 7043
    },
    {
      "epoch": 12.997229916897506,
      "grad_norm": 0.6707825064659119,
      "learning_rate": 2.1122999999999997e-05,
      "loss": 0.0173,
      "step": 7044
    },
    {
      "epoch": 12.999076638965835,
      "grad_norm": 0.6804872751235962,
      "learning_rate": 2.1126e-05,
      "loss": 0.0282,
      "step": 7045
    },
    {
      "epoch": 13.0,
      "grad_norm": 0.5780600309371948,
      "learning_rate": 2.1129e-05,
      "loss": 0.0378,
      "step": 7046
    },
    {
      "epoch": 13.00184672206833,
      "grad_norm": 1.1009926795959473,
      "learning_rate": 2.1132e-05,
      "loss": 0.2053,
      "step": 7047
    },
    {
      "epoch": 13.003693444136658,
      "grad_norm": 0.5999391078948975,
      "learning_rate": 2.1135e-05,
      "loss": 0.1755,
      "step": 7048
    },
    {
      "epoch": 13.005540166204986,
      "grad_norm": 0.556535542011261,
      "learning_rate": 2.1138e-05,
      "loss": 0.1356,
      "step": 7049
    },
    {
      "epoch": 13.007386888273315,
      "grad_norm": 0.600877583026886,
      "learning_rate": 2.1141e-05,
      "loss": 0.1368,
      "step": 7050
    },
    {
      "epoch": 13.009233610341644,
      "grad_norm": 0.7779603600502014,
      "learning_rate": 2.1144e-05,
      "loss": 0.1396,
      "step": 7051
    },
    {
      "epoch": 13.011080332409973,
      "grad_norm": 0.959456205368042,
      "learning_rate": 2.1147e-05,
      "loss": 0.1214,
      "step": 7052
    },
    {
      "epoch": 13.0129270544783,
      "grad_norm": 0.6287292242050171,
      "learning_rate": 2.115e-05,
      "loss": 0.1031,
      "step": 7053
    },
    {
      "epoch": 13.01477377654663,
      "grad_norm": 1.0254368782043457,
      "learning_rate": 2.1153e-05,
      "loss": 0.157,
      "step": 7054
    },
    {
      "epoch": 13.016620498614959,
      "grad_norm": 0.43327924609184265,
      "learning_rate": 2.1156000000000002e-05,
      "loss": 0.0638,
      "step": 7055
    },
    {
      "epoch": 13.018467220683288,
      "grad_norm": 0.6945347189903259,
      "learning_rate": 2.1159000000000002e-05,
      "loss": 0.1202,
      "step": 7056
    },
    {
      "epoch": 13.020313942751615,
      "grad_norm": 0.4801901578903198,
      "learning_rate": 2.1162e-05,
      "loss": 0.0553,
      "step": 7057
    },
    {
      "epoch": 13.022160664819944,
      "grad_norm": 0.6773169636726379,
      "learning_rate": 2.1165e-05,
      "loss": 0.0592,
      "step": 7058
    },
    {
      "epoch": 13.024007386888274,
      "grad_norm": 0.3857210874557495,
      "learning_rate": 2.1168e-05,
      "loss": 0.044,
      "step": 7059
    },
    {
      "epoch": 13.025854108956603,
      "grad_norm": 0.43382251262664795,
      "learning_rate": 2.1171e-05,
      "loss": 0.0577,
      "step": 7060
    },
    {
      "epoch": 13.02770083102493,
      "grad_norm": 0.5019907355308533,
      "learning_rate": 2.1174e-05,
      "loss": 0.0489,
      "step": 7061
    },
    {
      "epoch": 13.02954755309326,
      "grad_norm": 0.396818608045578,
      "learning_rate": 2.1177e-05,
      "loss": 0.045,
      "step": 7062
    },
    {
      "epoch": 13.031394275161588,
      "grad_norm": 0.3628042936325073,
      "learning_rate": 2.118e-05,
      "loss": 0.0311,
      "step": 7063
    },
    {
      "epoch": 13.033240997229917,
      "grad_norm": 0.3977918028831482,
      "learning_rate": 2.1183e-05,
      "loss": 0.0295,
      "step": 7064
    },
    {
      "epoch": 13.035087719298245,
      "grad_norm": 0.43366360664367676,
      "learning_rate": 2.1186000000000003e-05,
      "loss": 0.0235,
      "step": 7065
    },
    {
      "epoch": 13.036934441366574,
      "grad_norm": 0.20432913303375244,
      "learning_rate": 2.1189000000000003e-05,
      "loss": 0.0141,
      "step": 7066
    },
    {
      "epoch": 13.038781163434903,
      "grad_norm": 0.41155481338500977,
      "learning_rate": 2.1192e-05,
      "loss": 0.0175,
      "step": 7067
    },
    {
      "epoch": 13.040627885503232,
      "grad_norm": 0.5188421010971069,
      "learning_rate": 2.1195e-05,
      "loss": 0.0143,
      "step": 7068
    },
    {
      "epoch": 13.04247460757156,
      "grad_norm": 0.3438571095466614,
      "learning_rate": 2.1198e-05,
      "loss": 0.0173,
      "step": 7069
    },
    {
      "epoch": 13.044321329639889,
      "grad_norm": 0.38158324360847473,
      "learning_rate": 2.1201e-05,
      "loss": 0.0154,
      "step": 7070
    },
    {
      "epoch": 13.046168051708218,
      "grad_norm": 0.4931504726409912,
      "learning_rate": 2.1204e-05,
      "loss": 0.0211,
      "step": 7071
    },
    {
      "epoch": 13.048014773776547,
      "grad_norm": 0.6768643856048584,
      "learning_rate": 2.1207e-05,
      "loss": 0.0446,
      "step": 7072
    },
    {
      "epoch": 13.049861495844876,
      "grad_norm": 0.2771964967250824,
      "learning_rate": 2.121e-05,
      "loss": 0.0146,
      "step": 7073
    },
    {
      "epoch": 13.051708217913204,
      "grad_norm": 0.30338409543037415,
      "learning_rate": 2.1213e-05,
      "loss": 0.0146,
      "step": 7074
    },
    {
      "epoch": 13.053554939981533,
      "grad_norm": 0.22734993696212769,
      "learning_rate": 2.1216e-05,
      "loss": 0.012,
      "step": 7075
    },
    {
      "epoch": 13.055401662049862,
      "grad_norm": 0.43263518810272217,
      "learning_rate": 2.1219e-05,
      "loss": 0.0182,
      "step": 7076
    },
    {
      "epoch": 13.057248384118191,
      "grad_norm": 0.44592079520225525,
      "learning_rate": 2.1222e-05,
      "loss": 0.0124,
      "step": 7077
    },
    {
      "epoch": 13.059095106186518,
      "grad_norm": 0.2170630544424057,
      "learning_rate": 2.1225e-05,
      "loss": 0.0134,
      "step": 7078
    },
    {
      "epoch": 13.060941828254848,
      "grad_norm": 0.2715395987033844,
      "learning_rate": 2.1228e-05,
      "loss": 0.0108,
      "step": 7079
    },
    {
      "epoch": 13.062788550323177,
      "grad_norm": 0.3276883661746979,
      "learning_rate": 2.1231e-05,
      "loss": 0.0117,
      "step": 7080
    },
    {
      "epoch": 13.064635272391506,
      "grad_norm": 0.3050946593284607,
      "learning_rate": 2.1234e-05,
      "loss": 0.0141,
      "step": 7081
    },
    {
      "epoch": 13.066481994459833,
      "grad_norm": 0.3024367392063141,
      "learning_rate": 2.1237e-05,
      "loss": 0.0113,
      "step": 7082
    },
    {
      "epoch": 13.068328716528162,
      "grad_norm": 0.40349775552749634,
      "learning_rate": 2.124e-05,
      "loss": 0.0106,
      "step": 7083
    },
    {
      "epoch": 13.070175438596491,
      "grad_norm": 0.16834889352321625,
      "learning_rate": 2.1243e-05,
      "loss": 0.0054,
      "step": 7084
    },
    {
      "epoch": 13.07202216066482,
      "grad_norm": 0.3100760579109192,
      "learning_rate": 2.1246000000000003e-05,
      "loss": 0.0127,
      "step": 7085
    },
    {
      "epoch": 13.073868882733148,
      "grad_norm": 0.5087378621101379,
      "learning_rate": 2.1249000000000003e-05,
      "loss": 0.0306,
      "step": 7086
    },
    {
      "epoch": 13.075715604801477,
      "grad_norm": 0.5137645602226257,
      "learning_rate": 2.1252000000000003e-05,
      "loss": 0.0128,
      "step": 7087
    },
    {
      "epoch": 13.077562326869806,
      "grad_norm": 0.4320560395717621,
      "learning_rate": 2.1255000000000002e-05,
      "loss": 0.0168,
      "step": 7088
    },
    {
      "epoch": 13.079409048938135,
      "grad_norm": 0.3999880254268646,
      "learning_rate": 2.1258000000000002e-05,
      "loss": 0.0185,
      "step": 7089
    },
    {
      "epoch": 13.081255771006463,
      "grad_norm": 0.31367284059524536,
      "learning_rate": 2.1261000000000002e-05,
      "loss": 0.0126,
      "step": 7090
    },
    {
      "epoch": 13.083102493074792,
      "grad_norm": 0.49617084860801697,
      "learning_rate": 2.1264000000000002e-05,
      "loss": 0.0184,
      "step": 7091
    },
    {
      "epoch": 13.084949215143121,
      "grad_norm": 0.44734519720077515,
      "learning_rate": 2.1266999999999998e-05,
      "loss": 0.0105,
      "step": 7092
    },
    {
      "epoch": 13.08679593721145,
      "grad_norm": 0.5911281704902649,
      "learning_rate": 2.1269999999999998e-05,
      "loss": 0.0214,
      "step": 7093
    },
    {
      "epoch": 13.088642659279778,
      "grad_norm": 1.7160379886627197,
      "learning_rate": 2.1272999999999998e-05,
      "loss": 0.0247,
      "step": 7094
    },
    {
      "epoch": 13.090489381348107,
      "grad_norm": 0.3841411769390106,
      "learning_rate": 2.1276e-05,
      "loss": 0.0156,
      "step": 7095
    },
    {
      "epoch": 13.092336103416436,
      "grad_norm": 0.5357542037963867,
      "learning_rate": 2.1279e-05,
      "loss": 0.0416,
      "step": 7096
    },
    {
      "epoch": 13.094182825484765,
      "grad_norm": 0.9648202061653137,
      "learning_rate": 2.1282e-05,
      "loss": 0.3002,
      "step": 7097
    },
    {
      "epoch": 13.096029547553094,
      "grad_norm": 0.6870272159576416,
      "learning_rate": 2.1285e-05,
      "loss": 0.2282,
      "step": 7098
    },
    {
      "epoch": 13.097876269621421,
      "grad_norm": 0.5585189461708069,
      "learning_rate": 2.1288e-05,
      "loss": 0.156,
      "step": 7099
    },
    {
      "epoch": 13.09972299168975,
      "grad_norm": 0.7169678211212158,
      "learning_rate": 2.1291e-05,
      "loss": 0.1415,
      "step": 7100
    },
    {
      "epoch": 13.10156971375808,
      "grad_norm": 0.5396972298622131,
      "learning_rate": 2.1294e-05,
      "loss": 0.1254,
      "step": 7101
    },
    {
      "epoch": 13.103416435826409,
      "grad_norm": 0.6024976372718811,
      "learning_rate": 2.1297e-05,
      "loss": 0.1497,
      "step": 7102
    },
    {
      "epoch": 13.105263157894736,
      "grad_norm": 0.5685150027275085,
      "learning_rate": 2.13e-05,
      "loss": 0.116,
      "step": 7103
    },
    {
      "epoch": 13.107109879963065,
      "grad_norm": 0.4817744195461273,
      "learning_rate": 2.1303e-05,
      "loss": 0.1146,
      "step": 7104
    },
    {
      "epoch": 13.108956602031395,
      "grad_norm": 0.6075971126556396,
      "learning_rate": 2.1306000000000002e-05,
      "loss": 0.0902,
      "step": 7105
    },
    {
      "epoch": 13.110803324099724,
      "grad_norm": 0.4989315867424011,
      "learning_rate": 2.1309000000000002e-05,
      "loss": 0.0883,
      "step": 7106
    },
    {
      "epoch": 13.112650046168051,
      "grad_norm": 0.923866331577301,
      "learning_rate": 2.1312000000000002e-05,
      "loss": 0.0684,
      "step": 7107
    },
    {
      "epoch": 13.11449676823638,
      "grad_norm": 0.3913983106613159,
      "learning_rate": 2.1315000000000002e-05,
      "loss": 0.044,
      "step": 7108
    },
    {
      "epoch": 13.11634349030471,
      "grad_norm": 0.3887367248535156,
      "learning_rate": 2.1318e-05,
      "loss": 0.0533,
      "step": 7109
    },
    {
      "epoch": 13.118190212373039,
      "grad_norm": 0.36022719740867615,
      "learning_rate": 2.1321e-05,
      "loss": 0.0187,
      "step": 7110
    },
    {
      "epoch": 13.120036934441366,
      "grad_norm": 0.38603368401527405,
      "learning_rate": 2.1324e-05,
      "loss": 0.0578,
      "step": 7111
    },
    {
      "epoch": 13.121883656509695,
      "grad_norm": 0.75755375623703,
      "learning_rate": 2.1327e-05,
      "loss": 0.0812,
      "step": 7112
    },
    {
      "epoch": 13.123730378578024,
      "grad_norm": 0.31779199838638306,
      "learning_rate": 2.133e-05,
      "loss": 0.0181,
      "step": 7113
    },
    {
      "epoch": 13.125577100646353,
      "grad_norm": 0.6031693816184998,
      "learning_rate": 2.1333e-05,
      "loss": 0.0439,
      "step": 7114
    },
    {
      "epoch": 13.12742382271468,
      "grad_norm": 0.27690422534942627,
      "learning_rate": 2.1336000000000004e-05,
      "loss": 0.0171,
      "step": 7115
    },
    {
      "epoch": 13.12927054478301,
      "grad_norm": 0.3437330722808838,
      "learning_rate": 2.1339e-05,
      "loss": 0.013,
      "step": 7116
    },
    {
      "epoch": 13.131117266851339,
      "grad_norm": 0.5207774639129639,
      "learning_rate": 2.1342e-05,
      "loss": 0.0223,
      "step": 7117
    },
    {
      "epoch": 13.132963988919668,
      "grad_norm": 0.2619399130344391,
      "learning_rate": 2.1345e-05,
      "loss": 0.0232,
      "step": 7118
    },
    {
      "epoch": 13.134810710987995,
      "grad_norm": 0.24779574573040009,
      "learning_rate": 2.1348e-05,
      "loss": 0.0102,
      "step": 7119
    },
    {
      "epoch": 13.136657433056325,
      "grad_norm": 0.28220513463020325,
      "learning_rate": 2.1351e-05,
      "loss": 0.0174,
      "step": 7120
    },
    {
      "epoch": 13.138504155124654,
      "grad_norm": 0.3668232262134552,
      "learning_rate": 2.1354e-05,
      "loss": 0.02,
      "step": 7121
    },
    {
      "epoch": 13.140350877192983,
      "grad_norm": 0.27169662714004517,
      "learning_rate": 2.1357e-05,
      "loss": 0.0156,
      "step": 7122
    },
    {
      "epoch": 13.142197599261312,
      "grad_norm": 0.33091557025909424,
      "learning_rate": 2.136e-05,
      "loss": 0.015,
      "step": 7123
    },
    {
      "epoch": 13.14404432132964,
      "grad_norm": 0.5022887587547302,
      "learning_rate": 2.1363e-05,
      "loss": 0.0175,
      "step": 7124
    },
    {
      "epoch": 13.145891043397969,
      "grad_norm": 0.7188720107078552,
      "learning_rate": 2.1366000000000002e-05,
      "loss": 0.0187,
      "step": 7125
    },
    {
      "epoch": 13.147737765466298,
      "grad_norm": 0.5348430275917053,
      "learning_rate": 2.1369e-05,
      "loss": 0.019,
      "step": 7126
    },
    {
      "epoch": 13.149584487534627,
      "grad_norm": 0.34337300062179565,
      "learning_rate": 2.1372e-05,
      "loss": 0.0137,
      "step": 7127
    },
    {
      "epoch": 13.151431209602954,
      "grad_norm": 0.32131054997444153,
      "learning_rate": 2.1375e-05,
      "loss": 0.024,
      "step": 7128
    },
    {
      "epoch": 13.153277931671283,
      "grad_norm": 0.3087523877620697,
      "learning_rate": 2.1378e-05,
      "loss": 0.0113,
      "step": 7129
    },
    {
      "epoch": 13.155124653739612,
      "grad_norm": 0.23587988317012787,
      "learning_rate": 2.1381e-05,
      "loss": 0.0122,
      "step": 7130
    },
    {
      "epoch": 13.156971375807942,
      "grad_norm": 0.6149274110794067,
      "learning_rate": 2.1384e-05,
      "loss": 0.0207,
      "step": 7131
    },
    {
      "epoch": 13.158818097876269,
      "grad_norm": 0.366910457611084,
      "learning_rate": 2.1387e-05,
      "loss": 0.0134,
      "step": 7132
    },
    {
      "epoch": 13.160664819944598,
      "grad_norm": 0.4519480764865875,
      "learning_rate": 2.139e-05,
      "loss": 0.016,
      "step": 7133
    },
    {
      "epoch": 13.162511542012927,
      "grad_norm": 0.4220968186855316,
      "learning_rate": 2.1393e-05,
      "loss": 0.0148,
      "step": 7134
    },
    {
      "epoch": 13.164358264081256,
      "grad_norm": 0.5174040794372559,
      "learning_rate": 2.1396e-05,
      "loss": 0.0133,
      "step": 7135
    },
    {
      "epoch": 13.166204986149584,
      "grad_norm": 0.415140837430954,
      "learning_rate": 2.1399000000000003e-05,
      "loss": 0.0143,
      "step": 7136
    },
    {
      "epoch": 13.168051708217913,
      "grad_norm": 0.2746680974960327,
      "learning_rate": 2.1402000000000003e-05,
      "loss": 0.0097,
      "step": 7137
    },
    {
      "epoch": 13.169898430286242,
      "grad_norm": 0.5087571740150452,
      "learning_rate": 2.1405000000000003e-05,
      "loss": 0.0202,
      "step": 7138
    },
    {
      "epoch": 13.171745152354571,
      "grad_norm": 0.3475409746170044,
      "learning_rate": 2.1408000000000002e-05,
      "loss": 0.0182,
      "step": 7139
    },
    {
      "epoch": 13.173591874422899,
      "grad_norm": 0.6303392648696899,
      "learning_rate": 2.1411000000000002e-05,
      "loss": 0.0145,
      "step": 7140
    },
    {
      "epoch": 13.175438596491228,
      "grad_norm": 0.26322299242019653,
      "learning_rate": 2.1414e-05,
      "loss": 0.0111,
      "step": 7141
    },
    {
      "epoch": 13.177285318559557,
      "grad_norm": 0.3473210334777832,
      "learning_rate": 2.1417e-05,
      "loss": 0.0099,
      "step": 7142
    },
    {
      "epoch": 13.179132040627886,
      "grad_norm": 1.0462384223937988,
      "learning_rate": 2.1419999999999998e-05,
      "loss": 0.0269,
      "step": 7143
    },
    {
      "epoch": 13.180978762696213,
      "grad_norm": 0.7094504237174988,
      "learning_rate": 2.1422999999999998e-05,
      "loss": 0.0192,
      "step": 7144
    },
    {
      "epoch": 13.182825484764543,
      "grad_norm": 0.990999162197113,
      "learning_rate": 2.1425999999999998e-05,
      "loss": 0.0203,
      "step": 7145
    },
    {
      "epoch": 13.184672206832872,
      "grad_norm": 0.9668733477592468,
      "learning_rate": 2.1429e-05,
      "loss": 0.0425,
      "step": 7146
    },
    {
      "epoch": 13.1865189289012,
      "grad_norm": 0.7852587699890137,
      "learning_rate": 2.1432e-05,
      "loss": 0.2142,
      "step": 7147
    },
    {
      "epoch": 13.18836565096953,
      "grad_norm": 0.675623893737793,
      "learning_rate": 2.1435e-05,
      "loss": 0.162,
      "step": 7148
    },
    {
      "epoch": 13.190212373037857,
      "grad_norm": 0.6665027737617493,
      "learning_rate": 2.1438e-05,
      "loss": 0.1618,
      "step": 7149
    },
    {
      "epoch": 13.192059095106186,
      "grad_norm": 0.7854028940200806,
      "learning_rate": 2.1441e-05,
      "loss": 0.1313,
      "step": 7150
    },
    {
      "epoch": 13.193905817174516,
      "grad_norm": 0.775463342666626,
      "learning_rate": 2.1444e-05,
      "loss": 0.1387,
      "step": 7151
    },
    {
      "epoch": 13.195752539242845,
      "grad_norm": 3.532989501953125,
      "learning_rate": 2.1447e-05,
      "loss": 0.1338,
      "step": 7152
    },
    {
      "epoch": 13.197599261311172,
      "grad_norm": 0.5899909734725952,
      "learning_rate": 2.145e-05,
      "loss": 0.1463,
      "step": 7153
    },
    {
      "epoch": 13.199445983379501,
      "grad_norm": 0.4959987699985504,
      "learning_rate": 2.1453e-05,
      "loss": 0.0921,
      "step": 7154
    },
    {
      "epoch": 13.20129270544783,
      "grad_norm": 0.7340102195739746,
      "learning_rate": 2.1456e-05,
      "loss": 0.1113,
      "step": 7155
    },
    {
      "epoch": 13.20313942751616,
      "grad_norm": 0.8091896772384644,
      "learning_rate": 2.1459000000000002e-05,
      "loss": 0.0996,
      "step": 7156
    },
    {
      "epoch": 13.204986149584487,
      "grad_norm": 0.4246945381164551,
      "learning_rate": 2.1462000000000002e-05,
      "loss": 0.0573,
      "step": 7157
    },
    {
      "epoch": 13.206832871652816,
      "grad_norm": 0.4794583320617676,
      "learning_rate": 2.1465000000000002e-05,
      "loss": 0.0522,
      "step": 7158
    },
    {
      "epoch": 13.208679593721145,
      "grad_norm": 0.8014870882034302,
      "learning_rate": 2.1468000000000002e-05,
      "loss": 0.11,
      "step": 7159
    },
    {
      "epoch": 13.210526315789474,
      "grad_norm": 1.3814865350723267,
      "learning_rate": 2.1471e-05,
      "loss": 0.0724,
      "step": 7160
    },
    {
      "epoch": 13.212373037857802,
      "grad_norm": 1.1732836961746216,
      "learning_rate": 2.1474e-05,
      "loss": 0.0517,
      "step": 7161
    },
    {
      "epoch": 13.21421975992613,
      "grad_norm": 0.3207579553127289,
      "learning_rate": 2.1477e-05,
      "loss": 0.0429,
      "step": 7162
    },
    {
      "epoch": 13.21606648199446,
      "grad_norm": 0.4469239115715027,
      "learning_rate": 2.148e-05,
      "loss": 0.023,
      "step": 7163
    },
    {
      "epoch": 13.21791320406279,
      "grad_norm": 0.45449766516685486,
      "learning_rate": 2.1483e-05,
      "loss": 0.028,
      "step": 7164
    },
    {
      "epoch": 13.219759926131117,
      "grad_norm": 0.23852013051509857,
      "learning_rate": 2.1486e-05,
      "loss": 0.0117,
      "step": 7165
    },
    {
      "epoch": 13.221606648199446,
      "grad_norm": 0.20696599781513214,
      "learning_rate": 2.1489e-05,
      "loss": 0.0132,
      "step": 7166
    },
    {
      "epoch": 13.223453370267775,
      "grad_norm": 0.30197465419769287,
      "learning_rate": 2.1492e-05,
      "loss": 0.0111,
      "step": 7167
    },
    {
      "epoch": 13.225300092336104,
      "grad_norm": 0.2798866927623749,
      "learning_rate": 2.1495e-05,
      "loss": 0.0177,
      "step": 7168
    },
    {
      "epoch": 13.227146814404431,
      "grad_norm": 0.457897812128067,
      "learning_rate": 2.1498e-05,
      "loss": 0.0312,
      "step": 7169
    },
    {
      "epoch": 13.22899353647276,
      "grad_norm": 0.4140821397304535,
      "learning_rate": 2.1501e-05,
      "loss": 0.0123,
      "step": 7170
    },
    {
      "epoch": 13.23084025854109,
      "grad_norm": 0.6828473210334778,
      "learning_rate": 2.1504e-05,
      "loss": 0.025,
      "step": 7171
    },
    {
      "epoch": 13.232686980609419,
      "grad_norm": 0.23633523285388947,
      "learning_rate": 2.1507e-05,
      "loss": 0.0126,
      "step": 7172
    },
    {
      "epoch": 13.234533702677748,
      "grad_norm": 0.2687138319015503,
      "learning_rate": 2.151e-05,
      "loss": 0.0107,
      "step": 7173
    },
    {
      "epoch": 13.236380424746075,
      "grad_norm": 0.36586427688598633,
      "learning_rate": 2.1513e-05,
      "loss": 0.0112,
      "step": 7174
    },
    {
      "epoch": 13.238227146814404,
      "grad_norm": 0.40713468194007874,
      "learning_rate": 2.1516e-05,
      "loss": 0.0231,
      "step": 7175
    },
    {
      "epoch": 13.240073868882734,
      "grad_norm": 0.41447657346725464,
      "learning_rate": 2.1519000000000002e-05,
      "loss": 0.0454,
      "step": 7176
    },
    {
      "epoch": 13.241920590951063,
      "grad_norm": 0.6310411095619202,
      "learning_rate": 2.1522e-05,
      "loss": 0.023,
      "step": 7177
    },
    {
      "epoch": 13.24376731301939,
      "grad_norm": 0.3591703474521637,
      "learning_rate": 2.1525e-05,
      "loss": 0.0129,
      "step": 7178
    },
    {
      "epoch": 13.24561403508772,
      "grad_norm": 0.44197985529899597,
      "learning_rate": 2.1528e-05,
      "loss": 0.017,
      "step": 7179
    },
    {
      "epoch": 13.247460757156048,
      "grad_norm": 0.30777108669281006,
      "learning_rate": 2.1531e-05,
      "loss": 0.0164,
      "step": 7180
    },
    {
      "epoch": 13.249307479224377,
      "grad_norm": 0.3263159394264221,
      "learning_rate": 2.1534e-05,
      "loss": 0.0094,
      "step": 7181
    },
    {
      "epoch": 13.251154201292705,
      "grad_norm": 0.9163006544113159,
      "learning_rate": 2.1537e-05,
      "loss": 0.024,
      "step": 7182
    },
    {
      "epoch": 13.253000923361034,
      "grad_norm": 0.7877883911132812,
      "learning_rate": 2.154e-05,
      "loss": 0.0397,
      "step": 7183
    },
    {
      "epoch": 13.254847645429363,
      "grad_norm": 0.2887888252735138,
      "learning_rate": 2.1543e-05,
      "loss": 0.0111,
      "step": 7184
    },
    {
      "epoch": 13.256694367497692,
      "grad_norm": 0.4857921600341797,
      "learning_rate": 2.1546e-05,
      "loss": 0.0288,
      "step": 7185
    },
    {
      "epoch": 13.25854108956602,
      "grad_norm": 0.47236984968185425,
      "learning_rate": 2.1549000000000003e-05,
      "loss": 0.0161,
      "step": 7186
    },
    {
      "epoch": 13.260387811634349,
      "grad_norm": 0.30887550115585327,
      "learning_rate": 2.1552000000000003e-05,
      "loss": 0.0108,
      "step": 7187
    },
    {
      "epoch": 13.262234533702678,
      "grad_norm": 0.5033847689628601,
      "learning_rate": 2.1555000000000003e-05,
      "loss": 0.0173,
      "step": 7188
    },
    {
      "epoch": 13.264081255771007,
      "grad_norm": 0.3925565779209137,
      "learning_rate": 2.1558000000000003e-05,
      "loss": 0.0102,
      "step": 7189
    },
    {
      "epoch": 13.265927977839334,
      "grad_norm": 0.46677395701408386,
      "learning_rate": 2.1561e-05,
      "loss": 0.0185,
      "step": 7190
    },
    {
      "epoch": 13.267774699907664,
      "grad_norm": 0.3494510054588318,
      "learning_rate": 2.1564e-05,
      "loss": 0.0166,
      "step": 7191
    },
    {
      "epoch": 13.269621421975993,
      "grad_norm": 0.397417813539505,
      "learning_rate": 2.1567e-05,
      "loss": 0.0196,
      "step": 7192
    },
    {
      "epoch": 13.271468144044322,
      "grad_norm": 0.44307324290275574,
      "learning_rate": 2.157e-05,
      "loss": 0.0196,
      "step": 7193
    },
    {
      "epoch": 13.27331486611265,
      "grad_norm": 0.5888022184371948,
      "learning_rate": 2.1572999999999998e-05,
      "loss": 0.0492,
      "step": 7194
    },
    {
      "epoch": 13.275161588180978,
      "grad_norm": 0.4758075773715973,
      "learning_rate": 2.1575999999999998e-05,
      "loss": 0.0157,
      "step": 7195
    },
    {
      "epoch": 13.277008310249307,
      "grad_norm": 0.3339407742023468,
      "learning_rate": 2.1579e-05,
      "loss": 0.0157,
      "step": 7196
    },
    {
      "epoch": 13.278855032317637,
      "grad_norm": 1.0655595064163208,
      "learning_rate": 2.1582e-05,
      "loss": 0.2268,
      "step": 7197
    },
    {
      "epoch": 13.280701754385966,
      "grad_norm": 1.2859565019607544,
      "learning_rate": 2.1585e-05,
      "loss": 0.1929,
      "step": 7198
    },
    {
      "epoch": 13.282548476454293,
      "grad_norm": 0.6945502161979675,
      "learning_rate": 2.1588e-05,
      "loss": 0.1583,
      "step": 7199
    },
    {
      "epoch": 13.284395198522622,
      "grad_norm": 0.5635412335395813,
      "learning_rate": 2.1591e-05,
      "loss": 0.1227,
      "step": 7200
    },
    {
      "epoch": 13.286241920590951,
      "grad_norm": 0.710004448890686,
      "learning_rate": 2.1594e-05,
      "loss": 0.1405,
      "step": 7201
    },
    {
      "epoch": 13.28808864265928,
      "grad_norm": 0.8793792724609375,
      "learning_rate": 2.1597e-05,
      "loss": 0.1154,
      "step": 7202
    },
    {
      "epoch": 13.289935364727608,
      "grad_norm": 0.5695134401321411,
      "learning_rate": 2.16e-05,
      "loss": 0.0923,
      "step": 7203
    },
    {
      "epoch": 13.291782086795937,
      "grad_norm": 0.5201019048690796,
      "learning_rate": 2.1603e-05,
      "loss": 0.0729,
      "step": 7204
    },
    {
      "epoch": 13.293628808864266,
      "grad_norm": 0.8120662569999695,
      "learning_rate": 2.1606e-05,
      "loss": 0.1176,
      "step": 7205
    },
    {
      "epoch": 13.295475530932595,
      "grad_norm": 0.5279056429862976,
      "learning_rate": 2.1609000000000003e-05,
      "loss": 0.0907,
      "step": 7206
    },
    {
      "epoch": 13.297322253000923,
      "grad_norm": 0.9664171934127808,
      "learning_rate": 2.1612000000000002e-05,
      "loss": 0.1301,
      "step": 7207
    },
    {
      "epoch": 13.299168975069252,
      "grad_norm": 2.4879748821258545,
      "learning_rate": 2.1615000000000002e-05,
      "loss": 0.0676,
      "step": 7208
    },
    {
      "epoch": 13.301015697137581,
      "grad_norm": 0.6646405458450317,
      "learning_rate": 2.1618000000000002e-05,
      "loss": 0.1167,
      "step": 7209
    },
    {
      "epoch": 13.30286241920591,
      "grad_norm": 0.37683454155921936,
      "learning_rate": 2.1621000000000002e-05,
      "loss": 0.047,
      "step": 7210
    },
    {
      "epoch": 13.304709141274238,
      "grad_norm": 0.31387007236480713,
      "learning_rate": 2.1624e-05,
      "loss": 0.0184,
      "step": 7211
    },
    {
      "epoch": 13.306555863342567,
      "grad_norm": 0.2672097384929657,
      "learning_rate": 2.1627e-05,
      "loss": 0.0225,
      "step": 7212
    },
    {
      "epoch": 13.308402585410896,
      "grad_norm": 0.40251773595809937,
      "learning_rate": 2.163e-05,
      "loss": 0.0284,
      "step": 7213
    },
    {
      "epoch": 13.310249307479225,
      "grad_norm": 0.3544987142086029,
      "learning_rate": 2.1633e-05,
      "loss": 0.0207,
      "step": 7214
    },
    {
      "epoch": 13.312096029547552,
      "grad_norm": 0.33679115772247314,
      "learning_rate": 2.1635999999999997e-05,
      "loss": 0.0214,
      "step": 7215
    },
    {
      "epoch": 13.313942751615881,
      "grad_norm": 0.43973714113235474,
      "learning_rate": 2.1639e-05,
      "loss": 0.0154,
      "step": 7216
    },
    {
      "epoch": 13.31578947368421,
      "grad_norm": 0.3998846113681793,
      "learning_rate": 2.1642e-05,
      "loss": 0.0215,
      "step": 7217
    },
    {
      "epoch": 13.31763619575254,
      "grad_norm": 0.6689403057098389,
      "learning_rate": 2.1645e-05,
      "loss": 0.0149,
      "step": 7218
    },
    {
      "epoch": 13.319482917820867,
      "grad_norm": 0.6973395347595215,
      "learning_rate": 2.1648e-05,
      "loss": 0.023,
      "step": 7219
    },
    {
      "epoch": 13.321329639889196,
      "grad_norm": 0.6347224116325378,
      "learning_rate": 2.1651e-05,
      "loss": 0.0153,
      "step": 7220
    },
    {
      "epoch": 13.323176361957525,
      "grad_norm": 0.4243088662624359,
      "learning_rate": 2.1654e-05,
      "loss": 0.0161,
      "step": 7221
    },
    {
      "epoch": 13.325023084025855,
      "grad_norm": 0.4804117679595947,
      "learning_rate": 2.1657e-05,
      "loss": 0.0168,
      "step": 7222
    },
    {
      "epoch": 13.326869806094184,
      "grad_norm": 0.36143893003463745,
      "learning_rate": 2.166e-05,
      "loss": 0.0126,
      "step": 7223
    },
    {
      "epoch": 13.328716528162511,
      "grad_norm": 0.32996055483818054,
      "learning_rate": 2.1663e-05,
      "loss": 0.0167,
      "step": 7224
    },
    {
      "epoch": 13.33056325023084,
      "grad_norm": 0.2420320361852646,
      "learning_rate": 2.1666e-05,
      "loss": 0.011,
      "step": 7225
    },
    {
      "epoch": 13.33240997229917,
      "grad_norm": 0.42439308762550354,
      "learning_rate": 2.1669000000000002e-05,
      "loss": 0.0158,
      "step": 7226
    },
    {
      "epoch": 13.334256694367498,
      "grad_norm": 0.28058120608329773,
      "learning_rate": 2.1672000000000002e-05,
      "loss": 0.0122,
      "step": 7227
    },
    {
      "epoch": 13.336103416435826,
      "grad_norm": 0.2649085819721222,
      "learning_rate": 2.1675e-05,
      "loss": 0.0115,
      "step": 7228
    },
    {
      "epoch": 13.337950138504155,
      "grad_norm": 0.6683909893035889,
      "learning_rate": 2.1678e-05,
      "loss": 0.0249,
      "step": 7229
    },
    {
      "epoch": 13.339796860572484,
      "grad_norm": 0.40081146359443665,
      "learning_rate": 2.1681e-05,
      "loss": 0.0275,
      "step": 7230
    },
    {
      "epoch": 13.341643582640813,
      "grad_norm": 0.8051175475120544,
      "learning_rate": 2.1684e-05,
      "loss": 0.0199,
      "step": 7231
    },
    {
      "epoch": 13.34349030470914,
      "grad_norm": 0.21623077988624573,
      "learning_rate": 2.1687e-05,
      "loss": 0.0095,
      "step": 7232
    },
    {
      "epoch": 13.34533702677747,
      "grad_norm": 0.4252525269985199,
      "learning_rate": 2.169e-05,
      "loss": 0.0209,
      "step": 7233
    },
    {
      "epoch": 13.347183748845799,
      "grad_norm": 0.5248692035675049,
      "learning_rate": 2.1693e-05,
      "loss": 0.0206,
      "step": 7234
    },
    {
      "epoch": 13.349030470914128,
      "grad_norm": 16.779129028320312,
      "learning_rate": 2.1696e-05,
      "loss": 0.0261,
      "step": 7235
    },
    {
      "epoch": 13.350877192982455,
      "grad_norm": 0.3974554240703583,
      "learning_rate": 2.1699000000000003e-05,
      "loss": 0.0227,
      "step": 7236
    },
    {
      "epoch": 13.352723915050785,
      "grad_norm": 0.27056658267974854,
      "learning_rate": 2.1702000000000003e-05,
      "loss": 0.0111,
      "step": 7237
    },
    {
      "epoch": 13.354570637119114,
      "grad_norm": 0.42890921235084534,
      "learning_rate": 2.1705000000000003e-05,
      "loss": 0.0213,
      "step": 7238
    },
    {
      "epoch": 13.356417359187443,
      "grad_norm": 1.2918071746826172,
      "learning_rate": 2.1708e-05,
      "loss": 0.0128,
      "step": 7239
    },
    {
      "epoch": 13.35826408125577,
      "grad_norm": 0.5146956443786621,
      "learning_rate": 2.1711e-05,
      "loss": 0.0131,
      "step": 7240
    },
    {
      "epoch": 13.3601108033241,
      "grad_norm": 0.8489638566970825,
      "learning_rate": 2.1714e-05,
      "loss": 0.0167,
      "step": 7241
    },
    {
      "epoch": 13.361957525392429,
      "grad_norm": 1.5672495365142822,
      "learning_rate": 2.1717e-05,
      "loss": 0.0208,
      "step": 7242
    },
    {
      "epoch": 13.363804247460758,
      "grad_norm": 0.9432560801506042,
      "learning_rate": 2.172e-05,
      "loss": 0.0226,
      "step": 7243
    },
    {
      "epoch": 13.365650969529085,
      "grad_norm": 0.821049153804779,
      "learning_rate": 2.1723e-05,
      "loss": 0.0224,
      "step": 7244
    },
    {
      "epoch": 13.367497691597414,
      "grad_norm": 0.4392554759979248,
      "learning_rate": 2.1726e-05,
      "loss": 0.0179,
      "step": 7245
    },
    {
      "epoch": 13.369344413665743,
      "grad_norm": 0.5630433559417725,
      "learning_rate": 2.1729e-05,
      "loss": 0.0096,
      "step": 7246
    },
    {
      "epoch": 13.371191135734072,
      "grad_norm": 0.8121259212493896,
      "learning_rate": 2.1732e-05,
      "loss": 0.1936,
      "step": 7247
    },
    {
      "epoch": 13.373037857802402,
      "grad_norm": 2.499257802963257,
      "learning_rate": 2.1735e-05,
      "loss": 0.3012,
      "step": 7248
    },
    {
      "epoch": 13.374884579870729,
      "grad_norm": 0.7789328098297119,
      "learning_rate": 2.1738e-05,
      "loss": 0.1467,
      "step": 7249
    },
    {
      "epoch": 13.376731301939058,
      "grad_norm": 0.8124067783355713,
      "learning_rate": 2.1741e-05,
      "loss": 0.1901,
      "step": 7250
    },
    {
      "epoch": 13.378578024007387,
      "grad_norm": 0.9140483140945435,
      "learning_rate": 2.1744e-05,
      "loss": 0.1848,
      "step": 7251
    },
    {
      "epoch": 13.380424746075716,
      "grad_norm": 1.3085297346115112,
      "learning_rate": 2.1747e-05,
      "loss": 0.1444,
      "step": 7252
    },
    {
      "epoch": 13.382271468144044,
      "grad_norm": 0.7353108525276184,
      "learning_rate": 2.175e-05,
      "loss": 0.119,
      "step": 7253
    },
    {
      "epoch": 13.384118190212373,
      "grad_norm": 0.7093847990036011,
      "learning_rate": 2.1753e-05,
      "loss": 0.1582,
      "step": 7254
    },
    {
      "epoch": 13.385964912280702,
      "grad_norm": 0.660713791847229,
      "learning_rate": 2.1756e-05,
      "loss": 0.1042,
      "step": 7255
    },
    {
      "epoch": 13.387811634349031,
      "grad_norm": 0.53863126039505,
      "learning_rate": 2.1759e-05,
      "loss": 0.0789,
      "step": 7256
    },
    {
      "epoch": 13.389658356417359,
      "grad_norm": 0.437294602394104,
      "learning_rate": 2.1762000000000003e-05,
      "loss": 0.0701,
      "step": 7257
    },
    {
      "epoch": 13.391505078485688,
      "grad_norm": 0.5495612621307373,
      "learning_rate": 2.1765000000000003e-05,
      "loss": 0.0459,
      "step": 7258
    },
    {
      "epoch": 13.393351800554017,
      "grad_norm": 0.43721771240234375,
      "learning_rate": 2.1768000000000002e-05,
      "loss": 0.0419,
      "step": 7259
    },
    {
      "epoch": 13.395198522622346,
      "grad_norm": 0.65282142162323,
      "learning_rate": 2.1771000000000002e-05,
      "loss": 0.0607,
      "step": 7260
    },
    {
      "epoch": 13.397045244690673,
      "grad_norm": 0.4140559732913971,
      "learning_rate": 2.1774000000000002e-05,
      "loss": 0.0373,
      "step": 7261
    },
    {
      "epoch": 13.398891966759003,
      "grad_norm": 0.6380917429924011,
      "learning_rate": 2.1777000000000002e-05,
      "loss": 0.0588,
      "step": 7262
    },
    {
      "epoch": 13.400738688827332,
      "grad_norm": 0.4607631266117096,
      "learning_rate": 2.178e-05,
      "loss": 0.0628,
      "step": 7263
    },
    {
      "epoch": 13.40258541089566,
      "grad_norm": 0.7542992234230042,
      "learning_rate": 2.1782999999999998e-05,
      "loss": 0.0248,
      "step": 7264
    },
    {
      "epoch": 13.404432132963988,
      "grad_norm": 0.5851745009422302,
      "learning_rate": 2.1785999999999998e-05,
      "loss": 0.046,
      "step": 7265
    },
    {
      "epoch": 13.406278855032317,
      "grad_norm": 0.7884859442710876,
      "learning_rate": 2.1788999999999998e-05,
      "loss": 0.0225,
      "step": 7266
    },
    {
      "epoch": 13.408125577100646,
      "grad_norm": 0.38143467903137207,
      "learning_rate": 2.1792e-05,
      "loss": 0.0255,
      "step": 7267
    },
    {
      "epoch": 13.409972299168976,
      "grad_norm": 0.5632845759391785,
      "learning_rate": 2.1795e-05,
      "loss": 0.0206,
      "step": 7268
    },
    {
      "epoch": 13.411819021237303,
      "grad_norm": 0.4236559271812439,
      "learning_rate": 2.1798e-05,
      "loss": 0.038,
      "step": 7269
    },
    {
      "epoch": 13.413665743305632,
      "grad_norm": 0.43563246726989746,
      "learning_rate": 2.1801e-05,
      "loss": 0.0261,
      "step": 7270
    },
    {
      "epoch": 13.415512465373961,
      "grad_norm": 0.2808643579483032,
      "learning_rate": 2.1804e-05,
      "loss": 0.0161,
      "step": 7271
    },
    {
      "epoch": 13.41735918744229,
      "grad_norm": 0.45650675892829895,
      "learning_rate": 2.1807e-05,
      "loss": 0.0196,
      "step": 7272
    },
    {
      "epoch": 13.41920590951062,
      "grad_norm": 0.24618905782699585,
      "learning_rate": 2.181e-05,
      "loss": 0.0124,
      "step": 7273
    },
    {
      "epoch": 13.421052631578947,
      "grad_norm": 0.2942012846469879,
      "learning_rate": 2.1813e-05,
      "loss": 0.0202,
      "step": 7274
    },
    {
      "epoch": 13.422899353647276,
      "grad_norm": 0.26131290197372437,
      "learning_rate": 2.1816e-05,
      "loss": 0.0192,
      "step": 7275
    },
    {
      "epoch": 13.424746075715605,
      "grad_norm": 0.3145621418952942,
      "learning_rate": 2.1819e-05,
      "loss": 0.013,
      "step": 7276
    },
    {
      "epoch": 13.426592797783934,
      "grad_norm": 0.48726406693458557,
      "learning_rate": 2.1822000000000002e-05,
      "loss": 0.0152,
      "step": 7277
    },
    {
      "epoch": 13.428439519852262,
      "grad_norm": 0.4565317928791046,
      "learning_rate": 2.1825000000000002e-05,
      "loss": 0.0123,
      "step": 7278
    },
    {
      "epoch": 13.43028624192059,
      "grad_norm": 0.48445793986320496,
      "learning_rate": 2.1828000000000002e-05,
      "loss": 0.0154,
      "step": 7279
    },
    {
      "epoch": 13.43213296398892,
      "grad_norm": 0.3091747760772705,
      "learning_rate": 2.1831e-05,
      "loss": 0.0192,
      "step": 7280
    },
    {
      "epoch": 13.43397968605725,
      "grad_norm": 1.0693906545639038,
      "learning_rate": 2.1834e-05,
      "loss": 0.0188,
      "step": 7281
    },
    {
      "epoch": 13.435826408125576,
      "grad_norm": 0.3439318537712097,
      "learning_rate": 2.1837e-05,
      "loss": 0.0174,
      "step": 7282
    },
    {
      "epoch": 13.437673130193906,
      "grad_norm": 0.8398624658584595,
      "learning_rate": 2.184e-05,
      "loss": 0.0232,
      "step": 7283
    },
    {
      "epoch": 13.439519852262235,
      "grad_norm": 0.5656768679618835,
      "learning_rate": 2.1843e-05,
      "loss": 0.0239,
      "step": 7284
    },
    {
      "epoch": 13.441366574330564,
      "grad_norm": 0.8627192974090576,
      "learning_rate": 2.1846e-05,
      "loss": 0.0174,
      "step": 7285
    },
    {
      "epoch": 13.443213296398891,
      "grad_norm": 0.40234649181365967,
      "learning_rate": 2.1849e-05,
      "loss": 0.0186,
      "step": 7286
    },
    {
      "epoch": 13.44506001846722,
      "grad_norm": 0.26516568660736084,
      "learning_rate": 2.1852000000000004e-05,
      "loss": 0.0147,
      "step": 7287
    },
    {
      "epoch": 13.44690674053555,
      "grad_norm": 0.7071975469589233,
      "learning_rate": 2.1855e-05,
      "loss": 0.0153,
      "step": 7288
    },
    {
      "epoch": 13.448753462603879,
      "grad_norm": 1.1829360723495483,
      "learning_rate": 2.1858e-05,
      "loss": 0.025,
      "step": 7289
    },
    {
      "epoch": 13.450600184672206,
      "grad_norm": 0.37632063031196594,
      "learning_rate": 2.1861e-05,
      "loss": 0.023,
      "step": 7290
    },
    {
      "epoch": 13.452446906740535,
      "grad_norm": 0.29822513461112976,
      "learning_rate": 2.1864e-05,
      "loss": 0.013,
      "step": 7291
    },
    {
      "epoch": 13.454293628808864,
      "grad_norm": 0.3867357671260834,
      "learning_rate": 2.1867e-05,
      "loss": 0.0174,
      "step": 7292
    },
    {
      "epoch": 13.456140350877194,
      "grad_norm": 0.380867600440979,
      "learning_rate": 2.187e-05,
      "loss": 0.0137,
      "step": 7293
    },
    {
      "epoch": 13.45798707294552,
      "grad_norm": 0.4950351417064667,
      "learning_rate": 2.1873e-05,
      "loss": 0.0166,
      "step": 7294
    },
    {
      "epoch": 13.45983379501385,
      "grad_norm": 1.1106423139572144,
      "learning_rate": 2.1876e-05,
      "loss": 0.0274,
      "step": 7295
    },
    {
      "epoch": 13.46168051708218,
      "grad_norm": 0.5773835778236389,
      "learning_rate": 2.1879e-05,
      "loss": 0.0375,
      "step": 7296
    },
    {
      "epoch": 13.463527239150508,
      "grad_norm": 1.07211434841156,
      "learning_rate": 2.1882e-05,
      "loss": 0.232,
      "step": 7297
    },
    {
      "epoch": 13.465373961218837,
      "grad_norm": 0.7582486867904663,
      "learning_rate": 2.1885e-05,
      "loss": 0.1851,
      "step": 7298
    },
    {
      "epoch": 13.467220683287165,
      "grad_norm": 0.8717134594917297,
      "learning_rate": 2.1888e-05,
      "loss": 0.1718,
      "step": 7299
    },
    {
      "epoch": 13.469067405355494,
      "grad_norm": 0.5966388583183289,
      "learning_rate": 2.1891e-05,
      "loss": 0.1342,
      "step": 7300
    },
    {
      "epoch": 13.470914127423823,
      "grad_norm": 0.7327893972396851,
      "learning_rate": 2.1894e-05,
      "loss": 0.1264,
      "step": 7301
    },
    {
      "epoch": 13.472760849492152,
      "grad_norm": 0.8322132229804993,
      "learning_rate": 2.1897e-05,
      "loss": 0.1263,
      "step": 7302
    },
    {
      "epoch": 13.47460757156048,
      "grad_norm": 0.6498303413391113,
      "learning_rate": 2.19e-05,
      "loss": 0.0919,
      "step": 7303
    },
    {
      "epoch": 13.476454293628809,
      "grad_norm": 0.9275102019309998,
      "learning_rate": 2.1903e-05,
      "loss": 0.1055,
      "step": 7304
    },
    {
      "epoch": 13.478301015697138,
      "grad_norm": 0.5299592614173889,
      "learning_rate": 2.1906e-05,
      "loss": 0.084,
      "step": 7305
    },
    {
      "epoch": 13.480147737765467,
      "grad_norm": 0.4677112400531769,
      "learning_rate": 2.1909e-05,
      "loss": 0.0772,
      "step": 7306
    },
    {
      "epoch": 13.481994459833794,
      "grad_norm": 1.0250064134597778,
      "learning_rate": 2.1912000000000003e-05,
      "loss": 0.0993,
      "step": 7307
    },
    {
      "epoch": 13.483841181902124,
      "grad_norm": 0.40030497312545776,
      "learning_rate": 2.1915000000000003e-05,
      "loss": 0.0496,
      "step": 7308
    },
    {
      "epoch": 13.485687903970453,
      "grad_norm": 0.639595091342926,
      "learning_rate": 2.1918000000000003e-05,
      "loss": 0.1111,
      "step": 7309
    },
    {
      "epoch": 13.487534626038782,
      "grad_norm": 0.5788593292236328,
      "learning_rate": 2.1921000000000002e-05,
      "loss": 0.0682,
      "step": 7310
    },
    {
      "epoch": 13.48938134810711,
      "grad_norm": 1.2469546794891357,
      "learning_rate": 2.1924000000000002e-05,
      "loss": 0.0278,
      "step": 7311
    },
    {
      "epoch": 13.491228070175438,
      "grad_norm": 0.4102165699005127,
      "learning_rate": 2.1927000000000002e-05,
      "loss": 0.025,
      "step": 7312
    },
    {
      "epoch": 13.493074792243767,
      "grad_norm": 0.489040344953537,
      "learning_rate": 2.193e-05,
      "loss": 0.0358,
      "step": 7313
    },
    {
      "epoch": 13.494921514312097,
      "grad_norm": 0.269023597240448,
      "learning_rate": 2.1932999999999998e-05,
      "loss": 0.0198,
      "step": 7314
    },
    {
      "epoch": 13.496768236380424,
      "grad_norm": 0.7333555221557617,
      "learning_rate": 2.1935999999999998e-05,
      "loss": 0.0973,
      "step": 7315
    },
    {
      "epoch": 13.498614958448753,
      "grad_norm": 0.6423579454421997,
      "learning_rate": 2.1938999999999998e-05,
      "loss": 0.0403,
      "step": 7316
    },
    {
      "epoch": 13.500461680517082,
      "grad_norm": 0.356412410736084,
      "learning_rate": 2.1942e-05,
      "loss": 0.0143,
      "step": 7317
    },
    {
      "epoch": 13.502308402585411,
      "grad_norm": 0.7933915853500366,
      "learning_rate": 2.1945e-05,
      "loss": 0.0432,
      "step": 7318
    },
    {
      "epoch": 13.504155124653739,
      "grad_norm": 0.3011784553527832,
      "learning_rate": 2.1948e-05,
      "loss": 0.0215,
      "step": 7319
    },
    {
      "epoch": 13.506001846722068,
      "grad_norm": 0.2465125173330307,
      "learning_rate": 2.1951e-05,
      "loss": 0.0161,
      "step": 7320
    },
    {
      "epoch": 13.507848568790397,
      "grad_norm": 0.5328660607337952,
      "learning_rate": 2.1954e-05,
      "loss": 0.0158,
      "step": 7321
    },
    {
      "epoch": 13.509695290858726,
      "grad_norm": 0.33585166931152344,
      "learning_rate": 2.1957e-05,
      "loss": 0.0195,
      "step": 7322
    },
    {
      "epoch": 13.511542012927055,
      "grad_norm": 0.4674575924873352,
      "learning_rate": 2.196e-05,
      "loss": 0.016,
      "step": 7323
    },
    {
      "epoch": 13.513388734995383,
      "grad_norm": 0.629899263381958,
      "learning_rate": 2.1963e-05,
      "loss": 0.017,
      "step": 7324
    },
    {
      "epoch": 13.515235457063712,
      "grad_norm": 0.4062348008155823,
      "learning_rate": 2.1966e-05,
      "loss": 0.0181,
      "step": 7325
    },
    {
      "epoch": 13.517082179132041,
      "grad_norm": 0.2978869676589966,
      "learning_rate": 2.1969e-05,
      "loss": 0.0141,
      "step": 7326
    },
    {
      "epoch": 13.51892890120037,
      "grad_norm": 0.3354339003562927,
      "learning_rate": 2.1972000000000002e-05,
      "loss": 0.0153,
      "step": 7327
    },
    {
      "epoch": 13.520775623268698,
      "grad_norm": 1.723165512084961,
      "learning_rate": 2.1975000000000002e-05,
      "loss": 0.0139,
      "step": 7328
    },
    {
      "epoch": 13.522622345337027,
      "grad_norm": 0.6914976239204407,
      "learning_rate": 2.1978000000000002e-05,
      "loss": 0.0161,
      "step": 7329
    },
    {
      "epoch": 13.524469067405356,
      "grad_norm": 0.48744842410087585,
      "learning_rate": 2.1981000000000002e-05,
      "loss": 0.0153,
      "step": 7330
    },
    {
      "epoch": 13.526315789473685,
      "grad_norm": 0.22318950295448303,
      "learning_rate": 2.1984e-05,
      "loss": 0.0094,
      "step": 7331
    },
    {
      "epoch": 13.528162511542012,
      "grad_norm": 1.1957658529281616,
      "learning_rate": 2.1987e-05,
      "loss": 0.0422,
      "step": 7332
    },
    {
      "epoch": 13.530009233610341,
      "grad_norm": 0.39699581265449524,
      "learning_rate": 2.199e-05,
      "loss": 0.0111,
      "step": 7333
    },
    {
      "epoch": 13.53185595567867,
      "grad_norm": 0.30261993408203125,
      "learning_rate": 2.1993e-05,
      "loss": 0.0102,
      "step": 7334
    },
    {
      "epoch": 13.533702677747,
      "grad_norm": 0.46019795536994934,
      "learning_rate": 2.1996e-05,
      "loss": 0.0151,
      "step": 7335
    },
    {
      "epoch": 13.535549399815327,
      "grad_norm": 0.6777722835540771,
      "learning_rate": 2.1999e-05,
      "loss": 0.0186,
      "step": 7336
    },
    {
      "epoch": 13.537396121883656,
      "grad_norm": 0.44539082050323486,
      "learning_rate": 2.2002e-05,
      "loss": 0.0189,
      "step": 7337
    },
    {
      "epoch": 13.539242843951985,
      "grad_norm": 1.1245629787445068,
      "learning_rate": 2.2005e-05,
      "loss": 0.0343,
      "step": 7338
    },
    {
      "epoch": 13.541089566020315,
      "grad_norm": 0.4725266695022583,
      "learning_rate": 2.2008e-05,
      "loss": 0.0116,
      "step": 7339
    },
    {
      "epoch": 13.542936288088642,
      "grad_norm": 0.5914231538772583,
      "learning_rate": 2.2011e-05,
      "loss": 0.0257,
      "step": 7340
    },
    {
      "epoch": 13.544783010156971,
      "grad_norm": 0.8138396739959717,
      "learning_rate": 2.2014e-05,
      "loss": 0.0249,
      "step": 7341
    },
    {
      "epoch": 13.5466297322253,
      "grad_norm": 0.7865918874740601,
      "learning_rate": 2.2017e-05,
      "loss": 0.0177,
      "step": 7342
    },
    {
      "epoch": 13.54847645429363,
      "grad_norm": 0.5864379405975342,
      "learning_rate": 2.202e-05,
      "loss": 0.0211,
      "step": 7343
    },
    {
      "epoch": 13.550323176361957,
      "grad_norm": 0.5314502716064453,
      "learning_rate": 2.2023e-05,
      "loss": 0.0153,
      "step": 7344
    },
    {
      "epoch": 13.552169898430286,
      "grad_norm": 1.0157397985458374,
      "learning_rate": 2.2026e-05,
      "loss": 0.0318,
      "step": 7345
    },
    {
      "epoch": 13.554016620498615,
      "grad_norm": 0.9477611184120178,
      "learning_rate": 2.2029e-05,
      "loss": 0.0208,
      "step": 7346
    },
    {
      "epoch": 13.555863342566944,
      "grad_norm": 0.9941097497940063,
      "learning_rate": 2.2032000000000002e-05,
      "loss": 0.2425,
      "step": 7347
    },
    {
      "epoch": 13.557710064635273,
      "grad_norm": 0.6913464665412903,
      "learning_rate": 2.2035e-05,
      "loss": 0.2258,
      "step": 7348
    },
    {
      "epoch": 13.5595567867036,
      "grad_norm": 1.1100223064422607,
      "learning_rate": 2.2038e-05,
      "loss": 0.211,
      "step": 7349
    },
    {
      "epoch": 13.56140350877193,
      "grad_norm": 0.7175942063331604,
      "learning_rate": 2.2041e-05,
      "loss": 0.1512,
      "step": 7350
    },
    {
      "epoch": 13.563250230840259,
      "grad_norm": 0.6584893465042114,
      "learning_rate": 2.2044e-05,
      "loss": 0.1469,
      "step": 7351
    },
    {
      "epoch": 13.565096952908588,
      "grad_norm": 1.1889525651931763,
      "learning_rate": 2.2047e-05,
      "loss": 0.2033,
      "step": 7352
    },
    {
      "epoch": 13.566943674976915,
      "grad_norm": 0.49368175864219666,
      "learning_rate": 2.205e-05,
      "loss": 0.0866,
      "step": 7353
    },
    {
      "epoch": 13.568790397045245,
      "grad_norm": 0.514544665813446,
      "learning_rate": 2.2053e-05,
      "loss": 0.0779,
      "step": 7354
    },
    {
      "epoch": 13.570637119113574,
      "grad_norm": 1.3042136430740356,
      "learning_rate": 2.2056e-05,
      "loss": 0.1321,
      "step": 7355
    },
    {
      "epoch": 13.572483841181903,
      "grad_norm": 0.49586865305900574,
      "learning_rate": 2.2059e-05,
      "loss": 0.0815,
      "step": 7356
    },
    {
      "epoch": 13.57433056325023,
      "grad_norm": 0.5277937650680542,
      "learning_rate": 2.2062000000000003e-05,
      "loss": 0.0517,
      "step": 7357
    },
    {
      "epoch": 13.57617728531856,
      "grad_norm": 0.40392178297042847,
      "learning_rate": 2.2065000000000003e-05,
      "loss": 0.0841,
      "step": 7358
    },
    {
      "epoch": 13.578024007386889,
      "grad_norm": 0.5108973979949951,
      "learning_rate": 2.2068000000000003e-05,
      "loss": 0.0392,
      "step": 7359
    },
    {
      "epoch": 13.579870729455218,
      "grad_norm": 0.4702042043209076,
      "learning_rate": 2.2071000000000003e-05,
      "loss": 0.0296,
      "step": 7360
    },
    {
      "epoch": 13.581717451523545,
      "grad_norm": 0.4801340103149414,
      "learning_rate": 2.2074000000000002e-05,
      "loss": 0.0325,
      "step": 7361
    },
    {
      "epoch": 13.583564173591874,
      "grad_norm": 0.4173518419265747,
      "learning_rate": 2.2077e-05,
      "loss": 0.0225,
      "step": 7362
    },
    {
      "epoch": 13.585410895660203,
      "grad_norm": 0.4327594041824341,
      "learning_rate": 2.208e-05,
      "loss": 0.0191,
      "step": 7363
    },
    {
      "epoch": 13.587257617728532,
      "grad_norm": 0.5534989237785339,
      "learning_rate": 2.2083e-05,
      "loss": 0.038,
      "step": 7364
    },
    {
      "epoch": 13.58910433979686,
      "grad_norm": 0.19051194190979004,
      "learning_rate": 2.2085999999999998e-05,
      "loss": 0.0087,
      "step": 7365
    },
    {
      "epoch": 13.590951061865189,
      "grad_norm": 0.4202093482017517,
      "learning_rate": 2.2088999999999998e-05,
      "loss": 0.0308,
      "step": 7366
    },
    {
      "epoch": 13.592797783933518,
      "grad_norm": 0.7560642957687378,
      "learning_rate": 2.2092e-05,
      "loss": 0.02,
      "step": 7367
    },
    {
      "epoch": 13.594644506001847,
      "grad_norm": 0.2742268145084381,
      "learning_rate": 2.2095e-05,
      "loss": 0.0078,
      "step": 7368
    },
    {
      "epoch": 13.596491228070175,
      "grad_norm": 0.5704046487808228,
      "learning_rate": 2.2098e-05,
      "loss": 0.0199,
      "step": 7369
    },
    {
      "epoch": 13.598337950138504,
      "grad_norm": 0.6069647073745728,
      "learning_rate": 2.2101e-05,
      "loss": 0.0219,
      "step": 7370
    },
    {
      "epoch": 13.600184672206833,
      "grad_norm": 0.44164136052131653,
      "learning_rate": 2.2104e-05,
      "loss": 0.0419,
      "step": 7371
    },
    {
      "epoch": 13.602031394275162,
      "grad_norm": 0.48983797430992126,
      "learning_rate": 2.2107e-05,
      "loss": 0.026,
      "step": 7372
    },
    {
      "epoch": 13.603878116343491,
      "grad_norm": 0.32292410731315613,
      "learning_rate": 2.211e-05,
      "loss": 0.0121,
      "step": 7373
    },
    {
      "epoch": 13.605724838411819,
      "grad_norm": 1.0277533531188965,
      "learning_rate": 2.2113e-05,
      "loss": 0.0186,
      "step": 7374
    },
    {
      "epoch": 13.607571560480148,
      "grad_norm": 0.6106177568435669,
      "learning_rate": 2.2116e-05,
      "loss": 0.0234,
      "step": 7375
    },
    {
      "epoch": 13.609418282548477,
      "grad_norm": 0.7649809122085571,
      "learning_rate": 2.2119e-05,
      "loss": 0.0782,
      "step": 7376
    },
    {
      "epoch": 13.611265004616806,
      "grad_norm": 0.5042434930801392,
      "learning_rate": 2.2122000000000003e-05,
      "loss": 0.0448,
      "step": 7377
    },
    {
      "epoch": 13.613111726685133,
      "grad_norm": 0.4876888394355774,
      "learning_rate": 2.2125000000000002e-05,
      "loss": 0.0156,
      "step": 7378
    },
    {
      "epoch": 13.614958448753463,
      "grad_norm": 0.40460675954818726,
      "learning_rate": 2.2128000000000002e-05,
      "loss": 0.0153,
      "step": 7379
    },
    {
      "epoch": 13.616805170821792,
      "grad_norm": 0.21005268394947052,
      "learning_rate": 2.2131000000000002e-05,
      "loss": 0.0066,
      "step": 7380
    },
    {
      "epoch": 13.61865189289012,
      "grad_norm": 0.43871554732322693,
      "learning_rate": 2.2134000000000002e-05,
      "loss": 0.0267,
      "step": 7381
    },
    {
      "epoch": 13.620498614958448,
      "grad_norm": 0.3028872013092041,
      "learning_rate": 2.2137e-05,
      "loss": 0.0171,
      "step": 7382
    },
    {
      "epoch": 13.622345337026777,
      "grad_norm": 0.775253415107727,
      "learning_rate": 2.214e-05,
      "loss": 0.0185,
      "step": 7383
    },
    {
      "epoch": 13.624192059095106,
      "grad_norm": 1.5879731178283691,
      "learning_rate": 2.2143e-05,
      "loss": 0.0149,
      "step": 7384
    },
    {
      "epoch": 13.626038781163436,
      "grad_norm": 0.7068104147911072,
      "learning_rate": 2.2146e-05,
      "loss": 0.0302,
      "step": 7385
    },
    {
      "epoch": 13.627885503231763,
      "grad_norm": 0.5778458118438721,
      "learning_rate": 2.2149e-05,
      "loss": 0.0178,
      "step": 7386
    },
    {
      "epoch": 13.629732225300092,
      "grad_norm": 0.9886907935142517,
      "learning_rate": 2.2151999999999997e-05,
      "loss": 0.0136,
      "step": 7387
    },
    {
      "epoch": 13.631578947368421,
      "grad_norm": 0.5738271474838257,
      "learning_rate": 2.2155e-05,
      "loss": 0.0197,
      "step": 7388
    },
    {
      "epoch": 13.63342566943675,
      "grad_norm": 0.7942144870758057,
      "learning_rate": 2.2158e-05,
      "loss": 0.0268,
      "step": 7389
    },
    {
      "epoch": 13.635272391505078,
      "grad_norm": 0.7159650325775146,
      "learning_rate": 2.2161e-05,
      "loss": 0.0259,
      "step": 7390
    },
    {
      "epoch": 13.637119113573407,
      "grad_norm": 0.3914990723133087,
      "learning_rate": 2.2164e-05,
      "loss": 0.017,
      "step": 7391
    },
    {
      "epoch": 13.638965835641736,
      "grad_norm": 0.36542847752571106,
      "learning_rate": 2.2167e-05,
      "loss": 0.0195,
      "step": 7392
    },
    {
      "epoch": 13.640812557710065,
      "grad_norm": 0.48772743344306946,
      "learning_rate": 2.217e-05,
      "loss": 0.0335,
      "step": 7393
    },
    {
      "epoch": 13.642659279778393,
      "grad_norm": 0.36953237652778625,
      "learning_rate": 2.2173e-05,
      "loss": 0.0122,
      "step": 7394
    },
    {
      "epoch": 13.644506001846722,
      "grad_norm": 0.5144199132919312,
      "learning_rate": 2.2176e-05,
      "loss": 0.0246,
      "step": 7395
    },
    {
      "epoch": 13.64635272391505,
      "grad_norm": 0.517334520816803,
      "learning_rate": 2.2179e-05,
      "loss": 0.0179,
      "step": 7396
    },
    {
      "epoch": 13.64819944598338,
      "grad_norm": 1.3453407287597656,
      "learning_rate": 2.2182e-05,
      "loss": 0.2604,
      "step": 7397
    },
    {
      "epoch": 13.65004616805171,
      "grad_norm": 0.7100138068199158,
      "learning_rate": 2.2185000000000002e-05,
      "loss": 0.2018,
      "step": 7398
    },
    {
      "epoch": 13.651892890120036,
      "grad_norm": 0.8211055397987366,
      "learning_rate": 2.2188e-05,
      "loss": 0.233,
      "step": 7399
    },
    {
      "epoch": 13.653739612188366,
      "grad_norm": 0.592019259929657,
      "learning_rate": 2.2191e-05,
      "loss": 0.127,
      "step": 7400
    },
    {
      "epoch": 13.655586334256695,
      "grad_norm": 0.7646599411964417,
      "learning_rate": 2.2194e-05,
      "loss": 0.14,
      "step": 7401
    },
    {
      "epoch": 13.657433056325024,
      "grad_norm": 0.44219911098480225,
      "learning_rate": 2.2197e-05,
      "loss": 0.084,
      "step": 7402
    },
    {
      "epoch": 13.659279778393351,
      "grad_norm": 0.5861028432846069,
      "learning_rate": 2.22e-05,
      "loss": 0.0807,
      "step": 7403
    },
    {
      "epoch": 13.66112650046168,
      "grad_norm": 0.6860338449478149,
      "learning_rate": 2.2203e-05,
      "loss": 0.0879,
      "step": 7404
    },
    {
      "epoch": 13.66297322253001,
      "grad_norm": 0.733318567276001,
      "learning_rate": 2.2206e-05,
      "loss": 0.0745,
      "step": 7405
    },
    {
      "epoch": 13.664819944598339,
      "grad_norm": 0.4197256863117218,
      "learning_rate": 2.2209e-05,
      "loss": 0.052,
      "step": 7406
    },
    {
      "epoch": 13.666666666666666,
      "grad_norm": 0.4183002710342407,
      "learning_rate": 2.2212e-05,
      "loss": 0.0516,
      "step": 7407
    },
    {
      "epoch": 13.668513388734995,
      "grad_norm": 0.5573427677154541,
      "learning_rate": 2.2215000000000003e-05,
      "loss": 0.0903,
      "step": 7408
    },
    {
      "epoch": 13.670360110803324,
      "grad_norm": 0.6062901616096497,
      "learning_rate": 2.2218000000000003e-05,
      "loss": 0.0867,
      "step": 7409
    },
    {
      "epoch": 13.672206832871654,
      "grad_norm": 0.3610090911388397,
      "learning_rate": 2.2221000000000003e-05,
      "loss": 0.0313,
      "step": 7410
    },
    {
      "epoch": 13.67405355493998,
      "grad_norm": 0.45338574051856995,
      "learning_rate": 2.2224e-05,
      "loss": 0.0284,
      "step": 7411
    },
    {
      "epoch": 13.67590027700831,
      "grad_norm": 0.35504356026649475,
      "learning_rate": 2.2227e-05,
      "loss": 0.0506,
      "step": 7412
    },
    {
      "epoch": 13.67774699907664,
      "grad_norm": 0.5283374786376953,
      "learning_rate": 2.223e-05,
      "loss": 0.0264,
      "step": 7413
    },
    {
      "epoch": 13.679593721144968,
      "grad_norm": 0.9853783249855042,
      "learning_rate": 2.2233e-05,
      "loss": 0.0334,
      "step": 7414
    },
    {
      "epoch": 13.681440443213296,
      "grad_norm": 0.9584597945213318,
      "learning_rate": 2.2236e-05,
      "loss": 0.074,
      "step": 7415
    },
    {
      "epoch": 13.683287165281625,
      "grad_norm": 0.34112200140953064,
      "learning_rate": 2.2239e-05,
      "loss": 0.0306,
      "step": 7416
    },
    {
      "epoch": 13.685133887349954,
      "grad_norm": 0.7212913036346436,
      "learning_rate": 2.2241999999999998e-05,
      "loss": 0.0439,
      "step": 7417
    },
    {
      "epoch": 13.686980609418283,
      "grad_norm": 0.6430798768997192,
      "learning_rate": 2.2245e-05,
      "loss": 0.026,
      "step": 7418
    },
    {
      "epoch": 13.68882733148661,
      "grad_norm": 0.4628603756427765,
      "learning_rate": 2.2248e-05,
      "loss": 0.0142,
      "step": 7419
    },
    {
      "epoch": 13.69067405355494,
      "grad_norm": 1.0656933784484863,
      "learning_rate": 2.2251e-05,
      "loss": 0.0193,
      "step": 7420
    },
    {
      "epoch": 13.692520775623269,
      "grad_norm": 0.3730289041996002,
      "learning_rate": 2.2254e-05,
      "loss": 0.0253,
      "step": 7421
    },
    {
      "epoch": 13.694367497691598,
      "grad_norm": 0.3692324161529541,
      "learning_rate": 2.2257e-05,
      "loss": 0.0121,
      "step": 7422
    },
    {
      "epoch": 13.696214219759927,
      "grad_norm": 0.2735585570335388,
      "learning_rate": 2.226e-05,
      "loss": 0.0127,
      "step": 7423
    },
    {
      "epoch": 13.698060941828254,
      "grad_norm": 0.1949063241481781,
      "learning_rate": 2.2263e-05,
      "loss": 0.0086,
      "step": 7424
    },
    {
      "epoch": 13.699907663896584,
      "grad_norm": 0.4594740867614746,
      "learning_rate": 2.2266e-05,
      "loss": 0.018,
      "step": 7425
    },
    {
      "epoch": 13.701754385964913,
      "grad_norm": 0.3455602824687958,
      "learning_rate": 2.2269e-05,
      "loss": 0.0154,
      "step": 7426
    },
    {
      "epoch": 13.703601108033242,
      "grad_norm": 0.4273536503314972,
      "learning_rate": 2.2272e-05,
      "loss": 0.0141,
      "step": 7427
    },
    {
      "epoch": 13.70544783010157,
      "grad_norm": 0.5404307246208191,
      "learning_rate": 2.2275000000000003e-05,
      "loss": 0.0281,
      "step": 7428
    },
    {
      "epoch": 13.707294552169898,
      "grad_norm": 0.8004701137542725,
      "learning_rate": 2.2278000000000003e-05,
      "loss": 0.0256,
      "step": 7429
    },
    {
      "epoch": 13.709141274238227,
      "grad_norm": 0.8993343114852905,
      "learning_rate": 2.2281000000000002e-05,
      "loss": 0.0277,
      "step": 7430
    },
    {
      "epoch": 13.710987996306557,
      "grad_norm": 0.4021378755569458,
      "learning_rate": 2.2284000000000002e-05,
      "loss": 0.0253,
      "step": 7431
    },
    {
      "epoch": 13.712834718374884,
      "grad_norm": 0.37600457668304443,
      "learning_rate": 2.2287000000000002e-05,
      "loss": 0.0132,
      "step": 7432
    },
    {
      "epoch": 13.714681440443213,
      "grad_norm": 0.4592324495315552,
      "learning_rate": 2.2290000000000002e-05,
      "loss": 0.0146,
      "step": 7433
    },
    {
      "epoch": 13.716528162511542,
      "grad_norm": 0.3562804162502289,
      "learning_rate": 2.2293e-05,
      "loss": 0.0447,
      "step": 7434
    },
    {
      "epoch": 13.718374884579871,
      "grad_norm": 0.5064547657966614,
      "learning_rate": 2.2296e-05,
      "loss": 0.0174,
      "step": 7435
    },
    {
      "epoch": 13.720221606648199,
      "grad_norm": 0.300818532705307,
      "learning_rate": 2.2298999999999998e-05,
      "loss": 0.0202,
      "step": 7436
    },
    {
      "epoch": 13.722068328716528,
      "grad_norm": 0.4283485412597656,
      "learning_rate": 2.2301999999999998e-05,
      "loss": 0.0179,
      "step": 7437
    },
    {
      "epoch": 13.723915050784857,
      "grad_norm": 0.3496239483356476,
      "learning_rate": 2.2305e-05,
      "loss": 0.0069,
      "step": 7438
    },
    {
      "epoch": 13.725761772853186,
      "grad_norm": 0.7668992280960083,
      "learning_rate": 2.2308e-05,
      "loss": 0.0335,
      "step": 7439
    },
    {
      "epoch": 13.727608494921514,
      "grad_norm": 0.43262067437171936,
      "learning_rate": 2.2311e-05,
      "loss": 0.026,
      "step": 7440
    },
    {
      "epoch": 13.729455216989843,
      "grad_norm": 0.33174577355384827,
      "learning_rate": 2.2314e-05,
      "loss": 0.0163,
      "step": 7441
    },
    {
      "epoch": 13.731301939058172,
      "grad_norm": 0.4642624258995056,
      "learning_rate": 2.2317e-05,
      "loss": 0.0273,
      "step": 7442
    },
    {
      "epoch": 13.733148661126501,
      "grad_norm": 0.8481758236885071,
      "learning_rate": 2.232e-05,
      "loss": 0.0294,
      "step": 7443
    },
    {
      "epoch": 13.734995383194828,
      "grad_norm": 1.4036601781845093,
      "learning_rate": 2.2323e-05,
      "loss": 0.0194,
      "step": 7444
    },
    {
      "epoch": 13.736842105263158,
      "grad_norm": 0.8418857455253601,
      "learning_rate": 2.2326e-05,
      "loss": 0.0256,
      "step": 7445
    },
    {
      "epoch": 13.738688827331487,
      "grad_norm": 0.9245105981826782,
      "learning_rate": 2.2329e-05,
      "loss": 0.0291,
      "step": 7446
    },
    {
      "epoch": 13.740535549399816,
      "grad_norm": 0.9908247590065002,
      "learning_rate": 2.2332e-05,
      "loss": 0.2072,
      "step": 7447
    },
    {
      "epoch": 13.742382271468145,
      "grad_norm": 0.8401052355766296,
      "learning_rate": 2.2335000000000002e-05,
      "loss": 0.1964,
      "step": 7448
    },
    {
      "epoch": 13.744228993536472,
      "grad_norm": 0.6335350275039673,
      "learning_rate": 2.2338000000000002e-05,
      "loss": 0.1581,
      "step": 7449
    },
    {
      "epoch": 13.746075715604801,
      "grad_norm": 0.9489895701408386,
      "learning_rate": 2.2341000000000002e-05,
      "loss": 0.1505,
      "step": 7450
    },
    {
      "epoch": 13.74792243767313,
      "grad_norm": 0.5921657681465149,
      "learning_rate": 2.2344e-05,
      "loss": 0.148,
      "step": 7451
    },
    {
      "epoch": 13.749769159741458,
      "grad_norm": 0.674635112285614,
      "learning_rate": 2.2347e-05,
      "loss": 0.117,
      "step": 7452
    },
    {
      "epoch": 13.751615881809787,
      "grad_norm": 0.47981318831443787,
      "learning_rate": 2.235e-05,
      "loss": 0.0772,
      "step": 7453
    },
    {
      "epoch": 13.753462603878116,
      "grad_norm": 0.6606249809265137,
      "learning_rate": 2.2353e-05,
      "loss": 0.0903,
      "step": 7454
    },
    {
      "epoch": 13.755309325946445,
      "grad_norm": 0.8978612422943115,
      "learning_rate": 2.2356e-05,
      "loss": 0.1229,
      "step": 7455
    },
    {
      "epoch": 13.757156048014775,
      "grad_norm": 0.4498438239097595,
      "learning_rate": 2.2359e-05,
      "loss": 0.0753,
      "step": 7456
    },
    {
      "epoch": 13.759002770083102,
      "grad_norm": 1.2396608591079712,
      "learning_rate": 2.2362e-05,
      "loss": 0.1475,
      "step": 7457
    },
    {
      "epoch": 13.760849492151431,
      "grad_norm": 0.9009991884231567,
      "learning_rate": 2.2365000000000004e-05,
      "loss": 0.1124,
      "step": 7458
    },
    {
      "epoch": 13.76269621421976,
      "grad_norm": 0.5656141638755798,
      "learning_rate": 2.2368000000000003e-05,
      "loss": 0.0722,
      "step": 7459
    },
    {
      "epoch": 13.76454293628809,
      "grad_norm": 0.42123159766197205,
      "learning_rate": 2.2371e-05,
      "loss": 0.0566,
      "step": 7460
    },
    {
      "epoch": 13.766389658356417,
      "grad_norm": 0.2959066927433014,
      "learning_rate": 2.2374e-05,
      "loss": 0.0391,
      "step": 7461
    },
    {
      "epoch": 13.768236380424746,
      "grad_norm": 0.43474075198173523,
      "learning_rate": 2.2377e-05,
      "loss": 0.0366,
      "step": 7462
    },
    {
      "epoch": 13.770083102493075,
      "grad_norm": 0.5660763382911682,
      "learning_rate": 2.238e-05,
      "loss": 0.018,
      "step": 7463
    },
    {
      "epoch": 13.771929824561404,
      "grad_norm": 0.48695066571235657,
      "learning_rate": 2.2383e-05,
      "loss": 0.0308,
      "step": 7464
    },
    {
      "epoch": 13.773776546629731,
      "grad_norm": 0.21219317615032196,
      "learning_rate": 2.2386e-05,
      "loss": 0.0117,
      "step": 7465
    },
    {
      "epoch": 13.77562326869806,
      "grad_norm": 1.1667358875274658,
      "learning_rate": 2.2389e-05,
      "loss": 0.0448,
      "step": 7466
    },
    {
      "epoch": 13.77746999076639,
      "grad_norm": 0.47652381658554077,
      "learning_rate": 2.2392e-05,
      "loss": 0.0472,
      "step": 7467
    },
    {
      "epoch": 13.779316712834719,
      "grad_norm": 0.35283544659614563,
      "learning_rate": 2.2395e-05,
      "loss": 0.0243,
      "step": 7468
    },
    {
      "epoch": 13.781163434903046,
      "grad_norm": 0.2358464151620865,
      "learning_rate": 2.2398e-05,
      "loss": 0.0148,
      "step": 7469
    },
    {
      "epoch": 13.783010156971375,
      "grad_norm": 0.982874870300293,
      "learning_rate": 2.2401e-05,
      "loss": 0.0433,
      "step": 7470
    },
    {
      "epoch": 13.784856879039705,
      "grad_norm": 0.3252042829990387,
      "learning_rate": 2.2404e-05,
      "loss": 0.0167,
      "step": 7471
    },
    {
      "epoch": 13.786703601108034,
      "grad_norm": 0.3061884939670563,
      "learning_rate": 2.2407e-05,
      "loss": 0.0227,
      "step": 7472
    },
    {
      "epoch": 13.788550323176363,
      "grad_norm": 0.32022055983543396,
      "learning_rate": 2.241e-05,
      "loss": 0.047,
      "step": 7473
    },
    {
      "epoch": 13.79039704524469,
      "grad_norm": 0.31437739729881287,
      "learning_rate": 2.2413e-05,
      "loss": 0.017,
      "step": 7474
    },
    {
      "epoch": 13.79224376731302,
      "grad_norm": 0.17899277806282043,
      "learning_rate": 2.2416e-05,
      "loss": 0.0158,
      "step": 7475
    },
    {
      "epoch": 13.794090489381349,
      "grad_norm": 0.4037555158138275,
      "learning_rate": 2.2419e-05,
      "loss": 0.0195,
      "step": 7476
    },
    {
      "epoch": 13.795937211449676,
      "grad_norm": 3.9704298973083496,
      "learning_rate": 2.2422e-05,
      "loss": 0.0213,
      "step": 7477
    },
    {
      "epoch": 13.797783933518005,
      "grad_norm": 0.41569051146507263,
      "learning_rate": 2.2425000000000003e-05,
      "loss": 0.0202,
      "step": 7478
    },
    {
      "epoch": 13.799630655586334,
      "grad_norm": 0.3402164876461029,
      "learning_rate": 2.2428000000000003e-05,
      "loss": 0.0166,
      "step": 7479
    },
    {
      "epoch": 13.801477377654663,
      "grad_norm": 0.5518233180046082,
      "learning_rate": 2.2431000000000003e-05,
      "loss": 0.0183,
      "step": 7480
    },
    {
      "epoch": 13.803324099722992,
      "grad_norm": 0.28906118869781494,
      "learning_rate": 2.2434000000000002e-05,
      "loss": 0.016,
      "step": 7481
    },
    {
      "epoch": 13.80517082179132,
      "grad_norm": 0.4287084937095642,
      "learning_rate": 2.2437000000000002e-05,
      "loss": 0.016,
      "step": 7482
    },
    {
      "epoch": 13.807017543859649,
      "grad_norm": 0.455166757106781,
      "learning_rate": 2.2440000000000002e-05,
      "loss": 0.0202,
      "step": 7483
    },
    {
      "epoch": 13.808864265927978,
      "grad_norm": 0.2282412201166153,
      "learning_rate": 2.2443000000000002e-05,
      "loss": 0.0142,
      "step": 7484
    },
    {
      "epoch": 13.810710987996307,
      "grad_norm": 0.37432539463043213,
      "learning_rate": 2.2445999999999998e-05,
      "loss": 0.0164,
      "step": 7485
    },
    {
      "epoch": 13.812557710064635,
      "grad_norm": 0.870344877243042,
      "learning_rate": 2.2448999999999998e-05,
      "loss": 0.0241,
      "step": 7486
    },
    {
      "epoch": 13.814404432132964,
      "grad_norm": 0.2512337267398834,
      "learning_rate": 2.2451999999999998e-05,
      "loss": 0.012,
      "step": 7487
    },
    {
      "epoch": 13.816251154201293,
      "grad_norm": 0.25623369216918945,
      "learning_rate": 2.2455e-05,
      "loss": 0.0115,
      "step": 7488
    },
    {
      "epoch": 13.818097876269622,
      "grad_norm": 0.5129430890083313,
      "learning_rate": 2.2458e-05,
      "loss": 0.0146,
      "step": 7489
    },
    {
      "epoch": 13.81994459833795,
      "grad_norm": 0.5183201432228088,
      "learning_rate": 2.2461e-05,
      "loss": 0.0184,
      "step": 7490
    },
    {
      "epoch": 13.821791320406279,
      "grad_norm": 0.6669928431510925,
      "learning_rate": 2.2464e-05,
      "loss": 0.0239,
      "step": 7491
    },
    {
      "epoch": 13.823638042474608,
      "grad_norm": 0.3957689106464386,
      "learning_rate": 2.2467e-05,
      "loss": 0.011,
      "step": 7492
    },
    {
      "epoch": 13.825484764542937,
      "grad_norm": 0.4440816044807434,
      "learning_rate": 2.247e-05,
      "loss": 0.0125,
      "step": 7493
    },
    {
      "epoch": 13.827331486611264,
      "grad_norm": 0.4568639099597931,
      "learning_rate": 2.2473e-05,
      "loss": 0.0172,
      "step": 7494
    },
    {
      "epoch": 13.829178208679593,
      "grad_norm": 0.7566015720367432,
      "learning_rate": 2.2476e-05,
      "loss": 0.0202,
      "step": 7495
    },
    {
      "epoch": 13.831024930747922,
      "grad_norm": 0.6292212605476379,
      "learning_rate": 2.2479e-05,
      "loss": 0.0255,
      "step": 7496
    },
    {
      "epoch": 13.832871652816252,
      "grad_norm": 0.9029131531715393,
      "learning_rate": 2.2482e-05,
      "loss": 0.2179,
      "step": 7497
    },
    {
      "epoch": 13.83471837488458,
      "grad_norm": 0.6349799633026123,
      "learning_rate": 2.2485000000000002e-05,
      "loss": 0.1667,
      "step": 7498
    },
    {
      "epoch": 13.836565096952908,
      "grad_norm": 0.5616112351417542,
      "learning_rate": 2.2488000000000002e-05,
      "loss": 0.1714,
      "step": 7499
    },
    {
      "epoch": 13.838411819021237,
      "grad_norm": 0.6807947158813477,
      "learning_rate": 2.2491000000000002e-05,
      "loss": 0.1424,
      "step": 7500
    },
    {
      "epoch": 13.840258541089566,
      "grad_norm": 0.5123181343078613,
      "learning_rate": 2.2494000000000002e-05,
      "loss": 0.0932,
      "step": 7501
    },
    {
      "epoch": 13.842105263157894,
      "grad_norm": 0.6139744520187378,
      "learning_rate": 2.2497e-05,
      "loss": 0.096,
      "step": 7502
    },
    {
      "epoch": 13.843951985226223,
      "grad_norm": 0.7582729458808899,
      "learning_rate": 2.25e-05,
      "loss": 0.1449,
      "step": 7503
    },
    {
      "epoch": 13.845798707294552,
      "grad_norm": 0.5244767069816589,
      "learning_rate": 2.2503e-05,
      "loss": 0.0888,
      "step": 7504
    },
    {
      "epoch": 13.847645429362881,
      "grad_norm": 0.685250997543335,
      "learning_rate": 2.2506e-05,
      "loss": 0.0891,
      "step": 7505
    },
    {
      "epoch": 13.84949215143121,
      "grad_norm": 0.5729045271873474,
      "learning_rate": 2.2509e-05,
      "loss": 0.0945,
      "step": 7506
    },
    {
      "epoch": 13.851338873499538,
      "grad_norm": 0.76090407371521,
      "learning_rate": 2.2512e-05,
      "loss": 0.0704,
      "step": 7507
    },
    {
      "epoch": 13.853185595567867,
      "grad_norm": 0.9119696617126465,
      "learning_rate": 2.2515e-05,
      "loss": 0.0903,
      "step": 7508
    },
    {
      "epoch": 13.855032317636196,
      "grad_norm": 0.2945483922958374,
      "learning_rate": 2.2518e-05,
      "loss": 0.0369,
      "step": 7509
    },
    {
      "epoch": 13.856879039704525,
      "grad_norm": 0.7408194541931152,
      "learning_rate": 2.2521e-05,
      "loss": 0.133,
      "step": 7510
    },
    {
      "epoch": 13.858725761772853,
      "grad_norm": 0.37890946865081787,
      "learning_rate": 2.2524e-05,
      "loss": 0.0345,
      "step": 7511
    },
    {
      "epoch": 13.860572483841182,
      "grad_norm": 0.49662187695503235,
      "learning_rate": 2.2527e-05,
      "loss": 0.0414,
      "step": 7512
    },
    {
      "epoch": 13.86241920590951,
      "grad_norm": 0.4859124422073364,
      "learning_rate": 2.253e-05,
      "loss": 0.0302,
      "step": 7513
    },
    {
      "epoch": 13.86426592797784,
      "grad_norm": 0.34445253014564514,
      "learning_rate": 2.2533e-05,
      "loss": 0.0171,
      "step": 7514
    },
    {
      "epoch": 13.866112650046167,
      "grad_norm": 0.549643337726593,
      "learning_rate": 2.2536e-05,
      "loss": 0.0323,
      "step": 7515
    },
    {
      "epoch": 13.867959372114496,
      "grad_norm": 0.4903494417667389,
      "learning_rate": 2.2539e-05,
      "loss": 0.0471,
      "step": 7516
    },
    {
      "epoch": 13.869806094182826,
      "grad_norm": 0.3244134187698364,
      "learning_rate": 2.2542e-05,
      "loss": 0.0183,
      "step": 7517
    },
    {
      "epoch": 13.871652816251155,
      "grad_norm": 0.3773230016231537,
      "learning_rate": 2.2545e-05,
      "loss": 0.0282,
      "step": 7518
    },
    {
      "epoch": 13.873499538319482,
      "grad_norm": 0.5452808737754822,
      "learning_rate": 2.2548e-05,
      "loss": 0.0256,
      "step": 7519
    },
    {
      "epoch": 13.875346260387811,
      "grad_norm": 0.48886626958847046,
      "learning_rate": 2.2551e-05,
      "loss": 0.0276,
      "step": 7520
    },
    {
      "epoch": 13.87719298245614,
      "grad_norm": 0.47622591257095337,
      "learning_rate": 2.2554e-05,
      "loss": 0.0181,
      "step": 7521
    },
    {
      "epoch": 13.87903970452447,
      "grad_norm": 0.4958360195159912,
      "learning_rate": 2.2557e-05,
      "loss": 0.0228,
      "step": 7522
    },
    {
      "epoch": 13.880886426592799,
      "grad_norm": 0.3116703927516937,
      "learning_rate": 2.256e-05,
      "loss": 0.0171,
      "step": 7523
    },
    {
      "epoch": 13.882733148661126,
      "grad_norm": 0.7549593448638916,
      "learning_rate": 2.2563e-05,
      "loss": 0.0353,
      "step": 7524
    },
    {
      "epoch": 13.884579870729455,
      "grad_norm": 0.2600730359554291,
      "learning_rate": 2.2566e-05,
      "loss": 0.0107,
      "step": 7525
    },
    {
      "epoch": 13.886426592797784,
      "grad_norm": 0.1800379902124405,
      "learning_rate": 2.2569e-05,
      "loss": 0.0117,
      "step": 7526
    },
    {
      "epoch": 13.888273314866112,
      "grad_norm": 0.3467258810997009,
      "learning_rate": 2.2572e-05,
      "loss": 0.0173,
      "step": 7527
    },
    {
      "epoch": 13.89012003693444,
      "grad_norm": 0.45100924372673035,
      "learning_rate": 2.2575e-05,
      "loss": 0.0207,
      "step": 7528
    },
    {
      "epoch": 13.89196675900277,
      "grad_norm": 0.4383341670036316,
      "learning_rate": 2.2578000000000003e-05,
      "loss": 0.0171,
      "step": 7529
    },
    {
      "epoch": 13.8938134810711,
      "grad_norm": 0.2968244254589081,
      "learning_rate": 2.2581000000000003e-05,
      "loss": 0.016,
      "step": 7530
    },
    {
      "epoch": 13.895660203139428,
      "grad_norm": 0.7290030121803284,
      "learning_rate": 2.2584000000000003e-05,
      "loss": 0.0268,
      "step": 7531
    },
    {
      "epoch": 13.897506925207756,
      "grad_norm": 0.5313127636909485,
      "learning_rate": 2.2587000000000002e-05,
      "loss": 0.0235,
      "step": 7532
    },
    {
      "epoch": 13.899353647276085,
      "grad_norm": 0.7194654941558838,
      "learning_rate": 2.2590000000000002e-05,
      "loss": 0.0204,
      "step": 7533
    },
    {
      "epoch": 13.901200369344414,
      "grad_norm": 0.8485562801361084,
      "learning_rate": 2.2593e-05,
      "loss": 0.013,
      "step": 7534
    },
    {
      "epoch": 13.903047091412743,
      "grad_norm": 0.4126611649990082,
      "learning_rate": 2.2596e-05,
      "loss": 0.0223,
      "step": 7535
    },
    {
      "epoch": 13.90489381348107,
      "grad_norm": 0.5421310663223267,
      "learning_rate": 2.2598999999999998e-05,
      "loss": 0.0301,
      "step": 7536
    },
    {
      "epoch": 13.9067405355494,
      "grad_norm": 0.5941007137298584,
      "learning_rate": 2.2601999999999998e-05,
      "loss": 0.027,
      "step": 7537
    },
    {
      "epoch": 13.908587257617729,
      "grad_norm": 0.5690109729766846,
      "learning_rate": 2.2604999999999998e-05,
      "loss": 0.035,
      "step": 7538
    },
    {
      "epoch": 13.910433979686058,
      "grad_norm": 0.3790777027606964,
      "learning_rate": 2.2608e-05,
      "loss": 0.0122,
      "step": 7539
    },
    {
      "epoch": 13.912280701754385,
      "grad_norm": 0.5342270731925964,
      "learning_rate": 2.2611e-05,
      "loss": 0.028,
      "step": 7540
    },
    {
      "epoch": 13.914127423822714,
      "grad_norm": 0.4844706952571869,
      "learning_rate": 2.2614e-05,
      "loss": 0.0173,
      "step": 7541
    },
    {
      "epoch": 13.915974145891044,
      "grad_norm": 0.4223809838294983,
      "learning_rate": 2.2617e-05,
      "loss": 0.0214,
      "step": 7542
    },
    {
      "epoch": 13.917820867959373,
      "grad_norm": 0.4551135301589966,
      "learning_rate": 2.262e-05,
      "loss": 0.0239,
      "step": 7543
    },
    {
      "epoch": 13.9196675900277,
      "grad_norm": 0.4099441468715668,
      "learning_rate": 2.2623e-05,
      "loss": 0.0251,
      "step": 7544
    },
    {
      "epoch": 13.92151431209603,
      "grad_norm": 0.31305715441703796,
      "learning_rate": 2.2626e-05,
      "loss": 0.0124,
      "step": 7545
    },
    {
      "epoch": 13.923361034164358,
      "grad_norm": 0.6895094513893127,
      "learning_rate": 2.2629e-05,
      "loss": 0.0132,
      "step": 7546
    },
    {
      "epoch": 13.925207756232687,
      "grad_norm": 1.553663730621338,
      "learning_rate": 2.2632e-05,
      "loss": 0.3124,
      "step": 7547
    },
    {
      "epoch": 13.927054478301017,
      "grad_norm": 0.7344269752502441,
      "learning_rate": 2.2635e-05,
      "loss": 0.1667,
      "step": 7548
    },
    {
      "epoch": 13.928901200369344,
      "grad_norm": 0.867383599281311,
      "learning_rate": 2.2638000000000002e-05,
      "loss": 0.1546,
      "step": 7549
    },
    {
      "epoch": 13.930747922437673,
      "grad_norm": 0.5955139398574829,
      "learning_rate": 2.2641000000000002e-05,
      "loss": 0.1512,
      "step": 7550
    },
    {
      "epoch": 13.932594644506002,
      "grad_norm": 0.5191391110420227,
      "learning_rate": 2.2644000000000002e-05,
      "loss": 0.1248,
      "step": 7551
    },
    {
      "epoch": 13.93444136657433,
      "grad_norm": 0.9385023713111877,
      "learning_rate": 2.2647000000000002e-05,
      "loss": 0.1216,
      "step": 7552
    },
    {
      "epoch": 13.936288088642659,
      "grad_norm": 0.6964015960693359,
      "learning_rate": 2.265e-05,
      "loss": 0.1049,
      "step": 7553
    },
    {
      "epoch": 13.938134810710988,
      "grad_norm": 0.7118395566940308,
      "learning_rate": 2.2653e-05,
      "loss": 0.0931,
      "step": 7554
    },
    {
      "epoch": 13.939981532779317,
      "grad_norm": 0.630588948726654,
      "learning_rate": 2.2656e-05,
      "loss": 0.0625,
      "step": 7555
    },
    {
      "epoch": 13.941828254847646,
      "grad_norm": 0.5488806366920471,
      "learning_rate": 2.2659e-05,
      "loss": 0.1024,
      "step": 7556
    },
    {
      "epoch": 13.943674976915974,
      "grad_norm": 0.36687353253364563,
      "learning_rate": 2.2662e-05,
      "loss": 0.0363,
      "step": 7557
    },
    {
      "epoch": 13.945521698984303,
      "grad_norm": 0.7587217688560486,
      "learning_rate": 2.2665e-05,
      "loss": 0.0511,
      "step": 7558
    },
    {
      "epoch": 13.947368421052632,
      "grad_norm": 0.7604022026062012,
      "learning_rate": 2.2668e-05,
      "loss": 0.1023,
      "step": 7559
    },
    {
      "epoch": 13.949215143120961,
      "grad_norm": 0.37718528509140015,
      "learning_rate": 2.2671e-05,
      "loss": 0.0245,
      "step": 7560
    },
    {
      "epoch": 13.951061865189288,
      "grad_norm": 0.709337055683136,
      "learning_rate": 2.2674e-05,
      "loss": 0.0252,
      "step": 7561
    },
    {
      "epoch": 13.952908587257618,
      "grad_norm": 0.26128262281417847,
      "learning_rate": 2.2677e-05,
      "loss": 0.0245,
      "step": 7562
    },
    {
      "epoch": 13.954755309325947,
      "grad_norm": 0.4799002707004547,
      "learning_rate": 2.268e-05,
      "loss": 0.0175,
      "step": 7563
    },
    {
      "epoch": 13.956602031394276,
      "grad_norm": 0.3801664113998413,
      "learning_rate": 2.2683e-05,
      "loss": 0.0173,
      "step": 7564
    },
    {
      "epoch": 13.958448753462603,
      "grad_norm": 0.4788198173046112,
      "learning_rate": 2.2686e-05,
      "loss": 0.0225,
      "step": 7565
    },
    {
      "epoch": 13.960295475530932,
      "grad_norm": 0.3012051582336426,
      "learning_rate": 2.2689e-05,
      "loss": 0.0094,
      "step": 7566
    },
    {
      "epoch": 13.962142197599261,
      "grad_norm": 0.39352113008499146,
      "learning_rate": 2.2692e-05,
      "loss": 0.0176,
      "step": 7567
    },
    {
      "epoch": 13.96398891966759,
      "grad_norm": 0.3267962336540222,
      "learning_rate": 2.2695e-05,
      "loss": 0.0187,
      "step": 7568
    },
    {
      "epoch": 13.965835641735918,
      "grad_norm": 0.3652357757091522,
      "learning_rate": 2.2698000000000002e-05,
      "loss": 0.0163,
      "step": 7569
    },
    {
      "epoch": 13.967682363804247,
      "grad_norm": 0.29219409823417664,
      "learning_rate": 2.2701000000000002e-05,
      "loss": 0.0299,
      "step": 7570
    },
    {
      "epoch": 13.969529085872576,
      "grad_norm": 0.4231150150299072,
      "learning_rate": 2.2704e-05,
      "loss": 0.0173,
      "step": 7571
    },
    {
      "epoch": 13.971375807940905,
      "grad_norm": 0.8734736442565918,
      "learning_rate": 2.2707e-05,
      "loss": 0.0251,
      "step": 7572
    },
    {
      "epoch": 13.973222530009235,
      "grad_norm": 1.4855594635009766,
      "learning_rate": 2.271e-05,
      "loss": 0.0332,
      "step": 7573
    },
    {
      "epoch": 13.975069252077562,
      "grad_norm": 0.3971326947212219,
      "learning_rate": 2.2713e-05,
      "loss": 0.015,
      "step": 7574
    },
    {
      "epoch": 13.976915974145891,
      "grad_norm": 0.2669675052165985,
      "learning_rate": 2.2716e-05,
      "loss": 0.0083,
      "step": 7575
    },
    {
      "epoch": 13.97876269621422,
      "grad_norm": 0.33655616641044617,
      "learning_rate": 2.2719e-05,
      "loss": 0.0166,
      "step": 7576
    },
    {
      "epoch": 13.980609418282548,
      "grad_norm": 0.46512511372566223,
      "learning_rate": 2.2722e-05,
      "loss": 0.0172,
      "step": 7577
    },
    {
      "epoch": 13.982456140350877,
      "grad_norm": 0.7615096569061279,
      "learning_rate": 2.2725e-05,
      "loss": 0.0372,
      "step": 7578
    },
    {
      "epoch": 13.984302862419206,
      "grad_norm": 0.5171358585357666,
      "learning_rate": 2.2728000000000003e-05,
      "loss": 0.0172,
      "step": 7579
    },
    {
      "epoch": 13.986149584487535,
      "grad_norm": 0.46097612380981445,
      "learning_rate": 2.2731000000000003e-05,
      "loss": 0.0185,
      "step": 7580
    },
    {
      "epoch": 13.987996306555864,
      "grad_norm": 1.4962469339370728,
      "learning_rate": 2.2734000000000003e-05,
      "loss": 0.0318,
      "step": 7581
    },
    {
      "epoch": 13.989843028624191,
      "grad_norm": 0.5866150856018066,
      "learning_rate": 2.2737000000000003e-05,
      "loss": 0.0173,
      "step": 7582
    },
    {
      "epoch": 13.99168975069252,
      "grad_norm": 0.3140173554420471,
      "learning_rate": 2.274e-05,
      "loss": 0.0162,
      "step": 7583
    },
    {
      "epoch": 13.99353647276085,
      "grad_norm": 1.4404232501983643,
      "learning_rate": 2.2743e-05,
      "loss": 0.0216,
      "step": 7584
    },
    {
      "epoch": 13.995383194829179,
      "grad_norm": 0.34213533997535706,
      "learning_rate": 2.2746e-05,
      "loss": 0.0113,
      "step": 7585
    },
    {
      "epoch": 13.997229916897506,
      "grad_norm": 0.5488388538360596,
      "learning_rate": 2.2749e-05,
      "loss": 0.0235,
      "step": 7586
    },
    {
      "epoch": 13.999076638965835,
      "grad_norm": 0.4100989103317261,
      "learning_rate": 2.2752e-05,
      "loss": 0.0137,
      "step": 7587
    },
    {
      "epoch": 14.0,
      "grad_norm": 0.712295413017273,
      "learning_rate": 2.2754999999999998e-05,
      "loss": 0.0101,
      "step": 7588
    },
    {
      "epoch": 14.00184672206833,
      "grad_norm": 1.9446412324905396,
      "learning_rate": 2.2758e-05,
      "loss": 0.1777,
      "step": 7589
    },
    {
      "epoch": 14.003693444136658,
      "grad_norm": 2.1511576175689697,
      "learning_rate": 2.2761e-05,
      "loss": 0.1542,
      "step": 7590
    },
    {
      "epoch": 14.005540166204986,
      "grad_norm": 0.7647144794464111,
      "learning_rate": 2.2764e-05,
      "loss": 0.1752,
      "step": 7591
    },
    {
      "epoch": 14.007386888273315,
      "grad_norm": 1.0041519403457642,
      "learning_rate": 2.2767e-05,
      "loss": 0.1497,
      "step": 7592
    },
    {
      "epoch": 14.009233610341644,
      "grad_norm": 0.776584804058075,
      "learning_rate": 2.277e-05,
      "loss": 0.1156,
      "step": 7593
    },
    {
      "epoch": 14.011080332409973,
      "grad_norm": 4.055688381195068,
      "learning_rate": 2.2773e-05,
      "loss": 0.1211,
      "step": 7594
    },
    {
      "epoch": 14.0129270544783,
      "grad_norm": 0.8145771622657776,
      "learning_rate": 2.2776e-05,
      "loss": 0.1183,
      "step": 7595
    },
    {
      "epoch": 14.01477377654663,
      "grad_norm": 0.6137973070144653,
      "learning_rate": 2.2779e-05,
      "loss": 0.0915,
      "step": 7596
    },
    {
      "epoch": 14.016620498614959,
      "grad_norm": 0.6850339770317078,
      "learning_rate": 2.2782e-05,
      "loss": 0.1176,
      "step": 7597
    },
    {
      "epoch": 14.018467220683288,
      "grad_norm": 0.7108796238899231,
      "learning_rate": 2.2785e-05,
      "loss": 0.0925,
      "step": 7598
    },
    {
      "epoch": 14.020313942751615,
      "grad_norm": 0.8418132066726685,
      "learning_rate": 2.2788000000000003e-05,
      "loss": 0.1094,
      "step": 7599
    },
    {
      "epoch": 14.022160664819944,
      "grad_norm": 0.49468469619750977,
      "learning_rate": 2.2791000000000003e-05,
      "loss": 0.0733,
      "step": 7600
    },
    {
      "epoch": 14.024007386888274,
      "grad_norm": 0.4118848741054535,
      "learning_rate": 2.2794000000000002e-05,
      "loss": 0.0538,
      "step": 7601
    },
    {
      "epoch": 14.025854108956603,
      "grad_norm": 0.46933019161224365,
      "learning_rate": 2.2797000000000002e-05,
      "loss": 0.0378,
      "step": 7602
    },
    {
      "epoch": 14.02770083102493,
      "grad_norm": 0.514575183391571,
      "learning_rate": 2.2800000000000002e-05,
      "loss": 0.0528,
      "step": 7603
    },
    {
      "epoch": 14.02954755309326,
      "grad_norm": 0.41290318965911865,
      "learning_rate": 2.2803000000000002e-05,
      "loss": 0.0375,
      "step": 7604
    },
    {
      "epoch": 14.031394275161588,
      "grad_norm": 0.6624583005905151,
      "learning_rate": 2.2806e-05,
      "loss": 0.0243,
      "step": 7605
    },
    {
      "epoch": 14.033240997229917,
      "grad_norm": 0.4341171085834503,
      "learning_rate": 2.2809e-05,
      "loss": 0.0255,
      "step": 7606
    },
    {
      "epoch": 14.035087719298245,
      "grad_norm": 0.4499518871307373,
      "learning_rate": 2.2812e-05,
      "loss": 0.0431,
      "step": 7607
    },
    {
      "epoch": 14.036934441366574,
      "grad_norm": 2.0344090461730957,
      "learning_rate": 2.2814999999999998e-05,
      "loss": 0.0187,
      "step": 7608
    },
    {
      "epoch": 14.038781163434903,
      "grad_norm": 0.7375814914703369,
      "learning_rate": 2.2818e-05,
      "loss": 0.0182,
      "step": 7609
    },
    {
      "epoch": 14.040627885503232,
      "grad_norm": 0.45870843529701233,
      "learning_rate": 2.2821e-05,
      "loss": 0.0107,
      "step": 7610
    },
    {
      "epoch": 14.04247460757156,
      "grad_norm": 0.905041515827179,
      "learning_rate": 2.2824e-05,
      "loss": 0.0232,
      "step": 7611
    },
    {
      "epoch": 14.044321329639889,
      "grad_norm": 0.8704097867012024,
      "learning_rate": 2.2827e-05,
      "loss": 0.0249,
      "step": 7612
    },
    {
      "epoch": 14.046168051708218,
      "grad_norm": 0.29793083667755127,
      "learning_rate": 2.283e-05,
      "loss": 0.0115,
      "step": 7613
    },
    {
      "epoch": 14.048014773776547,
      "grad_norm": 0.43679356575012207,
      "learning_rate": 2.2833e-05,
      "loss": 0.0217,
      "step": 7614
    },
    {
      "epoch": 14.049861495844876,
      "grad_norm": 1.4915803670883179,
      "learning_rate": 2.2836e-05,
      "loss": 0.0183,
      "step": 7615
    },
    {
      "epoch": 14.051708217913204,
      "grad_norm": 0.692267119884491,
      "learning_rate": 2.2839e-05,
      "loss": 0.0362,
      "step": 7616
    },
    {
      "epoch": 14.053554939981533,
      "grad_norm": 0.4739607274532318,
      "learning_rate": 2.2842e-05,
      "loss": 0.0232,
      "step": 7617
    },
    {
      "epoch": 14.055401662049862,
      "grad_norm": 0.25771304965019226,
      "learning_rate": 2.2845e-05,
      "loss": 0.0114,
      "step": 7618
    },
    {
      "epoch": 14.057248384118191,
      "grad_norm": 0.2940288484096527,
      "learning_rate": 2.2848000000000002e-05,
      "loss": 0.0177,
      "step": 7619
    },
    {
      "epoch": 14.059095106186518,
      "grad_norm": 0.6190711855888367,
      "learning_rate": 2.2851000000000002e-05,
      "loss": 0.0162,
      "step": 7620
    },
    {
      "epoch": 14.060941828254848,
      "grad_norm": 0.4986996352672577,
      "learning_rate": 2.2854000000000002e-05,
      "loss": 0.0123,
      "step": 7621
    },
    {
      "epoch": 14.062788550323177,
      "grad_norm": 0.4452718496322632,
      "learning_rate": 2.2857e-05,
      "loss": 0.0198,
      "step": 7622
    },
    {
      "epoch": 14.064635272391506,
      "grad_norm": 0.269513338804245,
      "learning_rate": 2.286e-05,
      "loss": 0.011,
      "step": 7623
    },
    {
      "epoch": 14.066481994459833,
      "grad_norm": 0.3024609684944153,
      "learning_rate": 2.2863e-05,
      "loss": 0.0112,
      "step": 7624
    },
    {
      "epoch": 14.068328716528162,
      "grad_norm": 0.9202260375022888,
      "learning_rate": 2.2866e-05,
      "loss": 0.0176,
      "step": 7625
    },
    {
      "epoch": 14.070175438596491,
      "grad_norm": 0.44851964712142944,
      "learning_rate": 2.2869e-05,
      "loss": 0.0417,
      "step": 7626
    },
    {
      "epoch": 14.07202216066482,
      "grad_norm": 0.5856828093528748,
      "learning_rate": 2.2872e-05,
      "loss": 0.0176,
      "step": 7627
    },
    {
      "epoch": 14.073868882733148,
      "grad_norm": 0.2961845099925995,
      "learning_rate": 2.2875e-05,
      "loss": 0.0104,
      "step": 7628
    },
    {
      "epoch": 14.075715604801477,
      "grad_norm": 0.34900930523872375,
      "learning_rate": 2.2878e-05,
      "loss": 0.0187,
      "step": 7629
    },
    {
      "epoch": 14.077562326869806,
      "grad_norm": 0.7918524742126465,
      "learning_rate": 2.2881000000000003e-05,
      "loss": 0.0217,
      "step": 7630
    },
    {
      "epoch": 14.079409048938135,
      "grad_norm": 0.7197925448417664,
      "learning_rate": 2.2884000000000003e-05,
      "loss": 0.031,
      "step": 7631
    },
    {
      "epoch": 14.081255771006463,
      "grad_norm": 0.2891060709953308,
      "learning_rate": 2.2887e-05,
      "loss": 0.0096,
      "step": 7632
    },
    {
      "epoch": 14.083102493074792,
      "grad_norm": 0.6610003709793091,
      "learning_rate": 2.289e-05,
      "loss": 0.021,
      "step": 7633
    },
    {
      "epoch": 14.084949215143121,
      "grad_norm": 0.6482753157615662,
      "learning_rate": 2.2893e-05,
      "loss": 0.0185,
      "step": 7634
    },
    {
      "epoch": 14.08679593721145,
      "grad_norm": 0.3223477602005005,
      "learning_rate": 2.2896e-05,
      "loss": 0.0109,
      "step": 7635
    },
    {
      "epoch": 14.088642659279778,
      "grad_norm": 0.5384102463722229,
      "learning_rate": 2.2899e-05,
      "loss": 0.01,
      "step": 7636
    },
    {
      "epoch": 14.090489381348107,
      "grad_norm": 0.6918799877166748,
      "learning_rate": 2.2902e-05,
      "loss": 0.0252,
      "step": 7637
    },
    {
      "epoch": 14.092336103416436,
      "grad_norm": 0.864923894405365,
      "learning_rate": 2.2905e-05,
      "loss": 0.0249,
      "step": 7638
    },
    {
      "epoch": 14.094182825484765,
      "grad_norm": 0.9709484577178955,
      "learning_rate": 2.2907999999999998e-05,
      "loss": 0.2236,
      "step": 7639
    },
    {
      "epoch": 14.096029547553094,
      "grad_norm": 0.9872021079063416,
      "learning_rate": 2.2911e-05,
      "loss": 0.2536,
      "step": 7640
    },
    {
      "epoch": 14.097876269621421,
      "grad_norm": 0.7489765882492065,
      "learning_rate": 2.2914e-05,
      "loss": 0.1795,
      "step": 7641
    },
    {
      "epoch": 14.09972299168975,
      "grad_norm": 0.7499869465827942,
      "learning_rate": 2.2917e-05,
      "loss": 0.1119,
      "step": 7642
    },
    {
      "epoch": 14.10156971375808,
      "grad_norm": 0.6760444045066833,
      "learning_rate": 2.292e-05,
      "loss": 0.1,
      "step": 7643
    },
    {
      "epoch": 14.103416435826409,
      "grad_norm": 1.1884748935699463,
      "learning_rate": 2.2923e-05,
      "loss": 0.0716,
      "step": 7644
    },
    {
      "epoch": 14.105263157894736,
      "grad_norm": 0.6572530269622803,
      "learning_rate": 2.2926e-05,
      "loss": 0.0893,
      "step": 7645
    },
    {
      "epoch": 14.107109879963065,
      "grad_norm": 0.5504191517829895,
      "learning_rate": 2.2929e-05,
      "loss": 0.098,
      "step": 7646
    },
    {
      "epoch": 14.108956602031395,
      "grad_norm": 0.5794324278831482,
      "learning_rate": 2.2932e-05,
      "loss": 0.0736,
      "step": 7647
    },
    {
      "epoch": 14.110803324099724,
      "grad_norm": 0.39542296528816223,
      "learning_rate": 2.2935e-05,
      "loss": 0.0534,
      "step": 7648
    },
    {
      "epoch": 14.112650046168051,
      "grad_norm": 0.3459565043449402,
      "learning_rate": 2.2938e-05,
      "loss": 0.0387,
      "step": 7649
    },
    {
      "epoch": 14.11449676823638,
      "grad_norm": 0.9930928349494934,
      "learning_rate": 2.2941000000000003e-05,
      "loss": 0.047,
      "step": 7650
    },
    {
      "epoch": 14.11634349030471,
      "grad_norm": 0.4960794150829315,
      "learning_rate": 2.2944000000000003e-05,
      "loss": 0.0401,
      "step": 7651
    },
    {
      "epoch": 14.118190212373039,
      "grad_norm": 0.4362480342388153,
      "learning_rate": 2.2947000000000002e-05,
      "loss": 0.0395,
      "step": 7652
    },
    {
      "epoch": 14.120036934441366,
      "grad_norm": 0.43307340145111084,
      "learning_rate": 2.2950000000000002e-05,
      "loss": 0.0225,
      "step": 7653
    },
    {
      "epoch": 14.121883656509695,
      "grad_norm": 0.5992100834846497,
      "learning_rate": 2.2953000000000002e-05,
      "loss": 0.046,
      "step": 7654
    },
    {
      "epoch": 14.123730378578024,
      "grad_norm": 0.4235561490058899,
      "learning_rate": 2.2956000000000002e-05,
      "loss": 0.0209,
      "step": 7655
    },
    {
      "epoch": 14.125577100646353,
      "grad_norm": 0.3926125764846802,
      "learning_rate": 2.2959e-05,
      "loss": 0.0187,
      "step": 7656
    },
    {
      "epoch": 14.12742382271468,
      "grad_norm": 0.7339117527008057,
      "learning_rate": 2.2961999999999998e-05,
      "loss": 0.0595,
      "step": 7657
    },
    {
      "epoch": 14.12927054478301,
      "grad_norm": 0.4378303289413452,
      "learning_rate": 2.2964999999999998e-05,
      "loss": 0.0212,
      "step": 7658
    },
    {
      "epoch": 14.131117266851339,
      "grad_norm": 0.47204580903053284,
      "learning_rate": 2.2967999999999998e-05,
      "loss": 0.0407,
      "step": 7659
    },
    {
      "epoch": 14.132963988919668,
      "grad_norm": 0.31966009736061096,
      "learning_rate": 2.2971e-05,
      "loss": 0.0175,
      "step": 7660
    },
    {
      "epoch": 14.134810710987995,
      "grad_norm": 0.3838869333267212,
      "learning_rate": 2.2974e-05,
      "loss": 0.0186,
      "step": 7661
    },
    {
      "epoch": 14.136657433056325,
      "grad_norm": 0.19909749925136566,
      "learning_rate": 2.2977e-05,
      "loss": 0.0114,
      "step": 7662
    },
    {
      "epoch": 14.138504155124654,
      "grad_norm": 0.3598874807357788,
      "learning_rate": 2.298e-05,
      "loss": 0.0343,
      "step": 7663
    },
    {
      "epoch": 14.140350877192983,
      "grad_norm": 0.18918880820274353,
      "learning_rate": 2.2983e-05,
      "loss": 0.0096,
      "step": 7664
    },
    {
      "epoch": 14.142197599261312,
      "grad_norm": 0.2890123724937439,
      "learning_rate": 2.2986e-05,
      "loss": 0.0129,
      "step": 7665
    },
    {
      "epoch": 14.14404432132964,
      "grad_norm": 0.27688318490982056,
      "learning_rate": 2.2989e-05,
      "loss": 0.0144,
      "step": 7666
    },
    {
      "epoch": 14.145891043397969,
      "grad_norm": 0.46568000316619873,
      "learning_rate": 2.2992e-05,
      "loss": 0.0704,
      "step": 7667
    },
    {
      "epoch": 14.147737765466298,
      "grad_norm": 0.5880398750305176,
      "learning_rate": 2.2995e-05,
      "loss": 0.0179,
      "step": 7668
    },
    {
      "epoch": 14.149584487534627,
      "grad_norm": 0.28892356157302856,
      "learning_rate": 2.2998e-05,
      "loss": 0.011,
      "step": 7669
    },
    {
      "epoch": 14.151431209602954,
      "grad_norm": 0.5236003994941711,
      "learning_rate": 2.3001000000000002e-05,
      "loss": 0.0365,
      "step": 7670
    },
    {
      "epoch": 14.153277931671283,
      "grad_norm": 0.17960208654403687,
      "learning_rate": 2.3004000000000002e-05,
      "loss": 0.0081,
      "step": 7671
    },
    {
      "epoch": 14.155124653739612,
      "grad_norm": 0.45658397674560547,
      "learning_rate": 2.3007000000000002e-05,
      "loss": 0.0189,
      "step": 7672
    },
    {
      "epoch": 14.156971375807942,
      "grad_norm": 0.3621665835380554,
      "learning_rate": 2.301e-05,
      "loss": 0.0202,
      "step": 7673
    },
    {
      "epoch": 14.158818097876269,
      "grad_norm": 0.30046194791793823,
      "learning_rate": 2.3013e-05,
      "loss": 0.0108,
      "step": 7674
    },
    {
      "epoch": 14.160664819944598,
      "grad_norm": 0.33425599336624146,
      "learning_rate": 2.3016e-05,
      "loss": 0.0173,
      "step": 7675
    },
    {
      "epoch": 14.162511542012927,
      "grad_norm": 0.20881663262844086,
      "learning_rate": 2.3019e-05,
      "loss": 0.0091,
      "step": 7676
    },
    {
      "epoch": 14.164358264081256,
      "grad_norm": 0.8621287941932678,
      "learning_rate": 2.3022e-05,
      "loss": 0.027,
      "step": 7677
    },
    {
      "epoch": 14.166204986149584,
      "grad_norm": 0.21208317577838898,
      "learning_rate": 2.3025e-05,
      "loss": 0.0079,
      "step": 7678
    },
    {
      "epoch": 14.168051708217913,
      "grad_norm": 1.173541784286499,
      "learning_rate": 2.3028e-05,
      "loss": 0.0178,
      "step": 7679
    },
    {
      "epoch": 14.169898430286242,
      "grad_norm": 0.2508636713027954,
      "learning_rate": 2.3031000000000004e-05,
      "loss": 0.0074,
      "step": 7680
    },
    {
      "epoch": 14.171745152354571,
      "grad_norm": 0.30045756697654724,
      "learning_rate": 2.3034e-05,
      "loss": 0.0083,
      "step": 7681
    },
    {
      "epoch": 14.173591874422899,
      "grad_norm": 0.38787662982940674,
      "learning_rate": 2.3037e-05,
      "loss": 0.0241,
      "step": 7682
    },
    {
      "epoch": 14.175438596491228,
      "grad_norm": 0.7570717930793762,
      "learning_rate": 2.304e-05,
      "loss": 0.0133,
      "step": 7683
    },
    {
      "epoch": 14.177285318559557,
      "grad_norm": 0.885297954082489,
      "learning_rate": 2.3043e-05,
      "loss": 0.0244,
      "step": 7684
    },
    {
      "epoch": 14.179132040627886,
      "grad_norm": 1.052407145500183,
      "learning_rate": 2.3046e-05,
      "loss": 0.023,
      "step": 7685
    },
    {
      "epoch": 14.180978762696213,
      "grad_norm": 0.8462616801261902,
      "learning_rate": 2.3049e-05,
      "loss": 0.038,
      "step": 7686
    },
    {
      "epoch": 14.182825484764543,
      "grad_norm": 0.41136690974235535,
      "learning_rate": 2.3052e-05,
      "loss": 0.0126,
      "step": 7687
    },
    {
      "epoch": 14.184672206832872,
      "grad_norm": 0.42739972472190857,
      "learning_rate": 2.3055e-05,
      "loss": 0.0146,
      "step": 7688
    },
    {
      "epoch": 14.1865189289012,
      "grad_norm": 0.8237746953964233,
      "learning_rate": 2.3058e-05,
      "loss": 0.2013,
      "step": 7689
    },
    {
      "epoch": 14.18836565096953,
      "grad_norm": 0.7207422852516174,
      "learning_rate": 2.3061e-05,
      "loss": 0.208,
      "step": 7690
    },
    {
      "epoch": 14.190212373037857,
      "grad_norm": 0.9960768222808838,
      "learning_rate": 2.3064e-05,
      "loss": 0.1293,
      "step": 7691
    },
    {
      "epoch": 14.192059095106186,
      "grad_norm": 0.8917984962463379,
      "learning_rate": 2.3067e-05,
      "loss": 0.1429,
      "step": 7692
    },
    {
      "epoch": 14.193905817174516,
      "grad_norm": 0.5600876808166504,
      "learning_rate": 2.307e-05,
      "loss": 0.1572,
      "step": 7693
    },
    {
      "epoch": 14.195752539242845,
      "grad_norm": 0.4889839291572571,
      "learning_rate": 2.3073e-05,
      "loss": 0.1045,
      "step": 7694
    },
    {
      "epoch": 14.197599261311172,
      "grad_norm": 0.44030889868736267,
      "learning_rate": 2.3076e-05,
      "loss": 0.0875,
      "step": 7695
    },
    {
      "epoch": 14.199445983379501,
      "grad_norm": 0.48708662390708923,
      "learning_rate": 2.3079e-05,
      "loss": 0.0865,
      "step": 7696
    },
    {
      "epoch": 14.20129270544783,
      "grad_norm": 0.5878287553787231,
      "learning_rate": 2.3082e-05,
      "loss": 0.0882,
      "step": 7697
    },
    {
      "epoch": 14.20313942751616,
      "grad_norm": 0.6096134185791016,
      "learning_rate": 2.3085e-05,
      "loss": 0.1273,
      "step": 7698
    },
    {
      "epoch": 14.204986149584487,
      "grad_norm": 0.41568949818611145,
      "learning_rate": 2.3088e-05,
      "loss": 0.0412,
      "step": 7699
    },
    {
      "epoch": 14.206832871652816,
      "grad_norm": 0.8116822242736816,
      "learning_rate": 2.3091000000000003e-05,
      "loss": 0.0527,
      "step": 7700
    },
    {
      "epoch": 14.208679593721145,
      "grad_norm": 0.5746693015098572,
      "learning_rate": 2.3094000000000003e-05,
      "loss": 0.0963,
      "step": 7701
    },
    {
      "epoch": 14.210526315789474,
      "grad_norm": 0.4817805290222168,
      "learning_rate": 2.3097000000000003e-05,
      "loss": 0.0589,
      "step": 7702
    },
    {
      "epoch": 14.212373037857802,
      "grad_norm": 0.4487169086933136,
      "learning_rate": 2.3100000000000002e-05,
      "loss": 0.0365,
      "step": 7703
    },
    {
      "epoch": 14.21421975992613,
      "grad_norm": 0.5302636027336121,
      "learning_rate": 2.3103000000000002e-05,
      "loss": 0.0307,
      "step": 7704
    },
    {
      "epoch": 14.21606648199446,
      "grad_norm": 0.22351866960525513,
      "learning_rate": 2.3106000000000002e-05,
      "loss": 0.0133,
      "step": 7705
    },
    {
      "epoch": 14.21791320406279,
      "grad_norm": 0.41186782717704773,
      "learning_rate": 2.3109e-05,
      "loss": 0.0186,
      "step": 7706
    },
    {
      "epoch": 14.219759926131117,
      "grad_norm": 0.40214699506759644,
      "learning_rate": 2.3111999999999998e-05,
      "loss": 0.0177,
      "step": 7707
    },
    {
      "epoch": 14.221606648199446,
      "grad_norm": 0.24294708669185638,
      "learning_rate": 2.3114999999999998e-05,
      "loss": 0.0137,
      "step": 7708
    },
    {
      "epoch": 14.223453370267775,
      "grad_norm": 0.5617495775222778,
      "learning_rate": 2.3117999999999998e-05,
      "loss": 0.0234,
      "step": 7709
    },
    {
      "epoch": 14.225300092336104,
      "grad_norm": 0.8240264654159546,
      "learning_rate": 2.3121e-05,
      "loss": 0.0257,
      "step": 7710
    },
    {
      "epoch": 14.227146814404431,
      "grad_norm": 0.5336347818374634,
      "learning_rate": 2.3124e-05,
      "loss": 0.039,
      "step": 7711
    },
    {
      "epoch": 14.22899353647276,
      "grad_norm": 0.34149596095085144,
      "learning_rate": 2.3127e-05,
      "loss": 0.0169,
      "step": 7712
    },
    {
      "epoch": 14.23084025854109,
      "grad_norm": 0.24113301932811737,
      "learning_rate": 2.313e-05,
      "loss": 0.0133,
      "step": 7713
    },
    {
      "epoch": 14.232686980609419,
      "grad_norm": 0.2770822048187256,
      "learning_rate": 2.3133e-05,
      "loss": 0.0076,
      "step": 7714
    },
    {
      "epoch": 14.234533702677748,
      "grad_norm": 0.37785083055496216,
      "learning_rate": 2.3136e-05,
      "loss": 0.0303,
      "step": 7715
    },
    {
      "epoch": 14.236380424746075,
      "grad_norm": 0.36604055762290955,
      "learning_rate": 2.3139e-05,
      "loss": 0.0237,
      "step": 7716
    },
    {
      "epoch": 14.238227146814404,
      "grad_norm": 0.316237211227417,
      "learning_rate": 2.3142e-05,
      "loss": 0.0194,
      "step": 7717
    },
    {
      "epoch": 14.240073868882734,
      "grad_norm": 0.3500533103942871,
      "learning_rate": 2.3145e-05,
      "loss": 0.0121,
      "step": 7718
    },
    {
      "epoch": 14.241920590951063,
      "grad_norm": 0.3890174329280853,
      "learning_rate": 2.3148e-05,
      "loss": 0.0092,
      "step": 7719
    },
    {
      "epoch": 14.24376731301939,
      "grad_norm": 1.027275562286377,
      "learning_rate": 2.3151000000000002e-05,
      "loss": 0.0269,
      "step": 7720
    },
    {
      "epoch": 14.24561403508772,
      "grad_norm": 0.4223162531852722,
      "learning_rate": 2.3154000000000002e-05,
      "loss": 0.014,
      "step": 7721
    },
    {
      "epoch": 14.247460757156048,
      "grad_norm": 0.37757349014282227,
      "learning_rate": 2.3157000000000002e-05,
      "loss": 0.0101,
      "step": 7722
    },
    {
      "epoch": 14.249307479224377,
      "grad_norm": 0.34131377935409546,
      "learning_rate": 2.3160000000000002e-05,
      "loss": 0.0143,
      "step": 7723
    },
    {
      "epoch": 14.251154201292705,
      "grad_norm": 0.4734109044075012,
      "learning_rate": 2.3163e-05,
      "loss": 0.017,
      "step": 7724
    },
    {
      "epoch": 14.253000923361034,
      "grad_norm": 0.5335391759872437,
      "learning_rate": 2.3166e-05,
      "loss": 0.016,
      "step": 7725
    },
    {
      "epoch": 14.254847645429363,
      "grad_norm": 0.25241535902023315,
      "learning_rate": 2.3169e-05,
      "loss": 0.0092,
      "step": 7726
    },
    {
      "epoch": 14.256694367497692,
      "grad_norm": 0.6295625567436218,
      "learning_rate": 2.3172e-05,
      "loss": 0.0169,
      "step": 7727
    },
    {
      "epoch": 14.25854108956602,
      "grad_norm": 0.7412222623825073,
      "learning_rate": 2.3175e-05,
      "loss": 0.0148,
      "step": 7728
    },
    {
      "epoch": 14.260387811634349,
      "grad_norm": 0.3130491375923157,
      "learning_rate": 2.3178e-05,
      "loss": 0.0114,
      "step": 7729
    },
    {
      "epoch": 14.262234533702678,
      "grad_norm": 0.31834471225738525,
      "learning_rate": 2.3181000000000004e-05,
      "loss": 0.0106,
      "step": 7730
    },
    {
      "epoch": 14.264081255771007,
      "grad_norm": 1.08151376247406,
      "learning_rate": 2.3184e-05,
      "loss": 0.0213,
      "step": 7731
    },
    {
      "epoch": 14.265927977839334,
      "grad_norm": 0.7608094215393066,
      "learning_rate": 2.3187e-05,
      "loss": 0.0191,
      "step": 7732
    },
    {
      "epoch": 14.267774699907664,
      "grad_norm": 0.23599159717559814,
      "learning_rate": 2.319e-05,
      "loss": 0.0148,
      "step": 7733
    },
    {
      "epoch": 14.269621421975993,
      "grad_norm": 0.6872748136520386,
      "learning_rate": 2.3193e-05,
      "loss": 0.0133,
      "step": 7734
    },
    {
      "epoch": 14.271468144044322,
      "grad_norm": 0.7782183885574341,
      "learning_rate": 2.3196e-05,
      "loss": 0.0184,
      "step": 7735
    },
    {
      "epoch": 14.27331486611265,
      "grad_norm": 0.5043119788169861,
      "learning_rate": 2.3199e-05,
      "loss": 0.016,
      "step": 7736
    },
    {
      "epoch": 14.275161588180978,
      "grad_norm": 0.7329913973808289,
      "learning_rate": 2.3202e-05,
      "loss": 0.0162,
      "step": 7737
    },
    {
      "epoch": 14.277008310249307,
      "grad_norm": 0.37354522943496704,
      "learning_rate": 2.3205e-05,
      "loss": 0.0131,
      "step": 7738
    },
    {
      "epoch": 14.278855032317637,
      "grad_norm": 1.0684969425201416,
      "learning_rate": 2.3208e-05,
      "loss": 0.2945,
      "step": 7739
    },
    {
      "epoch": 14.280701754385966,
      "grad_norm": 0.5377521514892578,
      "learning_rate": 2.3211000000000002e-05,
      "loss": 0.1278,
      "step": 7740
    },
    {
      "epoch": 14.282548476454293,
      "grad_norm": 0.5380083322525024,
      "learning_rate": 2.3214000000000002e-05,
      "loss": 0.1373,
      "step": 7741
    },
    {
      "epoch": 14.284395198522622,
      "grad_norm": 1.1396440267562866,
      "learning_rate": 2.3217e-05,
      "loss": 0.1496,
      "step": 7742
    },
    {
      "epoch": 14.286241920590951,
      "grad_norm": 0.8189786672592163,
      "learning_rate": 2.322e-05,
      "loss": 0.1442,
      "step": 7743
    },
    {
      "epoch": 14.28808864265928,
      "grad_norm": 0.5815137028694153,
      "learning_rate": 2.3223e-05,
      "loss": 0.1651,
      "step": 7744
    },
    {
      "epoch": 14.289935364727608,
      "grad_norm": 0.5112689137458801,
      "learning_rate": 2.3226e-05,
      "loss": 0.0993,
      "step": 7745
    },
    {
      "epoch": 14.291782086795937,
      "grad_norm": 0.8791669607162476,
      "learning_rate": 2.3229e-05,
      "loss": 0.1104,
      "step": 7746
    },
    {
      "epoch": 14.293628808864266,
      "grad_norm": 0.9186159372329712,
      "learning_rate": 2.3232e-05,
      "loss": 0.0836,
      "step": 7747
    },
    {
      "epoch": 14.295475530932595,
      "grad_norm": 0.6723326444625854,
      "learning_rate": 2.3235e-05,
      "loss": 0.0736,
      "step": 7748
    },
    {
      "epoch": 14.297322253000923,
      "grad_norm": 0.5478449463844299,
      "learning_rate": 2.3238e-05,
      "loss": 0.1027,
      "step": 7749
    },
    {
      "epoch": 14.299168975069252,
      "grad_norm": 0.38028210401535034,
      "learning_rate": 2.3241000000000003e-05,
      "loss": 0.0377,
      "step": 7750
    },
    {
      "epoch": 14.301015697137581,
      "grad_norm": 0.47234466671943665,
      "learning_rate": 2.3244000000000003e-05,
      "loss": 0.0761,
      "step": 7751
    },
    {
      "epoch": 14.30286241920591,
      "grad_norm": 0.47043994069099426,
      "learning_rate": 2.3247000000000003e-05,
      "loss": 0.0569,
      "step": 7752
    },
    {
      "epoch": 14.304709141274238,
      "grad_norm": 0.44927072525024414,
      "learning_rate": 2.3250000000000003e-05,
      "loss": 0.0366,
      "step": 7753
    },
    {
      "epoch": 14.306555863342567,
      "grad_norm": 0.3382532596588135,
      "learning_rate": 2.3253000000000003e-05,
      "loss": 0.0298,
      "step": 7754
    },
    {
      "epoch": 14.308402585410896,
      "grad_norm": 0.30027419328689575,
      "learning_rate": 2.3256e-05,
      "loss": 0.0199,
      "step": 7755
    },
    {
      "epoch": 14.310249307479225,
      "grad_norm": 0.44675812125205994,
      "learning_rate": 2.3259e-05,
      "loss": 0.0302,
      "step": 7756
    },
    {
      "epoch": 14.312096029547552,
      "grad_norm": 0.803776204586029,
      "learning_rate": 2.3262e-05,
      "loss": 0.0209,
      "step": 7757
    },
    {
      "epoch": 14.313942751615881,
      "grad_norm": 0.32607215642929077,
      "learning_rate": 2.3265e-05,
      "loss": 0.0135,
      "step": 7758
    },
    {
      "epoch": 14.31578947368421,
      "grad_norm": 0.30982378125190735,
      "learning_rate": 2.3267999999999998e-05,
      "loss": 0.0143,
      "step": 7759
    },
    {
      "epoch": 14.31763619575254,
      "grad_norm": 0.38955262303352356,
      "learning_rate": 2.3270999999999998e-05,
      "loss": 0.0189,
      "step": 7760
    },
    {
      "epoch": 14.319482917820867,
      "grad_norm": 0.34186357259750366,
      "learning_rate": 2.3274e-05,
      "loss": 0.0128,
      "step": 7761
    },
    {
      "epoch": 14.321329639889196,
      "grad_norm": 0.29279205203056335,
      "learning_rate": 2.3277e-05,
      "loss": 0.0154,
      "step": 7762
    },
    {
      "epoch": 14.323176361957525,
      "grad_norm": 0.37621551752090454,
      "learning_rate": 2.328e-05,
      "loss": 0.012,
      "step": 7763
    },
    {
      "epoch": 14.325023084025855,
      "grad_norm": 0.4087648093700409,
      "learning_rate": 2.3283e-05,
      "loss": 0.0186,
      "step": 7764
    },
    {
      "epoch": 14.326869806094184,
      "grad_norm": 0.303737998008728,
      "learning_rate": 2.3286e-05,
      "loss": 0.0136,
      "step": 7765
    },
    {
      "epoch": 14.328716528162511,
      "grad_norm": 0.5038633942604065,
      "learning_rate": 2.3289e-05,
      "loss": 0.016,
      "step": 7766
    },
    {
      "epoch": 14.33056325023084,
      "grad_norm": 0.8184044361114502,
      "learning_rate": 2.3292e-05,
      "loss": 0.0283,
      "step": 7767
    },
    {
      "epoch": 14.33240997229917,
      "grad_norm": 1.2842833995819092,
      "learning_rate": 2.3295e-05,
      "loss": 0.0478,
      "step": 7768
    },
    {
      "epoch": 14.334256694367498,
      "grad_norm": 0.36860495805740356,
      "learning_rate": 2.3298e-05,
      "loss": 0.0141,
      "step": 7769
    },
    {
      "epoch": 14.336103416435826,
      "grad_norm": 0.6347278952598572,
      "learning_rate": 2.3301e-05,
      "loss": 0.0214,
      "step": 7770
    },
    {
      "epoch": 14.337950138504155,
      "grad_norm": 0.37546491622924805,
      "learning_rate": 2.3304000000000003e-05,
      "loss": 0.0141,
      "step": 7771
    },
    {
      "epoch": 14.339796860572484,
      "grad_norm": 0.39249086380004883,
      "learning_rate": 2.3307000000000002e-05,
      "loss": 0.0153,
      "step": 7772
    },
    {
      "epoch": 14.341643582640813,
      "grad_norm": 0.6554223895072937,
      "learning_rate": 2.3310000000000002e-05,
      "loss": 0.026,
      "step": 7773
    },
    {
      "epoch": 14.34349030470914,
      "grad_norm": 1.0025122165679932,
      "learning_rate": 2.3313000000000002e-05,
      "loss": 0.0111,
      "step": 7774
    },
    {
      "epoch": 14.34533702677747,
      "grad_norm": 0.6032886505126953,
      "learning_rate": 2.3316000000000002e-05,
      "loss": 0.0159,
      "step": 7775
    },
    {
      "epoch": 14.347183748845799,
      "grad_norm": 0.31834331154823303,
      "learning_rate": 2.3319e-05,
      "loss": 0.0152,
      "step": 7776
    },
    {
      "epoch": 14.349030470914128,
      "grad_norm": 0.7992693781852722,
      "learning_rate": 2.3322e-05,
      "loss": 0.0303,
      "step": 7777
    },
    {
      "epoch": 14.350877192982455,
      "grad_norm": 0.3116844892501831,
      "learning_rate": 2.3325e-05,
      "loss": 0.0156,
      "step": 7778
    },
    {
      "epoch": 14.352723915050785,
      "grad_norm": 0.2716032564640045,
      "learning_rate": 2.3328e-05,
      "loss": 0.0093,
      "step": 7779
    },
    {
      "epoch": 14.354570637119114,
      "grad_norm": 0.2736084759235382,
      "learning_rate": 2.3330999999999997e-05,
      "loss": 0.0083,
      "step": 7780
    },
    {
      "epoch": 14.356417359187443,
      "grad_norm": 0.6578453183174133,
      "learning_rate": 2.3334e-05,
      "loss": 0.0212,
      "step": 7781
    },
    {
      "epoch": 14.35826408125577,
      "grad_norm": 0.33356401324272156,
      "learning_rate": 2.3337e-05,
      "loss": 0.0104,
      "step": 7782
    },
    {
      "epoch": 14.3601108033241,
      "grad_norm": 0.37421756982803345,
      "learning_rate": 2.334e-05,
      "loss": 0.016,
      "step": 7783
    },
    {
      "epoch": 14.361957525392429,
      "grad_norm": 0.48051688075065613,
      "learning_rate": 2.3343e-05,
      "loss": 0.0171,
      "step": 7784
    },
    {
      "epoch": 14.363804247460758,
      "grad_norm": 0.5618257522583008,
      "learning_rate": 2.3346e-05,
      "loss": 0.0261,
      "step": 7785
    },
    {
      "epoch": 14.365650969529085,
      "grad_norm": 0.3171500861644745,
      "learning_rate": 2.3349e-05,
      "loss": 0.01,
      "step": 7786
    },
    {
      "epoch": 14.367497691597414,
      "grad_norm": 0.531436562538147,
      "learning_rate": 2.3352e-05,
      "loss": 0.0131,
      "step": 7787
    },
    {
      "epoch": 14.369344413665743,
      "grad_norm": 0.9091717600822449,
      "learning_rate": 2.3355e-05,
      "loss": 0.0163,
      "step": 7788
    },
    {
      "epoch": 14.371191135734072,
      "grad_norm": 0.6591691374778748,
      "learning_rate": 2.3358e-05,
      "loss": 0.191,
      "step": 7789
    },
    {
      "epoch": 14.373037857802402,
      "grad_norm": 0.5781044363975525,
      "learning_rate": 2.3361e-05,
      "loss": 0.1545,
      "step": 7790
    },
    {
      "epoch": 14.374884579870729,
      "grad_norm": 0.7448709607124329,
      "learning_rate": 2.3364000000000002e-05,
      "loss": 0.174,
      "step": 7791
    },
    {
      "epoch": 14.376731301939058,
      "grad_norm": 0.6972495913505554,
      "learning_rate": 2.3367000000000002e-05,
      "loss": 0.2,
      "step": 7792
    },
    {
      "epoch": 14.378578024007387,
      "grad_norm": 0.7102606296539307,
      "learning_rate": 2.337e-05,
      "loss": 0.1773,
      "step": 7793
    },
    {
      "epoch": 14.380424746075716,
      "grad_norm": 0.4949325621128082,
      "learning_rate": 2.3373e-05,
      "loss": 0.1013,
      "step": 7794
    },
    {
      "epoch": 14.382271468144044,
      "grad_norm": 1.218247652053833,
      "learning_rate": 2.3376e-05,
      "loss": 0.1204,
      "step": 7795
    },
    {
      "epoch": 14.384118190212373,
      "grad_norm": 0.5108643770217896,
      "learning_rate": 2.3379e-05,
      "loss": 0.0815,
      "step": 7796
    },
    {
      "epoch": 14.385964912280702,
      "grad_norm": 0.569215714931488,
      "learning_rate": 2.3382e-05,
      "loss": 0.0851,
      "step": 7797
    },
    {
      "epoch": 14.387811634349031,
      "grad_norm": 0.6716434359550476,
      "learning_rate": 2.3385e-05,
      "loss": 0.0753,
      "step": 7798
    },
    {
      "epoch": 14.389658356417359,
      "grad_norm": 0.32701218128204346,
      "learning_rate": 2.3388e-05,
      "loss": 0.0476,
      "step": 7799
    },
    {
      "epoch": 14.391505078485688,
      "grad_norm": 0.6348007321357727,
      "learning_rate": 2.3391e-05,
      "loss": 0.0478,
      "step": 7800
    },
    {
      "epoch": 14.393351800554017,
      "grad_norm": 0.423272967338562,
      "learning_rate": 2.3394000000000003e-05,
      "loss": 0.0638,
      "step": 7801
    },
    {
      "epoch": 14.395198522622346,
      "grad_norm": 0.4461739659309387,
      "learning_rate": 2.3397000000000003e-05,
      "loss": 0.0546,
      "step": 7802
    },
    {
      "epoch": 14.397045244690673,
      "grad_norm": 0.42833179235458374,
      "learning_rate": 2.3400000000000003e-05,
      "loss": 0.0362,
      "step": 7803
    },
    {
      "epoch": 14.398891966759003,
      "grad_norm": 0.32603225111961365,
      "learning_rate": 2.3403e-05,
      "loss": 0.0235,
      "step": 7804
    },
    {
      "epoch": 14.400738688827332,
      "grad_norm": 0.3583817780017853,
      "learning_rate": 2.3406e-05,
      "loss": 0.0227,
      "step": 7805
    },
    {
      "epoch": 14.40258541089566,
      "grad_norm": 0.3128528594970703,
      "learning_rate": 2.3409e-05,
      "loss": 0.0217,
      "step": 7806
    },
    {
      "epoch": 14.404432132963988,
      "grad_norm": 0.37281113862991333,
      "learning_rate": 2.3412e-05,
      "loss": 0.0135,
      "step": 7807
    },
    {
      "epoch": 14.406278855032317,
      "grad_norm": 0.5349627137184143,
      "learning_rate": 2.3415e-05,
      "loss": 0.0169,
      "step": 7808
    },
    {
      "epoch": 14.408125577100646,
      "grad_norm": 0.5006557703018188,
      "learning_rate": 2.3418e-05,
      "loss": 0.0434,
      "step": 7809
    },
    {
      "epoch": 14.409972299168976,
      "grad_norm": 0.5111393332481384,
      "learning_rate": 2.3420999999999998e-05,
      "loss": 0.0145,
      "step": 7810
    },
    {
      "epoch": 14.411819021237303,
      "grad_norm": 0.3418116867542267,
      "learning_rate": 2.3424e-05,
      "loss": 0.0188,
      "step": 7811
    },
    {
      "epoch": 14.413665743305632,
      "grad_norm": 0.33915090560913086,
      "learning_rate": 2.3427e-05,
      "loss": 0.0185,
      "step": 7812
    },
    {
      "epoch": 14.415512465373961,
      "grad_norm": 0.36737731099128723,
      "learning_rate": 2.343e-05,
      "loss": 0.0212,
      "step": 7813
    },
    {
      "epoch": 14.41735918744229,
      "grad_norm": 0.14595548808574677,
      "learning_rate": 2.3433e-05,
      "loss": 0.0051,
      "step": 7814
    },
    {
      "epoch": 14.41920590951062,
      "grad_norm": 0.3603948950767517,
      "learning_rate": 2.3436e-05,
      "loss": 0.0158,
      "step": 7815
    },
    {
      "epoch": 14.421052631578947,
      "grad_norm": 0.35886791348457336,
      "learning_rate": 2.3439e-05,
      "loss": 0.011,
      "step": 7816
    },
    {
      "epoch": 14.422899353647276,
      "grad_norm": 0.4424641728401184,
      "learning_rate": 2.3442e-05,
      "loss": 0.0139,
      "step": 7817
    },
    {
      "epoch": 14.424746075715605,
      "grad_norm": 0.21980373561382294,
      "learning_rate": 2.3445e-05,
      "loss": 0.0076,
      "step": 7818
    },
    {
      "epoch": 14.426592797783934,
      "grad_norm": 0.2293301224708557,
      "learning_rate": 2.3448e-05,
      "loss": 0.0113,
      "step": 7819
    },
    {
      "epoch": 14.428439519852262,
      "grad_norm": 0.4105096161365509,
      "learning_rate": 2.3451e-05,
      "loss": 0.0114,
      "step": 7820
    },
    {
      "epoch": 14.43028624192059,
      "grad_norm": 0.5419213175773621,
      "learning_rate": 2.3454000000000003e-05,
      "loss": 0.0182,
      "step": 7821
    },
    {
      "epoch": 14.43213296398892,
      "grad_norm": 0.21369367837905884,
      "learning_rate": 2.3457000000000003e-05,
      "loss": 0.0086,
      "step": 7822
    },
    {
      "epoch": 14.43397968605725,
      "grad_norm": 0.39795318245887756,
      "learning_rate": 2.3460000000000002e-05,
      "loss": 0.0152,
      "step": 7823
    },
    {
      "epoch": 14.435826408125576,
      "grad_norm": 0.6262432932853699,
      "learning_rate": 2.3463000000000002e-05,
      "loss": 0.0244,
      "step": 7824
    },
    {
      "epoch": 14.437673130193906,
      "grad_norm": 0.5282998085021973,
      "learning_rate": 2.3466000000000002e-05,
      "loss": 0.0165,
      "step": 7825
    },
    {
      "epoch": 14.439519852262235,
      "grad_norm": 0.7346487045288086,
      "learning_rate": 2.3469000000000002e-05,
      "loss": 0.0245,
      "step": 7826
    },
    {
      "epoch": 14.441366574330564,
      "grad_norm": 0.45807477831840515,
      "learning_rate": 2.3472e-05,
      "loss": 0.0092,
      "step": 7827
    },
    {
      "epoch": 14.443213296398891,
      "grad_norm": 0.5049607753753662,
      "learning_rate": 2.3475e-05,
      "loss": 0.0169,
      "step": 7828
    },
    {
      "epoch": 14.44506001846722,
      "grad_norm": 0.5809155702590942,
      "learning_rate": 2.3477999999999998e-05,
      "loss": 0.0192,
      "step": 7829
    },
    {
      "epoch": 14.44690674053555,
      "grad_norm": 0.45647743344306946,
      "learning_rate": 2.3480999999999998e-05,
      "loss": 0.0137,
      "step": 7830
    },
    {
      "epoch": 14.448753462603879,
      "grad_norm": 0.5045316815376282,
      "learning_rate": 2.3484e-05,
      "loss": 0.0242,
      "step": 7831
    },
    {
      "epoch": 14.450600184672206,
      "grad_norm": 0.742444634437561,
      "learning_rate": 2.3487e-05,
      "loss": 0.0121,
      "step": 7832
    },
    {
      "epoch": 14.452446906740535,
      "grad_norm": 0.8437967896461487,
      "learning_rate": 2.349e-05,
      "loss": 0.0217,
      "step": 7833
    },
    {
      "epoch": 14.454293628808864,
      "grad_norm": 0.2819928228855133,
      "learning_rate": 2.3493e-05,
      "loss": 0.0103,
      "step": 7834
    },
    {
      "epoch": 14.456140350877194,
      "grad_norm": 0.2498871237039566,
      "learning_rate": 2.3496e-05,
      "loss": 0.0074,
      "step": 7835
    },
    {
      "epoch": 14.45798707294552,
      "grad_norm": 0.6369275450706482,
      "learning_rate": 2.3499e-05,
      "loss": 0.025,
      "step": 7836
    },
    {
      "epoch": 14.45983379501385,
      "grad_norm": 0.5697141885757446,
      "learning_rate": 2.3502e-05,
      "loss": 0.0207,
      "step": 7837
    },
    {
      "epoch": 14.46168051708218,
      "grad_norm": 0.7965555191040039,
      "learning_rate": 2.3505e-05,
      "loss": 0.0286,
      "step": 7838
    },
    {
      "epoch": 14.463527239150508,
      "grad_norm": 1.235569715499878,
      "learning_rate": 2.3508e-05,
      "loss": 0.2713,
      "step": 7839
    },
    {
      "epoch": 14.465373961218837,
      "grad_norm": 0.6825758218765259,
      "learning_rate": 2.3511e-05,
      "loss": 0.1966,
      "step": 7840
    },
    {
      "epoch": 14.467220683287165,
      "grad_norm": 0.6262037754058838,
      "learning_rate": 2.3514000000000002e-05,
      "loss": 0.1178,
      "step": 7841
    },
    {
      "epoch": 14.469067405355494,
      "grad_norm": 1.8128679990768433,
      "learning_rate": 2.3517000000000002e-05,
      "loss": 0.1999,
      "step": 7842
    },
    {
      "epoch": 14.470914127423823,
      "grad_norm": 0.5639660954475403,
      "learning_rate": 2.3520000000000002e-05,
      "loss": 0.1532,
      "step": 7843
    },
    {
      "epoch": 14.472760849492152,
      "grad_norm": 0.5506031513214111,
      "learning_rate": 2.3523e-05,
      "loss": 0.1075,
      "step": 7844
    },
    {
      "epoch": 14.47460757156048,
      "grad_norm": 0.49700742959976196,
      "learning_rate": 2.3526e-05,
      "loss": 0.1151,
      "step": 7845
    },
    {
      "epoch": 14.476454293628809,
      "grad_norm": 0.5941647887229919,
      "learning_rate": 2.3529e-05,
      "loss": 0.115,
      "step": 7846
    },
    {
      "epoch": 14.478301015697138,
      "grad_norm": 0.6988394260406494,
      "learning_rate": 2.3532e-05,
      "loss": 0.0892,
      "step": 7847
    },
    {
      "epoch": 14.480147737765467,
      "grad_norm": 0.5593818426132202,
      "learning_rate": 2.3535e-05,
      "loss": 0.0557,
      "step": 7848
    },
    {
      "epoch": 14.481994459833794,
      "grad_norm": 0.41339749097824097,
      "learning_rate": 2.3538e-05,
      "loss": 0.0509,
      "step": 7849
    },
    {
      "epoch": 14.483841181902124,
      "grad_norm": 1.1100635528564453,
      "learning_rate": 2.3541e-05,
      "loss": 0.0631,
      "step": 7850
    },
    {
      "epoch": 14.485687903970453,
      "grad_norm": 0.7645638585090637,
      "learning_rate": 2.3544000000000004e-05,
      "loss": 0.1023,
      "step": 7851
    },
    {
      "epoch": 14.487534626038782,
      "grad_norm": 0.7224724888801575,
      "learning_rate": 2.3547000000000003e-05,
      "loss": 0.0816,
      "step": 7852
    },
    {
      "epoch": 14.48938134810711,
      "grad_norm": 0.26063552498817444,
      "learning_rate": 2.3550000000000003e-05,
      "loss": 0.0259,
      "step": 7853
    },
    {
      "epoch": 14.491228070175438,
      "grad_norm": 0.6297289729118347,
      "learning_rate": 2.3553e-05,
      "loss": 0.0648,
      "step": 7854
    },
    {
      "epoch": 14.493074792243767,
      "grad_norm": 0.5482573509216309,
      "learning_rate": 2.3556e-05,
      "loss": 0.0513,
      "step": 7855
    },
    {
      "epoch": 14.494921514312097,
      "grad_norm": 0.5147392153739929,
      "learning_rate": 2.3559e-05,
      "loss": 0.0398,
      "step": 7856
    },
    {
      "epoch": 14.496768236380424,
      "grad_norm": 0.398120641708374,
      "learning_rate": 2.3562e-05,
      "loss": 0.0416,
      "step": 7857
    },
    {
      "epoch": 14.498614958448753,
      "grad_norm": 0.3588913679122925,
      "learning_rate": 2.3565e-05,
      "loss": 0.0191,
      "step": 7858
    },
    {
      "epoch": 14.500461680517082,
      "grad_norm": 0.28668269515037537,
      "learning_rate": 2.3568e-05,
      "loss": 0.0169,
      "step": 7859
    },
    {
      "epoch": 14.502308402585411,
      "grad_norm": 0.3392992913722992,
      "learning_rate": 2.3571e-05,
      "loss": 0.0221,
      "step": 7860
    },
    {
      "epoch": 14.504155124653739,
      "grad_norm": 0.2852858901023865,
      "learning_rate": 2.3574e-05,
      "loss": 0.0154,
      "step": 7861
    },
    {
      "epoch": 14.506001846722068,
      "grad_norm": 0.27077481150627136,
      "learning_rate": 2.3577e-05,
      "loss": 0.013,
      "step": 7862
    },
    {
      "epoch": 14.507848568790397,
      "grad_norm": 0.3104352355003357,
      "learning_rate": 2.358e-05,
      "loss": 0.0121,
      "step": 7863
    },
    {
      "epoch": 14.509695290858726,
      "grad_norm": 0.32454049587249756,
      "learning_rate": 2.3583e-05,
      "loss": 0.0218,
      "step": 7864
    },
    {
      "epoch": 14.511542012927055,
      "grad_norm": 0.7309545278549194,
      "learning_rate": 2.3586e-05,
      "loss": 0.0169,
      "step": 7865
    },
    {
      "epoch": 14.513388734995383,
      "grad_norm": 0.20314842462539673,
      "learning_rate": 2.3589e-05,
      "loss": 0.0071,
      "step": 7866
    },
    {
      "epoch": 14.515235457063712,
      "grad_norm": 0.6824448108673096,
      "learning_rate": 2.3592e-05,
      "loss": 0.0474,
      "step": 7867
    },
    {
      "epoch": 14.517082179132041,
      "grad_norm": 0.6702181696891785,
      "learning_rate": 2.3595e-05,
      "loss": 0.0143,
      "step": 7868
    },
    {
      "epoch": 14.51892890120037,
      "grad_norm": 0.5272572636604309,
      "learning_rate": 2.3598e-05,
      "loss": 0.0121,
      "step": 7869
    },
    {
      "epoch": 14.520775623268698,
      "grad_norm": 0.5622138381004333,
      "learning_rate": 2.3601e-05,
      "loss": 0.0488,
      "step": 7870
    },
    {
      "epoch": 14.522622345337027,
      "grad_norm": 0.26477548480033875,
      "learning_rate": 2.3604000000000003e-05,
      "loss": 0.0078,
      "step": 7871
    },
    {
      "epoch": 14.524469067405356,
      "grad_norm": 0.45857515931129456,
      "learning_rate": 2.3607000000000003e-05,
      "loss": 0.0196,
      "step": 7872
    },
    {
      "epoch": 14.526315789473685,
      "grad_norm": 0.6599807739257812,
      "learning_rate": 2.3610000000000003e-05,
      "loss": 0.0134,
      "step": 7873
    },
    {
      "epoch": 14.528162511542012,
      "grad_norm": 0.3410775363445282,
      "learning_rate": 2.3613000000000002e-05,
      "loss": 0.0098,
      "step": 7874
    },
    {
      "epoch": 14.530009233610341,
      "grad_norm": 0.5473353266716003,
      "learning_rate": 2.3616000000000002e-05,
      "loss": 0.0162,
      "step": 7875
    },
    {
      "epoch": 14.53185595567867,
      "grad_norm": 0.626075267791748,
      "learning_rate": 2.3619000000000002e-05,
      "loss": 0.0172,
      "step": 7876
    },
    {
      "epoch": 14.533702677747,
      "grad_norm": 0.507577121257782,
      "learning_rate": 2.3622000000000002e-05,
      "loss": 0.0161,
      "step": 7877
    },
    {
      "epoch": 14.535549399815327,
      "grad_norm": 0.44902390241622925,
      "learning_rate": 2.3624999999999998e-05,
      "loss": 0.0278,
      "step": 7878
    },
    {
      "epoch": 14.537396121883656,
      "grad_norm": 0.4249652922153473,
      "learning_rate": 2.3627999999999998e-05,
      "loss": 0.0154,
      "step": 7879
    },
    {
      "epoch": 14.539242843951985,
      "grad_norm": 0.5330266356468201,
      "learning_rate": 2.3630999999999998e-05,
      "loss": 0.0079,
      "step": 7880
    },
    {
      "epoch": 14.541089566020315,
      "grad_norm": 0.5065855979919434,
      "learning_rate": 2.3633999999999998e-05,
      "loss": 0.0167,
      "step": 7881
    },
    {
      "epoch": 14.542936288088642,
      "grad_norm": 0.4754953384399414,
      "learning_rate": 2.3637e-05,
      "loss": 0.0159,
      "step": 7882
    },
    {
      "epoch": 14.544783010156971,
      "grad_norm": 0.4979807138442993,
      "learning_rate": 2.364e-05,
      "loss": 0.0181,
      "step": 7883
    },
    {
      "epoch": 14.5466297322253,
      "grad_norm": 0.5053218007087708,
      "learning_rate": 2.3643e-05,
      "loss": 0.0133,
      "step": 7884
    },
    {
      "epoch": 14.54847645429363,
      "grad_norm": 0.4692944586277008,
      "learning_rate": 2.3646e-05,
      "loss": 0.0168,
      "step": 7885
    },
    {
      "epoch": 14.550323176361957,
      "grad_norm": 0.521899938583374,
      "learning_rate": 2.3649e-05,
      "loss": 0.0204,
      "step": 7886
    },
    {
      "epoch": 14.552169898430286,
      "grad_norm": 0.8429057002067566,
      "learning_rate": 2.3652e-05,
      "loss": 0.0214,
      "step": 7887
    },
    {
      "epoch": 14.554016620498615,
      "grad_norm": 0.37914028763771057,
      "learning_rate": 2.3655e-05,
      "loss": 0.0169,
      "step": 7888
    },
    {
      "epoch": 14.555863342566944,
      "grad_norm": 0.636502742767334,
      "learning_rate": 2.3658e-05,
      "loss": 0.184,
      "step": 7889
    },
    {
      "epoch": 14.557710064635273,
      "grad_norm": 0.7152822017669678,
      "learning_rate": 2.3661e-05,
      "loss": 0.1724,
      "step": 7890
    },
    {
      "epoch": 14.5595567867036,
      "grad_norm": 0.49054843187332153,
      "learning_rate": 2.3664e-05,
      "loss": 0.1158,
      "step": 7891
    },
    {
      "epoch": 14.56140350877193,
      "grad_norm": 0.5772072076797485,
      "learning_rate": 2.3667000000000002e-05,
      "loss": 0.1338,
      "step": 7892
    },
    {
      "epoch": 14.563250230840259,
      "grad_norm": 0.5550270676612854,
      "learning_rate": 2.3670000000000002e-05,
      "loss": 0.1215,
      "step": 7893
    },
    {
      "epoch": 14.565096952908588,
      "grad_norm": 0.5168792605400085,
      "learning_rate": 2.3673000000000002e-05,
      "loss": 0.0839,
      "step": 7894
    },
    {
      "epoch": 14.566943674976915,
      "grad_norm": 0.47602367401123047,
      "learning_rate": 2.3676e-05,
      "loss": 0.0906,
      "step": 7895
    },
    {
      "epoch": 14.568790397045245,
      "grad_norm": 0.6356981992721558,
      "learning_rate": 2.3679e-05,
      "loss": 0.0955,
      "step": 7896
    },
    {
      "epoch": 14.570637119113574,
      "grad_norm": 0.5250639319419861,
      "learning_rate": 2.3682e-05,
      "loss": 0.0971,
      "step": 7897
    },
    {
      "epoch": 14.572483841181903,
      "grad_norm": 0.41396352648735046,
      "learning_rate": 2.3685e-05,
      "loss": 0.0668,
      "step": 7898
    },
    {
      "epoch": 14.57433056325023,
      "grad_norm": 0.5052721500396729,
      "learning_rate": 2.3688e-05,
      "loss": 0.0602,
      "step": 7899
    },
    {
      "epoch": 14.57617728531856,
      "grad_norm": 0.4817100763320923,
      "learning_rate": 2.3691e-05,
      "loss": 0.1045,
      "step": 7900
    },
    {
      "epoch": 14.578024007386889,
      "grad_norm": 0.33338767290115356,
      "learning_rate": 2.3694e-05,
      "loss": 0.0351,
      "step": 7901
    },
    {
      "epoch": 14.579870729455218,
      "grad_norm": 0.41364586353302,
      "learning_rate": 2.3697000000000004e-05,
      "loss": 0.0441,
      "step": 7902
    },
    {
      "epoch": 14.581717451523545,
      "grad_norm": 0.47825878858566284,
      "learning_rate": 2.37e-05,
      "loss": 0.0475,
      "step": 7903
    },
    {
      "epoch": 14.583564173591874,
      "grad_norm": 0.5705379247665405,
      "learning_rate": 2.3703e-05,
      "loss": 0.0455,
      "step": 7904
    },
    {
      "epoch": 14.585410895660203,
      "grad_norm": 0.4665312170982361,
      "learning_rate": 2.3706e-05,
      "loss": 0.0301,
      "step": 7905
    },
    {
      "epoch": 14.587257617728532,
      "grad_norm": 0.2677137553691864,
      "learning_rate": 2.3709e-05,
      "loss": 0.0145,
      "step": 7906
    },
    {
      "epoch": 14.58910433979686,
      "grad_norm": 0.34597480297088623,
      "learning_rate": 2.3712e-05,
      "loss": 0.0105,
      "step": 7907
    },
    {
      "epoch": 14.590951061865189,
      "grad_norm": 0.346442848443985,
      "learning_rate": 2.3715e-05,
      "loss": 0.0527,
      "step": 7908
    },
    {
      "epoch": 14.592797783933518,
      "grad_norm": 0.4659012258052826,
      "learning_rate": 2.3718e-05,
      "loss": 0.0168,
      "step": 7909
    },
    {
      "epoch": 14.594644506001847,
      "grad_norm": 0.3821612000465393,
      "learning_rate": 2.3721e-05,
      "loss": 0.0266,
      "step": 7910
    },
    {
      "epoch": 14.596491228070175,
      "grad_norm": 0.4945525825023651,
      "learning_rate": 2.3724e-05,
      "loss": 0.0283,
      "step": 7911
    },
    {
      "epoch": 14.598337950138504,
      "grad_norm": 0.37642475962638855,
      "learning_rate": 2.3727000000000002e-05,
      "loss": 0.0213,
      "step": 7912
    },
    {
      "epoch": 14.600184672206833,
      "grad_norm": 0.5657537579536438,
      "learning_rate": 2.373e-05,
      "loss": 0.0288,
      "step": 7913
    },
    {
      "epoch": 14.602031394275162,
      "grad_norm": 0.5660479664802551,
      "learning_rate": 2.3733e-05,
      "loss": 0.0282,
      "step": 7914
    },
    {
      "epoch": 14.603878116343491,
      "grad_norm": 0.4243585467338562,
      "learning_rate": 2.3736e-05,
      "loss": 0.0186,
      "step": 7915
    },
    {
      "epoch": 14.605724838411819,
      "grad_norm": 0.3085576295852661,
      "learning_rate": 2.3739e-05,
      "loss": 0.0117,
      "step": 7916
    },
    {
      "epoch": 14.607571560480148,
      "grad_norm": 0.223404198884964,
      "learning_rate": 2.3742e-05,
      "loss": 0.0106,
      "step": 7917
    },
    {
      "epoch": 14.609418282548477,
      "grad_norm": 0.4205861985683441,
      "learning_rate": 2.3745e-05,
      "loss": 0.0199,
      "step": 7918
    },
    {
      "epoch": 14.611265004616806,
      "grad_norm": 0.35712581872940063,
      "learning_rate": 2.3748e-05,
      "loss": 0.0156,
      "step": 7919
    },
    {
      "epoch": 14.613111726685133,
      "grad_norm": 0.37940770387649536,
      "learning_rate": 2.3751e-05,
      "loss": 0.0313,
      "step": 7920
    },
    {
      "epoch": 14.614958448753463,
      "grad_norm": 0.314486026763916,
      "learning_rate": 2.3754e-05,
      "loss": 0.0168,
      "step": 7921
    },
    {
      "epoch": 14.616805170821792,
      "grad_norm": 0.37937620282173157,
      "learning_rate": 2.3757000000000003e-05,
      "loss": 0.0139,
      "step": 7922
    },
    {
      "epoch": 14.61865189289012,
      "grad_norm": 0.36704617738723755,
      "learning_rate": 2.3760000000000003e-05,
      "loss": 0.0252,
      "step": 7923
    },
    {
      "epoch": 14.620498614958448,
      "grad_norm": 0.25925007462501526,
      "learning_rate": 2.3763000000000003e-05,
      "loss": 0.0096,
      "step": 7924
    },
    {
      "epoch": 14.622345337026777,
      "grad_norm": 0.38265901803970337,
      "learning_rate": 2.3766000000000003e-05,
      "loss": 0.0115,
      "step": 7925
    },
    {
      "epoch": 14.624192059095106,
      "grad_norm": 0.5699060559272766,
      "learning_rate": 2.3769000000000002e-05,
      "loss": 0.0143,
      "step": 7926
    },
    {
      "epoch": 14.626038781163436,
      "grad_norm": 0.369829386472702,
      "learning_rate": 2.3772e-05,
      "loss": 0.0146,
      "step": 7927
    },
    {
      "epoch": 14.627885503231763,
      "grad_norm": 0.5048293471336365,
      "learning_rate": 2.3775e-05,
      "loss": 0.0138,
      "step": 7928
    },
    {
      "epoch": 14.629732225300092,
      "grad_norm": 0.26562240719795227,
      "learning_rate": 2.3778e-05,
      "loss": 0.0103,
      "step": 7929
    },
    {
      "epoch": 14.631578947368421,
      "grad_norm": 0.28638893365859985,
      "learning_rate": 2.3780999999999998e-05,
      "loss": 0.0119,
      "step": 7930
    },
    {
      "epoch": 14.63342566943675,
      "grad_norm": 0.5839011073112488,
      "learning_rate": 2.3783999999999998e-05,
      "loss": 0.0164,
      "step": 7931
    },
    {
      "epoch": 14.635272391505078,
      "grad_norm": 0.3103630840778351,
      "learning_rate": 2.3787e-05,
      "loss": 0.0116,
      "step": 7932
    },
    {
      "epoch": 14.637119113573407,
      "grad_norm": 1.3576741218566895,
      "learning_rate": 2.379e-05,
      "loss": 0.014,
      "step": 7933
    },
    {
      "epoch": 14.638965835641736,
      "grad_norm": 1.0917797088623047,
      "learning_rate": 2.3793e-05,
      "loss": 0.0227,
      "step": 7934
    },
    {
      "epoch": 14.640812557710065,
      "grad_norm": 0.32580968737602234,
      "learning_rate": 2.3796e-05,
      "loss": 0.0073,
      "step": 7935
    },
    {
      "epoch": 14.642659279778393,
      "grad_norm": 0.285483181476593,
      "learning_rate": 2.3799e-05,
      "loss": 0.0104,
      "step": 7936
    },
    {
      "epoch": 14.644506001846722,
      "grad_norm": 0.40575113892555237,
      "learning_rate": 2.3802e-05,
      "loss": 0.0133,
      "step": 7937
    },
    {
      "epoch": 14.64635272391505,
      "grad_norm": 0.3636173605918884,
      "learning_rate": 2.3805e-05,
      "loss": 0.0177,
      "step": 7938
    },
    {
      "epoch": 14.64819944598338,
      "grad_norm": 0.8212683200836182,
      "learning_rate": 2.3808e-05,
      "loss": 0.2281,
      "step": 7939
    },
    {
      "epoch": 14.65004616805171,
      "grad_norm": 0.9566912651062012,
      "learning_rate": 2.3811e-05,
      "loss": 0.2628,
      "step": 7940
    },
    {
      "epoch": 14.651892890120036,
      "grad_norm": 0.6159496903419495,
      "learning_rate": 2.3814e-05,
      "loss": 0.1634,
      "step": 7941
    },
    {
      "epoch": 14.653739612188366,
      "grad_norm": 0.5956009030342102,
      "learning_rate": 2.3817000000000003e-05,
      "loss": 0.1371,
      "step": 7942
    },
    {
      "epoch": 14.655586334256695,
      "grad_norm": 0.7215802669525146,
      "learning_rate": 2.3820000000000002e-05,
      "loss": 0.1136,
      "step": 7943
    },
    {
      "epoch": 14.657433056325024,
      "grad_norm": 0.41611161828041077,
      "learning_rate": 2.3823000000000002e-05,
      "loss": 0.0957,
      "step": 7944
    },
    {
      "epoch": 14.659279778393351,
      "grad_norm": 0.5038338899612427,
      "learning_rate": 2.3826000000000002e-05,
      "loss": 0.0958,
      "step": 7945
    },
    {
      "epoch": 14.66112650046168,
      "grad_norm": 0.5414931774139404,
      "learning_rate": 2.3829000000000002e-05,
      "loss": 0.1228,
      "step": 7946
    },
    {
      "epoch": 14.66297322253001,
      "grad_norm": 0.5237676501274109,
      "learning_rate": 2.3832e-05,
      "loss": 0.0741,
      "step": 7947
    },
    {
      "epoch": 14.664819944598339,
      "grad_norm": 1.1172302961349487,
      "learning_rate": 2.3835e-05,
      "loss": 0.1129,
      "step": 7948
    },
    {
      "epoch": 14.666666666666666,
      "grad_norm": 0.6457228660583496,
      "learning_rate": 2.3838e-05,
      "loss": 0.0944,
      "step": 7949
    },
    {
      "epoch": 14.668513388734995,
      "grad_norm": 0.576670229434967,
      "learning_rate": 2.3841e-05,
      "loss": 0.0673,
      "step": 7950
    },
    {
      "epoch": 14.670360110803324,
      "grad_norm": 0.3982366919517517,
      "learning_rate": 2.3844e-05,
      "loss": 0.0644,
      "step": 7951
    },
    {
      "epoch": 14.672206832871654,
      "grad_norm": 1.198979139328003,
      "learning_rate": 2.3847e-05,
      "loss": 0.0439,
      "step": 7952
    },
    {
      "epoch": 14.67405355493998,
      "grad_norm": 0.3780513107776642,
      "learning_rate": 2.385e-05,
      "loss": 0.046,
      "step": 7953
    },
    {
      "epoch": 14.67590027700831,
      "grad_norm": 0.3405984938144684,
      "learning_rate": 2.3853e-05,
      "loss": 0.0152,
      "step": 7954
    },
    {
      "epoch": 14.67774699907664,
      "grad_norm": 0.3067670166492462,
      "learning_rate": 2.3856e-05,
      "loss": 0.0197,
      "step": 7955
    },
    {
      "epoch": 14.679593721144968,
      "grad_norm": 0.6061209440231323,
      "learning_rate": 2.3859e-05,
      "loss": 0.0246,
      "step": 7956
    },
    {
      "epoch": 14.681440443213296,
      "grad_norm": 0.4019205868244171,
      "learning_rate": 2.3862e-05,
      "loss": 0.0198,
      "step": 7957
    },
    {
      "epoch": 14.683287165281625,
      "grad_norm": 0.49981725215911865,
      "learning_rate": 2.3865e-05,
      "loss": 0.0207,
      "step": 7958
    },
    {
      "epoch": 14.685133887349954,
      "grad_norm": 0.3462250828742981,
      "learning_rate": 2.3868e-05,
      "loss": 0.0238,
      "step": 7959
    },
    {
      "epoch": 14.686980609418283,
      "grad_norm": 0.381491094827652,
      "learning_rate": 2.3871e-05,
      "loss": 0.0231,
      "step": 7960
    },
    {
      "epoch": 14.68882733148661,
      "grad_norm": 0.42520666122436523,
      "learning_rate": 2.3874e-05,
      "loss": 0.0258,
      "step": 7961
    },
    {
      "epoch": 14.69067405355494,
      "grad_norm": 0.417881578207016,
      "learning_rate": 2.3877000000000002e-05,
      "loss": 0.0287,
      "step": 7962
    },
    {
      "epoch": 14.692520775623269,
      "grad_norm": 0.43953225016593933,
      "learning_rate": 2.3880000000000002e-05,
      "loss": 0.0199,
      "step": 7963
    },
    {
      "epoch": 14.694367497691598,
      "grad_norm": 0.3966647982597351,
      "learning_rate": 2.3883e-05,
      "loss": 0.0178,
      "step": 7964
    },
    {
      "epoch": 14.696214219759927,
      "grad_norm": 0.9427505731582642,
      "learning_rate": 2.3886e-05,
      "loss": 0.036,
      "step": 7965
    },
    {
      "epoch": 14.698060941828254,
      "grad_norm": 0.461514949798584,
      "learning_rate": 2.3889e-05,
      "loss": 0.019,
      "step": 7966
    },
    {
      "epoch": 14.699907663896584,
      "grad_norm": 0.3639925718307495,
      "learning_rate": 2.3892e-05,
      "loss": 0.0103,
      "step": 7967
    },
    {
      "epoch": 14.701754385964913,
      "grad_norm": 0.3159520924091339,
      "learning_rate": 2.3895e-05,
      "loss": 0.0201,
      "step": 7968
    },
    {
      "epoch": 14.703601108033242,
      "grad_norm": 0.44609788060188293,
      "learning_rate": 2.3898e-05,
      "loss": 0.0196,
      "step": 7969
    },
    {
      "epoch": 14.70544783010157,
      "grad_norm": 0.4839285612106323,
      "learning_rate": 2.3901e-05,
      "loss": 0.0253,
      "step": 7970
    },
    {
      "epoch": 14.707294552169898,
      "grad_norm": 0.2293621152639389,
      "learning_rate": 2.3904e-05,
      "loss": 0.0074,
      "step": 7971
    },
    {
      "epoch": 14.709141274238227,
      "grad_norm": 0.26273688673973083,
      "learning_rate": 2.3907000000000003e-05,
      "loss": 0.0133,
      "step": 7972
    },
    {
      "epoch": 14.710987996306557,
      "grad_norm": 0.24619637429714203,
      "learning_rate": 2.3910000000000003e-05,
      "loss": 0.0125,
      "step": 7973
    },
    {
      "epoch": 14.712834718374884,
      "grad_norm": 0.8715478777885437,
      "learning_rate": 2.3913000000000003e-05,
      "loss": 0.0371,
      "step": 7974
    },
    {
      "epoch": 14.714681440443213,
      "grad_norm": 0.334750235080719,
      "learning_rate": 2.3916000000000003e-05,
      "loss": 0.0129,
      "step": 7975
    },
    {
      "epoch": 14.716528162511542,
      "grad_norm": 0.27692723274230957,
      "learning_rate": 2.3919e-05,
      "loss": 0.0161,
      "step": 7976
    },
    {
      "epoch": 14.718374884579871,
      "grad_norm": 0.5121621489524841,
      "learning_rate": 2.3922e-05,
      "loss": 0.0272,
      "step": 7977
    },
    {
      "epoch": 14.720221606648199,
      "grad_norm": 0.46436607837677,
      "learning_rate": 2.3925e-05,
      "loss": 0.0216,
      "step": 7978
    },
    {
      "epoch": 14.722068328716528,
      "grad_norm": 0.683923065662384,
      "learning_rate": 2.3928e-05,
      "loss": 0.0215,
      "step": 7979
    },
    {
      "epoch": 14.723915050784857,
      "grad_norm": 0.283103346824646,
      "learning_rate": 2.3931e-05,
      "loss": 0.0199,
      "step": 7980
    },
    {
      "epoch": 14.725761772853186,
      "grad_norm": 0.31182169914245605,
      "learning_rate": 2.3933999999999998e-05,
      "loss": 0.0139,
      "step": 7981
    },
    {
      "epoch": 14.727608494921514,
      "grad_norm": 0.36937060952186584,
      "learning_rate": 2.3937e-05,
      "loss": 0.0166,
      "step": 7982
    },
    {
      "epoch": 14.729455216989843,
      "grad_norm": 0.44678324460983276,
      "learning_rate": 2.394e-05,
      "loss": 0.0181,
      "step": 7983
    },
    {
      "epoch": 14.731301939058172,
      "grad_norm": 0.41206908226013184,
      "learning_rate": 2.3943e-05,
      "loss": 0.0191,
      "step": 7984
    },
    {
      "epoch": 14.733148661126501,
      "grad_norm": 0.5686429142951965,
      "learning_rate": 2.3946e-05,
      "loss": 0.0221,
      "step": 7985
    },
    {
      "epoch": 14.734995383194828,
      "grad_norm": 0.5409742593765259,
      "learning_rate": 2.3949e-05,
      "loss": 0.0241,
      "step": 7986
    },
    {
      "epoch": 14.736842105263158,
      "grad_norm": 0.6201287508010864,
      "learning_rate": 2.3952e-05,
      "loss": 0.0248,
      "step": 7987
    },
    {
      "epoch": 14.738688827331487,
      "grad_norm": 2.362740993499756,
      "learning_rate": 2.3955e-05,
      "loss": 0.0808,
      "step": 7988
    },
    {
      "epoch": 14.740535549399816,
      "grad_norm": 0.7609465718269348,
      "learning_rate": 2.3958e-05,
      "loss": 0.2072,
      "step": 7989
    },
    {
      "epoch": 14.742382271468145,
      "grad_norm": 0.48303288221359253,
      "learning_rate": 2.3961e-05,
      "loss": 0.1277,
      "step": 7990
    },
    {
      "epoch": 14.744228993536472,
      "grad_norm": 0.8768311142921448,
      "learning_rate": 2.3964e-05,
      "loss": 0.1706,
      "step": 7991
    },
    {
      "epoch": 14.746075715604801,
      "grad_norm": 0.6291162967681885,
      "learning_rate": 2.3967000000000003e-05,
      "loss": 0.1365,
      "step": 7992
    },
    {
      "epoch": 14.74792243767313,
      "grad_norm": 0.5517061948776245,
      "learning_rate": 2.3970000000000003e-05,
      "loss": 0.1203,
      "step": 7993
    },
    {
      "epoch": 14.749769159741458,
      "grad_norm": 0.5301569104194641,
      "learning_rate": 2.3973000000000002e-05,
      "loss": 0.13,
      "step": 7994
    },
    {
      "epoch": 14.751615881809787,
      "grad_norm": 0.4237849712371826,
      "learning_rate": 2.3976000000000002e-05,
      "loss": 0.0786,
      "step": 7995
    },
    {
      "epoch": 14.753462603878116,
      "grad_norm": 0.7277824282646179,
      "learning_rate": 2.3979000000000002e-05,
      "loss": 0.0815,
      "step": 7996
    },
    {
      "epoch": 14.755309325946445,
      "grad_norm": 0.4558088779449463,
      "learning_rate": 2.3982000000000002e-05,
      "loss": 0.0836,
      "step": 7997
    },
    {
      "epoch": 14.757156048014775,
      "grad_norm": 0.3935530185699463,
      "learning_rate": 2.3985e-05,
      "loss": 0.0501,
      "step": 7998
    },
    {
      "epoch": 14.759002770083102,
      "grad_norm": 0.48821765184402466,
      "learning_rate": 2.3988e-05,
      "loss": 0.0783,
      "step": 7999
    },
    {
      "epoch": 14.760849492151431,
      "grad_norm": 0.8362130522727966,
      "learning_rate": 2.3991e-05,
      "loss": 0.0436,
      "step": 8000
    },
    {
      "epoch": 14.760849492151431,
      "eval_cer": 0.1098841278481997,
      "eval_loss": 0.31252992153167725,
      "eval_runtime": 16.3111,
      "eval_samples_per_second": 18.638,
      "eval_steps_per_second": 0.613,
      "eval_wer": 0.38468917881811204,
      "step": 8000
    },
    {
      "epoch": 14.76269621421976,
      "grad_norm": 0.43701910972595215,
      "learning_rate": 2.3993999999999998e-05,
      "loss": 0.0576,
      "step": 8001
    },
    {
      "epoch": 14.76454293628809,
      "grad_norm": 0.7060490250587463,
      "learning_rate": 2.3997e-05,
      "loss": 0.0971,
      "step": 8002
    },
    {
      "epoch": 14.766389658356417,
      "grad_norm": 0.3635128140449524,
      "learning_rate": 2.4e-05,
      "loss": 0.0305,
      "step": 8003
    },
    {
      "epoch": 14.768236380424746,
      "grad_norm": 0.3958315849304199,
      "learning_rate": 2.4003e-05,
      "loss": 0.0637,
      "step": 8004
    },
    {
      "epoch": 14.770083102493075,
      "grad_norm": 0.3374282121658325,
      "learning_rate": 2.4006e-05,
      "loss": 0.0186,
      "step": 8005
    },
    {
      "epoch": 14.771929824561404,
      "grad_norm": 0.3075411915779114,
      "learning_rate": 2.4009e-05,
      "loss": 0.0136,
      "step": 8006
    },
    {
      "epoch": 14.773776546629731,
      "grad_norm": 0.6738809943199158,
      "learning_rate": 2.4012e-05,
      "loss": 0.0266,
      "step": 8007
    },
    {
      "epoch": 14.77562326869806,
      "grad_norm": 0.1647556722164154,
      "learning_rate": 2.4015e-05,
      "loss": 0.012,
      "step": 8008
    },
    {
      "epoch": 14.77746999076639,
      "grad_norm": 0.3219056725502014,
      "learning_rate": 2.4018e-05,
      "loss": 0.0206,
      "step": 8009
    },
    {
      "epoch": 14.779316712834719,
      "grad_norm": 0.341482937335968,
      "learning_rate": 2.4021e-05,
      "loss": 0.0144,
      "step": 8010
    },
    {
      "epoch": 14.781163434903046,
      "grad_norm": 0.24657687544822693,
      "learning_rate": 2.4024e-05,
      "loss": 0.0141,
      "step": 8011
    },
    {
      "epoch": 14.783010156971375,
      "grad_norm": 0.4916836619377136,
      "learning_rate": 2.4027e-05,
      "loss": 0.0137,
      "step": 8012
    },
    {
      "epoch": 14.784856879039705,
      "grad_norm": 1.4418329000473022,
      "learning_rate": 2.4030000000000002e-05,
      "loss": 0.0184,
      "step": 8013
    },
    {
      "epoch": 14.786703601108034,
      "grad_norm": 0.2875135540962219,
      "learning_rate": 2.4033000000000002e-05,
      "loss": 0.0108,
      "step": 8014
    },
    {
      "epoch": 14.788550323176363,
      "grad_norm": 0.3412550389766693,
      "learning_rate": 2.4036e-05,
      "loss": 0.0152,
      "step": 8015
    },
    {
      "epoch": 14.79039704524469,
      "grad_norm": 0.5497750043869019,
      "learning_rate": 2.4039e-05,
      "loss": 0.0202,
      "step": 8016
    },
    {
      "epoch": 14.79224376731302,
      "grad_norm": 0.32832539081573486,
      "learning_rate": 2.4042e-05,
      "loss": 0.0145,
      "step": 8017
    },
    {
      "epoch": 14.794090489381349,
      "grad_norm": 0.2167942225933075,
      "learning_rate": 2.4045e-05,
      "loss": 0.0151,
      "step": 8018
    },
    {
      "epoch": 14.795937211449676,
      "grad_norm": 0.36289235949516296,
      "learning_rate": 2.4048e-05,
      "loss": 0.0123,
      "step": 8019
    },
    {
      "epoch": 14.797783933518005,
      "grad_norm": 0.3763875961303711,
      "learning_rate": 2.4051e-05,
      "loss": 0.0097,
      "step": 8020
    },
    {
      "epoch": 14.799630655586334,
      "grad_norm": 0.5189347863197327,
      "learning_rate": 2.4054e-05,
      "loss": 0.0125,
      "step": 8021
    },
    {
      "epoch": 14.801477377654663,
      "grad_norm": 0.2630421817302704,
      "learning_rate": 2.4057e-05,
      "loss": 0.0184,
      "step": 8022
    },
    {
      "epoch": 14.803324099722992,
      "grad_norm": 0.6870790123939514,
      "learning_rate": 2.4060000000000003e-05,
      "loss": 0.0161,
      "step": 8023
    },
    {
      "epoch": 14.80517082179132,
      "grad_norm": 0.65828537940979,
      "learning_rate": 2.4063000000000003e-05,
      "loss": 0.0193,
      "step": 8024
    },
    {
      "epoch": 14.807017543859649,
      "grad_norm": 0.34641534090042114,
      "learning_rate": 2.4066000000000003e-05,
      "loss": 0.0123,
      "step": 8025
    },
    {
      "epoch": 14.808864265927978,
      "grad_norm": 0.24606692790985107,
      "learning_rate": 2.4069e-05,
      "loss": 0.0111,
      "step": 8026
    },
    {
      "epoch": 14.810710987996307,
      "grad_norm": 0.5390129685401917,
      "learning_rate": 2.4072e-05,
      "loss": 0.0243,
      "step": 8027
    },
    {
      "epoch": 14.812557710064635,
      "grad_norm": 0.5146270394325256,
      "learning_rate": 2.4075e-05,
      "loss": 0.021,
      "step": 8028
    },
    {
      "epoch": 14.814404432132964,
      "grad_norm": 0.6502516865730286,
      "learning_rate": 2.4078e-05,
      "loss": 0.0277,
      "step": 8029
    },
    {
      "epoch": 14.816251154201293,
      "grad_norm": 0.4489519000053406,
      "learning_rate": 2.4081e-05,
      "loss": 0.0205,
      "step": 8030
    },
    {
      "epoch": 14.818097876269622,
      "grad_norm": 0.9653266072273254,
      "learning_rate": 2.4084e-05,
      "loss": 0.0226,
      "step": 8031
    },
    {
      "epoch": 14.81994459833795,
      "grad_norm": 0.6530783772468567,
      "learning_rate": 2.4086999999999998e-05,
      "loss": 0.0192,
      "step": 8032
    },
    {
      "epoch": 14.821791320406279,
      "grad_norm": 0.5835822224617004,
      "learning_rate": 2.409e-05,
      "loss": 0.0247,
      "step": 8033
    },
    {
      "epoch": 14.823638042474608,
      "grad_norm": 0.6050820350646973,
      "learning_rate": 2.4093e-05,
      "loss": 0.031,
      "step": 8034
    },
    {
      "epoch": 14.825484764542937,
      "grad_norm": 0.5342754125595093,
      "learning_rate": 2.4096e-05,
      "loss": 0.0263,
      "step": 8035
    },
    {
      "epoch": 14.827331486611264,
      "grad_norm": 0.792512059211731,
      "learning_rate": 2.4099e-05,
      "loss": 0.0365,
      "step": 8036
    },
    {
      "epoch": 14.829178208679593,
      "grad_norm": 0.5477274656295776,
      "learning_rate": 2.4102e-05,
      "loss": 0.0271,
      "step": 8037
    },
    {
      "epoch": 14.831024930747922,
      "grad_norm": 0.6358621120452881,
      "learning_rate": 2.4105e-05,
      "loss": 0.0362,
      "step": 8038
    },
    {
      "epoch": 14.832871652816252,
      "grad_norm": 0.8861833810806274,
      "learning_rate": 2.4108e-05,
      "loss": 0.1902,
      "step": 8039
    },
    {
      "epoch": 14.83471837488458,
      "grad_norm": 0.7224254608154297,
      "learning_rate": 2.4111e-05,
      "loss": 0.1701,
      "step": 8040
    },
    {
      "epoch": 14.836565096952908,
      "grad_norm": 0.6825087070465088,
      "learning_rate": 2.4114e-05,
      "loss": 0.1845,
      "step": 8041
    },
    {
      "epoch": 14.838411819021237,
      "grad_norm": 0.6417450904846191,
      "learning_rate": 2.4117e-05,
      "loss": 0.1565,
      "step": 8042
    },
    {
      "epoch": 14.840258541089566,
      "grad_norm": 0.884933352470398,
      "learning_rate": 2.4120000000000003e-05,
      "loss": 0.1841,
      "step": 8043
    },
    {
      "epoch": 14.842105263157894,
      "grad_norm": 0.6482926607131958,
      "learning_rate": 2.4123000000000003e-05,
      "loss": 0.0914,
      "step": 8044
    },
    {
      "epoch": 14.843951985226223,
      "grad_norm": 0.5458600521087646,
      "learning_rate": 2.4126000000000002e-05,
      "loss": 0.0925,
      "step": 8045
    },
    {
      "epoch": 14.845798707294552,
      "grad_norm": 0.7142218947410583,
      "learning_rate": 2.4129000000000002e-05,
      "loss": 0.0796,
      "step": 8046
    },
    {
      "epoch": 14.847645429362881,
      "grad_norm": 0.5783588290214539,
      "learning_rate": 2.4132000000000002e-05,
      "loss": 0.0939,
      "step": 8047
    },
    {
      "epoch": 14.84949215143121,
      "grad_norm": 0.6058603525161743,
      "learning_rate": 2.4135000000000002e-05,
      "loss": 0.0602,
      "step": 8048
    },
    {
      "epoch": 14.851338873499538,
      "grad_norm": 0.743747889995575,
      "learning_rate": 2.4138e-05,
      "loss": 0.0614,
      "step": 8049
    },
    {
      "epoch": 14.853185595567867,
      "grad_norm": 0.6152564287185669,
      "learning_rate": 2.4140999999999998e-05,
      "loss": 0.0833,
      "step": 8050
    },
    {
      "epoch": 14.855032317636196,
      "grad_norm": 0.5742655992507935,
      "learning_rate": 2.4143999999999998e-05,
      "loss": 0.0513,
      "step": 8051
    },
    {
      "epoch": 14.856879039704525,
      "grad_norm": 0.29754096269607544,
      "learning_rate": 2.4146999999999998e-05,
      "loss": 0.0243,
      "step": 8052
    },
    {
      "epoch": 14.858725761772853,
      "grad_norm": 0.3988395929336548,
      "learning_rate": 2.415e-05,
      "loss": 0.0397,
      "step": 8053
    },
    {
      "epoch": 14.860572483841182,
      "grad_norm": 0.6038896441459656,
      "learning_rate": 2.4153e-05,
      "loss": 0.0843,
      "step": 8054
    },
    {
      "epoch": 14.86241920590951,
      "grad_norm": 0.4392443597316742,
      "learning_rate": 2.4156e-05,
      "loss": 0.0268,
      "step": 8055
    },
    {
      "epoch": 14.86426592797784,
      "grad_norm": 0.35345086455345154,
      "learning_rate": 2.4159e-05,
      "loss": 0.0246,
      "step": 8056
    },
    {
      "epoch": 14.866112650046167,
      "grad_norm": 0.40793344378471375,
      "learning_rate": 2.4162e-05,
      "loss": 0.0501,
      "step": 8057
    },
    {
      "epoch": 14.867959372114496,
      "grad_norm": 0.3084394633769989,
      "learning_rate": 2.4165e-05,
      "loss": 0.0521,
      "step": 8058
    },
    {
      "epoch": 14.869806094182826,
      "grad_norm": 0.26706284284591675,
      "learning_rate": 2.4168e-05,
      "loss": 0.0158,
      "step": 8059
    },
    {
      "epoch": 14.871652816251155,
      "grad_norm": 0.19276896119117737,
      "learning_rate": 2.4171e-05,
      "loss": 0.0191,
      "step": 8060
    },
    {
      "epoch": 14.873499538319482,
      "grad_norm": 0.3058687746524811,
      "learning_rate": 2.4174e-05,
      "loss": 0.0155,
      "step": 8061
    },
    {
      "epoch": 14.875346260387811,
      "grad_norm": 0.35725489258766174,
      "learning_rate": 2.4177e-05,
      "loss": 0.0197,
      "step": 8062
    },
    {
      "epoch": 14.87719298245614,
      "grad_norm": 0.26820647716522217,
      "learning_rate": 2.4180000000000002e-05,
      "loss": 0.0093,
      "step": 8063
    },
    {
      "epoch": 14.87903970452447,
      "grad_norm": 0.2957378625869751,
      "learning_rate": 2.4183000000000002e-05,
      "loss": 0.0145,
      "step": 8064
    },
    {
      "epoch": 14.880886426592799,
      "grad_norm": 0.3385465145111084,
      "learning_rate": 2.4186000000000002e-05,
      "loss": 0.0141,
      "step": 8065
    },
    {
      "epoch": 14.882733148661126,
      "grad_norm": 0.40991589426994324,
      "learning_rate": 2.4189e-05,
      "loss": 0.0143,
      "step": 8066
    },
    {
      "epoch": 14.884579870729455,
      "grad_norm": 0.43255510926246643,
      "learning_rate": 2.4192e-05,
      "loss": 0.0063,
      "step": 8067
    },
    {
      "epoch": 14.886426592797784,
      "grad_norm": 0.34434038400650024,
      "learning_rate": 2.4195e-05,
      "loss": 0.0127,
      "step": 8068
    },
    {
      "epoch": 14.888273314866112,
      "grad_norm": 0.506460964679718,
      "learning_rate": 2.4198e-05,
      "loss": 0.0153,
      "step": 8069
    },
    {
      "epoch": 14.89012003693444,
      "grad_norm": 0.7858155369758606,
      "learning_rate": 2.4201e-05,
      "loss": 0.0255,
      "step": 8070
    },
    {
      "epoch": 14.89196675900277,
      "grad_norm": 0.48727911710739136,
      "learning_rate": 2.4204e-05,
      "loss": 0.012,
      "step": 8071
    },
    {
      "epoch": 14.8938134810711,
      "grad_norm": 0.26856595277786255,
      "learning_rate": 2.4207e-05,
      "loss": 0.0118,
      "step": 8072
    },
    {
      "epoch": 14.895660203139428,
      "grad_norm": 0.4287573993206024,
      "learning_rate": 2.4210000000000004e-05,
      "loss": 0.017,
      "step": 8073
    },
    {
      "epoch": 14.897506925207756,
      "grad_norm": 0.3845706284046173,
      "learning_rate": 2.4213000000000003e-05,
      "loss": 0.0061,
      "step": 8074
    },
    {
      "epoch": 14.899353647276085,
      "grad_norm": 0.9846393465995789,
      "learning_rate": 2.4216e-05,
      "loss": 0.0209,
      "step": 8075
    },
    {
      "epoch": 14.901200369344414,
      "grad_norm": 0.48798713088035583,
      "learning_rate": 2.4219e-05,
      "loss": 0.037,
      "step": 8076
    },
    {
      "epoch": 14.903047091412743,
      "grad_norm": 0.658402144908905,
      "learning_rate": 2.4222e-05,
      "loss": 0.0244,
      "step": 8077
    },
    {
      "epoch": 14.90489381348107,
      "grad_norm": 0.47966620326042175,
      "learning_rate": 2.4225e-05,
      "loss": 0.0245,
      "step": 8078
    },
    {
      "epoch": 14.9067405355494,
      "grad_norm": 0.7510401606559753,
      "learning_rate": 2.4228e-05,
      "loss": 0.0347,
      "step": 8079
    },
    {
      "epoch": 14.908587257617729,
      "grad_norm": 0.24847644567489624,
      "learning_rate": 2.4231e-05,
      "loss": 0.0092,
      "step": 8080
    },
    {
      "epoch": 14.910433979686058,
      "grad_norm": 0.6264692544937134,
      "learning_rate": 2.4234e-05,
      "loss": 0.0249,
      "step": 8081
    },
    {
      "epoch": 14.912280701754385,
      "grad_norm": 0.5359593033790588,
      "learning_rate": 2.4237e-05,
      "loss": 0.0277,
      "step": 8082
    },
    {
      "epoch": 14.914127423822714,
      "grad_norm": 0.4312509000301361,
      "learning_rate": 2.4240000000000002e-05,
      "loss": 0.0192,
      "step": 8083
    },
    {
      "epoch": 14.915974145891044,
      "grad_norm": 0.44593143463134766,
      "learning_rate": 2.4243e-05,
      "loss": 0.014,
      "step": 8084
    },
    {
      "epoch": 14.917820867959373,
      "grad_norm": 0.4584122598171234,
      "learning_rate": 2.4246e-05,
      "loss": 0.0323,
      "step": 8085
    },
    {
      "epoch": 14.9196675900277,
      "grad_norm": 0.5289592742919922,
      "learning_rate": 2.4249e-05,
      "loss": 0.0281,
      "step": 8086
    },
    {
      "epoch": 14.92151431209603,
      "grad_norm": 0.32490670680999756,
      "learning_rate": 2.4252e-05,
      "loss": 0.0157,
      "step": 8087
    },
    {
      "epoch": 14.923361034164358,
      "grad_norm": 0.43771255016326904,
      "learning_rate": 2.4255e-05,
      "loss": 0.028,
      "step": 8088
    },
    {
      "epoch": 14.925207756232687,
      "grad_norm": 1.3917237520217896,
      "learning_rate": 2.4258e-05,
      "loss": 0.2859,
      "step": 8089
    },
    {
      "epoch": 14.927054478301017,
      "grad_norm": 0.5033819079399109,
      "learning_rate": 2.4261e-05,
      "loss": 0.1646,
      "step": 8090
    },
    {
      "epoch": 14.928901200369344,
      "grad_norm": 0.5540380477905273,
      "learning_rate": 2.4264e-05,
      "loss": 0.1251,
      "step": 8091
    },
    {
      "epoch": 14.930747922437673,
      "grad_norm": 0.6526817083358765,
      "learning_rate": 2.4267e-05,
      "loss": 0.0993,
      "step": 8092
    },
    {
      "epoch": 14.932594644506002,
      "grad_norm": 0.8004580140113831,
      "learning_rate": 2.4270000000000003e-05,
      "loss": 0.1988,
      "step": 8093
    },
    {
      "epoch": 14.93444136657433,
      "grad_norm": 0.46787190437316895,
      "learning_rate": 2.4273000000000003e-05,
      "loss": 0.0718,
      "step": 8094
    },
    {
      "epoch": 14.936288088642659,
      "grad_norm": 0.6435708999633789,
      "learning_rate": 2.4276000000000003e-05,
      "loss": 0.1238,
      "step": 8095
    },
    {
      "epoch": 14.938134810710988,
      "grad_norm": 0.506477952003479,
      "learning_rate": 2.4279000000000003e-05,
      "loss": 0.0728,
      "step": 8096
    },
    {
      "epoch": 14.939981532779317,
      "grad_norm": 0.6455765962600708,
      "learning_rate": 2.4282000000000002e-05,
      "loss": 0.0505,
      "step": 8097
    },
    {
      "epoch": 14.941828254847646,
      "grad_norm": 0.7211839556694031,
      "learning_rate": 2.4285000000000002e-05,
      "loss": 0.1329,
      "step": 8098
    },
    {
      "epoch": 14.943674976915974,
      "grad_norm": 0.4941684305667877,
      "learning_rate": 2.4288e-05,
      "loss": 0.0469,
      "step": 8099
    },
    {
      "epoch": 14.945521698984303,
      "grad_norm": 0.8883525133132935,
      "learning_rate": 2.4291e-05,
      "loss": 0.0281,
      "step": 8100
    },
    {
      "epoch": 14.947368421052632,
      "grad_norm": 0.5026273131370544,
      "learning_rate": 2.4293999999999998e-05,
      "loss": 0.0234,
      "step": 8101
    },
    {
      "epoch": 14.949215143120961,
      "grad_norm": 0.3770243227481842,
      "learning_rate": 2.4296999999999998e-05,
      "loss": 0.0339,
      "step": 8102
    },
    {
      "epoch": 14.951061865189288,
      "grad_norm": 0.2530992031097412,
      "learning_rate": 2.43e-05,
      "loss": 0.0166,
      "step": 8103
    },
    {
      "epoch": 14.952908587257618,
      "grad_norm": 0.3264380395412445,
      "learning_rate": 2.4303e-05,
      "loss": 0.0198,
      "step": 8104
    },
    {
      "epoch": 14.954755309325947,
      "grad_norm": 0.6021093130111694,
      "learning_rate": 2.4306e-05,
      "loss": 0.0234,
      "step": 8105
    },
    {
      "epoch": 14.956602031394276,
      "grad_norm": 0.39788442850112915,
      "learning_rate": 2.4309e-05,
      "loss": 0.0212,
      "step": 8106
    },
    {
      "epoch": 14.958448753462603,
      "grad_norm": 0.39071908593177795,
      "learning_rate": 2.4312e-05,
      "loss": 0.019,
      "step": 8107
    },
    {
      "epoch": 14.960295475530932,
      "grad_norm": 0.3165871202945709,
      "learning_rate": 2.4315e-05,
      "loss": 0.0152,
      "step": 8108
    },
    {
      "epoch": 14.962142197599261,
      "grad_norm": 0.3403398096561432,
      "learning_rate": 2.4318e-05,
      "loss": 0.0207,
      "step": 8109
    },
    {
      "epoch": 14.96398891966759,
      "grad_norm": 0.5380444526672363,
      "learning_rate": 2.4321e-05,
      "loss": 0.0117,
      "step": 8110
    },
    {
      "epoch": 14.965835641735918,
      "grad_norm": 0.43799179792404175,
      "learning_rate": 2.4324e-05,
      "loss": 0.022,
      "step": 8111
    },
    {
      "epoch": 14.967682363804247,
      "grad_norm": 1.1583491563796997,
      "learning_rate": 2.4327e-05,
      "loss": 0.0156,
      "step": 8112
    },
    {
      "epoch": 14.969529085872576,
      "grad_norm": 0.3655841052532196,
      "learning_rate": 2.4330000000000003e-05,
      "loss": 0.0137,
      "step": 8113
    },
    {
      "epoch": 14.971375807940905,
      "grad_norm": 0.5397698283195496,
      "learning_rate": 2.4333000000000002e-05,
      "loss": 0.0192,
      "step": 8114
    },
    {
      "epoch": 14.973222530009235,
      "grad_norm": 0.33571380376815796,
      "learning_rate": 2.4336000000000002e-05,
      "loss": 0.0102,
      "step": 8115
    },
    {
      "epoch": 14.975069252077562,
      "grad_norm": 0.3407282829284668,
      "learning_rate": 2.4339000000000002e-05,
      "loss": 0.0122,
      "step": 8116
    },
    {
      "epoch": 14.976915974145891,
      "grad_norm": 0.6962429285049438,
      "learning_rate": 2.4342000000000002e-05,
      "loss": 0.0198,
      "step": 8117
    },
    {
      "epoch": 14.97876269621422,
      "grad_norm": 0.5999586582183838,
      "learning_rate": 2.4345e-05,
      "loss": 0.0195,
      "step": 8118
    },
    {
      "epoch": 14.980609418282548,
      "grad_norm": 0.6077821254730225,
      "learning_rate": 2.4348e-05,
      "loss": 0.027,
      "step": 8119
    },
    {
      "epoch": 14.982456140350877,
      "grad_norm": 0.6994907259941101,
      "learning_rate": 2.4351e-05,
      "loss": 0.0174,
      "step": 8120
    },
    {
      "epoch": 14.984302862419206,
      "grad_norm": 0.5564958453178406,
      "learning_rate": 2.4354e-05,
      "loss": 0.0163,
      "step": 8121
    },
    {
      "epoch": 14.986149584487535,
      "grad_norm": 0.3990434408187866,
      "learning_rate": 2.4357e-05,
      "loss": 0.0126,
      "step": 8122
    },
    {
      "epoch": 14.987996306555864,
      "grad_norm": 0.608258843421936,
      "learning_rate": 2.4360000000000004e-05,
      "loss": 0.0144,
      "step": 8123
    },
    {
      "epoch": 14.989843028624191,
      "grad_norm": 0.524206280708313,
      "learning_rate": 2.4363e-05,
      "loss": 0.0243,
      "step": 8124
    },
    {
      "epoch": 14.99168975069252,
      "grad_norm": 0.4362499713897705,
      "learning_rate": 2.4366e-05,
      "loss": 0.0215,
      "step": 8125
    },
    {
      "epoch": 14.99353647276085,
      "grad_norm": 0.8296584486961365,
      "learning_rate": 2.4369e-05,
      "loss": 0.0185,
      "step": 8126
    },
    {
      "epoch": 14.995383194829179,
      "grad_norm": 0.41815313696861267,
      "learning_rate": 2.4372e-05,
      "loss": 0.0166,
      "step": 8127
    },
    {
      "epoch": 14.997229916897506,
      "grad_norm": 0.3036484718322754,
      "learning_rate": 2.4375e-05,
      "loss": 0.01,
      "step": 8128
    },
    {
      "epoch": 14.999076638965835,
      "grad_norm": 0.6441636681556702,
      "learning_rate": 2.4378e-05,
      "loss": 0.0215,
      "step": 8129
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.900398313999176,
      "learning_rate": 2.4381e-05,
      "loss": 0.0189,
      "step": 8130
    },
    {
      "epoch": 15.00184672206833,
      "grad_norm": 1.1221733093261719,
      "learning_rate": 2.4384e-05,
      "loss": 0.213,
      "step": 8131
    },
    {
      "epoch": 15.003693444136658,
      "grad_norm": 1.017161250114441,
      "learning_rate": 2.4387e-05,
      "loss": 0.2084,
      "step": 8132
    },
    {
      "epoch": 15.005540166204986,
      "grad_norm": 0.7330501675605774,
      "learning_rate": 2.439e-05,
      "loss": 0.1478,
      "step": 8133
    },
    {
      "epoch": 15.007386888273315,
      "grad_norm": 0.568026065826416,
      "learning_rate": 2.4393000000000002e-05,
      "loss": 0.1205,
      "step": 8134
    },
    {
      "epoch": 15.009233610341644,
      "grad_norm": 0.6616817116737366,
      "learning_rate": 2.4396e-05,
      "loss": 0.1764,
      "step": 8135
    },
    {
      "epoch": 15.011080332409973,
      "grad_norm": 0.5726966857910156,
      "learning_rate": 2.4399e-05,
      "loss": 0.0941,
      "step": 8136
    },
    {
      "epoch": 15.0129270544783,
      "grad_norm": 0.6629717350006104,
      "learning_rate": 2.4402e-05,
      "loss": 0.1011,
      "step": 8137
    },
    {
      "epoch": 15.01477377654663,
      "grad_norm": 0.5155584216117859,
      "learning_rate": 2.4405e-05,
      "loss": 0.099,
      "step": 8138
    },
    {
      "epoch": 15.016620498614959,
      "grad_norm": 0.6650702953338623,
      "learning_rate": 2.4408e-05,
      "loss": 0.0914,
      "step": 8139
    },
    {
      "epoch": 15.018467220683288,
      "grad_norm": 0.6222261190414429,
      "learning_rate": 2.4411e-05,
      "loss": 0.0997,
      "step": 8140
    },
    {
      "epoch": 15.020313942751615,
      "grad_norm": 0.6313560009002686,
      "learning_rate": 2.4414e-05,
      "loss": 0.0588,
      "step": 8141
    },
    {
      "epoch": 15.022160664819944,
      "grad_norm": 0.46788403391838074,
      "learning_rate": 2.4417e-05,
      "loss": 0.0533,
      "step": 8142
    },
    {
      "epoch": 15.024007386888274,
      "grad_norm": 0.569367527961731,
      "learning_rate": 2.442e-05,
      "loss": 0.0355,
      "step": 8143
    },
    {
      "epoch": 15.025854108956603,
      "grad_norm": 0.5730012059211731,
      "learning_rate": 2.4423000000000003e-05,
      "loss": 0.062,
      "step": 8144
    },
    {
      "epoch": 15.02770083102493,
      "grad_norm": 0.4984351396560669,
      "learning_rate": 2.4426000000000003e-05,
      "loss": 0.0679,
      "step": 8145
    },
    {
      "epoch": 15.02954755309326,
      "grad_norm": 1.6580547094345093,
      "learning_rate": 2.4429000000000003e-05,
      "loss": 0.038,
      "step": 8146
    },
    {
      "epoch": 15.031394275161588,
      "grad_norm": 0.21034662425518036,
      "learning_rate": 2.4432000000000003e-05,
      "loss": 0.0143,
      "step": 8147
    },
    {
      "epoch": 15.033240997229917,
      "grad_norm": 0.3205229938030243,
      "learning_rate": 2.4435e-05,
      "loss": 0.0212,
      "step": 8148
    },
    {
      "epoch": 15.035087719298245,
      "grad_norm": 0.41856059432029724,
      "learning_rate": 2.4438e-05,
      "loss": 0.0437,
      "step": 8149
    },
    {
      "epoch": 15.036934441366574,
      "grad_norm": 0.4145803451538086,
      "learning_rate": 2.4441e-05,
      "loss": 0.0165,
      "step": 8150
    },
    {
      "epoch": 15.038781163434903,
      "grad_norm": 0.18975673615932465,
      "learning_rate": 2.4444e-05,
      "loss": 0.0091,
      "step": 8151
    },
    {
      "epoch": 15.040627885503232,
      "grad_norm": 0.3147350251674652,
      "learning_rate": 2.4446999999999998e-05,
      "loss": 0.0151,
      "step": 8152
    },
    {
      "epoch": 15.04247460757156,
      "grad_norm": 0.27458664774894714,
      "learning_rate": 2.4449999999999998e-05,
      "loss": 0.0121,
      "step": 8153
    },
    {
      "epoch": 15.044321329639889,
      "grad_norm": 0.35295307636260986,
      "learning_rate": 2.4453e-05,
      "loss": 0.0168,
      "step": 8154
    },
    {
      "epoch": 15.046168051708218,
      "grad_norm": 0.5425369143486023,
      "learning_rate": 2.4456e-05,
      "loss": 0.0208,
      "step": 8155
    },
    {
      "epoch": 15.048014773776547,
      "grad_norm": 0.4190623462200165,
      "learning_rate": 2.4459e-05,
      "loss": 0.0103,
      "step": 8156
    },
    {
      "epoch": 15.049861495844876,
      "grad_norm": 1.073426604270935,
      "learning_rate": 2.4462e-05,
      "loss": 0.0246,
      "step": 8157
    },
    {
      "epoch": 15.051708217913204,
      "grad_norm": 0.29241880774497986,
      "learning_rate": 2.4465e-05,
      "loss": 0.0177,
      "step": 8158
    },
    {
      "epoch": 15.053554939981533,
      "grad_norm": 0.3127516806125641,
      "learning_rate": 2.4468e-05,
      "loss": 0.0212,
      "step": 8159
    },
    {
      "epoch": 15.055401662049862,
      "grad_norm": 0.5754639506340027,
      "learning_rate": 2.4471e-05,
      "loss": 0.0134,
      "step": 8160
    },
    {
      "epoch": 15.057248384118191,
      "grad_norm": 0.7174379825592041,
      "learning_rate": 2.4474e-05,
      "loss": 0.0163,
      "step": 8161
    },
    {
      "epoch": 15.059095106186518,
      "grad_norm": 0.4467093348503113,
      "learning_rate": 2.4477e-05,
      "loss": 0.0129,
      "step": 8162
    },
    {
      "epoch": 15.060941828254848,
      "grad_norm": 0.3584469258785248,
      "learning_rate": 2.448e-05,
      "loss": 0.0077,
      "step": 8163
    },
    {
      "epoch": 15.062788550323177,
      "grad_norm": 0.26156914234161377,
      "learning_rate": 2.4483000000000003e-05,
      "loss": 0.0103,
      "step": 8164
    },
    {
      "epoch": 15.064635272391506,
      "grad_norm": 0.4861030578613281,
      "learning_rate": 2.4486000000000002e-05,
      "loss": 0.0195,
      "step": 8165
    },
    {
      "epoch": 15.066481994459833,
      "grad_norm": 0.31749069690704346,
      "learning_rate": 2.4489000000000002e-05,
      "loss": 0.0056,
      "step": 8166
    },
    {
      "epoch": 15.068328716528162,
      "grad_norm": 0.4234580993652344,
      "learning_rate": 2.4492000000000002e-05,
      "loss": 0.0115,
      "step": 8167
    },
    {
      "epoch": 15.070175438596491,
      "grad_norm": 0.3904325067996979,
      "learning_rate": 2.4495000000000002e-05,
      "loss": 0.0101,
      "step": 8168
    },
    {
      "epoch": 15.07202216066482,
      "grad_norm": 0.7674003839492798,
      "learning_rate": 2.4498e-05,
      "loss": 0.0173,
      "step": 8169
    },
    {
      "epoch": 15.073868882733148,
      "grad_norm": 0.21063567698001862,
      "learning_rate": 2.4501e-05,
      "loss": 0.012,
      "step": 8170
    },
    {
      "epoch": 15.075715604801477,
      "grad_norm": 0.3684234023094177,
      "learning_rate": 2.4504e-05,
      "loss": 0.0196,
      "step": 8171
    },
    {
      "epoch": 15.077562326869806,
      "grad_norm": 1.3102104663848877,
      "learning_rate": 2.4507e-05,
      "loss": 0.0079,
      "step": 8172
    },
    {
      "epoch": 15.079409048938135,
      "grad_norm": 0.29982709884643555,
      "learning_rate": 2.4509999999999997e-05,
      "loss": 0.013,
      "step": 8173
    },
    {
      "epoch": 15.081255771006463,
      "grad_norm": 0.40088796615600586,
      "learning_rate": 2.4513e-05,
      "loss": 0.0137,
      "step": 8174
    },
    {
      "epoch": 15.083102493074792,
      "grad_norm": 0.5310261249542236,
      "learning_rate": 2.4516e-05,
      "loss": 0.0173,
      "step": 8175
    },
    {
      "epoch": 15.084949215143121,
      "grad_norm": 0.54770427942276,
      "learning_rate": 2.4519e-05,
      "loss": 0.0104,
      "step": 8176
    },
    {
      "epoch": 15.08679593721145,
      "grad_norm": 0.6323532462120056,
      "learning_rate": 2.4522e-05,
      "loss": 0.0115,
      "step": 8177
    },
    {
      "epoch": 15.088642659279778,
      "grad_norm": 0.6644243597984314,
      "learning_rate": 2.4525e-05,
      "loss": 0.0084,
      "step": 8178
    },
    {
      "epoch": 15.090489381348107,
      "grad_norm": 0.450868159532547,
      "learning_rate": 2.4528e-05,
      "loss": 0.0122,
      "step": 8179
    },
    {
      "epoch": 15.092336103416436,
      "grad_norm": 0.47782060503959656,
      "learning_rate": 2.4531e-05,
      "loss": 0.0095,
      "step": 8180
    },
    {
      "epoch": 15.094182825484765,
      "grad_norm": 0.7100746035575867,
      "learning_rate": 2.4534e-05,
      "loss": 0.1596,
      "step": 8181
    },
    {
      "epoch": 15.096029547553094,
      "grad_norm": 0.7997457981109619,
      "learning_rate": 2.4537e-05,
      "loss": 0.2233,
      "step": 8182
    },
    {
      "epoch": 15.097876269621421,
      "grad_norm": 0.6452318429946899,
      "learning_rate": 2.454e-05,
      "loss": 0.1357,
      "step": 8183
    },
    {
      "epoch": 15.09972299168975,
      "grad_norm": 0.780385434627533,
      "learning_rate": 2.4543000000000002e-05,
      "loss": 0.1706,
      "step": 8184
    },
    {
      "epoch": 15.10156971375808,
      "grad_norm": 0.5169774889945984,
      "learning_rate": 2.4546000000000002e-05,
      "loss": 0.1229,
      "step": 8185
    },
    {
      "epoch": 15.103416435826409,
      "grad_norm": 0.630163848400116,
      "learning_rate": 2.4549e-05,
      "loss": 0.153,
      "step": 8186
    },
    {
      "epoch": 15.105263157894736,
      "grad_norm": 0.8105989098548889,
      "learning_rate": 2.4552e-05,
      "loss": 0.1634,
      "step": 8187
    },
    {
      "epoch": 15.107109879963065,
      "grad_norm": 0.4336480498313904,
      "learning_rate": 2.4555e-05,
      "loss": 0.0631,
      "step": 8188
    },
    {
      "epoch": 15.108956602031395,
      "grad_norm": 0.5514599680900574,
      "learning_rate": 2.4558e-05,
      "loss": 0.1063,
      "step": 8189
    },
    {
      "epoch": 15.110803324099724,
      "grad_norm": 0.37652090191841125,
      "learning_rate": 2.4561e-05,
      "loss": 0.0588,
      "step": 8190
    },
    {
      "epoch": 15.112650046168051,
      "grad_norm": 0.5351255536079407,
      "learning_rate": 2.4564e-05,
      "loss": 0.0622,
      "step": 8191
    },
    {
      "epoch": 15.11449676823638,
      "grad_norm": 0.35973745584487915,
      "learning_rate": 2.4567e-05,
      "loss": 0.041,
      "step": 8192
    },
    {
      "epoch": 15.11634349030471,
      "grad_norm": 2.9132304191589355,
      "learning_rate": 2.457e-05,
      "loss": 0.0948,
      "step": 8193
    },
    {
      "epoch": 15.118190212373039,
      "grad_norm": 0.2675500512123108,
      "learning_rate": 2.4573000000000003e-05,
      "loss": 0.027,
      "step": 8194
    },
    {
      "epoch": 15.120036934441366,
      "grad_norm": 0.38470208644866943,
      "learning_rate": 2.4576000000000003e-05,
      "loss": 0.0556,
      "step": 8195
    },
    {
      "epoch": 15.121883656509695,
      "grad_norm": 0.5402805209159851,
      "learning_rate": 2.4579000000000003e-05,
      "loss": 0.049,
      "step": 8196
    },
    {
      "epoch": 15.123730378578024,
      "grad_norm": 0.2357839196920395,
      "learning_rate": 2.4582000000000003e-05,
      "loss": 0.0499,
      "step": 8197
    },
    {
      "epoch": 15.125577100646353,
      "grad_norm": 0.40822380781173706,
      "learning_rate": 2.4585e-05,
      "loss": 0.0178,
      "step": 8198
    },
    {
      "epoch": 15.12742382271468,
      "grad_norm": 0.2533171772956848,
      "learning_rate": 2.4588e-05,
      "loss": 0.0195,
      "step": 8199
    },
    {
      "epoch": 15.12927054478301,
      "grad_norm": 0.305999755859375,
      "learning_rate": 2.4591e-05,
      "loss": 0.0238,
      "step": 8200
    },
    {
      "epoch": 15.131117266851339,
      "grad_norm": 0.470662385225296,
      "learning_rate": 2.4594e-05,
      "loss": 0.0509,
      "step": 8201
    },
    {
      "epoch": 15.132963988919668,
      "grad_norm": 0.26189985871315,
      "learning_rate": 2.4597e-05,
      "loss": 0.0165,
      "step": 8202
    },
    {
      "epoch": 15.134810710987995,
      "grad_norm": 1.0813108682632446,
      "learning_rate": 2.4599999999999998e-05,
      "loss": 0.0425,
      "step": 8203
    },
    {
      "epoch": 15.136657433056325,
      "grad_norm": 0.33218902349472046,
      "learning_rate": 2.4603e-05,
      "loss": 0.0088,
      "step": 8204
    },
    {
      "epoch": 15.138504155124654,
      "grad_norm": 0.22756680846214294,
      "learning_rate": 2.4606e-05,
      "loss": 0.0087,
      "step": 8205
    },
    {
      "epoch": 15.140350877192983,
      "grad_norm": 0.34461233019828796,
      "learning_rate": 2.4609e-05,
      "loss": 0.0082,
      "step": 8206
    },
    {
      "epoch": 15.142197599261312,
      "grad_norm": 0.5454093217849731,
      "learning_rate": 2.4612e-05,
      "loss": 0.0142,
      "step": 8207
    },
    {
      "epoch": 15.14404432132964,
      "grad_norm": 0.30798277258872986,
      "learning_rate": 2.4615e-05,
      "loss": 0.0128,
      "step": 8208
    },
    {
      "epoch": 15.145891043397969,
      "grad_norm": 0.4576432704925537,
      "learning_rate": 2.4618e-05,
      "loss": 0.0336,
      "step": 8209
    },
    {
      "epoch": 15.147737765466298,
      "grad_norm": 0.247364342212677,
      "learning_rate": 2.4621e-05,
      "loss": 0.0112,
      "step": 8210
    },
    {
      "epoch": 15.149584487534627,
      "grad_norm": 0.614842414855957,
      "learning_rate": 2.4624e-05,
      "loss": 0.0118,
      "step": 8211
    },
    {
      "epoch": 15.151431209602954,
      "grad_norm": 0.5782668590545654,
      "learning_rate": 2.4627e-05,
      "loss": 0.0135,
      "step": 8212
    },
    {
      "epoch": 15.153277931671283,
      "grad_norm": 0.4947219491004944,
      "learning_rate": 2.463e-05,
      "loss": 0.016,
      "step": 8213
    },
    {
      "epoch": 15.155124653739612,
      "grad_norm": 0.33534643054008484,
      "learning_rate": 2.4633000000000003e-05,
      "loss": 0.012,
      "step": 8214
    },
    {
      "epoch": 15.156971375807942,
      "grad_norm": 0.2798976004123688,
      "learning_rate": 2.4636000000000003e-05,
      "loss": 0.0101,
      "step": 8215
    },
    {
      "epoch": 15.158818097876269,
      "grad_norm": 0.5524477362632751,
      "learning_rate": 2.4639000000000002e-05,
      "loss": 0.0207,
      "step": 8216
    },
    {
      "epoch": 15.160664819944598,
      "grad_norm": 0.3597121238708496,
      "learning_rate": 2.4642000000000002e-05,
      "loss": 0.0116,
      "step": 8217
    },
    {
      "epoch": 15.162511542012927,
      "grad_norm": 0.5381467938423157,
      "learning_rate": 2.4645000000000002e-05,
      "loss": 0.0115,
      "step": 8218
    },
    {
      "epoch": 15.164358264081256,
      "grad_norm": 0.48072612285614014,
      "learning_rate": 2.4648000000000002e-05,
      "loss": 0.0178,
      "step": 8219
    },
    {
      "epoch": 15.166204986149584,
      "grad_norm": 0.9097961187362671,
      "learning_rate": 2.4651e-05,
      "loss": 0.0215,
      "step": 8220
    },
    {
      "epoch": 15.168051708217913,
      "grad_norm": 0.6664448976516724,
      "learning_rate": 2.4654e-05,
      "loss": 0.0224,
      "step": 8221
    },
    {
      "epoch": 15.169898430286242,
      "grad_norm": 0.20867213606834412,
      "learning_rate": 2.4656999999999998e-05,
      "loss": 0.0062,
      "step": 8222
    },
    {
      "epoch": 15.171745152354571,
      "grad_norm": 0.3724096715450287,
      "learning_rate": 2.4659999999999998e-05,
      "loss": 0.0171,
      "step": 8223
    },
    {
      "epoch": 15.173591874422899,
      "grad_norm": 0.5794903635978699,
      "learning_rate": 2.4663e-05,
      "loss": 0.0147,
      "step": 8224
    },
    {
      "epoch": 15.175438596491228,
      "grad_norm": 0.9206926822662354,
      "learning_rate": 2.4666e-05,
      "loss": 0.0137,
      "step": 8225
    },
    {
      "epoch": 15.177285318559557,
      "grad_norm": 0.4412376880645752,
      "learning_rate": 2.4669e-05,
      "loss": 0.014,
      "step": 8226
    },
    {
      "epoch": 15.179132040627886,
      "grad_norm": 0.5383925437927246,
      "learning_rate": 2.4672e-05,
      "loss": 0.0171,
      "step": 8227
    },
    {
      "epoch": 15.180978762696213,
      "grad_norm": 0.5624282360076904,
      "learning_rate": 2.4675e-05,
      "loss": 0.0146,
      "step": 8228
    },
    {
      "epoch": 15.182825484764543,
      "grad_norm": 0.6234703660011292,
      "learning_rate": 2.4678e-05,
      "loss": 0.0174,
      "step": 8229
    },
    {
      "epoch": 15.184672206832872,
      "grad_norm": 0.9138628840446472,
      "learning_rate": 2.4681e-05,
      "loss": 0.0164,
      "step": 8230
    },
    {
      "epoch": 15.1865189289012,
      "grad_norm": 1.2404528856277466,
      "learning_rate": 2.4684e-05,
      "loss": 0.2855,
      "step": 8231
    },
    {
      "epoch": 15.18836565096953,
      "grad_norm": 0.7322246432304382,
      "learning_rate": 2.4687e-05,
      "loss": 0.1642,
      "step": 8232
    },
    {
      "epoch": 15.190212373037857,
      "grad_norm": 0.6022248864173889,
      "learning_rate": 2.469e-05,
      "loss": 0.1234,
      "step": 8233
    },
    {
      "epoch": 15.192059095106186,
      "grad_norm": 0.5909581780433655,
      "learning_rate": 2.4693000000000002e-05,
      "loss": 0.1583,
      "step": 8234
    },
    {
      "epoch": 15.193905817174516,
      "grad_norm": 0.560417652130127,
      "learning_rate": 2.4696000000000002e-05,
      "loss": 0.1112,
      "step": 8235
    },
    {
      "epoch": 15.195752539242845,
      "grad_norm": 0.5888610482215881,
      "learning_rate": 2.4699000000000002e-05,
      "loss": 0.105,
      "step": 8236
    },
    {
      "epoch": 15.197599261311172,
      "grad_norm": 0.945635199546814,
      "learning_rate": 2.4702e-05,
      "loss": 0.107,
      "step": 8237
    },
    {
      "epoch": 15.199445983379501,
      "grad_norm": 0.6924152374267578,
      "learning_rate": 2.4705e-05,
      "loss": 0.0957,
      "step": 8238
    },
    {
      "epoch": 15.20129270544783,
      "grad_norm": 0.4013981521129608,
      "learning_rate": 2.4708e-05,
      "loss": 0.0823,
      "step": 8239
    },
    {
      "epoch": 15.20313942751616,
      "grad_norm": 0.5373607277870178,
      "learning_rate": 2.4711e-05,
      "loss": 0.0763,
      "step": 8240
    },
    {
      "epoch": 15.204986149584487,
      "grad_norm": 0.7776703238487244,
      "learning_rate": 2.4714e-05,
      "loss": 0.087,
      "step": 8241
    },
    {
      "epoch": 15.206832871652816,
      "grad_norm": 0.4090069532394409,
      "learning_rate": 2.4717e-05,
      "loss": 0.0799,
      "step": 8242
    },
    {
      "epoch": 15.208679593721145,
      "grad_norm": 0.43735113739967346,
      "learning_rate": 2.472e-05,
      "loss": 0.0332,
      "step": 8243
    },
    {
      "epoch": 15.210526315789474,
      "grad_norm": 0.3009633421897888,
      "learning_rate": 2.4723000000000004e-05,
      "loss": 0.0261,
      "step": 8244
    },
    {
      "epoch": 15.212373037857802,
      "grad_norm": 0.42820805311203003,
      "learning_rate": 2.4726000000000003e-05,
      "loss": 0.0317,
      "step": 8245
    },
    {
      "epoch": 15.21421975992613,
      "grad_norm": 0.4590342044830322,
      "learning_rate": 2.4729000000000003e-05,
      "loss": 0.0215,
      "step": 8246
    },
    {
      "epoch": 15.21606648199446,
      "grad_norm": 0.5674740076065063,
      "learning_rate": 2.4732e-05,
      "loss": 0.0199,
      "step": 8247
    },
    {
      "epoch": 15.21791320406279,
      "grad_norm": 0.677531898021698,
      "learning_rate": 2.4735e-05,
      "loss": 0.0201,
      "step": 8248
    },
    {
      "epoch": 15.219759926131117,
      "grad_norm": 0.3279813826084137,
      "learning_rate": 2.4738e-05,
      "loss": 0.0202,
      "step": 8249
    },
    {
      "epoch": 15.221606648199446,
      "grad_norm": 0.3743285536766052,
      "learning_rate": 2.4741e-05,
      "loss": 0.0164,
      "step": 8250
    },
    {
      "epoch": 15.223453370267775,
      "grad_norm": 0.5781236290931702,
      "learning_rate": 2.4744e-05,
      "loss": 0.0164,
      "step": 8251
    },
    {
      "epoch": 15.225300092336104,
      "grad_norm": 0.47307130694389343,
      "learning_rate": 2.4747e-05,
      "loss": 0.0157,
      "step": 8252
    },
    {
      "epoch": 15.227146814404431,
      "grad_norm": 0.3261812925338745,
      "learning_rate": 2.475e-05,
      "loss": 0.0109,
      "step": 8253
    },
    {
      "epoch": 15.22899353647276,
      "grad_norm": 0.6554425954818726,
      "learning_rate": 2.4753e-05,
      "loss": 0.0216,
      "step": 8254
    },
    {
      "epoch": 15.23084025854109,
      "grad_norm": 0.3593559265136719,
      "learning_rate": 2.4756e-05,
      "loss": 0.0221,
      "step": 8255
    },
    {
      "epoch": 15.232686980609419,
      "grad_norm": 0.3336569368839264,
      "learning_rate": 2.4759e-05,
      "loss": 0.0117,
      "step": 8256
    },
    {
      "epoch": 15.234533702677748,
      "grad_norm": 1.3743199110031128,
      "learning_rate": 2.4762e-05,
      "loss": 0.0199,
      "step": 8257
    },
    {
      "epoch": 15.236380424746075,
      "grad_norm": 0.25178849697113037,
      "learning_rate": 2.4765e-05,
      "loss": 0.0083,
      "step": 8258
    },
    {
      "epoch": 15.238227146814404,
      "grad_norm": 0.37484949827194214,
      "learning_rate": 2.4768e-05,
      "loss": 0.0118,
      "step": 8259
    },
    {
      "epoch": 15.240073868882734,
      "grad_norm": 0.46891501545906067,
      "learning_rate": 2.4771e-05,
      "loss": 0.0663,
      "step": 8260
    },
    {
      "epoch": 15.241920590951063,
      "grad_norm": 0.4010298252105713,
      "learning_rate": 2.4774e-05,
      "loss": 0.0222,
      "step": 8261
    },
    {
      "epoch": 15.24376731301939,
      "grad_norm": 0.4214724600315094,
      "learning_rate": 2.4777e-05,
      "loss": 0.0121,
      "step": 8262
    },
    {
      "epoch": 15.24561403508772,
      "grad_norm": 0.29846876859664917,
      "learning_rate": 2.478e-05,
      "loss": 0.0178,
      "step": 8263
    },
    {
      "epoch": 15.247460757156048,
      "grad_norm": 0.2897872030735016,
      "learning_rate": 2.4783e-05,
      "loss": 0.0109,
      "step": 8264
    },
    {
      "epoch": 15.249307479224377,
      "grad_norm": 0.5751253366470337,
      "learning_rate": 2.4786000000000003e-05,
      "loss": 0.0195,
      "step": 8265
    },
    {
      "epoch": 15.251154201292705,
      "grad_norm": 0.3264547288417816,
      "learning_rate": 2.4789000000000003e-05,
      "loss": 0.0142,
      "step": 8266
    },
    {
      "epoch": 15.253000923361034,
      "grad_norm": 0.5448217988014221,
      "learning_rate": 2.4792000000000003e-05,
      "loss": 0.0166,
      "step": 8267
    },
    {
      "epoch": 15.254847645429363,
      "grad_norm": 0.4921300411224365,
      "learning_rate": 2.4795000000000002e-05,
      "loss": 0.0176,
      "step": 8268
    },
    {
      "epoch": 15.256694367497692,
      "grad_norm": 0.522863507270813,
      "learning_rate": 2.4798000000000002e-05,
      "loss": 0.0248,
      "step": 8269
    },
    {
      "epoch": 15.25854108956602,
      "grad_norm": 0.6211817264556885,
      "learning_rate": 2.4801000000000002e-05,
      "loss": 0.0174,
      "step": 8270
    },
    {
      "epoch": 15.260387811634349,
      "grad_norm": 0.48236003518104553,
      "learning_rate": 2.4804e-05,
      "loss": 0.0082,
      "step": 8271
    },
    {
      "epoch": 15.262234533702678,
      "grad_norm": 0.27671054005622864,
      "learning_rate": 2.4806999999999998e-05,
      "loss": 0.0151,
      "step": 8272
    },
    {
      "epoch": 15.264081255771007,
      "grad_norm": 0.6454673409461975,
      "learning_rate": 2.4809999999999998e-05,
      "loss": 0.0204,
      "step": 8273
    },
    {
      "epoch": 15.265927977839334,
      "grad_norm": 0.5842491388320923,
      "learning_rate": 2.4812999999999998e-05,
      "loss": 0.0149,
      "step": 8274
    },
    {
      "epoch": 15.267774699907664,
      "grad_norm": 0.42065128684043884,
      "learning_rate": 2.4816e-05,
      "loss": 0.0142,
      "step": 8275
    },
    {
      "epoch": 15.269621421975993,
      "grad_norm": 0.5586716532707214,
      "learning_rate": 2.4819e-05,
      "loss": 0.0245,
      "step": 8276
    },
    {
      "epoch": 15.271468144044322,
      "grad_norm": 0.6155346632003784,
      "learning_rate": 2.4822e-05,
      "loss": 0.0293,
      "step": 8277
    },
    {
      "epoch": 15.27331486611265,
      "grad_norm": 0.33992999792099,
      "learning_rate": 2.4825e-05,
      "loss": 0.0101,
      "step": 8278
    },
    {
      "epoch": 15.275161588180978,
      "grad_norm": 0.25734278559684753,
      "learning_rate": 2.4828e-05,
      "loss": 0.0123,
      "step": 8279
    },
    {
      "epoch": 15.277008310249307,
      "grad_norm": 2.350486993789673,
      "learning_rate": 2.4831e-05,
      "loss": 0.0196,
      "step": 8280
    },
    {
      "epoch": 15.278855032317637,
      "grad_norm": 0.7939901947975159,
      "learning_rate": 2.4834e-05,
      "loss": 0.1937,
      "step": 8281
    },
    {
      "epoch": 15.280701754385966,
      "grad_norm": 1.1575547456741333,
      "learning_rate": 2.4837e-05,
      "loss": 0.2262,
      "step": 8282
    },
    {
      "epoch": 15.282548476454293,
      "grad_norm": 0.9486008882522583,
      "learning_rate": 2.484e-05,
      "loss": 0.2102,
      "step": 8283
    },
    {
      "epoch": 15.284395198522622,
      "grad_norm": 0.4723045229911804,
      "learning_rate": 2.4843e-05,
      "loss": 0.1015,
      "step": 8284
    },
    {
      "epoch": 15.286241920590951,
      "grad_norm": 0.8471699953079224,
      "learning_rate": 2.4846000000000002e-05,
      "loss": 0.1094,
      "step": 8285
    },
    {
      "epoch": 15.28808864265928,
      "grad_norm": 0.7257476449012756,
      "learning_rate": 2.4849000000000002e-05,
      "loss": 0.0808,
      "step": 8286
    },
    {
      "epoch": 15.289935364727608,
      "grad_norm": 0.6892672777175903,
      "learning_rate": 2.4852000000000002e-05,
      "loss": 0.0707,
      "step": 8287
    },
    {
      "epoch": 15.291782086795937,
      "grad_norm": 0.5333134531974792,
      "learning_rate": 2.4855000000000002e-05,
      "loss": 0.0887,
      "step": 8288
    },
    {
      "epoch": 15.293628808864266,
      "grad_norm": 0.4352606534957886,
      "learning_rate": 2.4858e-05,
      "loss": 0.0717,
      "step": 8289
    },
    {
      "epoch": 15.295475530932595,
      "grad_norm": 0.5483019948005676,
      "learning_rate": 2.4861e-05,
      "loss": 0.0879,
      "step": 8290
    },
    {
      "epoch": 15.297322253000923,
      "grad_norm": 0.37380048632621765,
      "learning_rate": 2.4864e-05,
      "loss": 0.043,
      "step": 8291
    },
    {
      "epoch": 15.299168975069252,
      "grad_norm": 0.4453401565551758,
      "learning_rate": 2.4867e-05,
      "loss": 0.0515,
      "step": 8292
    },
    {
      "epoch": 15.301015697137581,
      "grad_norm": 0.45170173048973083,
      "learning_rate": 2.487e-05,
      "loss": 0.0563,
      "step": 8293
    },
    {
      "epoch": 15.30286241920591,
      "grad_norm": 0.4668281674385071,
      "learning_rate": 2.4873e-05,
      "loss": 0.0672,
      "step": 8294
    },
    {
      "epoch": 15.304709141274238,
      "grad_norm": 0.3134375810623169,
      "learning_rate": 2.4876000000000004e-05,
      "loss": 0.0483,
      "step": 8295
    },
    {
      "epoch": 15.306555863342567,
      "grad_norm": 0.4253406524658203,
      "learning_rate": 2.4879e-05,
      "loss": 0.0176,
      "step": 8296
    },
    {
      "epoch": 15.308402585410896,
      "grad_norm": 0.6559887528419495,
      "learning_rate": 2.4882e-05,
      "loss": 0.0227,
      "step": 8297
    },
    {
      "epoch": 15.310249307479225,
      "grad_norm": 0.4013559818267822,
      "learning_rate": 2.4885e-05,
      "loss": 0.0298,
      "step": 8298
    },
    {
      "epoch": 15.312096029547552,
      "grad_norm": 0.5104976296424866,
      "learning_rate": 2.4888e-05,
      "loss": 0.0187,
      "step": 8299
    },
    {
      "epoch": 15.313942751615881,
      "grad_norm": 0.3234640657901764,
      "learning_rate": 2.4891e-05,
      "loss": 0.0326,
      "step": 8300
    },
    {
      "epoch": 15.31578947368421,
      "grad_norm": 0.3353601396083832,
      "learning_rate": 2.4894e-05,
      "loss": 0.0245,
      "step": 8301
    },
    {
      "epoch": 15.31763619575254,
      "grad_norm": 0.4961949288845062,
      "learning_rate": 2.4897e-05,
      "loss": 0.0093,
      "step": 8302
    },
    {
      "epoch": 15.319482917820867,
      "grad_norm": 0.5354464054107666,
      "learning_rate": 2.49e-05,
      "loss": 0.0193,
      "step": 8303
    },
    {
      "epoch": 15.321329639889196,
      "grad_norm": 0.53691166639328,
      "learning_rate": 2.4903e-05,
      "loss": 0.0211,
      "step": 8304
    },
    {
      "epoch": 15.323176361957525,
      "grad_norm": 0.22957701981067657,
      "learning_rate": 2.4906000000000002e-05,
      "loss": 0.0073,
      "step": 8305
    },
    {
      "epoch": 15.325023084025855,
      "grad_norm": 0.2727436125278473,
      "learning_rate": 2.4909e-05,
      "loss": 0.0119,
      "step": 8306
    },
    {
      "epoch": 15.326869806094184,
      "grad_norm": 0.25763967633247375,
      "learning_rate": 2.4912e-05,
      "loss": 0.0152,
      "step": 8307
    },
    {
      "epoch": 15.328716528162511,
      "grad_norm": 0.36578795313835144,
      "learning_rate": 2.4915e-05,
      "loss": 0.0519,
      "step": 8308
    },
    {
      "epoch": 15.33056325023084,
      "grad_norm": 0.45741453766822815,
      "learning_rate": 2.4918e-05,
      "loss": 0.0198,
      "step": 8309
    },
    {
      "epoch": 15.33240997229917,
      "grad_norm": 0.2624121606349945,
      "learning_rate": 2.4921e-05,
      "loss": 0.0141,
      "step": 8310
    },
    {
      "epoch": 15.334256694367498,
      "grad_norm": 3.8751049041748047,
      "learning_rate": 2.4924e-05,
      "loss": 0.0151,
      "step": 8311
    },
    {
      "epoch": 15.336103416435826,
      "grad_norm": 0.37224772572517395,
      "learning_rate": 2.4927e-05,
      "loss": 0.0326,
      "step": 8312
    },
    {
      "epoch": 15.337950138504155,
      "grad_norm": 1.0172438621520996,
      "learning_rate": 2.493e-05,
      "loss": 0.0277,
      "step": 8313
    },
    {
      "epoch": 15.339796860572484,
      "grad_norm": 5.484036922454834,
      "learning_rate": 2.4933e-05,
      "loss": 0.029,
      "step": 8314
    },
    {
      "epoch": 15.341643582640813,
      "grad_norm": 0.3800578713417053,
      "learning_rate": 2.4936000000000003e-05,
      "loss": 0.009,
      "step": 8315
    },
    {
      "epoch": 15.34349030470914,
      "grad_norm": 0.3240329921245575,
      "learning_rate": 2.4939000000000003e-05,
      "loss": 0.0097,
      "step": 8316
    },
    {
      "epoch": 15.34533702677747,
      "grad_norm": 0.6319631338119507,
      "learning_rate": 2.4942000000000003e-05,
      "loss": 0.0205,
      "step": 8317
    },
    {
      "epoch": 15.347183748845799,
      "grad_norm": 0.4132469892501831,
      "learning_rate": 2.4945000000000003e-05,
      "loss": 0.0226,
      "step": 8318
    },
    {
      "epoch": 15.349030470914128,
      "grad_norm": 0.41254034638404846,
      "learning_rate": 2.4948000000000002e-05,
      "loss": 0.0103,
      "step": 8319
    },
    {
      "epoch": 15.350877192982455,
      "grad_norm": 0.705245852470398,
      "learning_rate": 2.4951e-05,
      "loss": 0.0195,
      "step": 8320
    },
    {
      "epoch": 15.352723915050785,
      "grad_norm": 0.988831102848053,
      "learning_rate": 2.4954e-05,
      "loss": 0.0222,
      "step": 8321
    },
    {
      "epoch": 15.354570637119114,
      "grad_norm": 0.5189151167869568,
      "learning_rate": 2.4957e-05,
      "loss": 0.0228,
      "step": 8322
    },
    {
      "epoch": 15.356417359187443,
      "grad_norm": 0.41819465160369873,
      "learning_rate": 2.4959999999999998e-05,
      "loss": 0.0218,
      "step": 8323
    },
    {
      "epoch": 15.35826408125577,
      "grad_norm": 0.2617114186286926,
      "learning_rate": 2.4962999999999998e-05,
      "loss": 0.0126,
      "step": 8324
    },
    {
      "epoch": 15.3601108033241,
      "grad_norm": 0.5680693984031677,
      "learning_rate": 2.4966e-05,
      "loss": 0.0208,
      "step": 8325
    },
    {
      "epoch": 15.361957525392429,
      "grad_norm": 3.6656782627105713,
      "learning_rate": 2.4969e-05,
      "loss": 0.0187,
      "step": 8326
    },
    {
      "epoch": 15.363804247460758,
      "grad_norm": 0.5825662612915039,
      "learning_rate": 2.4972e-05,
      "loss": 0.0178,
      "step": 8327
    },
    {
      "epoch": 15.365650969529085,
      "grad_norm": 0.4422478973865509,
      "learning_rate": 2.4975e-05,
      "loss": 0.0168,
      "step": 8328
    },
    {
      "epoch": 15.367497691597414,
      "grad_norm": 0.7485823035240173,
      "learning_rate": 2.4978e-05,
      "loss": 0.0232,
      "step": 8329
    },
    {
      "epoch": 15.369344413665743,
      "grad_norm": 0.7661342620849609,
      "learning_rate": 2.4981e-05,
      "loss": 0.0386,
      "step": 8330
    },
    {
      "epoch": 15.371191135734072,
      "grad_norm": 0.7451789975166321,
      "learning_rate": 2.4984e-05,
      "loss": 0.1895,
      "step": 8331
    },
    {
      "epoch": 15.373037857802402,
      "grad_norm": 0.7402056455612183,
      "learning_rate": 2.4987e-05,
      "loss": 0.1275,
      "step": 8332
    },
    {
      "epoch": 15.374884579870729,
      "grad_norm": 0.7276471257209778,
      "learning_rate": 2.499e-05,
      "loss": 0.2129,
      "step": 8333
    },
    {
      "epoch": 15.376731301939058,
      "grad_norm": 0.6210782527923584,
      "learning_rate": 2.4993e-05,
      "loss": 0.1199,
      "step": 8334
    },
    {
      "epoch": 15.378578024007387,
      "grad_norm": 0.7629683017730713,
      "learning_rate": 2.4996000000000003e-05,
      "loss": 0.1289,
      "step": 8335
    },
    {
      "epoch": 15.380424746075716,
      "grad_norm": 0.5822878479957581,
      "learning_rate": 2.4999000000000002e-05,
      "loss": 0.1173,
      "step": 8336
    },
    {
      "epoch": 15.382271468144044,
      "grad_norm": 0.551585853099823,
      "learning_rate": 2.5002000000000002e-05,
      "loss": 0.0899,
      "step": 8337
    },
    {
      "epoch": 15.384118190212373,
      "grad_norm": 0.48120298981666565,
      "learning_rate": 2.5005000000000002e-05,
      "loss": 0.0821,
      "step": 8338
    },
    {
      "epoch": 15.385964912280702,
      "grad_norm": 0.6678263545036316,
      "learning_rate": 2.5008000000000002e-05,
      "loss": 0.0939,
      "step": 8339
    },
    {
      "epoch": 15.387811634349031,
      "grad_norm": 0.4962681233882904,
      "learning_rate": 2.5011e-05,
      "loss": 0.068,
      "step": 8340
    },
    {
      "epoch": 15.389658356417359,
      "grad_norm": 0.4213404059410095,
      "learning_rate": 2.5014e-05,
      "loss": 0.0482,
      "step": 8341
    },
    {
      "epoch": 15.391505078485688,
      "grad_norm": 0.5263739228248596,
      "learning_rate": 2.5017e-05,
      "loss": 0.066,
      "step": 8342
    },
    {
      "epoch": 15.393351800554017,
      "grad_norm": 0.45648154616355896,
      "learning_rate": 2.502e-05,
      "loss": 0.0465,
      "step": 8343
    },
    {
      "epoch": 15.395198522622346,
      "grad_norm": 0.33768972754478455,
      "learning_rate": 2.5023e-05,
      "loss": 0.0305,
      "step": 8344
    },
    {
      "epoch": 15.397045244690673,
      "grad_norm": 0.42431968450546265,
      "learning_rate": 2.5026e-05,
      "loss": 0.0504,
      "step": 8345
    },
    {
      "epoch": 15.398891966759003,
      "grad_norm": 0.25444427132606506,
      "learning_rate": 2.5029e-05,
      "loss": 0.0166,
      "step": 8346
    },
    {
      "epoch": 15.400738688827332,
      "grad_norm": 0.41962170600891113,
      "learning_rate": 2.5032e-05,
      "loss": 0.0172,
      "step": 8347
    },
    {
      "epoch": 15.40258541089566,
      "grad_norm": 0.7231307029724121,
      "learning_rate": 2.5035e-05,
      "loss": 0.0227,
      "step": 8348
    },
    {
      "epoch": 15.404432132963988,
      "grad_norm": 0.4448593556880951,
      "learning_rate": 2.5038e-05,
      "loss": 0.0197,
      "step": 8349
    },
    {
      "epoch": 15.406278855032317,
      "grad_norm": 3.519976854324341,
      "learning_rate": 2.5041e-05,
      "loss": 0.0768,
      "step": 8350
    },
    {
      "epoch": 15.408125577100646,
      "grad_norm": 0.3873467743396759,
      "learning_rate": 2.5044e-05,
      "loss": 0.0219,
      "step": 8351
    },
    {
      "epoch": 15.409972299168976,
      "grad_norm": 0.2840051054954529,
      "learning_rate": 2.5047e-05,
      "loss": 0.0239,
      "step": 8352
    },
    {
      "epoch": 15.411819021237303,
      "grad_norm": 0.43799832463264465,
      "learning_rate": 2.505e-05,
      "loss": 0.0164,
      "step": 8353
    },
    {
      "epoch": 15.413665743305632,
      "grad_norm": 0.4558172821998596,
      "learning_rate": 2.5053e-05,
      "loss": 0.0175,
      "step": 8354
    },
    {
      "epoch": 15.415512465373961,
      "grad_norm": 0.2507631778717041,
      "learning_rate": 2.5056000000000002e-05,
      "loss": 0.0132,
      "step": 8355
    },
    {
      "epoch": 15.41735918744229,
      "grad_norm": 0.2601400315761566,
      "learning_rate": 2.5059000000000002e-05,
      "loss": 0.0132,
      "step": 8356
    },
    {
      "epoch": 15.41920590951062,
      "grad_norm": 0.3279007077217102,
      "learning_rate": 2.5062e-05,
      "loss": 0.0118,
      "step": 8357
    },
    {
      "epoch": 15.421052631578947,
      "grad_norm": 0.36395788192749023,
      "learning_rate": 2.5065e-05,
      "loss": 0.0134,
      "step": 8358
    },
    {
      "epoch": 15.422899353647276,
      "grad_norm": 0.6187348365783691,
      "learning_rate": 2.5068e-05,
      "loss": 0.0233,
      "step": 8359
    },
    {
      "epoch": 15.424746075715605,
      "grad_norm": 0.6799266338348389,
      "learning_rate": 2.5071e-05,
      "loss": 0.0133,
      "step": 8360
    },
    {
      "epoch": 15.426592797783934,
      "grad_norm": 0.9328656196594238,
      "learning_rate": 2.5074e-05,
      "loss": 0.0264,
      "step": 8361
    },
    {
      "epoch": 15.428439519852262,
      "grad_norm": 0.4898659288883209,
      "learning_rate": 2.5077e-05,
      "loss": 0.0151,
      "step": 8362
    },
    {
      "epoch": 15.43028624192059,
      "grad_norm": 0.6099836230278015,
      "learning_rate": 2.508e-05,
      "loss": 0.0189,
      "step": 8363
    },
    {
      "epoch": 15.43213296398892,
      "grad_norm": 0.23199789226055145,
      "learning_rate": 2.5083e-05,
      "loss": 0.0081,
      "step": 8364
    },
    {
      "epoch": 15.43397968605725,
      "grad_norm": 0.24271805584430695,
      "learning_rate": 2.5086000000000003e-05,
      "loss": 0.0095,
      "step": 8365
    },
    {
      "epoch": 15.435826408125576,
      "grad_norm": 0.37805840373039246,
      "learning_rate": 2.5089000000000003e-05,
      "loss": 0.0161,
      "step": 8366
    },
    {
      "epoch": 15.437673130193906,
      "grad_norm": 0.49491918087005615,
      "learning_rate": 2.5092000000000003e-05,
      "loss": 0.0079,
      "step": 8367
    },
    {
      "epoch": 15.439519852262235,
      "grad_norm": 0.7112870216369629,
      "learning_rate": 2.5095000000000003e-05,
      "loss": 0.0227,
      "step": 8368
    },
    {
      "epoch": 15.441366574330564,
      "grad_norm": 0.4112222492694855,
      "learning_rate": 2.5098000000000003e-05,
      "loss": 0.01,
      "step": 8369
    },
    {
      "epoch": 15.443213296398891,
      "grad_norm": 0.6659647226333618,
      "learning_rate": 2.5101e-05,
      "loss": 0.0222,
      "step": 8370
    },
    {
      "epoch": 15.44506001846722,
      "grad_norm": 0.4516090452671051,
      "learning_rate": 2.5104e-05,
      "loss": 0.0133,
      "step": 8371
    },
    {
      "epoch": 15.44690674053555,
      "grad_norm": 0.7286649942398071,
      "learning_rate": 2.5107e-05,
      "loss": 0.018,
      "step": 8372
    },
    {
      "epoch": 15.448753462603879,
      "grad_norm": 0.6150336265563965,
      "learning_rate": 2.511e-05,
      "loss": 0.0234,
      "step": 8373
    },
    {
      "epoch": 15.450600184672206,
      "grad_norm": 0.266416072845459,
      "learning_rate": 2.5112999999999998e-05,
      "loss": 0.0122,
      "step": 8374
    },
    {
      "epoch": 15.452446906740535,
      "grad_norm": 0.26499465107917786,
      "learning_rate": 2.5116e-05,
      "loss": 0.0112,
      "step": 8375
    },
    {
      "epoch": 15.454293628808864,
      "grad_norm": 0.5799499154090881,
      "learning_rate": 2.5119e-05,
      "loss": 0.0175,
      "step": 8376
    },
    {
      "epoch": 15.456140350877194,
      "grad_norm": 0.625869631767273,
      "learning_rate": 2.5122e-05,
      "loss": 0.0204,
      "step": 8377
    },
    {
      "epoch": 15.45798707294552,
      "grad_norm": 0.6527268886566162,
      "learning_rate": 2.5125e-05,
      "loss": 0.039,
      "step": 8378
    },
    {
      "epoch": 15.45983379501385,
      "grad_norm": 1.0619781017303467,
      "learning_rate": 2.5128e-05,
      "loss": 0.0231,
      "step": 8379
    },
    {
      "epoch": 15.46168051708218,
      "grad_norm": 0.5146917700767517,
      "learning_rate": 2.5131e-05,
      "loss": 0.0252,
      "step": 8380
    },
    {
      "epoch": 15.463527239150508,
      "grad_norm": 0.8048033118247986,
      "learning_rate": 2.5134e-05,
      "loss": 0.2758,
      "step": 8381
    },
    {
      "epoch": 15.465373961218837,
      "grad_norm": 0.56715989112854,
      "learning_rate": 2.5137e-05,
      "loss": 0.1736,
      "step": 8382
    },
    {
      "epoch": 15.467220683287165,
      "grad_norm": 1.2161662578582764,
      "learning_rate": 2.514e-05,
      "loss": 0.1227,
      "step": 8383
    },
    {
      "epoch": 15.469067405355494,
      "grad_norm": 0.6002994775772095,
      "learning_rate": 2.5143e-05,
      "loss": 0.1131,
      "step": 8384
    },
    {
      "epoch": 15.470914127423823,
      "grad_norm": 0.6885210871696472,
      "learning_rate": 2.5146e-05,
      "loss": 0.1846,
      "step": 8385
    },
    {
      "epoch": 15.472760849492152,
      "grad_norm": 0.5307503342628479,
      "learning_rate": 2.5149000000000003e-05,
      "loss": 0.0912,
      "step": 8386
    },
    {
      "epoch": 15.47460757156048,
      "grad_norm": 0.35235410928726196,
      "learning_rate": 2.5152000000000002e-05,
      "loss": 0.0671,
      "step": 8387
    },
    {
      "epoch": 15.476454293628809,
      "grad_norm": 0.6050505638122559,
      "learning_rate": 2.5155000000000002e-05,
      "loss": 0.1021,
      "step": 8388
    },
    {
      "epoch": 15.478301015697138,
      "grad_norm": 0.526045024394989,
      "learning_rate": 2.5158000000000002e-05,
      "loss": 0.0765,
      "step": 8389
    },
    {
      "epoch": 15.480147737765467,
      "grad_norm": 0.36944779753685,
      "learning_rate": 2.5161000000000002e-05,
      "loss": 0.0586,
      "step": 8390
    },
    {
      "epoch": 15.481994459833794,
      "grad_norm": 0.41731560230255127,
      "learning_rate": 2.5164e-05,
      "loss": 0.052,
      "step": 8391
    },
    {
      "epoch": 15.483841181902124,
      "grad_norm": 0.8687494397163391,
      "learning_rate": 2.5167e-05,
      "loss": 0.0673,
      "step": 8392
    },
    {
      "epoch": 15.485687903970453,
      "grad_norm": 0.6858538389205933,
      "learning_rate": 2.517e-05,
      "loss": 0.0602,
      "step": 8393
    },
    {
      "epoch": 15.487534626038782,
      "grad_norm": 1.1235566139221191,
      "learning_rate": 2.5172999999999998e-05,
      "loss": 0.0459,
      "step": 8394
    },
    {
      "epoch": 15.48938134810711,
      "grad_norm": 0.41303691267967224,
      "learning_rate": 2.5175999999999997e-05,
      "loss": 0.0571,
      "step": 8395
    },
    {
      "epoch": 15.491228070175438,
      "grad_norm": 0.5385510325431824,
      "learning_rate": 2.5179e-05,
      "loss": 0.0399,
      "step": 8396
    },
    {
      "epoch": 15.493074792243767,
      "grad_norm": 0.3182043731212616,
      "learning_rate": 2.5182e-05,
      "loss": 0.0363,
      "step": 8397
    },
    {
      "epoch": 15.494921514312097,
      "grad_norm": 0.24598966538906097,
      "learning_rate": 2.5185e-05,
      "loss": 0.0111,
      "step": 8398
    },
    {
      "epoch": 15.496768236380424,
      "grad_norm": 0.40207529067993164,
      "learning_rate": 2.5188e-05,
      "loss": 0.0207,
      "step": 8399
    },
    {
      "epoch": 15.498614958448753,
      "grad_norm": 0.5250248312950134,
      "learning_rate": 2.5191e-05,
      "loss": 0.0587,
      "step": 8400
    },
    {
      "epoch": 15.500461680517082,
      "grad_norm": 0.33213678002357483,
      "learning_rate": 2.5194e-05,
      "loss": 0.0199,
      "step": 8401
    },
    {
      "epoch": 15.502308402585411,
      "grad_norm": 0.17228974401950836,
      "learning_rate": 2.5197e-05,
      "loss": 0.0081,
      "step": 8402
    },
    {
      "epoch": 15.504155124653739,
      "grad_norm": 0.3369840383529663,
      "learning_rate": 2.52e-05,
      "loss": 0.0109,
      "step": 8403
    },
    {
      "epoch": 15.506001846722068,
      "grad_norm": 0.31128376722335815,
      "learning_rate": 2.5203e-05,
      "loss": 0.0113,
      "step": 8404
    },
    {
      "epoch": 15.507848568790397,
      "grad_norm": 0.36564984917640686,
      "learning_rate": 2.5206e-05,
      "loss": 0.0221,
      "step": 8405
    },
    {
      "epoch": 15.509695290858726,
      "grad_norm": 0.6158779859542847,
      "learning_rate": 2.5209000000000002e-05,
      "loss": 0.0188,
      "step": 8406
    },
    {
      "epoch": 15.511542012927055,
      "grad_norm": 0.4490515887737274,
      "learning_rate": 2.5212000000000002e-05,
      "loss": 0.0254,
      "step": 8407
    },
    {
      "epoch": 15.513388734995383,
      "grad_norm": 0.4023129642009735,
      "learning_rate": 2.5215e-05,
      "loss": 0.0178,
      "step": 8408
    },
    {
      "epoch": 15.515235457063712,
      "grad_norm": 0.7977317571640015,
      "learning_rate": 2.5218e-05,
      "loss": 0.0323,
      "step": 8409
    },
    {
      "epoch": 15.517082179132041,
      "grad_norm": 0.8370609283447266,
      "learning_rate": 2.5221e-05,
      "loss": 0.016,
      "step": 8410
    },
    {
      "epoch": 15.51892890120037,
      "grad_norm": 0.512177586555481,
      "learning_rate": 2.5224e-05,
      "loss": 0.0407,
      "step": 8411
    },
    {
      "epoch": 15.520775623268698,
      "grad_norm": 0.4275936186313629,
      "learning_rate": 2.5227e-05,
      "loss": 0.0131,
      "step": 8412
    },
    {
      "epoch": 15.522622345337027,
      "grad_norm": 0.22574132680892944,
      "learning_rate": 2.523e-05,
      "loss": 0.0088,
      "step": 8413
    },
    {
      "epoch": 15.524469067405356,
      "grad_norm": 0.2435065507888794,
      "learning_rate": 2.5233e-05,
      "loss": 0.0085,
      "step": 8414
    },
    {
      "epoch": 15.526315789473685,
      "grad_norm": 0.6562572717666626,
      "learning_rate": 2.5236e-05,
      "loss": 0.0204,
      "step": 8415
    },
    {
      "epoch": 15.528162511542012,
      "grad_norm": 0.3493426442146301,
      "learning_rate": 2.5239000000000003e-05,
      "loss": 0.006,
      "step": 8416
    },
    {
      "epoch": 15.530009233610341,
      "grad_norm": 0.3053175210952759,
      "learning_rate": 2.5242000000000003e-05,
      "loss": 0.0384,
      "step": 8417
    },
    {
      "epoch": 15.53185595567867,
      "grad_norm": 0.4377683699131012,
      "learning_rate": 2.5245000000000003e-05,
      "loss": 0.0208,
      "step": 8418
    },
    {
      "epoch": 15.533702677747,
      "grad_norm": 0.35627028346061707,
      "learning_rate": 2.5248e-05,
      "loss": 0.0177,
      "step": 8419
    },
    {
      "epoch": 15.535549399815327,
      "grad_norm": 0.33622390031814575,
      "learning_rate": 2.5251e-05,
      "loss": 0.0101,
      "step": 8420
    },
    {
      "epoch": 15.537396121883656,
      "grad_norm": 0.40122103691101074,
      "learning_rate": 2.5254e-05,
      "loss": 0.0137,
      "step": 8421
    },
    {
      "epoch": 15.539242843951985,
      "grad_norm": 0.44825655221939087,
      "learning_rate": 2.5257e-05,
      "loss": 0.0162,
      "step": 8422
    },
    {
      "epoch": 15.541089566020315,
      "grad_norm": 0.32165566086769104,
      "learning_rate": 2.526e-05,
      "loss": 0.0148,
      "step": 8423
    },
    {
      "epoch": 15.542936288088642,
      "grad_norm": 0.43938520550727844,
      "learning_rate": 2.5263e-05,
      "loss": 0.0132,
      "step": 8424
    },
    {
      "epoch": 15.544783010156971,
      "grad_norm": 0.36822405457496643,
      "learning_rate": 2.5266e-05,
      "loss": 0.0166,
      "step": 8425
    },
    {
      "epoch": 15.5466297322253,
      "grad_norm": 0.7597044706344604,
      "learning_rate": 2.5269e-05,
      "loss": 0.0244,
      "step": 8426
    },
    {
      "epoch": 15.54847645429363,
      "grad_norm": 0.6303825378417969,
      "learning_rate": 2.5272e-05,
      "loss": 0.0175,
      "step": 8427
    },
    {
      "epoch": 15.550323176361957,
      "grad_norm": 0.38170045614242554,
      "learning_rate": 2.5275e-05,
      "loss": 0.0105,
      "step": 8428
    },
    {
      "epoch": 15.552169898430286,
      "grad_norm": 0.5360559225082397,
      "learning_rate": 2.5278e-05,
      "loss": 0.0149,
      "step": 8429
    },
    {
      "epoch": 15.554016620498615,
      "grad_norm": 0.6139945983886719,
      "learning_rate": 2.5281e-05,
      "loss": 0.0233,
      "step": 8430
    },
    {
      "epoch": 15.555863342566944,
      "grad_norm": 0.8562393188476562,
      "learning_rate": 2.5284e-05,
      "loss": 0.2325,
      "step": 8431
    },
    {
      "epoch": 15.557710064635273,
      "grad_norm": 1.1304420232772827,
      "learning_rate": 2.5287e-05,
      "loss": 0.1774,
      "step": 8432
    },
    {
      "epoch": 15.5595567867036,
      "grad_norm": 1.1057580709457397,
      "learning_rate": 2.529e-05,
      "loss": 0.166,
      "step": 8433
    },
    {
      "epoch": 15.56140350877193,
      "grad_norm": 0.7534903287887573,
      "learning_rate": 2.5293e-05,
      "loss": 0.1355,
      "step": 8434
    },
    {
      "epoch": 15.563250230840259,
      "grad_norm": 0.6858900785446167,
      "learning_rate": 2.5296e-05,
      "loss": 0.1177,
      "step": 8435
    },
    {
      "epoch": 15.565096952908588,
      "grad_norm": 0.4988061785697937,
      "learning_rate": 2.5299000000000003e-05,
      "loss": 0.098,
      "step": 8436
    },
    {
      "epoch": 15.566943674976915,
      "grad_norm": 0.6958936452865601,
      "learning_rate": 2.5302000000000003e-05,
      "loss": 0.1319,
      "step": 8437
    },
    {
      "epoch": 15.568790397045245,
      "grad_norm": 0.7454564571380615,
      "learning_rate": 2.5305000000000003e-05,
      "loss": 0.1212,
      "step": 8438
    },
    {
      "epoch": 15.570637119113574,
      "grad_norm": 0.7997215390205383,
      "learning_rate": 2.5308000000000002e-05,
      "loss": 0.0711,
      "step": 8439
    },
    {
      "epoch": 15.572483841181903,
      "grad_norm": 0.5632338523864746,
      "learning_rate": 2.5311000000000002e-05,
      "loss": 0.113,
      "step": 8440
    },
    {
      "epoch": 15.57433056325023,
      "grad_norm": 0.759938657283783,
      "learning_rate": 2.5314000000000002e-05,
      "loss": 0.0797,
      "step": 8441
    },
    {
      "epoch": 15.57617728531856,
      "grad_norm": 1.036684513092041,
      "learning_rate": 2.5317000000000002e-05,
      "loss": 0.1144,
      "step": 8442
    },
    {
      "epoch": 15.578024007386889,
      "grad_norm": 0.31157785654067993,
      "learning_rate": 2.5319999999999998e-05,
      "loss": 0.0285,
      "step": 8443
    },
    {
      "epoch": 15.579870729455218,
      "grad_norm": 0.3606410324573517,
      "learning_rate": 2.5322999999999998e-05,
      "loss": 0.0345,
      "step": 8444
    },
    {
      "epoch": 15.581717451523545,
      "grad_norm": 0.3788275420665741,
      "learning_rate": 2.5325999999999998e-05,
      "loss": 0.026,
      "step": 8445
    },
    {
      "epoch": 15.583564173591874,
      "grad_norm": 0.30666303634643555,
      "learning_rate": 2.5329e-05,
      "loss": 0.0198,
      "step": 8446
    },
    {
      "epoch": 15.585410895660203,
      "grad_norm": 0.7083009481430054,
      "learning_rate": 2.5332e-05,
      "loss": 0.0227,
      "step": 8447
    },
    {
      "epoch": 15.587257617728532,
      "grad_norm": 0.21037326753139496,
      "learning_rate": 2.5335e-05,
      "loss": 0.014,
      "step": 8448
    },
    {
      "epoch": 15.58910433979686,
      "grad_norm": 0.3480874300003052,
      "learning_rate": 2.5338e-05,
      "loss": 0.0089,
      "step": 8449
    },
    {
      "epoch": 15.590951061865189,
      "grad_norm": 0.843414306640625,
      "learning_rate": 2.5341e-05,
      "loss": 0.0191,
      "step": 8450
    },
    {
      "epoch": 15.592797783933518,
      "grad_norm": 0.8120933175086975,
      "learning_rate": 2.5344e-05,
      "loss": 0.0163,
      "step": 8451
    },
    {
      "epoch": 15.594644506001847,
      "grad_norm": 0.38233235478401184,
      "learning_rate": 2.5347e-05,
      "loss": 0.0201,
      "step": 8452
    },
    {
      "epoch": 15.596491228070175,
      "grad_norm": 0.289878785610199,
      "learning_rate": 2.535e-05,
      "loss": 0.0105,
      "step": 8453
    },
    {
      "epoch": 15.598337950138504,
      "grad_norm": 0.35070326924324036,
      "learning_rate": 2.5353e-05,
      "loss": 0.0175,
      "step": 8454
    },
    {
      "epoch": 15.600184672206833,
      "grad_norm": 0.40063390135765076,
      "learning_rate": 2.5356e-05,
      "loss": 0.0099,
      "step": 8455
    },
    {
      "epoch": 15.602031394275162,
      "grad_norm": 0.18869224190711975,
      "learning_rate": 2.5359000000000002e-05,
      "loss": 0.0083,
      "step": 8456
    },
    {
      "epoch": 15.603878116343491,
      "grad_norm": 0.39934617280960083,
      "learning_rate": 2.5362000000000002e-05,
      "loss": 0.0146,
      "step": 8457
    },
    {
      "epoch": 15.605724838411819,
      "grad_norm": 0.4403945803642273,
      "learning_rate": 2.5365000000000002e-05,
      "loss": 0.0167,
      "step": 8458
    },
    {
      "epoch": 15.607571560480148,
      "grad_norm": 0.39084070920944214,
      "learning_rate": 2.5368000000000002e-05,
      "loss": 0.0118,
      "step": 8459
    },
    {
      "epoch": 15.609418282548477,
      "grad_norm": 0.7717267870903015,
      "learning_rate": 2.5371e-05,
      "loss": 0.0245,
      "step": 8460
    },
    {
      "epoch": 15.611265004616806,
      "grad_norm": 0.38657015562057495,
      "learning_rate": 2.5374e-05,
      "loss": 0.0112,
      "step": 8461
    },
    {
      "epoch": 15.613111726685133,
      "grad_norm": 0.28539496660232544,
      "learning_rate": 2.5377e-05,
      "loss": 0.0134,
      "step": 8462
    },
    {
      "epoch": 15.614958448753463,
      "grad_norm": 0.3543245494365692,
      "learning_rate": 2.538e-05,
      "loss": 0.0148,
      "step": 8463
    },
    {
      "epoch": 15.616805170821792,
      "grad_norm": 0.3170192241668701,
      "learning_rate": 2.5383e-05,
      "loss": 0.0177,
      "step": 8464
    },
    {
      "epoch": 15.61865189289012,
      "grad_norm": 0.5334312915802002,
      "learning_rate": 2.5386e-05,
      "loss": 0.0139,
      "step": 8465
    },
    {
      "epoch": 15.620498614958448,
      "grad_norm": 0.700171947479248,
      "learning_rate": 2.5389000000000004e-05,
      "loss": 0.0322,
      "step": 8466
    },
    {
      "epoch": 15.622345337026777,
      "grad_norm": 0.6469569206237793,
      "learning_rate": 2.5392000000000004e-05,
      "loss": 0.0183,
      "step": 8467
    },
    {
      "epoch": 15.624192059095106,
      "grad_norm": 0.9562910199165344,
      "learning_rate": 2.5395e-05,
      "loss": 0.0204,
      "step": 8468
    },
    {
      "epoch": 15.626038781163436,
      "grad_norm": 0.4306713938713074,
      "learning_rate": 2.5398e-05,
      "loss": 0.012,
      "step": 8469
    },
    {
      "epoch": 15.627885503231763,
      "grad_norm": 1.4701281785964966,
      "learning_rate": 2.5401e-05,
      "loss": 0.0265,
      "step": 8470
    },
    {
      "epoch": 15.629732225300092,
      "grad_norm": 0.3720110058784485,
      "learning_rate": 2.5404e-05,
      "loss": 0.0154,
      "step": 8471
    },
    {
      "epoch": 15.631578947368421,
      "grad_norm": 0.4950038492679596,
      "learning_rate": 2.5407e-05,
      "loss": 0.0166,
      "step": 8472
    },
    {
      "epoch": 15.63342566943675,
      "grad_norm": 0.5569928288459778,
      "learning_rate": 2.541e-05,
      "loss": 0.0178,
      "step": 8473
    },
    {
      "epoch": 15.635272391505078,
      "grad_norm": 0.33441534638404846,
      "learning_rate": 2.5413e-05,
      "loss": 0.0121,
      "step": 8474
    },
    {
      "epoch": 15.637119113573407,
      "grad_norm": 0.2959272563457489,
      "learning_rate": 2.5416e-05,
      "loss": 0.0163,
      "step": 8475
    },
    {
      "epoch": 15.638965835641736,
      "grad_norm": 0.4156274199485779,
      "learning_rate": 2.5419000000000002e-05,
      "loss": 0.0141,
      "step": 8476
    },
    {
      "epoch": 15.640812557710065,
      "grad_norm": 0.4267812669277191,
      "learning_rate": 2.5422e-05,
      "loss": 0.0313,
      "step": 8477
    },
    {
      "epoch": 15.642659279778393,
      "grad_norm": 0.5098170042037964,
      "learning_rate": 2.5425e-05,
      "loss": 0.0209,
      "step": 8478
    },
    {
      "epoch": 15.644506001846722,
      "grad_norm": 0.3128315806388855,
      "learning_rate": 2.5428e-05,
      "loss": 0.014,
      "step": 8479
    },
    {
      "epoch": 15.64635272391505,
      "grad_norm": 1.0015007257461548,
      "learning_rate": 2.5431e-05,
      "loss": 0.0429,
      "step": 8480
    },
    {
      "epoch": 15.64819944598338,
      "grad_norm": 1.4465745687484741,
      "learning_rate": 2.5434e-05,
      "loss": 0.2054,
      "step": 8481
    },
    {
      "epoch": 15.65004616805171,
      "grad_norm": 0.8323967456817627,
      "learning_rate": 2.5437e-05,
      "loss": 0.2016,
      "step": 8482
    },
    {
      "epoch": 15.651892890120036,
      "grad_norm": 0.5159797668457031,
      "learning_rate": 2.544e-05,
      "loss": 0.0936,
      "step": 8483
    },
    {
      "epoch": 15.653739612188366,
      "grad_norm": 0.5815988183021545,
      "learning_rate": 2.5443e-05,
      "loss": 0.1431,
      "step": 8484
    },
    {
      "epoch": 15.655586334256695,
      "grad_norm": 0.6828600168228149,
      "learning_rate": 2.5446e-05,
      "loss": 0.1107,
      "step": 8485
    },
    {
      "epoch": 15.657433056325024,
      "grad_norm": 0.6418094635009766,
      "learning_rate": 2.5449000000000003e-05,
      "loss": 0.1151,
      "step": 8486
    },
    {
      "epoch": 15.659279778393351,
      "grad_norm": 0.5052107572555542,
      "learning_rate": 2.5452000000000003e-05,
      "loss": 0.1149,
      "step": 8487
    },
    {
      "epoch": 15.66112650046168,
      "grad_norm": 0.4315268099308014,
      "learning_rate": 2.5455000000000003e-05,
      "loss": 0.1143,
      "step": 8488
    },
    {
      "epoch": 15.66297322253001,
      "grad_norm": 0.49266499280929565,
      "learning_rate": 2.5458000000000003e-05,
      "loss": 0.0622,
      "step": 8489
    },
    {
      "epoch": 15.664819944598339,
      "grad_norm": 1.084386944770813,
      "learning_rate": 2.5461000000000002e-05,
      "loss": 0.0626,
      "step": 8490
    },
    {
      "epoch": 15.666666666666666,
      "grad_norm": 0.5839085578918457,
      "learning_rate": 2.5464000000000002e-05,
      "loss": 0.0729,
      "step": 8491
    },
    {
      "epoch": 15.668513388734995,
      "grad_norm": 0.5455254316329956,
      "learning_rate": 2.5467e-05,
      "loss": 0.0823,
      "step": 8492
    },
    {
      "epoch": 15.670360110803324,
      "grad_norm": 1.681470513343811,
      "learning_rate": 2.547e-05,
      "loss": 0.1319,
      "step": 8493
    },
    {
      "epoch": 15.672206832871654,
      "grad_norm": 0.481160432100296,
      "learning_rate": 2.5472999999999998e-05,
      "loss": 0.0468,
      "step": 8494
    },
    {
      "epoch": 15.67405355493998,
      "grad_norm": 0.4013616442680359,
      "learning_rate": 2.5475999999999998e-05,
      "loss": 0.0352,
      "step": 8495
    },
    {
      "epoch": 15.67590027700831,
      "grad_norm": 2.3175511360168457,
      "learning_rate": 2.5479e-05,
      "loss": 0.046,
      "step": 8496
    },
    {
      "epoch": 15.67774699907664,
      "grad_norm": 0.27709218859672546,
      "learning_rate": 2.5482e-05,
      "loss": 0.0217,
      "step": 8497
    },
    {
      "epoch": 15.679593721144968,
      "grad_norm": 0.4663790762424469,
      "learning_rate": 2.5485e-05,
      "loss": 0.0275,
      "step": 8498
    },
    {
      "epoch": 15.681440443213296,
      "grad_norm": 0.40022438764572144,
      "learning_rate": 2.5488e-05,
      "loss": 0.0344,
      "step": 8499
    },
    {
      "epoch": 15.683287165281625,
      "grad_norm": 0.29696258902549744,
      "learning_rate": 2.5491e-05,
      "loss": 0.0146,
      "step": 8500
    },
    {
      "epoch": 15.685133887349954,
      "grad_norm": 0.22244930267333984,
      "learning_rate": 2.5494e-05,
      "loss": 0.0199,
      "step": 8501
    },
    {
      "epoch": 15.686980609418283,
      "grad_norm": 0.2986993193626404,
      "learning_rate": 2.5497e-05,
      "loss": 0.0203,
      "step": 8502
    },
    {
      "epoch": 15.68882733148661,
      "grad_norm": 0.7852336764335632,
      "learning_rate": 2.55e-05,
      "loss": 0.0247,
      "step": 8503
    },
    {
      "epoch": 15.69067405355494,
      "grad_norm": 0.6283233761787415,
      "learning_rate": 2.5503e-05,
      "loss": 0.037,
      "step": 8504
    },
    {
      "epoch": 15.692520775623269,
      "grad_norm": 0.544381320476532,
      "learning_rate": 2.5506e-05,
      "loss": 0.0193,
      "step": 8505
    },
    {
      "epoch": 15.694367497691598,
      "grad_norm": 0.7549434304237366,
      "learning_rate": 2.5509e-05,
      "loss": 0.0537,
      "step": 8506
    },
    {
      "epoch": 15.696214219759927,
      "grad_norm": 0.407317191362381,
      "learning_rate": 2.5512000000000002e-05,
      "loss": 0.0351,
      "step": 8507
    },
    {
      "epoch": 15.698060941828254,
      "grad_norm": 0.3477088510990143,
      "learning_rate": 2.5515000000000002e-05,
      "loss": 0.0156,
      "step": 8508
    },
    {
      "epoch": 15.699907663896584,
      "grad_norm": 0.2679806351661682,
      "learning_rate": 2.5518000000000002e-05,
      "loss": 0.0121,
      "step": 8509
    },
    {
      "epoch": 15.701754385964913,
      "grad_norm": 0.42538949847221375,
      "learning_rate": 2.5521000000000002e-05,
      "loss": 0.013,
      "step": 8510
    },
    {
      "epoch": 15.703601108033242,
      "grad_norm": 0.5099429488182068,
      "learning_rate": 2.5524e-05,
      "loss": 0.0166,
      "step": 8511
    },
    {
      "epoch": 15.70544783010157,
      "grad_norm": 0.26436755061149597,
      "learning_rate": 2.5527e-05,
      "loss": 0.0109,
      "step": 8512
    },
    {
      "epoch": 15.707294552169898,
      "grad_norm": 0.28746193647384644,
      "learning_rate": 2.553e-05,
      "loss": 0.0149,
      "step": 8513
    },
    {
      "epoch": 15.709141274238227,
      "grad_norm": 0.48255273699760437,
      "learning_rate": 2.5533e-05,
      "loss": 0.014,
      "step": 8514
    },
    {
      "epoch": 15.710987996306557,
      "grad_norm": 0.7732357978820801,
      "learning_rate": 2.5536e-05,
      "loss": 0.0217,
      "step": 8515
    },
    {
      "epoch": 15.712834718374884,
      "grad_norm": 0.40901118516921997,
      "learning_rate": 2.5539e-05,
      "loss": 0.0213,
      "step": 8516
    },
    {
      "epoch": 15.714681440443213,
      "grad_norm": 0.4318062365055084,
      "learning_rate": 2.5542e-05,
      "loss": 0.0158,
      "step": 8517
    },
    {
      "epoch": 15.716528162511542,
      "grad_norm": 0.5730619430541992,
      "learning_rate": 2.5545e-05,
      "loss": 0.0102,
      "step": 8518
    },
    {
      "epoch": 15.718374884579871,
      "grad_norm": 0.5702316761016846,
      "learning_rate": 2.5548e-05,
      "loss": 0.0291,
      "step": 8519
    },
    {
      "epoch": 15.720221606648199,
      "grad_norm": 0.4010271728038788,
      "learning_rate": 2.5551e-05,
      "loss": 0.016,
      "step": 8520
    },
    {
      "epoch": 15.722068328716528,
      "grad_norm": 0.5954778790473938,
      "learning_rate": 2.5554e-05,
      "loss": 0.0295,
      "step": 8521
    },
    {
      "epoch": 15.723915050784857,
      "grad_norm": 0.2628062963485718,
      "learning_rate": 2.5557e-05,
      "loss": 0.0097,
      "step": 8522
    },
    {
      "epoch": 15.725761772853186,
      "grad_norm": 0.8188832998275757,
      "learning_rate": 2.556e-05,
      "loss": 0.0128,
      "step": 8523
    },
    {
      "epoch": 15.727608494921514,
      "grad_norm": 0.6020427346229553,
      "learning_rate": 2.5563e-05,
      "loss": 0.0184,
      "step": 8524
    },
    {
      "epoch": 15.729455216989843,
      "grad_norm": 0.31666770577430725,
      "learning_rate": 2.5566e-05,
      "loss": 0.0136,
      "step": 8525
    },
    {
      "epoch": 15.731301939058172,
      "grad_norm": 0.6543294787406921,
      "learning_rate": 2.5569e-05,
      "loss": 0.0218,
      "step": 8526
    },
    {
      "epoch": 15.733148661126501,
      "grad_norm": 0.3938964903354645,
      "learning_rate": 2.5572000000000002e-05,
      "loss": 0.0089,
      "step": 8527
    },
    {
      "epoch": 15.734995383194828,
      "grad_norm": 0.43330466747283936,
      "learning_rate": 2.5575e-05,
      "loss": 0.0204,
      "step": 8528
    },
    {
      "epoch": 15.736842105263158,
      "grad_norm": 0.834074079990387,
      "learning_rate": 2.5578e-05,
      "loss": 0.02,
      "step": 8529
    },
    {
      "epoch": 15.738688827331487,
      "grad_norm": 1.3514752388000488,
      "learning_rate": 2.5581e-05,
      "loss": 0.0343,
      "step": 8530
    },
    {
      "epoch": 15.740535549399816,
      "grad_norm": 1.174109935760498,
      "learning_rate": 2.5584e-05,
      "loss": 0.2363,
      "step": 8531
    },
    {
      "epoch": 15.742382271468145,
      "grad_norm": 0.7513344883918762,
      "learning_rate": 2.5587e-05,
      "loss": 0.2025,
      "step": 8532
    },
    {
      "epoch": 15.744228993536472,
      "grad_norm": 0.631834089756012,
      "learning_rate": 2.559e-05,
      "loss": 0.1719,
      "step": 8533
    },
    {
      "epoch": 15.746075715604801,
      "grad_norm": 0.7519591450691223,
      "learning_rate": 2.5593e-05,
      "loss": 0.1275,
      "step": 8534
    },
    {
      "epoch": 15.74792243767313,
      "grad_norm": 0.5671206712722778,
      "learning_rate": 2.5596e-05,
      "loss": 0.1259,
      "step": 8535
    },
    {
      "epoch": 15.749769159741458,
      "grad_norm": 0.6367517113685608,
      "learning_rate": 2.5599e-05,
      "loss": 0.1023,
      "step": 8536
    },
    {
      "epoch": 15.751615881809787,
      "grad_norm": 0.4786352515220642,
      "learning_rate": 2.5602000000000003e-05,
      "loss": 0.0637,
      "step": 8537
    },
    {
      "epoch": 15.753462603878116,
      "grad_norm": 0.5861334204673767,
      "learning_rate": 2.5605000000000003e-05,
      "loss": 0.0849,
      "step": 8538
    },
    {
      "epoch": 15.755309325946445,
      "grad_norm": 0.9659382700920105,
      "learning_rate": 2.5608000000000003e-05,
      "loss": 0.1039,
      "step": 8539
    },
    {
      "epoch": 15.757156048014775,
      "grad_norm": 0.48477768898010254,
      "learning_rate": 2.5611000000000003e-05,
      "loss": 0.061,
      "step": 8540
    },
    {
      "epoch": 15.759002770083102,
      "grad_norm": 0.4459925591945648,
      "learning_rate": 2.5614000000000002e-05,
      "loss": 0.0569,
      "step": 8541
    },
    {
      "epoch": 15.760849492151431,
      "grad_norm": 0.7846724390983582,
      "learning_rate": 2.5617e-05,
      "loss": 0.0975,
      "step": 8542
    },
    {
      "epoch": 15.76269621421976,
      "grad_norm": 0.4942132830619812,
      "learning_rate": 2.562e-05,
      "loss": 0.0602,
      "step": 8543
    },
    {
      "epoch": 15.76454293628809,
      "grad_norm": 0.47594746947288513,
      "learning_rate": 2.5623e-05,
      "loss": 0.0484,
      "step": 8544
    },
    {
      "epoch": 15.766389658356417,
      "grad_norm": 0.2541988492012024,
      "learning_rate": 2.5625999999999998e-05,
      "loss": 0.031,
      "step": 8545
    },
    {
      "epoch": 15.768236380424746,
      "grad_norm": 0.37014615535736084,
      "learning_rate": 2.5628999999999998e-05,
      "loss": 0.0276,
      "step": 8546
    },
    {
      "epoch": 15.770083102493075,
      "grad_norm": 0.47405698895454407,
      "learning_rate": 2.5632e-05,
      "loss": 0.0331,
      "step": 8547
    },
    {
      "epoch": 15.771929824561404,
      "grad_norm": 0.3453981578350067,
      "learning_rate": 2.5635e-05,
      "loss": 0.0446,
      "step": 8548
    },
    {
      "epoch": 15.773776546629731,
      "grad_norm": 0.5440641045570374,
      "learning_rate": 2.5638e-05,
      "loss": 0.0215,
      "step": 8549
    },
    {
      "epoch": 15.77562326869806,
      "grad_norm": 0.20974309742450714,
      "learning_rate": 2.5641e-05,
      "loss": 0.0206,
      "step": 8550
    },
    {
      "epoch": 15.77746999076639,
      "grad_norm": 0.6189224123954773,
      "learning_rate": 2.5644e-05,
      "loss": 0.028,
      "step": 8551
    },
    {
      "epoch": 15.779316712834719,
      "grad_norm": 0.48690396547317505,
      "learning_rate": 2.5647e-05,
      "loss": 0.0184,
      "step": 8552
    },
    {
      "epoch": 15.781163434903046,
      "grad_norm": 0.593185544013977,
      "learning_rate": 2.565e-05,
      "loss": 0.025,
      "step": 8553
    },
    {
      "epoch": 15.783010156971375,
      "grad_norm": 1.684084415435791,
      "learning_rate": 2.5653e-05,
      "loss": 0.0218,
      "step": 8554
    },
    {
      "epoch": 15.784856879039705,
      "grad_norm": 0.21022409200668335,
      "learning_rate": 2.5656e-05,
      "loss": 0.0134,
      "step": 8555
    },
    {
      "epoch": 15.786703601108034,
      "grad_norm": 0.37620681524276733,
      "learning_rate": 2.5659e-05,
      "loss": 0.0152,
      "step": 8556
    },
    {
      "epoch": 15.788550323176363,
      "grad_norm": 0.2804875671863556,
      "learning_rate": 2.5662000000000003e-05,
      "loss": 0.0231,
      "step": 8557
    },
    {
      "epoch": 15.79039704524469,
      "grad_norm": 0.25551798939704895,
      "learning_rate": 2.5665000000000002e-05,
      "loss": 0.0104,
      "step": 8558
    },
    {
      "epoch": 15.79224376731302,
      "grad_norm": 0.3816471993923187,
      "learning_rate": 2.5668000000000002e-05,
      "loss": 0.0161,
      "step": 8559
    },
    {
      "epoch": 15.794090489381349,
      "grad_norm": 0.6349307298660278,
      "learning_rate": 2.5671000000000002e-05,
      "loss": 0.0188,
      "step": 8560
    },
    {
      "epoch": 15.795937211449676,
      "grad_norm": 0.37047889828681946,
      "learning_rate": 2.5674000000000002e-05,
      "loss": 0.041,
      "step": 8561
    },
    {
      "epoch": 15.797783933518005,
      "grad_norm": 0.6681861877441406,
      "learning_rate": 2.5677e-05,
      "loss": 0.0179,
      "step": 8562
    },
    {
      "epoch": 15.799630655586334,
      "grad_norm": 0.661729633808136,
      "learning_rate": 2.568e-05,
      "loss": 0.0259,
      "step": 8563
    },
    {
      "epoch": 15.801477377654663,
      "grad_norm": 0.3159634470939636,
      "learning_rate": 2.5683e-05,
      "loss": 0.0194,
      "step": 8564
    },
    {
      "epoch": 15.803324099722992,
      "grad_norm": 0.30529919266700745,
      "learning_rate": 2.5686e-05,
      "loss": 0.0103,
      "step": 8565
    },
    {
      "epoch": 15.80517082179132,
      "grad_norm": 0.4781964421272278,
      "learning_rate": 2.5688999999999997e-05,
      "loss": 0.0195,
      "step": 8566
    },
    {
      "epoch": 15.807017543859649,
      "grad_norm": 0.3040353059768677,
      "learning_rate": 2.5692e-05,
      "loss": 0.0169,
      "step": 8567
    },
    {
      "epoch": 15.808864265927978,
      "grad_norm": 0.6087676882743835,
      "learning_rate": 2.5695e-05,
      "loss": 0.0193,
      "step": 8568
    },
    {
      "epoch": 15.810710987996307,
      "grad_norm": 0.5178345441818237,
      "learning_rate": 2.5698e-05,
      "loss": 0.0268,
      "step": 8569
    },
    {
      "epoch": 15.812557710064635,
      "grad_norm": 0.3992609679698944,
      "learning_rate": 2.5701e-05,
      "loss": 0.0255,
      "step": 8570
    },
    {
      "epoch": 15.814404432132964,
      "grad_norm": 0.404962956905365,
      "learning_rate": 2.5704e-05,
      "loss": 0.0186,
      "step": 8571
    },
    {
      "epoch": 15.816251154201293,
      "grad_norm": 0.4101698696613312,
      "learning_rate": 2.5707e-05,
      "loss": 0.0143,
      "step": 8572
    },
    {
      "epoch": 15.818097876269622,
      "grad_norm": 0.34589701890945435,
      "learning_rate": 2.571e-05,
      "loss": 0.0072,
      "step": 8573
    },
    {
      "epoch": 15.81994459833795,
      "grad_norm": 0.33388593792915344,
      "learning_rate": 2.5713e-05,
      "loss": 0.0127,
      "step": 8574
    },
    {
      "epoch": 15.821791320406279,
      "grad_norm": 0.5369555354118347,
      "learning_rate": 2.5716e-05,
      "loss": 0.0251,
      "step": 8575
    },
    {
      "epoch": 15.823638042474608,
      "grad_norm": 0.7632960677146912,
      "learning_rate": 2.5719e-05,
      "loss": 0.0163,
      "step": 8576
    },
    {
      "epoch": 15.825484764542937,
      "grad_norm": 0.38925424218177795,
      "learning_rate": 2.5722000000000002e-05,
      "loss": 0.013,
      "step": 8577
    },
    {
      "epoch": 15.827331486611264,
      "grad_norm": 0.43634846806526184,
      "learning_rate": 2.5725000000000002e-05,
      "loss": 0.026,
      "step": 8578
    },
    {
      "epoch": 15.829178208679593,
      "grad_norm": 0.60167396068573,
      "learning_rate": 2.5728e-05,
      "loss": 0.0209,
      "step": 8579
    },
    {
      "epoch": 15.831024930747922,
      "grad_norm": 2.016740322113037,
      "learning_rate": 2.5731e-05,
      "loss": 0.0264,
      "step": 8580
    },
    {
      "epoch": 15.832871652816252,
      "grad_norm": 0.6113871932029724,
      "learning_rate": 2.5734e-05,
      "loss": 0.1981,
      "step": 8581
    },
    {
      "epoch": 15.83471837488458,
      "grad_norm": 0.5366085767745972,
      "learning_rate": 2.5737e-05,
      "loss": 0.1967,
      "step": 8582
    },
    {
      "epoch": 15.836565096952908,
      "grad_norm": 0.720666229724884,
      "learning_rate": 2.574e-05,
      "loss": 0.1343,
      "step": 8583
    },
    {
      "epoch": 15.838411819021237,
      "grad_norm": 0.5490646958351135,
      "learning_rate": 2.5743e-05,
      "loss": 0.1563,
      "step": 8584
    },
    {
      "epoch": 15.840258541089566,
      "grad_norm": 1.0201573371887207,
      "learning_rate": 2.5746e-05,
      "loss": 0.1194,
      "step": 8585
    },
    {
      "epoch": 15.842105263157894,
      "grad_norm": 0.49989888072013855,
      "learning_rate": 2.5749e-05,
      "loss": 0.1354,
      "step": 8586
    },
    {
      "epoch": 15.843951985226223,
      "grad_norm": 0.3904455602169037,
      "learning_rate": 2.5752000000000003e-05,
      "loss": 0.0918,
      "step": 8587
    },
    {
      "epoch": 15.845798707294552,
      "grad_norm": 0.5409448146820068,
      "learning_rate": 2.5755000000000003e-05,
      "loss": 0.0916,
      "step": 8588
    },
    {
      "epoch": 15.847645429362881,
      "grad_norm": 0.6536295413970947,
      "learning_rate": 2.5758000000000003e-05,
      "loss": 0.113,
      "step": 8589
    },
    {
      "epoch": 15.84949215143121,
      "grad_norm": 0.5758427381515503,
      "learning_rate": 2.5761000000000003e-05,
      "loss": 0.0632,
      "step": 8590
    },
    {
      "epoch": 15.851338873499538,
      "grad_norm": 0.7237721085548401,
      "learning_rate": 2.5764e-05,
      "loss": 0.0726,
      "step": 8591
    },
    {
      "epoch": 15.853185595567867,
      "grad_norm": 0.30059942603111267,
      "learning_rate": 2.5767e-05,
      "loss": 0.0357,
      "step": 8592
    },
    {
      "epoch": 15.855032317636196,
      "grad_norm": 0.26866260170936584,
      "learning_rate": 2.577e-05,
      "loss": 0.0456,
      "step": 8593
    },
    {
      "epoch": 15.856879039704525,
      "grad_norm": 0.506643533706665,
      "learning_rate": 2.5773e-05,
      "loss": 0.0887,
      "step": 8594
    },
    {
      "epoch": 15.858725761772853,
      "grad_norm": 0.31971853971481323,
      "learning_rate": 2.5776e-05,
      "loss": 0.0334,
      "step": 8595
    },
    {
      "epoch": 15.860572483841182,
      "grad_norm": 0.30274417996406555,
      "learning_rate": 2.5779e-05,
      "loss": 0.045,
      "step": 8596
    },
    {
      "epoch": 15.86241920590951,
      "grad_norm": 0.44081827998161316,
      "learning_rate": 2.5782e-05,
      "loss": 0.0305,
      "step": 8597
    },
    {
      "epoch": 15.86426592797784,
      "grad_norm": 0.6209914684295654,
      "learning_rate": 2.5785e-05,
      "loss": 0.0405,
      "step": 8598
    },
    {
      "epoch": 15.866112650046167,
      "grad_norm": 1.5701161623001099,
      "learning_rate": 2.5788e-05,
      "loss": 0.0168,
      "step": 8599
    },
    {
      "epoch": 15.867959372114496,
      "grad_norm": 0.6633787155151367,
      "learning_rate": 2.5791e-05,
      "loss": 0.0203,
      "step": 8600
    },
    {
      "epoch": 15.869806094182826,
      "grad_norm": 0.3315002918243408,
      "learning_rate": 2.5794e-05,
      "loss": 0.0113,
      "step": 8601
    },
    {
      "epoch": 15.871652816251155,
      "grad_norm": 0.44118741154670715,
      "learning_rate": 2.5797e-05,
      "loss": 0.0359,
      "step": 8602
    },
    {
      "epoch": 15.873499538319482,
      "grad_norm": 0.8586814999580383,
      "learning_rate": 2.58e-05,
      "loss": 0.0212,
      "step": 8603
    },
    {
      "epoch": 15.875346260387811,
      "grad_norm": 0.6964927911758423,
      "learning_rate": 2.5803e-05,
      "loss": 0.0188,
      "step": 8604
    },
    {
      "epoch": 15.87719298245614,
      "grad_norm": 0.4353424608707428,
      "learning_rate": 2.5806e-05,
      "loss": 0.0134,
      "step": 8605
    },
    {
      "epoch": 15.87903970452447,
      "grad_norm": 0.43677806854248047,
      "learning_rate": 2.5809e-05,
      "loss": 0.0218,
      "step": 8606
    },
    {
      "epoch": 15.880886426592799,
      "grad_norm": 0.41412249207496643,
      "learning_rate": 2.5812000000000003e-05,
      "loss": 0.0115,
      "step": 8607
    },
    {
      "epoch": 15.882733148661126,
      "grad_norm": 0.3291482627391815,
      "learning_rate": 2.5815000000000003e-05,
      "loss": 0.0131,
      "step": 8608
    },
    {
      "epoch": 15.884579870729455,
      "grad_norm": 0.683460533618927,
      "learning_rate": 2.5818000000000003e-05,
      "loss": 0.0141,
      "step": 8609
    },
    {
      "epoch": 15.886426592797784,
      "grad_norm": 0.6479490995407104,
      "learning_rate": 2.5821000000000002e-05,
      "loss": 0.0231,
      "step": 8610
    },
    {
      "epoch": 15.888273314866112,
      "grad_norm": 0.259980171918869,
      "learning_rate": 2.5824000000000002e-05,
      "loss": 0.0112,
      "step": 8611
    },
    {
      "epoch": 15.89012003693444,
      "grad_norm": 0.2572813630104065,
      "learning_rate": 2.5827000000000002e-05,
      "loss": 0.0096,
      "step": 8612
    },
    {
      "epoch": 15.89196675900277,
      "grad_norm": 0.45563334226608276,
      "learning_rate": 2.5830000000000002e-05,
      "loss": 0.0119,
      "step": 8613
    },
    {
      "epoch": 15.8938134810711,
      "grad_norm": 0.31853538751602173,
      "learning_rate": 2.5833e-05,
      "loss": 0.0119,
      "step": 8614
    },
    {
      "epoch": 15.895660203139428,
      "grad_norm": 0.2511363923549652,
      "learning_rate": 2.5835999999999998e-05,
      "loss": 0.014,
      "step": 8615
    },
    {
      "epoch": 15.897506925207756,
      "grad_norm": 0.24003857374191284,
      "learning_rate": 2.5838999999999998e-05,
      "loss": 0.0103,
      "step": 8616
    },
    {
      "epoch": 15.899353647276085,
      "grad_norm": 0.3936110734939575,
      "learning_rate": 2.5842e-05,
      "loss": 0.0221,
      "step": 8617
    },
    {
      "epoch": 15.901200369344414,
      "grad_norm": 0.519484281539917,
      "learning_rate": 2.5845e-05,
      "loss": 0.0199,
      "step": 8618
    },
    {
      "epoch": 15.903047091412743,
      "grad_norm": 0.9927149415016174,
      "learning_rate": 2.5848e-05,
      "loss": 0.022,
      "step": 8619
    },
    {
      "epoch": 15.90489381348107,
      "grad_norm": 0.8863538503646851,
      "learning_rate": 2.5851e-05,
      "loss": 0.0233,
      "step": 8620
    },
    {
      "epoch": 15.9067405355494,
      "grad_norm": 0.2927561402320862,
      "learning_rate": 2.5854e-05,
      "loss": 0.012,
      "step": 8621
    },
    {
      "epoch": 15.908587257617729,
      "grad_norm": 0.6133517026901245,
      "learning_rate": 2.5857e-05,
      "loss": 0.0304,
      "step": 8622
    },
    {
      "epoch": 15.910433979686058,
      "grad_norm": 0.5118392109870911,
      "learning_rate": 2.586e-05,
      "loss": 0.0228,
      "step": 8623
    },
    {
      "epoch": 15.912280701754385,
      "grad_norm": 0.42478421330451965,
      "learning_rate": 2.5863e-05,
      "loss": 0.0119,
      "step": 8624
    },
    {
      "epoch": 15.914127423822714,
      "grad_norm": 0.46227601170539856,
      "learning_rate": 2.5866e-05,
      "loss": 0.0203,
      "step": 8625
    },
    {
      "epoch": 15.915974145891044,
      "grad_norm": 0.6137253642082214,
      "learning_rate": 2.5869e-05,
      "loss": 0.014,
      "step": 8626
    },
    {
      "epoch": 15.917820867959373,
      "grad_norm": 0.7873435616493225,
      "learning_rate": 2.5872000000000002e-05,
      "loss": 0.0196,
      "step": 8627
    },
    {
      "epoch": 15.9196675900277,
      "grad_norm": 1.0679621696472168,
      "learning_rate": 2.5875000000000002e-05,
      "loss": 0.0282,
      "step": 8628
    },
    {
      "epoch": 15.92151431209603,
      "grad_norm": 0.3143375813961029,
      "learning_rate": 2.5878000000000002e-05,
      "loss": 0.0101,
      "step": 8629
    },
    {
      "epoch": 15.923361034164358,
      "grad_norm": 1.0865501165390015,
      "learning_rate": 2.5881000000000002e-05,
      "loss": 0.0225,
      "step": 8630
    },
    {
      "epoch": 15.925207756232687,
      "grad_norm": 1.2255518436431885,
      "learning_rate": 2.5884e-05,
      "loss": 0.2034,
      "step": 8631
    },
    {
      "epoch": 15.927054478301017,
      "grad_norm": 0.8532062768936157,
      "learning_rate": 2.5887e-05,
      "loss": 0.1659,
      "step": 8632
    },
    {
      "epoch": 15.928901200369344,
      "grad_norm": 0.46837377548217773,
      "learning_rate": 2.589e-05,
      "loss": 0.1113,
      "step": 8633
    },
    {
      "epoch": 15.930747922437673,
      "grad_norm": 0.45477795600891113,
      "learning_rate": 2.5893e-05,
      "loss": 0.1057,
      "step": 8634
    },
    {
      "epoch": 15.932594644506002,
      "grad_norm": 0.715966522693634,
      "learning_rate": 2.5896e-05,
      "loss": 0.1335,
      "step": 8635
    },
    {
      "epoch": 15.93444136657433,
      "grad_norm": 0.5230908989906311,
      "learning_rate": 2.5899e-05,
      "loss": 0.0844,
      "step": 8636
    },
    {
      "epoch": 15.936288088642659,
      "grad_norm": 0.49790942668914795,
      "learning_rate": 2.5902e-05,
      "loss": 0.0792,
      "step": 8637
    },
    {
      "epoch": 15.938134810710988,
      "grad_norm": 0.6336291432380676,
      "learning_rate": 2.5905000000000004e-05,
      "loss": 0.0816,
      "step": 8638
    },
    {
      "epoch": 15.939981532779317,
      "grad_norm": 0.4619077444076538,
      "learning_rate": 2.5908000000000003e-05,
      "loss": 0.0695,
      "step": 8639
    },
    {
      "epoch": 15.941828254847646,
      "grad_norm": 0.4522983431816101,
      "learning_rate": 2.5911e-05,
      "loss": 0.0567,
      "step": 8640
    },
    {
      "epoch": 15.943674976915974,
      "grad_norm": 0.4374760687351227,
      "learning_rate": 2.5914e-05,
      "loss": 0.0483,
      "step": 8641
    },
    {
      "epoch": 15.945521698984303,
      "grad_norm": 0.36170950531959534,
      "learning_rate": 2.5917e-05,
      "loss": 0.0285,
      "step": 8642
    },
    {
      "epoch": 15.947368421052632,
      "grad_norm": 0.39567574858665466,
      "learning_rate": 2.592e-05,
      "loss": 0.0634,
      "step": 8643
    },
    {
      "epoch": 15.949215143120961,
      "grad_norm": 0.4324873089790344,
      "learning_rate": 2.5923e-05,
      "loss": 0.0238,
      "step": 8644
    },
    {
      "epoch": 15.951061865189288,
      "grad_norm": 0.4158211648464203,
      "learning_rate": 2.5926e-05,
      "loss": 0.0155,
      "step": 8645
    },
    {
      "epoch": 15.952908587257618,
      "grad_norm": 0.45500287413597107,
      "learning_rate": 2.5929e-05,
      "loss": 0.0335,
      "step": 8646
    },
    {
      "epoch": 15.954755309325947,
      "grad_norm": 0.3046945035457611,
      "learning_rate": 2.5932e-05,
      "loss": 0.0179,
      "step": 8647
    },
    {
      "epoch": 15.956602031394276,
      "grad_norm": 0.4032018780708313,
      "learning_rate": 2.5935e-05,
      "loss": 0.0238,
      "step": 8648
    },
    {
      "epoch": 15.958448753462603,
      "grad_norm": 0.3440537750720978,
      "learning_rate": 2.5938e-05,
      "loss": 0.0231,
      "step": 8649
    },
    {
      "epoch": 15.960295475530932,
      "grad_norm": 0.5092036724090576,
      "learning_rate": 2.5941e-05,
      "loss": 0.0209,
      "step": 8650
    },
    {
      "epoch": 15.962142197599261,
      "grad_norm": 0.3283311128616333,
      "learning_rate": 2.5944e-05,
      "loss": 0.0182,
      "step": 8651
    },
    {
      "epoch": 15.96398891966759,
      "grad_norm": 0.3121730387210846,
      "learning_rate": 2.5947e-05,
      "loss": 0.0148,
      "step": 8652
    },
    {
      "epoch": 15.965835641735918,
      "grad_norm": 0.9534308314323425,
      "learning_rate": 2.595e-05,
      "loss": 0.0242,
      "step": 8653
    },
    {
      "epoch": 15.967682363804247,
      "grad_norm": 0.47783422470092773,
      "learning_rate": 2.5953e-05,
      "loss": 0.0173,
      "step": 8654
    },
    {
      "epoch": 15.969529085872576,
      "grad_norm": 0.43015190958976746,
      "learning_rate": 2.5956e-05,
      "loss": 0.0154,
      "step": 8655
    },
    {
      "epoch": 15.971375807940905,
      "grad_norm": 0.25951239466667175,
      "learning_rate": 2.5959e-05,
      "loss": 0.0139,
      "step": 8656
    },
    {
      "epoch": 15.973222530009235,
      "grad_norm": 0.45808809995651245,
      "learning_rate": 2.5962e-05,
      "loss": 0.0122,
      "step": 8657
    },
    {
      "epoch": 15.975069252077562,
      "grad_norm": 0.4710130989551544,
      "learning_rate": 2.5965000000000003e-05,
      "loss": 0.0117,
      "step": 8658
    },
    {
      "epoch": 15.976915974145891,
      "grad_norm": 0.3743680715560913,
      "learning_rate": 2.5968000000000003e-05,
      "loss": 0.0205,
      "step": 8659
    },
    {
      "epoch": 15.97876269621422,
      "grad_norm": 0.24802172183990479,
      "learning_rate": 2.5971000000000003e-05,
      "loss": 0.0124,
      "step": 8660
    },
    {
      "epoch": 15.980609418282548,
      "grad_norm": 0.5540706515312195,
      "learning_rate": 2.5974000000000002e-05,
      "loss": 0.018,
      "step": 8661
    },
    {
      "epoch": 15.982456140350877,
      "grad_norm": 0.45025646686553955,
      "learning_rate": 2.5977000000000002e-05,
      "loss": 0.0117,
      "step": 8662
    },
    {
      "epoch": 15.984302862419206,
      "grad_norm": 0.5681506991386414,
      "learning_rate": 2.5980000000000002e-05,
      "loss": 0.0216,
      "step": 8663
    },
    {
      "epoch": 15.986149584487535,
      "grad_norm": 0.23212771117687225,
      "learning_rate": 2.5983000000000002e-05,
      "loss": 0.007,
      "step": 8664
    },
    {
      "epoch": 15.987996306555864,
      "grad_norm": 0.7229043841362,
      "learning_rate": 2.5985999999999998e-05,
      "loss": 0.021,
      "step": 8665
    },
    {
      "epoch": 15.989843028624191,
      "grad_norm": 0.3772416412830353,
      "learning_rate": 2.5988999999999998e-05,
      "loss": 0.0144,
      "step": 8666
    },
    {
      "epoch": 15.99168975069252,
      "grad_norm": 0.4555790424346924,
      "learning_rate": 2.5991999999999998e-05,
      "loss": 0.0222,
      "step": 8667
    },
    {
      "epoch": 15.99353647276085,
      "grad_norm": 0.5141981244087219,
      "learning_rate": 2.5995e-05,
      "loss": 0.0156,
      "step": 8668
    },
    {
      "epoch": 15.995383194829179,
      "grad_norm": 0.7101907730102539,
      "learning_rate": 2.5998e-05,
      "loss": 0.0207,
      "step": 8669
    },
    {
      "epoch": 15.997229916897506,
      "grad_norm": 0.2490791529417038,
      "learning_rate": 2.6001e-05,
      "loss": 0.0159,
      "step": 8670
    },
    {
      "epoch": 15.999076638965835,
      "grad_norm": 0.40662047266960144,
      "learning_rate": 2.6004e-05,
      "loss": 0.0144,
      "step": 8671
    },
    {
      "epoch": 16.0,
      "grad_norm": 0.4492111802101135,
      "learning_rate": 2.6007e-05,
      "loss": 0.0078,
      "step": 8672
    },
    {
      "epoch": 16.00184672206833,
      "grad_norm": 0.644540548324585,
      "learning_rate": 2.601e-05,
      "loss": 0.2018,
      "step": 8673
    },
    {
      "epoch": 16.00369344413666,
      "grad_norm": 0.6267561316490173,
      "learning_rate": 2.6013e-05,
      "loss": 0.1662,
      "step": 8674
    },
    {
      "epoch": 16.005540166204987,
      "grad_norm": 0.4369864761829376,
      "learning_rate": 2.6016e-05,
      "loss": 0.1051,
      "step": 8675
    },
    {
      "epoch": 16.007386888273317,
      "grad_norm": 0.6001611948013306,
      "learning_rate": 2.6019e-05,
      "loss": 0.119,
      "step": 8676
    },
    {
      "epoch": 16.009233610341642,
      "grad_norm": 0.46866223216056824,
      "learning_rate": 2.6022e-05,
      "loss": 0.0915,
      "step": 8677
    },
    {
      "epoch": 16.01108033240997,
      "grad_norm": 0.42183881998062134,
      "learning_rate": 2.6025000000000002e-05,
      "loss": 0.0853,
      "step": 8678
    },
    {
      "epoch": 16.0129270544783,
      "grad_norm": 0.6449214816093445,
      "learning_rate": 2.6028000000000002e-05,
      "loss": 0.1073,
      "step": 8679
    },
    {
      "epoch": 16.01477377654663,
      "grad_norm": 0.7335696816444397,
      "learning_rate": 2.6031000000000002e-05,
      "loss": 0.0924,
      "step": 8680
    },
    {
      "epoch": 16.01662049861496,
      "grad_norm": 0.7378032803535461,
      "learning_rate": 2.6034000000000002e-05,
      "loss": 0.1016,
      "step": 8681
    },
    {
      "epoch": 16.018467220683288,
      "grad_norm": 0.6013514399528503,
      "learning_rate": 2.6037e-05,
      "loss": 0.045,
      "step": 8682
    },
    {
      "epoch": 16.020313942751617,
      "grad_norm": 0.36441826820373535,
      "learning_rate": 2.604e-05,
      "loss": 0.0536,
      "step": 8683
    },
    {
      "epoch": 16.022160664819946,
      "grad_norm": 0.41581907868385315,
      "learning_rate": 2.6043e-05,
      "loss": 0.0506,
      "step": 8684
    },
    {
      "epoch": 16.02400738688827,
      "grad_norm": 0.29539382457733154,
      "learning_rate": 2.6046e-05,
      "loss": 0.0389,
      "step": 8685
    },
    {
      "epoch": 16.0258541089566,
      "grad_norm": 0.45296069979667664,
      "learning_rate": 2.6049e-05,
      "loss": 0.0536,
      "step": 8686
    },
    {
      "epoch": 16.02770083102493,
      "grad_norm": 0.5271998643875122,
      "learning_rate": 2.6052e-05,
      "loss": 0.0438,
      "step": 8687
    },
    {
      "epoch": 16.02954755309326,
      "grad_norm": 0.3643032908439636,
      "learning_rate": 2.6055000000000004e-05,
      "loss": 0.0438,
      "step": 8688
    },
    {
      "epoch": 16.03139427516159,
      "grad_norm": 0.3770063817501068,
      "learning_rate": 2.6058e-05,
      "loss": 0.0265,
      "step": 8689
    },
    {
      "epoch": 16.033240997229917,
      "grad_norm": 0.33927151560783386,
      "learning_rate": 2.6061e-05,
      "loss": 0.0203,
      "step": 8690
    },
    {
      "epoch": 16.035087719298247,
      "grad_norm": 0.34185677766799927,
      "learning_rate": 2.6064e-05,
      "loss": 0.0275,
      "step": 8691
    },
    {
      "epoch": 16.036934441366576,
      "grad_norm": 0.26765990257263184,
      "learning_rate": 2.6067e-05,
      "loss": 0.035,
      "step": 8692
    },
    {
      "epoch": 16.0387811634349,
      "grad_norm": 0.64670729637146,
      "learning_rate": 2.607e-05,
      "loss": 0.0165,
      "step": 8693
    },
    {
      "epoch": 16.04062788550323,
      "grad_norm": 0.32306674122810364,
      "learning_rate": 2.6073e-05,
      "loss": 0.0191,
      "step": 8694
    },
    {
      "epoch": 16.04247460757156,
      "grad_norm": 0.283189982175827,
      "learning_rate": 2.6076e-05,
      "loss": 0.0151,
      "step": 8695
    },
    {
      "epoch": 16.04432132963989,
      "grad_norm": 0.21966862678527832,
      "learning_rate": 2.6079e-05,
      "loss": 0.0103,
      "step": 8696
    },
    {
      "epoch": 16.046168051708218,
      "grad_norm": 0.30473917722702026,
      "learning_rate": 2.6082e-05,
      "loss": 0.0099,
      "step": 8697
    },
    {
      "epoch": 16.048014773776547,
      "grad_norm": 0.2754177451133728,
      "learning_rate": 2.6085000000000002e-05,
      "loss": 0.0066,
      "step": 8698
    },
    {
      "epoch": 16.049861495844876,
      "grad_norm": 0.2849455773830414,
      "learning_rate": 2.6088e-05,
      "loss": 0.0134,
      "step": 8699
    },
    {
      "epoch": 16.051708217913205,
      "grad_norm": 0.519327700138092,
      "learning_rate": 2.6091e-05,
      "loss": 0.0099,
      "step": 8700
    },
    {
      "epoch": 16.053554939981534,
      "grad_norm": 0.5305798649787903,
      "learning_rate": 2.6094e-05,
      "loss": 0.0102,
      "step": 8701
    },
    {
      "epoch": 16.05540166204986,
      "grad_norm": 0.7305449843406677,
      "learning_rate": 2.6097e-05,
      "loss": 0.0127,
      "step": 8702
    },
    {
      "epoch": 16.05724838411819,
      "grad_norm": 0.3311886787414551,
      "learning_rate": 2.61e-05,
      "loss": 0.0167,
      "step": 8703
    },
    {
      "epoch": 16.05909510618652,
      "grad_norm": 0.20189516246318817,
      "learning_rate": 2.6103e-05,
      "loss": 0.0069,
      "step": 8704
    },
    {
      "epoch": 16.060941828254848,
      "grad_norm": 0.2906367778778076,
      "learning_rate": 2.6106e-05,
      "loss": 0.0057,
      "step": 8705
    },
    {
      "epoch": 16.062788550323177,
      "grad_norm": 0.3915349543094635,
      "learning_rate": 2.6109e-05,
      "loss": 0.0093,
      "step": 8706
    },
    {
      "epoch": 16.064635272391506,
      "grad_norm": 0.3892596364021301,
      "learning_rate": 2.6112e-05,
      "loss": 0.0176,
      "step": 8707
    },
    {
      "epoch": 16.066481994459835,
      "grad_norm": 0.5112737417221069,
      "learning_rate": 2.6115000000000003e-05,
      "loss": 0.0191,
      "step": 8708
    },
    {
      "epoch": 16.068328716528164,
      "grad_norm": 0.5281062722206116,
      "learning_rate": 2.6118000000000003e-05,
      "loss": 0.0106,
      "step": 8709
    },
    {
      "epoch": 16.07017543859649,
      "grad_norm": 0.7477819919586182,
      "learning_rate": 2.6121000000000003e-05,
      "loss": 0.0182,
      "step": 8710
    },
    {
      "epoch": 16.07202216066482,
      "grad_norm": 0.43619078397750854,
      "learning_rate": 2.6124000000000003e-05,
      "loss": 0.0159,
      "step": 8711
    },
    {
      "epoch": 16.073868882733148,
      "grad_norm": 0.44052207469940186,
      "learning_rate": 2.6127000000000002e-05,
      "loss": 0.0111,
      "step": 8712
    },
    {
      "epoch": 16.075715604801477,
      "grad_norm": 0.3810107111930847,
      "learning_rate": 2.6130000000000002e-05,
      "loss": 0.0176,
      "step": 8713
    },
    {
      "epoch": 16.077562326869806,
      "grad_norm": 0.2601052224636078,
      "learning_rate": 2.6133e-05,
      "loss": 0.0149,
      "step": 8714
    },
    {
      "epoch": 16.079409048938135,
      "grad_norm": 0.40293899178504944,
      "learning_rate": 2.6136e-05,
      "loss": 0.0082,
      "step": 8715
    },
    {
      "epoch": 16.081255771006465,
      "grad_norm": 0.4989589750766754,
      "learning_rate": 2.6138999999999998e-05,
      "loss": 0.0122,
      "step": 8716
    },
    {
      "epoch": 16.083102493074794,
      "grad_norm": 0.9980294108390808,
      "learning_rate": 2.6141999999999998e-05,
      "loss": 0.0185,
      "step": 8717
    },
    {
      "epoch": 16.08494921514312,
      "grad_norm": 0.363986611366272,
      "learning_rate": 2.6145e-05,
      "loss": 0.0106,
      "step": 8718
    },
    {
      "epoch": 16.08679593721145,
      "grad_norm": 0.21164149045944214,
      "learning_rate": 2.6148e-05,
      "loss": 0.0043,
      "step": 8719
    },
    {
      "epoch": 16.088642659279778,
      "grad_norm": 0.41821521520614624,
      "learning_rate": 2.6151e-05,
      "loss": 0.0127,
      "step": 8720
    },
    {
      "epoch": 16.090489381348107,
      "grad_norm": 0.3259578049182892,
      "learning_rate": 2.6154e-05,
      "loss": 0.0104,
      "step": 8721
    },
    {
      "epoch": 16.092336103416436,
      "grad_norm": 0.6169150471687317,
      "learning_rate": 2.6157e-05,
      "loss": 0.0095,
      "step": 8722
    },
    {
      "epoch": 16.094182825484765,
      "grad_norm": 0.8425424695014954,
      "learning_rate": 2.616e-05,
      "loss": 0.2042,
      "step": 8723
    },
    {
      "epoch": 16.096029547553094,
      "grad_norm": 0.7024513483047485,
      "learning_rate": 2.6163e-05,
      "loss": 0.1881,
      "step": 8724
    },
    {
      "epoch": 16.097876269621423,
      "grad_norm": 0.7537822723388672,
      "learning_rate": 2.6166e-05,
      "loss": 0.1639,
      "step": 8725
    },
    {
      "epoch": 16.099722991689752,
      "grad_norm": 0.47779718041419983,
      "learning_rate": 2.6169e-05,
      "loss": 0.1255,
      "step": 8726
    },
    {
      "epoch": 16.101569713758078,
      "grad_norm": 0.6427140235900879,
      "learning_rate": 2.6172e-05,
      "loss": 0.1821,
      "step": 8727
    },
    {
      "epoch": 16.103416435826407,
      "grad_norm": 0.7359801530838013,
      "learning_rate": 2.6175000000000003e-05,
      "loss": 0.0879,
      "step": 8728
    },
    {
      "epoch": 16.105263157894736,
      "grad_norm": 0.5765436887741089,
      "learning_rate": 2.6178000000000002e-05,
      "loss": 0.0853,
      "step": 8729
    },
    {
      "epoch": 16.107109879963065,
      "grad_norm": 0.7798746824264526,
      "learning_rate": 2.6181000000000002e-05,
      "loss": 0.0957,
      "step": 8730
    },
    {
      "epoch": 16.108956602031395,
      "grad_norm": 0.49390435218811035,
      "learning_rate": 2.6184000000000002e-05,
      "loss": 0.081,
      "step": 8731
    },
    {
      "epoch": 16.110803324099724,
      "grad_norm": 0.403861403465271,
      "learning_rate": 2.6187000000000002e-05,
      "loss": 0.0444,
      "step": 8732
    },
    {
      "epoch": 16.112650046168053,
      "grad_norm": 0.43862980604171753,
      "learning_rate": 2.619e-05,
      "loss": 0.0694,
      "step": 8733
    },
    {
      "epoch": 16.114496768236382,
      "grad_norm": 0.6070998311042786,
      "learning_rate": 2.6193e-05,
      "loss": 0.049,
      "step": 8734
    },
    {
      "epoch": 16.116343490304708,
      "grad_norm": 0.32739898562431335,
      "learning_rate": 2.6196e-05,
      "loss": 0.0496,
      "step": 8735
    },
    {
      "epoch": 16.118190212373037,
      "grad_norm": 0.46069347858428955,
      "learning_rate": 2.6199e-05,
      "loss": 0.0236,
      "step": 8736
    },
    {
      "epoch": 16.120036934441366,
      "grad_norm": 0.3739839196205139,
      "learning_rate": 2.6202e-05,
      "loss": 0.0266,
      "step": 8737
    },
    {
      "epoch": 16.121883656509695,
      "grad_norm": 0.27839231491088867,
      "learning_rate": 2.6205e-05,
      "loss": 0.0486,
      "step": 8738
    },
    {
      "epoch": 16.123730378578024,
      "grad_norm": 0.3901972472667694,
      "learning_rate": 2.6208e-05,
      "loss": 0.0198,
      "step": 8739
    },
    {
      "epoch": 16.125577100646353,
      "grad_norm": 0.25915515422821045,
      "learning_rate": 2.6211e-05,
      "loss": 0.0236,
      "step": 8740
    },
    {
      "epoch": 16.127423822714682,
      "grad_norm": 0.29264405369758606,
      "learning_rate": 2.6214e-05,
      "loss": 0.0165,
      "step": 8741
    },
    {
      "epoch": 16.12927054478301,
      "grad_norm": 0.2875673472881317,
      "learning_rate": 2.6217e-05,
      "loss": 0.0152,
      "step": 8742
    },
    {
      "epoch": 16.131117266851337,
      "grad_norm": 0.38900938630104065,
      "learning_rate": 2.622e-05,
      "loss": 0.0146,
      "step": 8743
    },
    {
      "epoch": 16.132963988919666,
      "grad_norm": 0.23090846836566925,
      "learning_rate": 2.6223e-05,
      "loss": 0.024,
      "step": 8744
    },
    {
      "epoch": 16.134810710987995,
      "grad_norm": 0.34287866950035095,
      "learning_rate": 2.6226e-05,
      "loss": 0.0101,
      "step": 8745
    },
    {
      "epoch": 16.136657433056325,
      "grad_norm": 0.398605078458786,
      "learning_rate": 2.6229e-05,
      "loss": 0.0152,
      "step": 8746
    },
    {
      "epoch": 16.138504155124654,
      "grad_norm": 0.4421793520450592,
      "learning_rate": 2.6232e-05,
      "loss": 0.011,
      "step": 8747
    },
    {
      "epoch": 16.140350877192983,
      "grad_norm": 0.42159032821655273,
      "learning_rate": 2.6235000000000002e-05,
      "loss": 0.0113,
      "step": 8748
    },
    {
      "epoch": 16.142197599261312,
      "grad_norm": 0.5255663990974426,
      "learning_rate": 2.6238000000000002e-05,
      "loss": 0.0111,
      "step": 8749
    },
    {
      "epoch": 16.14404432132964,
      "grad_norm": 0.593675971031189,
      "learning_rate": 2.6241e-05,
      "loss": 0.0276,
      "step": 8750
    },
    {
      "epoch": 16.14589104339797,
      "grad_norm": 0.4048978090286255,
      "learning_rate": 2.6244e-05,
      "loss": 0.0132,
      "step": 8751
    },
    {
      "epoch": 16.147737765466296,
      "grad_norm": 0.252643883228302,
      "learning_rate": 2.6247e-05,
      "loss": 0.0096,
      "step": 8752
    },
    {
      "epoch": 16.149584487534625,
      "grad_norm": 0.5537533760070801,
      "learning_rate": 2.625e-05,
      "loss": 0.0143,
      "step": 8753
    },
    {
      "epoch": 16.151431209602954,
      "grad_norm": 0.4042913019657135,
      "learning_rate": 2.6253e-05,
      "loss": 0.0175,
      "step": 8754
    },
    {
      "epoch": 16.153277931671283,
      "grad_norm": 0.2454013079404831,
      "learning_rate": 2.6256e-05,
      "loss": 0.0095,
      "step": 8755
    },
    {
      "epoch": 16.155124653739612,
      "grad_norm": 0.3569791913032532,
      "learning_rate": 2.6259e-05,
      "loss": 0.0148,
      "step": 8756
    },
    {
      "epoch": 16.15697137580794,
      "grad_norm": 0.2660028040409088,
      "learning_rate": 2.6262e-05,
      "loss": 0.0113,
      "step": 8757
    },
    {
      "epoch": 16.15881809787627,
      "grad_norm": 0.2081262171268463,
      "learning_rate": 2.6265e-05,
      "loss": 0.0132,
      "step": 8758
    },
    {
      "epoch": 16.1606648199446,
      "grad_norm": 0.42625632882118225,
      "learning_rate": 2.6268000000000003e-05,
      "loss": 0.0068,
      "step": 8759
    },
    {
      "epoch": 16.162511542012926,
      "grad_norm": 0.20090582966804504,
      "learning_rate": 2.6271000000000003e-05,
      "loss": 0.0067,
      "step": 8760
    },
    {
      "epoch": 16.164358264081255,
      "grad_norm": 0.2835046947002411,
      "learning_rate": 2.6274000000000003e-05,
      "loss": 0.0099,
      "step": 8761
    },
    {
      "epoch": 16.166204986149584,
      "grad_norm": 0.8398069739341736,
      "learning_rate": 2.6277000000000003e-05,
      "loss": 0.0142,
      "step": 8762
    },
    {
      "epoch": 16.168051708217913,
      "grad_norm": 0.25367045402526855,
      "learning_rate": 2.628e-05,
      "loss": 0.0073,
      "step": 8763
    },
    {
      "epoch": 16.169898430286242,
      "grad_norm": 0.30854055285453796,
      "learning_rate": 2.6283e-05,
      "loss": 0.0086,
      "step": 8764
    },
    {
      "epoch": 16.17174515235457,
      "grad_norm": 0.5980221033096313,
      "learning_rate": 2.6286e-05,
      "loss": 0.0124,
      "step": 8765
    },
    {
      "epoch": 16.1735918744229,
      "grad_norm": 0.38559502363204956,
      "learning_rate": 2.6289e-05,
      "loss": 0.0119,
      "step": 8766
    },
    {
      "epoch": 16.17543859649123,
      "grad_norm": 0.3606363534927368,
      "learning_rate": 2.6292e-05,
      "loss": 0.0109,
      "step": 8767
    },
    {
      "epoch": 16.177285318559555,
      "grad_norm": 0.4450051486492157,
      "learning_rate": 2.6294999999999998e-05,
      "loss": 0.0133,
      "step": 8768
    },
    {
      "epoch": 16.179132040627884,
      "grad_norm": 0.412380576133728,
      "learning_rate": 2.6298e-05,
      "loss": 0.0103,
      "step": 8769
    },
    {
      "epoch": 16.180978762696213,
      "grad_norm": 0.5443732738494873,
      "learning_rate": 2.6301e-05,
      "loss": 0.012,
      "step": 8770
    },
    {
      "epoch": 16.182825484764543,
      "grad_norm": 0.8527624011039734,
      "learning_rate": 2.6304e-05,
      "loss": 0.0083,
      "step": 8771
    },
    {
      "epoch": 16.18467220683287,
      "grad_norm": 0.6248682737350464,
      "learning_rate": 2.6307e-05,
      "loss": 0.016,
      "step": 8772
    },
    {
      "epoch": 16.1865189289012,
      "grad_norm": 0.609694242477417,
      "learning_rate": 2.631e-05,
      "loss": 0.2052,
      "step": 8773
    },
    {
      "epoch": 16.18836565096953,
      "grad_norm": 1.0320067405700684,
      "learning_rate": 2.6313e-05,
      "loss": 0.2287,
      "step": 8774
    },
    {
      "epoch": 16.19021237303786,
      "grad_norm": 0.5504393577575684,
      "learning_rate": 2.6316e-05,
      "loss": 0.1232,
      "step": 8775
    },
    {
      "epoch": 16.19205909510619,
      "grad_norm": 0.5152016878128052,
      "learning_rate": 2.6319e-05,
      "loss": 0.1213,
      "step": 8776
    },
    {
      "epoch": 16.193905817174514,
      "grad_norm": 0.5570312738418579,
      "learning_rate": 2.6322e-05,
      "loss": 0.1493,
      "step": 8777
    },
    {
      "epoch": 16.195752539242843,
      "grad_norm": 0.4364548623561859,
      "learning_rate": 2.6325e-05,
      "loss": 0.1012,
      "step": 8778
    },
    {
      "epoch": 16.197599261311172,
      "grad_norm": 0.4299827218055725,
      "learning_rate": 2.6328000000000003e-05,
      "loss": 0.1013,
      "step": 8779
    },
    {
      "epoch": 16.1994459833795,
      "grad_norm": 0.4067482054233551,
      "learning_rate": 2.6331000000000003e-05,
      "loss": 0.0621,
      "step": 8780
    },
    {
      "epoch": 16.20129270544783,
      "grad_norm": 0.5066156387329102,
      "learning_rate": 2.6334000000000002e-05,
      "loss": 0.0927,
      "step": 8781
    },
    {
      "epoch": 16.20313942751616,
      "grad_norm": 0.5903967618942261,
      "learning_rate": 2.6337000000000002e-05,
      "loss": 0.1129,
      "step": 8782
    },
    {
      "epoch": 16.20498614958449,
      "grad_norm": 0.3994077146053314,
      "learning_rate": 2.6340000000000002e-05,
      "loss": 0.0623,
      "step": 8783
    },
    {
      "epoch": 16.206832871652818,
      "grad_norm": 0.3687579333782196,
      "learning_rate": 2.6343000000000002e-05,
      "loss": 0.0591,
      "step": 8784
    },
    {
      "epoch": 16.208679593721143,
      "grad_norm": 0.4481465816497803,
      "learning_rate": 2.6346e-05,
      "loss": 0.0368,
      "step": 8785
    },
    {
      "epoch": 16.210526315789473,
      "grad_norm": 0.5493351221084595,
      "learning_rate": 2.6349e-05,
      "loss": 0.0353,
      "step": 8786
    },
    {
      "epoch": 16.2123730378578,
      "grad_norm": 0.47865021228790283,
      "learning_rate": 2.6351999999999998e-05,
      "loss": 0.0393,
      "step": 8787
    },
    {
      "epoch": 16.21421975992613,
      "grad_norm": 0.6238217353820801,
      "learning_rate": 2.6354999999999998e-05,
      "loss": 0.0197,
      "step": 8788
    },
    {
      "epoch": 16.21606648199446,
      "grad_norm": 0.4697655141353607,
      "learning_rate": 2.6358e-05,
      "loss": 0.0289,
      "step": 8789
    },
    {
      "epoch": 16.21791320406279,
      "grad_norm": 0.573111355304718,
      "learning_rate": 2.6361e-05,
      "loss": 0.0179,
      "step": 8790
    },
    {
      "epoch": 16.21975992613112,
      "grad_norm": 0.20533904433250427,
      "learning_rate": 2.6364e-05,
      "loss": 0.0134,
      "step": 8791
    },
    {
      "epoch": 16.221606648199447,
      "grad_norm": 0.37425920367240906,
      "learning_rate": 2.6367e-05,
      "loss": 0.0392,
      "step": 8792
    },
    {
      "epoch": 16.223453370267773,
      "grad_norm": 0.3609694540500641,
      "learning_rate": 2.637e-05,
      "loss": 0.0111,
      "step": 8793
    },
    {
      "epoch": 16.225300092336102,
      "grad_norm": 0.2295820564031601,
      "learning_rate": 2.6373e-05,
      "loss": 0.0086,
      "step": 8794
    },
    {
      "epoch": 16.22714681440443,
      "grad_norm": 0.43150201439857483,
      "learning_rate": 2.6376e-05,
      "loss": 0.015,
      "step": 8795
    },
    {
      "epoch": 16.22899353647276,
      "grad_norm": 0.2645535469055176,
      "learning_rate": 2.6379e-05,
      "loss": 0.0154,
      "step": 8796
    },
    {
      "epoch": 16.23084025854109,
      "grad_norm": 0.26752784848213196,
      "learning_rate": 2.6382e-05,
      "loss": 0.0157,
      "step": 8797
    },
    {
      "epoch": 16.23268698060942,
      "grad_norm": 0.34007754921913147,
      "learning_rate": 2.6385e-05,
      "loss": 0.0194,
      "step": 8798
    },
    {
      "epoch": 16.234533702677748,
      "grad_norm": 0.29169556498527527,
      "learning_rate": 2.6388000000000002e-05,
      "loss": 0.0145,
      "step": 8799
    },
    {
      "epoch": 16.236380424746077,
      "grad_norm": 0.234590083360672,
      "learning_rate": 2.6391000000000002e-05,
      "loss": 0.0115,
      "step": 8800
    },
    {
      "epoch": 16.238227146814406,
      "grad_norm": 0.684226930141449,
      "learning_rate": 2.6394000000000002e-05,
      "loss": 0.0135,
      "step": 8801
    },
    {
      "epoch": 16.24007386888273,
      "grad_norm": 0.3227801024913788,
      "learning_rate": 2.6397e-05,
      "loss": 0.0207,
      "step": 8802
    },
    {
      "epoch": 16.24192059095106,
      "grad_norm": 0.930953323841095,
      "learning_rate": 2.64e-05,
      "loss": 0.0195,
      "step": 8803
    },
    {
      "epoch": 16.24376731301939,
      "grad_norm": 0.6033096313476562,
      "learning_rate": 2.6403e-05,
      "loss": 0.0184,
      "step": 8804
    },
    {
      "epoch": 16.24561403508772,
      "grad_norm": 0.3899700343608856,
      "learning_rate": 2.6406e-05,
      "loss": 0.0099,
      "step": 8805
    },
    {
      "epoch": 16.24746075715605,
      "grad_norm": 0.3385612666606903,
      "learning_rate": 2.6409e-05,
      "loss": 0.0079,
      "step": 8806
    },
    {
      "epoch": 16.249307479224377,
      "grad_norm": 0.7180120348930359,
      "learning_rate": 2.6412e-05,
      "loss": 0.0113,
      "step": 8807
    },
    {
      "epoch": 16.251154201292707,
      "grad_norm": 0.42312514781951904,
      "learning_rate": 2.6415e-05,
      "loss": 0.0125,
      "step": 8808
    },
    {
      "epoch": 16.253000923361036,
      "grad_norm": 0.291399747133255,
      "learning_rate": 2.6418000000000004e-05,
      "loss": 0.0124,
      "step": 8809
    },
    {
      "epoch": 16.25484764542936,
      "grad_norm": 0.33661431074142456,
      "learning_rate": 2.6421000000000003e-05,
      "loss": 0.0105,
      "step": 8810
    },
    {
      "epoch": 16.25669436749769,
      "grad_norm": 0.24821096658706665,
      "learning_rate": 2.6424000000000003e-05,
      "loss": 0.0066,
      "step": 8811
    },
    {
      "epoch": 16.25854108956602,
      "grad_norm": 0.3174000680446625,
      "learning_rate": 2.6427e-05,
      "loss": 0.0142,
      "step": 8812
    },
    {
      "epoch": 16.26038781163435,
      "grad_norm": 0.5178481340408325,
      "learning_rate": 2.643e-05,
      "loss": 0.0081,
      "step": 8813
    },
    {
      "epoch": 16.262234533702678,
      "grad_norm": 0.6077439188957214,
      "learning_rate": 2.6433e-05,
      "loss": 0.0206,
      "step": 8814
    },
    {
      "epoch": 16.264081255771007,
      "grad_norm": 0.33663487434387207,
      "learning_rate": 2.6436e-05,
      "loss": 0.0069,
      "step": 8815
    },
    {
      "epoch": 16.265927977839336,
      "grad_norm": 0.5525741577148438,
      "learning_rate": 2.6439e-05,
      "loss": 0.0187,
      "step": 8816
    },
    {
      "epoch": 16.267774699907665,
      "grad_norm": 0.4057387709617615,
      "learning_rate": 2.6442e-05,
      "loss": 0.0064,
      "step": 8817
    },
    {
      "epoch": 16.26962142197599,
      "grad_norm": 0.36393001675605774,
      "learning_rate": 2.6445e-05,
      "loss": 0.0158,
      "step": 8818
    },
    {
      "epoch": 16.27146814404432,
      "grad_norm": 0.3587076663970947,
      "learning_rate": 2.6448e-05,
      "loss": 0.0094,
      "step": 8819
    },
    {
      "epoch": 16.27331486611265,
      "grad_norm": 0.24401932954788208,
      "learning_rate": 2.6451e-05,
      "loss": 0.011,
      "step": 8820
    },
    {
      "epoch": 16.27516158818098,
      "grad_norm": 0.5385698676109314,
      "learning_rate": 2.6454e-05,
      "loss": 0.017,
      "step": 8821
    },
    {
      "epoch": 16.277008310249307,
      "grad_norm": 0.22548331320285797,
      "learning_rate": 2.6457e-05,
      "loss": 0.0077,
      "step": 8822
    },
    {
      "epoch": 16.278855032317637,
      "grad_norm": 0.8914181590080261,
      "learning_rate": 2.646e-05,
      "loss": 0.2099,
      "step": 8823
    },
    {
      "epoch": 16.280701754385966,
      "grad_norm": 0.6201906204223633,
      "learning_rate": 2.6463e-05,
      "loss": 0.1274,
      "step": 8824
    },
    {
      "epoch": 16.282548476454295,
      "grad_norm": 0.7464452385902405,
      "learning_rate": 2.6466e-05,
      "loss": 0.1219,
      "step": 8825
    },
    {
      "epoch": 16.284395198522624,
      "grad_norm": 0.5396767854690552,
      "learning_rate": 2.6469e-05,
      "loss": 0.1038,
      "step": 8826
    },
    {
      "epoch": 16.28624192059095,
      "grad_norm": 0.7753463983535767,
      "learning_rate": 2.6472e-05,
      "loss": 0.1185,
      "step": 8827
    },
    {
      "epoch": 16.28808864265928,
      "grad_norm": 0.47809740900993347,
      "learning_rate": 2.6475e-05,
      "loss": 0.0847,
      "step": 8828
    },
    {
      "epoch": 16.289935364727608,
      "grad_norm": 0.614112913608551,
      "learning_rate": 2.6478000000000003e-05,
      "loss": 0.1215,
      "step": 8829
    },
    {
      "epoch": 16.291782086795937,
      "grad_norm": 0.6368290781974792,
      "learning_rate": 2.6481000000000003e-05,
      "loss": 0.0517,
      "step": 8830
    },
    {
      "epoch": 16.293628808864266,
      "grad_norm": 0.6140844225883484,
      "learning_rate": 2.6484000000000003e-05,
      "loss": 0.0946,
      "step": 8831
    },
    {
      "epoch": 16.295475530932595,
      "grad_norm": 0.5148943662643433,
      "learning_rate": 2.6487000000000002e-05,
      "loss": 0.09,
      "step": 8832
    },
    {
      "epoch": 16.297322253000925,
      "grad_norm": 0.5340912342071533,
      "learning_rate": 2.6490000000000002e-05,
      "loss": 0.0596,
      "step": 8833
    },
    {
      "epoch": 16.299168975069254,
      "grad_norm": 0.43576526641845703,
      "learning_rate": 2.6493000000000002e-05,
      "loss": 0.0451,
      "step": 8834
    },
    {
      "epoch": 16.30101569713758,
      "grad_norm": 0.5544477105140686,
      "learning_rate": 2.6496000000000002e-05,
      "loss": 0.0455,
      "step": 8835
    },
    {
      "epoch": 16.30286241920591,
      "grad_norm": 0.4253067672252655,
      "learning_rate": 2.6499e-05,
      "loss": 0.0667,
      "step": 8836
    },
    {
      "epoch": 16.304709141274238,
      "grad_norm": 0.36090901494026184,
      "learning_rate": 2.6501999999999998e-05,
      "loss": 0.0404,
      "step": 8837
    },
    {
      "epoch": 16.306555863342567,
      "grad_norm": 0.39279404282569885,
      "learning_rate": 2.6504999999999998e-05,
      "loss": 0.0379,
      "step": 8838
    },
    {
      "epoch": 16.308402585410896,
      "grad_norm": 0.3534673750400543,
      "learning_rate": 2.6508e-05,
      "loss": 0.0238,
      "step": 8839
    },
    {
      "epoch": 16.310249307479225,
      "grad_norm": 0.25616320967674255,
      "learning_rate": 2.6511e-05,
      "loss": 0.0186,
      "step": 8840
    },
    {
      "epoch": 16.312096029547554,
      "grad_norm": 0.16772395372390747,
      "learning_rate": 2.6514e-05,
      "loss": 0.0112,
      "step": 8841
    },
    {
      "epoch": 16.313942751615883,
      "grad_norm": 0.30720284581184387,
      "learning_rate": 2.6517e-05,
      "loss": 0.0131,
      "step": 8842
    },
    {
      "epoch": 16.31578947368421,
      "grad_norm": 0.25934576988220215,
      "learning_rate": 2.652e-05,
      "loss": 0.0222,
      "step": 8843
    },
    {
      "epoch": 16.317636195752538,
      "grad_norm": 0.4097827076911926,
      "learning_rate": 2.6523e-05,
      "loss": 0.0266,
      "step": 8844
    },
    {
      "epoch": 16.319482917820867,
      "grad_norm": 0.4845149517059326,
      "learning_rate": 2.6526e-05,
      "loss": 0.0156,
      "step": 8845
    },
    {
      "epoch": 16.321329639889196,
      "grad_norm": 0.2271205484867096,
      "learning_rate": 2.6529e-05,
      "loss": 0.0152,
      "step": 8846
    },
    {
      "epoch": 16.323176361957525,
      "grad_norm": 0.2507508397102356,
      "learning_rate": 2.6532e-05,
      "loss": 0.011,
      "step": 8847
    },
    {
      "epoch": 16.325023084025855,
      "grad_norm": 0.22607016563415527,
      "learning_rate": 2.6535e-05,
      "loss": 0.0114,
      "step": 8848
    },
    {
      "epoch": 16.326869806094184,
      "grad_norm": 0.2688290774822235,
      "learning_rate": 2.6538000000000002e-05,
      "loss": 0.0227,
      "step": 8849
    },
    {
      "epoch": 16.328716528162513,
      "grad_norm": 0.39142295718193054,
      "learning_rate": 2.6541000000000002e-05,
      "loss": 0.0137,
      "step": 8850
    },
    {
      "epoch": 16.330563250230842,
      "grad_norm": 0.20058947801589966,
      "learning_rate": 2.6544000000000002e-05,
      "loss": 0.0078,
      "step": 8851
    },
    {
      "epoch": 16.332409972299168,
      "grad_norm": 0.1827445924282074,
      "learning_rate": 2.6547000000000002e-05,
      "loss": 0.0075,
      "step": 8852
    },
    {
      "epoch": 16.334256694367497,
      "grad_norm": 0.2042992264032364,
      "learning_rate": 2.655e-05,
      "loss": 0.0049,
      "step": 8853
    },
    {
      "epoch": 16.336103416435826,
      "grad_norm": 0.7434996962547302,
      "learning_rate": 2.6553e-05,
      "loss": 0.0413,
      "step": 8854
    },
    {
      "epoch": 16.337950138504155,
      "grad_norm": 0.274636447429657,
      "learning_rate": 2.6556e-05,
      "loss": 0.0115,
      "step": 8855
    },
    {
      "epoch": 16.339796860572484,
      "grad_norm": 0.31662482023239136,
      "learning_rate": 2.6559e-05,
      "loss": 0.0116,
      "step": 8856
    },
    {
      "epoch": 16.341643582640813,
      "grad_norm": 0.3444797992706299,
      "learning_rate": 2.6562e-05,
      "loss": 0.0123,
      "step": 8857
    },
    {
      "epoch": 16.343490304709142,
      "grad_norm": 0.2661111652851105,
      "learning_rate": 2.6565e-05,
      "loss": 0.0088,
      "step": 8858
    },
    {
      "epoch": 16.34533702677747,
      "grad_norm": 0.6574168801307678,
      "learning_rate": 2.6568000000000004e-05,
      "loss": 0.0146,
      "step": 8859
    },
    {
      "epoch": 16.347183748845797,
      "grad_norm": 0.3464463949203491,
      "learning_rate": 2.6571000000000004e-05,
      "loss": 0.0088,
      "step": 8860
    },
    {
      "epoch": 16.349030470914126,
      "grad_norm": 0.35741615295410156,
      "learning_rate": 2.6574e-05,
      "loss": 0.0107,
      "step": 8861
    },
    {
      "epoch": 16.350877192982455,
      "grad_norm": 0.4415765702724457,
      "learning_rate": 2.6577e-05,
      "loss": 0.0141,
      "step": 8862
    },
    {
      "epoch": 16.352723915050785,
      "grad_norm": 0.23338951170444489,
      "learning_rate": 2.658e-05,
      "loss": 0.0076,
      "step": 8863
    },
    {
      "epoch": 16.354570637119114,
      "grad_norm": 0.3000146448612213,
      "learning_rate": 2.6583e-05,
      "loss": 0.0111,
      "step": 8864
    },
    {
      "epoch": 16.356417359187443,
      "grad_norm": 0.29810795187950134,
      "learning_rate": 2.6586e-05,
      "loss": 0.0096,
      "step": 8865
    },
    {
      "epoch": 16.358264081255772,
      "grad_norm": 0.21178984642028809,
      "learning_rate": 2.6589e-05,
      "loss": 0.0066,
      "step": 8866
    },
    {
      "epoch": 16.3601108033241,
      "grad_norm": 1.2656605243682861,
      "learning_rate": 2.6592e-05,
      "loss": 0.0226,
      "step": 8867
    },
    {
      "epoch": 16.361957525392427,
      "grad_norm": 0.2447841763496399,
      "learning_rate": 2.6595e-05,
      "loss": 0.01,
      "step": 8868
    },
    {
      "epoch": 16.363804247460756,
      "grad_norm": 0.34307822585105896,
      "learning_rate": 2.6598000000000002e-05,
      "loss": 0.0142,
      "step": 8869
    },
    {
      "epoch": 16.365650969529085,
      "grad_norm": 0.7280043363571167,
      "learning_rate": 2.6601e-05,
      "loss": 0.0089,
      "step": 8870
    },
    {
      "epoch": 16.367497691597414,
      "grad_norm": 0.504188060760498,
      "learning_rate": 2.6604e-05,
      "loss": 0.0217,
      "step": 8871
    },
    {
      "epoch": 16.369344413665743,
      "grad_norm": 0.6764096617698669,
      "learning_rate": 2.6607e-05,
      "loss": 0.0127,
      "step": 8872
    },
    {
      "epoch": 16.371191135734072,
      "grad_norm": 0.5507748126983643,
      "learning_rate": 2.661e-05,
      "loss": 0.1713,
      "step": 8873
    },
    {
      "epoch": 16.3730378578024,
      "grad_norm": 1.3963656425476074,
      "learning_rate": 2.6613e-05,
      "loss": 0.1514,
      "step": 8874
    },
    {
      "epoch": 16.37488457987073,
      "grad_norm": 0.575821042060852,
      "learning_rate": 2.6616e-05,
      "loss": 0.1285,
      "step": 8875
    },
    {
      "epoch": 16.37673130193906,
      "grad_norm": 0.543881893157959,
      "learning_rate": 2.6619e-05,
      "loss": 0.1011,
      "step": 8876
    },
    {
      "epoch": 16.378578024007385,
      "grad_norm": 0.5799001455307007,
      "learning_rate": 2.6622e-05,
      "loss": 0.143,
      "step": 8877
    },
    {
      "epoch": 16.380424746075715,
      "grad_norm": 0.5811880230903625,
      "learning_rate": 2.6625e-05,
      "loss": 0.0985,
      "step": 8878
    },
    {
      "epoch": 16.382271468144044,
      "grad_norm": 0.6220884919166565,
      "learning_rate": 2.6628e-05,
      "loss": 0.0882,
      "step": 8879
    },
    {
      "epoch": 16.384118190212373,
      "grad_norm": 0.8949848413467407,
      "learning_rate": 2.6631000000000003e-05,
      "loss": 0.0702,
      "step": 8880
    },
    {
      "epoch": 16.385964912280702,
      "grad_norm": 0.5241614580154419,
      "learning_rate": 2.6634000000000003e-05,
      "loss": 0.0712,
      "step": 8881
    },
    {
      "epoch": 16.38781163434903,
      "grad_norm": 0.703456461429596,
      "learning_rate": 2.6637000000000003e-05,
      "loss": 0.0976,
      "step": 8882
    },
    {
      "epoch": 16.38965835641736,
      "grad_norm": 0.3968183398246765,
      "learning_rate": 2.6640000000000002e-05,
      "loss": 0.0355,
      "step": 8883
    },
    {
      "epoch": 16.39150507848569,
      "grad_norm": 0.6287881135940552,
      "learning_rate": 2.6643000000000002e-05,
      "loss": 0.0638,
      "step": 8884
    },
    {
      "epoch": 16.393351800554015,
      "grad_norm": 0.44642600417137146,
      "learning_rate": 2.6646000000000002e-05,
      "loss": 0.0287,
      "step": 8885
    },
    {
      "epoch": 16.395198522622344,
      "grad_norm": 0.7361428737640381,
      "learning_rate": 2.6649e-05,
      "loss": 0.0444,
      "step": 8886
    },
    {
      "epoch": 16.397045244690673,
      "grad_norm": 0.3432881832122803,
      "learning_rate": 2.6651999999999998e-05,
      "loss": 0.0413,
      "step": 8887
    },
    {
      "epoch": 16.398891966759003,
      "grad_norm": 0.38371410965919495,
      "learning_rate": 2.6654999999999998e-05,
      "loss": 0.0439,
      "step": 8888
    },
    {
      "epoch": 16.40073868882733,
      "grad_norm": 0.45330309867858887,
      "learning_rate": 2.6657999999999998e-05,
      "loss": 0.0178,
      "step": 8889
    },
    {
      "epoch": 16.40258541089566,
      "grad_norm": 0.33713167905807495,
      "learning_rate": 2.6661e-05,
      "loss": 0.0224,
      "step": 8890
    },
    {
      "epoch": 16.40443213296399,
      "grad_norm": 0.2396007478237152,
      "learning_rate": 2.6664e-05,
      "loss": 0.0105,
      "step": 8891
    },
    {
      "epoch": 16.40627885503232,
      "grad_norm": 0.2867899537086487,
      "learning_rate": 2.6667e-05,
      "loss": 0.037,
      "step": 8892
    },
    {
      "epoch": 16.408125577100645,
      "grad_norm": 0.32196927070617676,
      "learning_rate": 2.667e-05,
      "loss": 0.0145,
      "step": 8893
    },
    {
      "epoch": 16.409972299168974,
      "grad_norm": 0.38088223338127136,
      "learning_rate": 2.6673e-05,
      "loss": 0.0302,
      "step": 8894
    },
    {
      "epoch": 16.411819021237303,
      "grad_norm": 0.22422203421592712,
      "learning_rate": 2.6676e-05,
      "loss": 0.0101,
      "step": 8895
    },
    {
      "epoch": 16.413665743305632,
      "grad_norm": 0.3829784393310547,
      "learning_rate": 2.6679e-05,
      "loss": 0.0165,
      "step": 8896
    },
    {
      "epoch": 16.41551246537396,
      "grad_norm": 0.44792550802230835,
      "learning_rate": 2.6682e-05,
      "loss": 0.0166,
      "step": 8897
    },
    {
      "epoch": 16.41735918744229,
      "grad_norm": 0.5492526292800903,
      "learning_rate": 2.6685e-05,
      "loss": 0.0354,
      "step": 8898
    },
    {
      "epoch": 16.41920590951062,
      "grad_norm": 0.3540531396865845,
      "learning_rate": 2.6688e-05,
      "loss": 0.0148,
      "step": 8899
    },
    {
      "epoch": 16.42105263157895,
      "grad_norm": 0.1499744951725006,
      "learning_rate": 2.6691000000000002e-05,
      "loss": 0.0078,
      "step": 8900
    },
    {
      "epoch": 16.422899353647278,
      "grad_norm": 0.14010149240493774,
      "learning_rate": 2.6694000000000002e-05,
      "loss": 0.0083,
      "step": 8901
    },
    {
      "epoch": 16.424746075715603,
      "grad_norm": 0.6157842874526978,
      "learning_rate": 2.6697000000000002e-05,
      "loss": 0.0167,
      "step": 8902
    },
    {
      "epoch": 16.426592797783933,
      "grad_norm": 0.20352981984615326,
      "learning_rate": 2.6700000000000002e-05,
      "loss": 0.0094,
      "step": 8903
    },
    {
      "epoch": 16.42843951985226,
      "grad_norm": 0.27032119035720825,
      "learning_rate": 2.6703e-05,
      "loss": 0.0177,
      "step": 8904
    },
    {
      "epoch": 16.43028624192059,
      "grad_norm": 0.5394765138626099,
      "learning_rate": 2.6706e-05,
      "loss": 0.0202,
      "step": 8905
    },
    {
      "epoch": 16.43213296398892,
      "grad_norm": 0.34230151772499084,
      "learning_rate": 2.6709e-05,
      "loss": 0.0122,
      "step": 8906
    },
    {
      "epoch": 16.43397968605725,
      "grad_norm": 0.8664721250534058,
      "learning_rate": 2.6712e-05,
      "loss": 0.0166,
      "step": 8907
    },
    {
      "epoch": 16.43582640812558,
      "grad_norm": 0.41157254576683044,
      "learning_rate": 2.6715e-05,
      "loss": 0.0111,
      "step": 8908
    },
    {
      "epoch": 16.437673130193907,
      "grad_norm": 0.38635289669036865,
      "learning_rate": 2.6718e-05,
      "loss": 0.0133,
      "step": 8909
    },
    {
      "epoch": 16.439519852262233,
      "grad_norm": 0.47712311148643494,
      "learning_rate": 2.6721e-05,
      "loss": 0.0139,
      "step": 8910
    },
    {
      "epoch": 16.441366574330562,
      "grad_norm": 0.2243277132511139,
      "learning_rate": 2.6724e-05,
      "loss": 0.0093,
      "step": 8911
    },
    {
      "epoch": 16.44321329639889,
      "grad_norm": 0.396609365940094,
      "learning_rate": 2.6727e-05,
      "loss": 0.0191,
      "step": 8912
    },
    {
      "epoch": 16.44506001846722,
      "grad_norm": 0.7725248336791992,
      "learning_rate": 2.673e-05,
      "loss": 0.0263,
      "step": 8913
    },
    {
      "epoch": 16.44690674053555,
      "grad_norm": 0.2902774512767792,
      "learning_rate": 2.6733e-05,
      "loss": 0.0075,
      "step": 8914
    },
    {
      "epoch": 16.44875346260388,
      "grad_norm": 0.31212446093559265,
      "learning_rate": 2.6736e-05,
      "loss": 0.0075,
      "step": 8915
    },
    {
      "epoch": 16.450600184672208,
      "grad_norm": 0.27319321036338806,
      "learning_rate": 2.6739e-05,
      "loss": 0.0095,
      "step": 8916
    },
    {
      "epoch": 16.452446906740537,
      "grad_norm": 0.42757007479667664,
      "learning_rate": 2.6742e-05,
      "loss": 0.0152,
      "step": 8917
    },
    {
      "epoch": 16.454293628808863,
      "grad_norm": 0.2784916162490845,
      "learning_rate": 2.6745e-05,
      "loss": 0.009,
      "step": 8918
    },
    {
      "epoch": 16.45614035087719,
      "grad_norm": 0.6861608028411865,
      "learning_rate": 2.6748e-05,
      "loss": 0.0113,
      "step": 8919
    },
    {
      "epoch": 16.45798707294552,
      "grad_norm": 0.4131728410720825,
      "learning_rate": 2.6751000000000002e-05,
      "loss": 0.0332,
      "step": 8920
    },
    {
      "epoch": 16.45983379501385,
      "grad_norm": 0.7632374167442322,
      "learning_rate": 2.6754e-05,
      "loss": 0.0233,
      "step": 8921
    },
    {
      "epoch": 16.46168051708218,
      "grad_norm": 0.908040463924408,
      "learning_rate": 2.6757e-05,
      "loss": 0.0177,
      "step": 8922
    },
    {
      "epoch": 16.46352723915051,
      "grad_norm": 1.1065903902053833,
      "learning_rate": 2.676e-05,
      "loss": 0.2018,
      "step": 8923
    },
    {
      "epoch": 16.465373961218837,
      "grad_norm": 0.8847578167915344,
      "learning_rate": 2.6763e-05,
      "loss": 0.2644,
      "step": 8924
    },
    {
      "epoch": 16.467220683287167,
      "grad_norm": 0.45100724697113037,
      "learning_rate": 2.6766e-05,
      "loss": 0.0949,
      "step": 8925
    },
    {
      "epoch": 16.469067405355496,
      "grad_norm": 0.5411179065704346,
      "learning_rate": 2.6769e-05,
      "loss": 0.1164,
      "step": 8926
    },
    {
      "epoch": 16.47091412742382,
      "grad_norm": 0.4566073417663574,
      "learning_rate": 2.6772e-05,
      "loss": 0.0827,
      "step": 8927
    },
    {
      "epoch": 16.47276084949215,
      "grad_norm": 0.539164662361145,
      "learning_rate": 2.6775e-05,
      "loss": 0.0678,
      "step": 8928
    },
    {
      "epoch": 16.47460757156048,
      "grad_norm": 1.2484465837478638,
      "learning_rate": 2.6778e-05,
      "loss": 0.1078,
      "step": 8929
    },
    {
      "epoch": 16.47645429362881,
      "grad_norm": 0.6017367243766785,
      "learning_rate": 2.6781000000000003e-05,
      "loss": 0.0881,
      "step": 8930
    },
    {
      "epoch": 16.478301015697138,
      "grad_norm": 0.5592476725578308,
      "learning_rate": 2.6784000000000003e-05,
      "loss": 0.067,
      "step": 8931
    },
    {
      "epoch": 16.480147737765467,
      "grad_norm": 0.46658194065093994,
      "learning_rate": 2.6787000000000003e-05,
      "loss": 0.0652,
      "step": 8932
    },
    {
      "epoch": 16.481994459833796,
      "grad_norm": 0.6025909781455994,
      "learning_rate": 2.6790000000000003e-05,
      "loss": 0.0648,
      "step": 8933
    },
    {
      "epoch": 16.483841181902125,
      "grad_norm": 0.3990728259086609,
      "learning_rate": 2.6793000000000002e-05,
      "loss": 0.0394,
      "step": 8934
    },
    {
      "epoch": 16.48568790397045,
      "grad_norm": 0.46543025970458984,
      "learning_rate": 2.6796e-05,
      "loss": 0.0596,
      "step": 8935
    },
    {
      "epoch": 16.48753462603878,
      "grad_norm": 0.49842751026153564,
      "learning_rate": 2.6799e-05,
      "loss": 0.0499,
      "step": 8936
    },
    {
      "epoch": 16.48938134810711,
      "grad_norm": 0.38525688648223877,
      "learning_rate": 2.6802e-05,
      "loss": 0.0232,
      "step": 8937
    },
    {
      "epoch": 16.49122807017544,
      "grad_norm": 0.2686164081096649,
      "learning_rate": 2.6805e-05,
      "loss": 0.0166,
      "step": 8938
    },
    {
      "epoch": 16.493074792243767,
      "grad_norm": 0.37435072660446167,
      "learning_rate": 2.6807999999999998e-05,
      "loss": 0.0144,
      "step": 8939
    },
    {
      "epoch": 16.494921514312097,
      "grad_norm": 0.37180137634277344,
      "learning_rate": 2.6811e-05,
      "loss": 0.02,
      "step": 8940
    },
    {
      "epoch": 16.496768236380426,
      "grad_norm": 0.36476054787635803,
      "learning_rate": 2.6814e-05,
      "loss": 0.0186,
      "step": 8941
    },
    {
      "epoch": 16.498614958448755,
      "grad_norm": 0.5643298625946045,
      "learning_rate": 2.6817e-05,
      "loss": 0.0202,
      "step": 8942
    },
    {
      "epoch": 16.50046168051708,
      "grad_norm": 0.3396724462509155,
      "learning_rate": 2.682e-05,
      "loss": 0.0094,
      "step": 8943
    },
    {
      "epoch": 16.50230840258541,
      "grad_norm": 0.29790958762168884,
      "learning_rate": 2.6823e-05,
      "loss": 0.0142,
      "step": 8944
    },
    {
      "epoch": 16.50415512465374,
      "grad_norm": 0.29155367612838745,
      "learning_rate": 2.6826e-05,
      "loss": 0.0153,
      "step": 8945
    },
    {
      "epoch": 16.506001846722068,
      "grad_norm": 0.5531447529792786,
      "learning_rate": 2.6829e-05,
      "loss": 0.0105,
      "step": 8946
    },
    {
      "epoch": 16.507848568790397,
      "grad_norm": 0.40035152435302734,
      "learning_rate": 2.6832e-05,
      "loss": 0.0206,
      "step": 8947
    },
    {
      "epoch": 16.509695290858726,
      "grad_norm": 0.3767993152141571,
      "learning_rate": 2.6835e-05,
      "loss": 0.0121,
      "step": 8948
    },
    {
      "epoch": 16.511542012927055,
      "grad_norm": 0.32723501324653625,
      "learning_rate": 2.6838e-05,
      "loss": 0.0166,
      "step": 8949
    },
    {
      "epoch": 16.513388734995385,
      "grad_norm": 0.2774938642978668,
      "learning_rate": 2.6841000000000003e-05,
      "loss": 0.0453,
      "step": 8950
    },
    {
      "epoch": 16.51523545706371,
      "grad_norm": 0.6788330674171448,
      "learning_rate": 2.6844000000000003e-05,
      "loss": 0.017,
      "step": 8951
    },
    {
      "epoch": 16.51708217913204,
      "grad_norm": 0.7191709280014038,
      "learning_rate": 2.6847000000000002e-05,
      "loss": 0.0142,
      "step": 8952
    },
    {
      "epoch": 16.51892890120037,
      "grad_norm": 0.2165871262550354,
      "learning_rate": 2.6850000000000002e-05,
      "loss": 0.0089,
      "step": 8953
    },
    {
      "epoch": 16.520775623268698,
      "grad_norm": 0.4514833092689514,
      "learning_rate": 2.6853000000000002e-05,
      "loss": 0.0141,
      "step": 8954
    },
    {
      "epoch": 16.522622345337027,
      "grad_norm": 0.245515376329422,
      "learning_rate": 2.6856000000000002e-05,
      "loss": 0.0066,
      "step": 8955
    },
    {
      "epoch": 16.524469067405356,
      "grad_norm": 0.600486159324646,
      "learning_rate": 2.6859e-05,
      "loss": 0.0116,
      "step": 8956
    },
    {
      "epoch": 16.526315789473685,
      "grad_norm": 0.27086642384529114,
      "learning_rate": 2.6862e-05,
      "loss": 0.0099,
      "step": 8957
    },
    {
      "epoch": 16.528162511542014,
      "grad_norm": 0.4400690197944641,
      "learning_rate": 2.6865e-05,
      "loss": 0.0167,
      "step": 8958
    },
    {
      "epoch": 16.530009233610343,
      "grad_norm": 0.661552369594574,
      "learning_rate": 2.6867999999999998e-05,
      "loss": 0.0111,
      "step": 8959
    },
    {
      "epoch": 16.53185595567867,
      "grad_norm": 0.4757016599178314,
      "learning_rate": 2.6871e-05,
      "loss": 0.0387,
      "step": 8960
    },
    {
      "epoch": 16.533702677746998,
      "grad_norm": 0.5109610557556152,
      "learning_rate": 2.6874e-05,
      "loss": 0.0171,
      "step": 8961
    },
    {
      "epoch": 16.535549399815327,
      "grad_norm": 0.5325073599815369,
      "learning_rate": 2.6877e-05,
      "loss": 0.0185,
      "step": 8962
    },
    {
      "epoch": 16.537396121883656,
      "grad_norm": 0.6453657150268555,
      "learning_rate": 2.688e-05,
      "loss": 0.0185,
      "step": 8963
    },
    {
      "epoch": 16.539242843951985,
      "grad_norm": 0.6359427571296692,
      "learning_rate": 2.6883e-05,
      "loss": 0.025,
      "step": 8964
    },
    {
      "epoch": 16.541089566020315,
      "grad_norm": 0.7388697862625122,
      "learning_rate": 2.6886e-05,
      "loss": 0.0146,
      "step": 8965
    },
    {
      "epoch": 16.542936288088644,
      "grad_norm": 0.21786294877529144,
      "learning_rate": 2.6889e-05,
      "loss": 0.0075,
      "step": 8966
    },
    {
      "epoch": 16.544783010156973,
      "grad_norm": 0.5282310843467712,
      "learning_rate": 2.6892e-05,
      "loss": 0.0137,
      "step": 8967
    },
    {
      "epoch": 16.5466297322253,
      "grad_norm": 0.3669258654117584,
      "learning_rate": 2.6895e-05,
      "loss": 0.0152,
      "step": 8968
    },
    {
      "epoch": 16.548476454293628,
      "grad_norm": 0.6105878353118896,
      "learning_rate": 2.6898e-05,
      "loss": 0.0108,
      "step": 8969
    },
    {
      "epoch": 16.550323176361957,
      "grad_norm": 0.23995943367481232,
      "learning_rate": 2.6901000000000002e-05,
      "loss": 0.0152,
      "step": 8970
    },
    {
      "epoch": 16.552169898430286,
      "grad_norm": 0.43827906250953674,
      "learning_rate": 2.6904000000000002e-05,
      "loss": 0.0079,
      "step": 8971
    },
    {
      "epoch": 16.554016620498615,
      "grad_norm": 0.7252397537231445,
      "learning_rate": 2.6907000000000002e-05,
      "loss": 0.0357,
      "step": 8972
    },
    {
      "epoch": 16.555863342566944,
      "grad_norm": 0.651902437210083,
      "learning_rate": 2.691e-05,
      "loss": 0.2199,
      "step": 8973
    },
    {
      "epoch": 16.557710064635273,
      "grad_norm": 0.4844575822353363,
      "learning_rate": 2.6913e-05,
      "loss": 0.1116,
      "step": 8974
    },
    {
      "epoch": 16.559556786703602,
      "grad_norm": 0.6385934352874756,
      "learning_rate": 2.6916e-05,
      "loss": 0.1731,
      "step": 8975
    },
    {
      "epoch": 16.56140350877193,
      "grad_norm": 0.5000412464141846,
      "learning_rate": 2.6919e-05,
      "loss": 0.141,
      "step": 8976
    },
    {
      "epoch": 16.563250230840257,
      "grad_norm": 0.7032834887504578,
      "learning_rate": 2.6922e-05,
      "loss": 0.0896,
      "step": 8977
    },
    {
      "epoch": 16.565096952908586,
      "grad_norm": 0.5850236415863037,
      "learning_rate": 2.6925e-05,
      "loss": 0.1414,
      "step": 8978
    },
    {
      "epoch": 16.566943674976915,
      "grad_norm": 0.44791483879089355,
      "learning_rate": 2.6928e-05,
      "loss": 0.0648,
      "step": 8979
    },
    {
      "epoch": 16.568790397045245,
      "grad_norm": 0.576742947101593,
      "learning_rate": 2.6931000000000004e-05,
      "loss": 0.082,
      "step": 8980
    },
    {
      "epoch": 16.570637119113574,
      "grad_norm": 0.6310506463050842,
      "learning_rate": 2.6934000000000003e-05,
      "loss": 0.0532,
      "step": 8981
    },
    {
      "epoch": 16.572483841181903,
      "grad_norm": 0.6527665257453918,
      "learning_rate": 2.6937000000000003e-05,
      "loss": 0.0754,
      "step": 8982
    },
    {
      "epoch": 16.574330563250232,
      "grad_norm": 0.7429980039596558,
      "learning_rate": 2.6940000000000003e-05,
      "loss": 0.0793,
      "step": 8983
    },
    {
      "epoch": 16.57617728531856,
      "grad_norm": 0.7954683303833008,
      "learning_rate": 2.6943e-05,
      "loss": 0.1071,
      "step": 8984
    },
    {
      "epoch": 16.578024007386887,
      "grad_norm": 0.3820796608924866,
      "learning_rate": 2.6946e-05,
      "loss": 0.0265,
      "step": 8985
    },
    {
      "epoch": 16.579870729455216,
      "grad_norm": 0.5825003981590271,
      "learning_rate": 2.6949e-05,
      "loss": 0.0859,
      "step": 8986
    },
    {
      "epoch": 16.581717451523545,
      "grad_norm": 0.45463883876800537,
      "learning_rate": 2.6952e-05,
      "loss": 0.0692,
      "step": 8987
    },
    {
      "epoch": 16.583564173591874,
      "grad_norm": 0.1894882619380951,
      "learning_rate": 2.6955e-05,
      "loss": 0.0134,
      "step": 8988
    },
    {
      "epoch": 16.585410895660203,
      "grad_norm": 0.3609786033630371,
      "learning_rate": 2.6958e-05,
      "loss": 0.0163,
      "step": 8989
    },
    {
      "epoch": 16.587257617728532,
      "grad_norm": 0.7221500277519226,
      "learning_rate": 2.6961e-05,
      "loss": 0.0289,
      "step": 8990
    },
    {
      "epoch": 16.58910433979686,
      "grad_norm": 0.5865622758865356,
      "learning_rate": 2.6964e-05,
      "loss": 0.0236,
      "step": 8991
    },
    {
      "epoch": 16.59095106186519,
      "grad_norm": 0.2695974111557007,
      "learning_rate": 2.6967e-05,
      "loss": 0.0117,
      "step": 8992
    },
    {
      "epoch": 16.592797783933516,
      "grad_norm": 0.30907437205314636,
      "learning_rate": 2.697e-05,
      "loss": 0.0111,
      "step": 8993
    },
    {
      "epoch": 16.594644506001845,
      "grad_norm": 0.3169298768043518,
      "learning_rate": 2.6973e-05,
      "loss": 0.0141,
      "step": 8994
    },
    {
      "epoch": 16.596491228070175,
      "grad_norm": 0.739950954914093,
      "learning_rate": 2.6976e-05,
      "loss": 0.0227,
      "step": 8995
    },
    {
      "epoch": 16.598337950138504,
      "grad_norm": 0.3569239377975464,
      "learning_rate": 2.6979e-05,
      "loss": 0.0109,
      "step": 8996
    },
    {
      "epoch": 16.600184672206833,
      "grad_norm": 0.5756157636642456,
      "learning_rate": 2.6982e-05,
      "loss": 0.0196,
      "step": 8997
    },
    {
      "epoch": 16.602031394275162,
      "grad_norm": 0.661070704460144,
      "learning_rate": 2.6985e-05,
      "loss": 0.0209,
      "step": 8998
    },
    {
      "epoch": 16.60387811634349,
      "grad_norm": 0.4226855933666229,
      "learning_rate": 2.6988e-05,
      "loss": 0.0138,
      "step": 8999
    },
    {
      "epoch": 16.60572483841182,
      "grad_norm": 0.5164405107498169,
      "learning_rate": 2.6991000000000003e-05,
      "loss": 0.0205,
      "step": 9000
    },
    {
      "epoch": 16.60572483841182,
      "eval_cer": 0.11260595691733416,
      "eval_loss": 0.3083859980106354,
      "eval_runtime": 15.9192,
      "eval_samples_per_second": 19.096,
      "eval_steps_per_second": 0.628,
      "eval_wer": 0.39102072141212585,
      "step": 9000
    },
    {
      "epoch": 16.607571560480146,
      "grad_norm": 0.32492971420288086,
      "learning_rate": 2.6994000000000003e-05,
      "loss": 0.0159,
      "step": 9001
    },
    {
      "epoch": 16.609418282548475,
      "grad_norm": 0.30137714743614197,
      "learning_rate": 2.6997000000000003e-05,
      "loss": 0.0065,
      "step": 9002
    },
    {
      "epoch": 16.611265004616804,
      "grad_norm": 0.6133913993835449,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.0154,
      "step": 9003
    },
    {
      "epoch": 16.613111726685133,
      "grad_norm": 0.3343926668167114,
      "learning_rate": 2.7003000000000002e-05,
      "loss": 0.0207,
      "step": 9004
    },
    {
      "epoch": 16.614958448753463,
      "grad_norm": 0.7353482246398926,
      "learning_rate": 2.7006000000000002e-05,
      "loss": 0.0181,
      "step": 9005
    },
    {
      "epoch": 16.61680517082179,
      "grad_norm": 0.6952447891235352,
      "learning_rate": 2.7009000000000002e-05,
      "loss": 0.0143,
      "step": 9006
    },
    {
      "epoch": 16.61865189289012,
      "grad_norm": 0.27672079205513,
      "learning_rate": 2.7012e-05,
      "loss": 0.0133,
      "step": 9007
    },
    {
      "epoch": 16.62049861495845,
      "grad_norm": 0.2262832075357437,
      "learning_rate": 2.7015e-05,
      "loss": 0.0108,
      "step": 9008
    },
    {
      "epoch": 16.62234533702678,
      "grad_norm": 0.7397500276565552,
      "learning_rate": 2.7017999999999998e-05,
      "loss": 0.0131,
      "step": 9009
    },
    {
      "epoch": 16.624192059095105,
      "grad_norm": 0.2488490790128708,
      "learning_rate": 2.7020999999999998e-05,
      "loss": 0.0147,
      "step": 9010
    },
    {
      "epoch": 16.626038781163434,
      "grad_norm": 0.7176915407180786,
      "learning_rate": 2.7024e-05,
      "loss": 0.0199,
      "step": 9011
    },
    {
      "epoch": 16.627885503231763,
      "grad_norm": 0.34909358620643616,
      "learning_rate": 2.7027e-05,
      "loss": 0.021,
      "step": 9012
    },
    {
      "epoch": 16.629732225300092,
      "grad_norm": 0.45952746272087097,
      "learning_rate": 2.703e-05,
      "loss": 0.0236,
      "step": 9013
    },
    {
      "epoch": 16.63157894736842,
      "grad_norm": 0.5121981501579285,
      "learning_rate": 2.7033e-05,
      "loss": 0.0224,
      "step": 9014
    },
    {
      "epoch": 16.63342566943675,
      "grad_norm": 0.3115614950656891,
      "learning_rate": 2.7036e-05,
      "loss": 0.0103,
      "step": 9015
    },
    {
      "epoch": 16.63527239150508,
      "grad_norm": 0.40834179520606995,
      "learning_rate": 2.7039e-05,
      "loss": 0.0198,
      "step": 9016
    },
    {
      "epoch": 16.63711911357341,
      "grad_norm": 0.29144415259361267,
      "learning_rate": 2.7042e-05,
      "loss": 0.015,
      "step": 9017
    },
    {
      "epoch": 16.638965835641734,
      "grad_norm": 0.535346269607544,
      "learning_rate": 2.7045e-05,
      "loss": 0.0197,
      "step": 9018
    },
    {
      "epoch": 16.640812557710063,
      "grad_norm": 0.6192907094955444,
      "learning_rate": 2.7048e-05,
      "loss": 0.0367,
      "step": 9019
    },
    {
      "epoch": 16.642659279778393,
      "grad_norm": 0.31909242272377014,
      "learning_rate": 2.7051e-05,
      "loss": 0.0191,
      "step": 9020
    },
    {
      "epoch": 16.64450600184672,
      "grad_norm": 0.43197062611579895,
      "learning_rate": 2.7054000000000002e-05,
      "loss": 0.0174,
      "step": 9021
    },
    {
      "epoch": 16.64635272391505,
      "grad_norm": 0.3951015770435333,
      "learning_rate": 2.7057000000000002e-05,
      "loss": 0.0128,
      "step": 9022
    },
    {
      "epoch": 16.64819944598338,
      "grad_norm": 0.775962233543396,
      "learning_rate": 2.7060000000000002e-05,
      "loss": 0.2058,
      "step": 9023
    },
    {
      "epoch": 16.65004616805171,
      "grad_norm": 0.6837390661239624,
      "learning_rate": 2.7063e-05,
      "loss": 0.1974,
      "step": 9024
    },
    {
      "epoch": 16.65189289012004,
      "grad_norm": 0.499403715133667,
      "learning_rate": 2.7066e-05,
      "loss": 0.1654,
      "step": 9025
    },
    {
      "epoch": 16.653739612188367,
      "grad_norm": 0.5308801531791687,
      "learning_rate": 2.7069e-05,
      "loss": 0.1186,
      "step": 9026
    },
    {
      "epoch": 16.655586334256693,
      "grad_norm": 0.5072598457336426,
      "learning_rate": 2.7072e-05,
      "loss": 0.1018,
      "step": 9027
    },
    {
      "epoch": 16.657433056325022,
      "grad_norm": 0.5138546824455261,
      "learning_rate": 2.7075e-05,
      "loss": 0.1102,
      "step": 9028
    },
    {
      "epoch": 16.65927977839335,
      "grad_norm": 0.49401623010635376,
      "learning_rate": 2.7078e-05,
      "loss": 0.0695,
      "step": 9029
    },
    {
      "epoch": 16.66112650046168,
      "grad_norm": 0.47299742698669434,
      "learning_rate": 2.7081e-05,
      "loss": 0.0837,
      "step": 9030
    },
    {
      "epoch": 16.66297322253001,
      "grad_norm": 0.39474567770957947,
      "learning_rate": 2.7084000000000004e-05,
      "loss": 0.0639,
      "step": 9031
    },
    {
      "epoch": 16.66481994459834,
      "grad_norm": 0.4763193726539612,
      "learning_rate": 2.7087000000000003e-05,
      "loss": 0.1129,
      "step": 9032
    },
    {
      "epoch": 16.666666666666668,
      "grad_norm": 0.6414193511009216,
      "learning_rate": 2.709e-05,
      "loss": 0.047,
      "step": 9033
    },
    {
      "epoch": 16.668513388734997,
      "grad_norm": 0.4468953609466553,
      "learning_rate": 2.7093e-05,
      "loss": 0.0435,
      "step": 9034
    },
    {
      "epoch": 16.670360110803323,
      "grad_norm": 0.6856594681739807,
      "learning_rate": 2.7096e-05,
      "loss": 0.0754,
      "step": 9035
    },
    {
      "epoch": 16.67220683287165,
      "grad_norm": 0.2718965709209442,
      "learning_rate": 2.7099e-05,
      "loss": 0.0332,
      "step": 9036
    },
    {
      "epoch": 16.67405355493998,
      "grad_norm": 0.28632283210754395,
      "learning_rate": 2.7102e-05,
      "loss": 0.0433,
      "step": 9037
    },
    {
      "epoch": 16.67590027700831,
      "grad_norm": 0.4210139214992523,
      "learning_rate": 2.7105e-05,
      "loss": 0.0345,
      "step": 9038
    },
    {
      "epoch": 16.67774699907664,
      "grad_norm": 0.3410778343677521,
      "learning_rate": 2.7108e-05,
      "loss": 0.0247,
      "step": 9039
    },
    {
      "epoch": 16.67959372114497,
      "grad_norm": 0.4439637362957001,
      "learning_rate": 2.7111e-05,
      "loss": 0.0153,
      "step": 9040
    },
    {
      "epoch": 16.681440443213297,
      "grad_norm": 0.9153331518173218,
      "learning_rate": 2.7114e-05,
      "loss": 0.0107,
      "step": 9041
    },
    {
      "epoch": 16.683287165281627,
      "grad_norm": 0.3366173505783081,
      "learning_rate": 2.7117e-05,
      "loss": 0.0122,
      "step": 9042
    },
    {
      "epoch": 16.685133887349952,
      "grad_norm": 0.20940084755420685,
      "learning_rate": 2.712e-05,
      "loss": 0.0175,
      "step": 9043
    },
    {
      "epoch": 16.68698060941828,
      "grad_norm": 0.861236572265625,
      "learning_rate": 2.7123e-05,
      "loss": 0.0342,
      "step": 9044
    },
    {
      "epoch": 16.68882733148661,
      "grad_norm": 0.41026780009269714,
      "learning_rate": 2.7126e-05,
      "loss": 0.0277,
      "step": 9045
    },
    {
      "epoch": 16.69067405355494,
      "grad_norm": 0.24577605724334717,
      "learning_rate": 2.7129e-05,
      "loss": 0.0119,
      "step": 9046
    },
    {
      "epoch": 16.69252077562327,
      "grad_norm": 0.33204078674316406,
      "learning_rate": 2.7132e-05,
      "loss": 0.0131,
      "step": 9047
    },
    {
      "epoch": 16.694367497691598,
      "grad_norm": 0.1572006791830063,
      "learning_rate": 2.7135e-05,
      "loss": 0.0093,
      "step": 9048
    },
    {
      "epoch": 16.696214219759927,
      "grad_norm": 0.6033346652984619,
      "learning_rate": 2.7138e-05,
      "loss": 0.0179,
      "step": 9049
    },
    {
      "epoch": 16.698060941828256,
      "grad_norm": 0.31534436345100403,
      "learning_rate": 2.7141e-05,
      "loss": 0.011,
      "step": 9050
    },
    {
      "epoch": 16.69990766389658,
      "grad_norm": 0.29961878061294556,
      "learning_rate": 2.7144000000000003e-05,
      "loss": 0.0254,
      "step": 9051
    },
    {
      "epoch": 16.70175438596491,
      "grad_norm": 0.263561487197876,
      "learning_rate": 2.7147000000000003e-05,
      "loss": 0.0121,
      "step": 9052
    },
    {
      "epoch": 16.70360110803324,
      "grad_norm": 0.28035175800323486,
      "learning_rate": 2.7150000000000003e-05,
      "loss": 0.0079,
      "step": 9053
    },
    {
      "epoch": 16.70544783010157,
      "grad_norm": 0.43491461873054504,
      "learning_rate": 2.7153000000000002e-05,
      "loss": 0.0226,
      "step": 9054
    },
    {
      "epoch": 16.7072945521699,
      "grad_norm": 0.2045207917690277,
      "learning_rate": 2.7156000000000002e-05,
      "loss": 0.0108,
      "step": 9055
    },
    {
      "epoch": 16.709141274238227,
      "grad_norm": 0.35303208231925964,
      "learning_rate": 2.7159000000000002e-05,
      "loss": 0.0138,
      "step": 9056
    },
    {
      "epoch": 16.710987996306557,
      "grad_norm": 0.24313557147979736,
      "learning_rate": 2.7162000000000002e-05,
      "loss": 0.0126,
      "step": 9057
    },
    {
      "epoch": 16.712834718374886,
      "grad_norm": 0.4155482351779938,
      "learning_rate": 2.7164999999999998e-05,
      "loss": 0.0184,
      "step": 9058
    },
    {
      "epoch": 16.714681440443215,
      "grad_norm": 0.35526689887046814,
      "learning_rate": 2.7167999999999998e-05,
      "loss": 0.0127,
      "step": 9059
    },
    {
      "epoch": 16.71652816251154,
      "grad_norm": 1.286755084991455,
      "learning_rate": 2.7170999999999998e-05,
      "loss": 0.0127,
      "step": 9060
    },
    {
      "epoch": 16.71837488457987,
      "grad_norm": 0.4949331283569336,
      "learning_rate": 2.7174e-05,
      "loss": 0.018,
      "step": 9061
    },
    {
      "epoch": 16.7202216066482,
      "grad_norm": 0.2951291501522064,
      "learning_rate": 2.7177e-05,
      "loss": 0.0107,
      "step": 9062
    },
    {
      "epoch": 16.722068328716528,
      "grad_norm": 0.29090291261672974,
      "learning_rate": 2.718e-05,
      "loss": 0.0178,
      "step": 9063
    },
    {
      "epoch": 16.723915050784857,
      "grad_norm": 0.4155007600784302,
      "learning_rate": 2.7183e-05,
      "loss": 0.0122,
      "step": 9064
    },
    {
      "epoch": 16.725761772853186,
      "grad_norm": 0.5140042901039124,
      "learning_rate": 2.7186e-05,
      "loss": 0.012,
      "step": 9065
    },
    {
      "epoch": 16.727608494921515,
      "grad_norm": 0.6684512495994568,
      "learning_rate": 2.7189e-05,
      "loss": 0.0278,
      "step": 9066
    },
    {
      "epoch": 16.729455216989845,
      "grad_norm": 0.5385333895683289,
      "learning_rate": 2.7192e-05,
      "loss": 0.0111,
      "step": 9067
    },
    {
      "epoch": 16.73130193905817,
      "grad_norm": 0.5201623439788818,
      "learning_rate": 2.7195e-05,
      "loss": 0.0127,
      "step": 9068
    },
    {
      "epoch": 16.7331486611265,
      "grad_norm": 0.353179395198822,
      "learning_rate": 2.7198e-05,
      "loss": 0.0092,
      "step": 9069
    },
    {
      "epoch": 16.73499538319483,
      "grad_norm": 0.3160616159439087,
      "learning_rate": 2.7201e-05,
      "loss": 0.0141,
      "step": 9070
    },
    {
      "epoch": 16.736842105263158,
      "grad_norm": 0.21332009136676788,
      "learning_rate": 2.7204000000000002e-05,
      "loss": 0.0079,
      "step": 9071
    },
    {
      "epoch": 16.738688827331487,
      "grad_norm": 0.6301090121269226,
      "learning_rate": 2.7207000000000002e-05,
      "loss": 0.029,
      "step": 9072
    },
    {
      "epoch": 16.740535549399816,
      "grad_norm": 0.8239292502403259,
      "learning_rate": 2.7210000000000002e-05,
      "loss": 0.2004,
      "step": 9073
    },
    {
      "epoch": 16.742382271468145,
      "grad_norm": 0.6537140011787415,
      "learning_rate": 2.7213000000000002e-05,
      "loss": 0.1969,
      "step": 9074
    },
    {
      "epoch": 16.744228993536474,
      "grad_norm": 0.6992922425270081,
      "learning_rate": 2.7216e-05,
      "loss": 0.1587,
      "step": 9075
    },
    {
      "epoch": 16.746075715604803,
      "grad_norm": 0.5989820957183838,
      "learning_rate": 2.7219e-05,
      "loss": 0.0969,
      "step": 9076
    },
    {
      "epoch": 16.74792243767313,
      "grad_norm": 0.440569132566452,
      "learning_rate": 2.7222e-05,
      "loss": 0.0913,
      "step": 9077
    },
    {
      "epoch": 16.749769159741458,
      "grad_norm": 0.41203105449676514,
      "learning_rate": 2.7225e-05,
      "loss": 0.0934,
      "step": 9078
    },
    {
      "epoch": 16.751615881809787,
      "grad_norm": 0.4103984534740448,
      "learning_rate": 2.7228e-05,
      "loss": 0.1059,
      "step": 9079
    },
    {
      "epoch": 16.753462603878116,
      "grad_norm": 0.4278160035610199,
      "learning_rate": 2.7231e-05,
      "loss": 0.0839,
      "step": 9080
    },
    {
      "epoch": 16.755309325946445,
      "grad_norm": 0.4500424563884735,
      "learning_rate": 2.7234000000000004e-05,
      "loss": 0.0509,
      "step": 9081
    },
    {
      "epoch": 16.757156048014775,
      "grad_norm": 0.4022199809551239,
      "learning_rate": 2.7237e-05,
      "loss": 0.0625,
      "step": 9082
    },
    {
      "epoch": 16.759002770083104,
      "grad_norm": 0.41468554735183716,
      "learning_rate": 2.724e-05,
      "loss": 0.0718,
      "step": 9083
    },
    {
      "epoch": 16.760849492151433,
      "grad_norm": 0.6440789699554443,
      "learning_rate": 2.7243e-05,
      "loss": 0.0631,
      "step": 9084
    },
    {
      "epoch": 16.76269621421976,
      "grad_norm": 0.6056049466133118,
      "learning_rate": 2.7246e-05,
      "loss": 0.0454,
      "step": 9085
    },
    {
      "epoch": 16.764542936288088,
      "grad_norm": 0.47346749901771545,
      "learning_rate": 2.7249e-05,
      "loss": 0.0418,
      "step": 9086
    },
    {
      "epoch": 16.766389658356417,
      "grad_norm": 0.3096546232700348,
      "learning_rate": 2.7252e-05,
      "loss": 0.0327,
      "step": 9087
    },
    {
      "epoch": 16.768236380424746,
      "grad_norm": 0.17871594429016113,
      "learning_rate": 2.7255e-05,
      "loss": 0.0147,
      "step": 9088
    },
    {
      "epoch": 16.770083102493075,
      "grad_norm": 0.3867637813091278,
      "learning_rate": 2.7258e-05,
      "loss": 0.0307,
      "step": 9089
    },
    {
      "epoch": 16.771929824561404,
      "grad_norm": 0.46113646030426025,
      "learning_rate": 2.7261e-05,
      "loss": 0.0482,
      "step": 9090
    },
    {
      "epoch": 16.773776546629733,
      "grad_norm": 0.5311370491981506,
      "learning_rate": 2.7264000000000002e-05,
      "loss": 0.0398,
      "step": 9091
    },
    {
      "epoch": 16.775623268698062,
      "grad_norm": 0.3156410753726959,
      "learning_rate": 2.7267e-05,
      "loss": 0.079,
      "step": 9092
    },
    {
      "epoch": 16.777469990766388,
      "grad_norm": 0.38590025901794434,
      "learning_rate": 2.727e-05,
      "loss": 0.0261,
      "step": 9093
    },
    {
      "epoch": 16.779316712834717,
      "grad_norm": 0.24432797729969025,
      "learning_rate": 2.7273e-05,
      "loss": 0.0094,
      "step": 9094
    },
    {
      "epoch": 16.781163434903046,
      "grad_norm": 0.2722245454788208,
      "learning_rate": 2.7276e-05,
      "loss": 0.0119,
      "step": 9095
    },
    {
      "epoch": 16.783010156971375,
      "grad_norm": 0.31428173184394836,
      "learning_rate": 2.7279e-05,
      "loss": 0.0155,
      "step": 9096
    },
    {
      "epoch": 16.784856879039705,
      "grad_norm": 0.41833361983299255,
      "learning_rate": 2.7282e-05,
      "loss": 0.026,
      "step": 9097
    },
    {
      "epoch": 16.786703601108034,
      "grad_norm": 0.21867933869361877,
      "learning_rate": 2.7285e-05,
      "loss": 0.0102,
      "step": 9098
    },
    {
      "epoch": 16.788550323176363,
      "grad_norm": 0.3054670989513397,
      "learning_rate": 2.7288e-05,
      "loss": 0.0162,
      "step": 9099
    },
    {
      "epoch": 16.790397045244692,
      "grad_norm": 0.26028329133987427,
      "learning_rate": 2.7291e-05,
      "loss": 0.0175,
      "step": 9100
    },
    {
      "epoch": 16.792243767313018,
      "grad_norm": 0.26109904050827026,
      "learning_rate": 2.7294000000000003e-05,
      "loss": 0.0083,
      "step": 9101
    },
    {
      "epoch": 16.794090489381347,
      "grad_norm": 0.3616829514503479,
      "learning_rate": 2.7297000000000003e-05,
      "loss": 0.0109,
      "step": 9102
    },
    {
      "epoch": 16.795937211449676,
      "grad_norm": 0.3457438349723816,
      "learning_rate": 2.7300000000000003e-05,
      "loss": 0.0127,
      "step": 9103
    },
    {
      "epoch": 16.797783933518005,
      "grad_norm": 0.32503554224967957,
      "learning_rate": 2.7303000000000003e-05,
      "loss": 0.0143,
      "step": 9104
    },
    {
      "epoch": 16.799630655586334,
      "grad_norm": 0.17920361459255219,
      "learning_rate": 2.7306000000000002e-05,
      "loss": 0.0073,
      "step": 9105
    },
    {
      "epoch": 16.801477377654663,
      "grad_norm": 0.0969751626253128,
      "learning_rate": 2.7309000000000002e-05,
      "loss": 0.0028,
      "step": 9106
    },
    {
      "epoch": 16.803324099722992,
      "grad_norm": 0.25758984684944153,
      "learning_rate": 2.7312e-05,
      "loss": 0.0101,
      "step": 9107
    },
    {
      "epoch": 16.80517082179132,
      "grad_norm": 0.33695298433303833,
      "learning_rate": 2.7315e-05,
      "loss": 0.0104,
      "step": 9108
    },
    {
      "epoch": 16.80701754385965,
      "grad_norm": 0.8208706974983215,
      "learning_rate": 2.7318e-05,
      "loss": 0.0117,
      "step": 9109
    },
    {
      "epoch": 16.808864265927976,
      "grad_norm": 0.3066198527812958,
      "learning_rate": 2.7320999999999998e-05,
      "loss": 0.0106,
      "step": 9110
    },
    {
      "epoch": 16.810710987996305,
      "grad_norm": 0.9490204453468323,
      "learning_rate": 2.7324e-05,
      "loss": 0.031,
      "step": 9111
    },
    {
      "epoch": 16.812557710064635,
      "grad_norm": 0.344614177942276,
      "learning_rate": 2.7327e-05,
      "loss": 0.0114,
      "step": 9112
    },
    {
      "epoch": 16.814404432132964,
      "grad_norm": 0.27744945883750916,
      "learning_rate": 2.733e-05,
      "loss": 0.0085,
      "step": 9113
    },
    {
      "epoch": 16.816251154201293,
      "grad_norm": 0.3573910593986511,
      "learning_rate": 2.7333e-05,
      "loss": 0.0129,
      "step": 9114
    },
    {
      "epoch": 16.818097876269622,
      "grad_norm": 0.44293442368507385,
      "learning_rate": 2.7336e-05,
      "loss": 0.0201,
      "step": 9115
    },
    {
      "epoch": 16.81994459833795,
      "grad_norm": 0.5297267436981201,
      "learning_rate": 2.7339e-05,
      "loss": 0.0124,
      "step": 9116
    },
    {
      "epoch": 16.82179132040628,
      "grad_norm": 0.33147096633911133,
      "learning_rate": 2.7342e-05,
      "loss": 0.0143,
      "step": 9117
    },
    {
      "epoch": 16.823638042474606,
      "grad_norm": 0.36776602268218994,
      "learning_rate": 2.7345e-05,
      "loss": 0.0164,
      "step": 9118
    },
    {
      "epoch": 16.825484764542935,
      "grad_norm": 0.3640115559101105,
      "learning_rate": 2.7348e-05,
      "loss": 0.011,
      "step": 9119
    },
    {
      "epoch": 16.827331486611264,
      "grad_norm": 0.4329957962036133,
      "learning_rate": 2.7351e-05,
      "loss": 0.0157,
      "step": 9120
    },
    {
      "epoch": 16.829178208679593,
      "grad_norm": 0.9924282431602478,
      "learning_rate": 2.7354000000000003e-05,
      "loss": 0.0122,
      "step": 9121
    },
    {
      "epoch": 16.831024930747922,
      "grad_norm": 0.8403156399726868,
      "learning_rate": 2.7357000000000003e-05,
      "loss": 0.0295,
      "step": 9122
    },
    {
      "epoch": 16.83287165281625,
      "grad_norm": 0.7272826433181763,
      "learning_rate": 2.7360000000000002e-05,
      "loss": 0.2257,
      "step": 9123
    },
    {
      "epoch": 16.83471837488458,
      "grad_norm": 0.512742280960083,
      "learning_rate": 2.7363000000000002e-05,
      "loss": 0.1599,
      "step": 9124
    },
    {
      "epoch": 16.83656509695291,
      "grad_norm": 0.5445821285247803,
      "learning_rate": 2.7366000000000002e-05,
      "loss": 0.1199,
      "step": 9125
    },
    {
      "epoch": 16.83841181902124,
      "grad_norm": 0.57929927110672,
      "learning_rate": 2.7369000000000002e-05,
      "loss": 0.1051,
      "step": 9126
    },
    {
      "epoch": 16.840258541089565,
      "grad_norm": 0.5976483225822449,
      "learning_rate": 2.7372e-05,
      "loss": 0.1466,
      "step": 9127
    },
    {
      "epoch": 16.842105263157894,
      "grad_norm": 0.6029165983200073,
      "learning_rate": 2.7375e-05,
      "loss": 0.0952,
      "step": 9128
    },
    {
      "epoch": 16.843951985226223,
      "grad_norm": 0.47818711400032043,
      "learning_rate": 2.7378e-05,
      "loss": 0.1068,
      "step": 9129
    },
    {
      "epoch": 16.845798707294552,
      "grad_norm": 0.351434588432312,
      "learning_rate": 2.7381e-05,
      "loss": 0.0641,
      "step": 9130
    },
    {
      "epoch": 16.84764542936288,
      "grad_norm": 0.4097804129123688,
      "learning_rate": 2.7383999999999997e-05,
      "loss": 0.0618,
      "step": 9131
    },
    {
      "epoch": 16.84949215143121,
      "grad_norm": 0.41000524163246155,
      "learning_rate": 2.7387e-05,
      "loss": 0.0526,
      "step": 9132
    },
    {
      "epoch": 16.85133887349954,
      "grad_norm": 0.4092511236667633,
      "learning_rate": 2.739e-05,
      "loss": 0.0473,
      "step": 9133
    },
    {
      "epoch": 16.85318559556787,
      "grad_norm": 0.5258705019950867,
      "learning_rate": 2.7393e-05,
      "loss": 0.0966,
      "step": 9134
    },
    {
      "epoch": 16.855032317636194,
      "grad_norm": 0.7498922944068909,
      "learning_rate": 2.7396e-05,
      "loss": 0.0614,
      "step": 9135
    },
    {
      "epoch": 16.856879039704523,
      "grad_norm": 0.4262339770793915,
      "learning_rate": 2.7399e-05,
      "loss": 0.0303,
      "step": 9136
    },
    {
      "epoch": 16.858725761772853,
      "grad_norm": 0.45660534501075745,
      "learning_rate": 2.7402e-05,
      "loss": 0.034,
      "step": 9137
    },
    {
      "epoch": 16.86057248384118,
      "grad_norm": 0.34982824325561523,
      "learning_rate": 2.7405e-05,
      "loss": 0.0176,
      "step": 9138
    },
    {
      "epoch": 16.86241920590951,
      "grad_norm": 0.6266517639160156,
      "learning_rate": 2.7408e-05,
      "loss": 0.0548,
      "step": 9139
    },
    {
      "epoch": 16.86426592797784,
      "grad_norm": 0.26492610573768616,
      "learning_rate": 2.7411e-05,
      "loss": 0.0214,
      "step": 9140
    },
    {
      "epoch": 16.86611265004617,
      "grad_norm": 0.3128187954425812,
      "learning_rate": 2.7414e-05,
      "loss": 0.0197,
      "step": 9141
    },
    {
      "epoch": 16.8679593721145,
      "grad_norm": 0.4308634102344513,
      "learning_rate": 2.7417000000000002e-05,
      "loss": 0.0205,
      "step": 9142
    },
    {
      "epoch": 16.869806094182824,
      "grad_norm": 0.319132924079895,
      "learning_rate": 2.7420000000000002e-05,
      "loss": 0.0178,
      "step": 9143
    },
    {
      "epoch": 16.871652816251153,
      "grad_norm": 0.2279515415430069,
      "learning_rate": 2.7423e-05,
      "loss": 0.0097,
      "step": 9144
    },
    {
      "epoch": 16.873499538319482,
      "grad_norm": 0.2076783925294876,
      "learning_rate": 2.7426e-05,
      "loss": 0.014,
      "step": 9145
    },
    {
      "epoch": 16.87534626038781,
      "grad_norm": 0.38339874148368835,
      "learning_rate": 2.7429e-05,
      "loss": 0.0168,
      "step": 9146
    },
    {
      "epoch": 16.87719298245614,
      "grad_norm": 0.476095974445343,
      "learning_rate": 2.7432e-05,
      "loss": 0.0216,
      "step": 9147
    },
    {
      "epoch": 16.87903970452447,
      "grad_norm": 0.25043198466300964,
      "learning_rate": 2.7435e-05,
      "loss": 0.0108,
      "step": 9148
    },
    {
      "epoch": 16.8808864265928,
      "grad_norm": 0.4922471344470978,
      "learning_rate": 2.7438e-05,
      "loss": 0.0111,
      "step": 9149
    },
    {
      "epoch": 16.882733148661128,
      "grad_norm": 0.2648068964481354,
      "learning_rate": 2.7441e-05,
      "loss": 0.0192,
      "step": 9150
    },
    {
      "epoch": 16.884579870729453,
      "grad_norm": 0.47534462809562683,
      "learning_rate": 2.7444e-05,
      "loss": 0.017,
      "step": 9151
    },
    {
      "epoch": 16.886426592797783,
      "grad_norm": 0.37938153743743896,
      "learning_rate": 2.7447000000000003e-05,
      "loss": 0.0418,
      "step": 9152
    },
    {
      "epoch": 16.88827331486611,
      "grad_norm": 0.3794781267642975,
      "learning_rate": 2.7450000000000003e-05,
      "loss": 0.0161,
      "step": 9153
    },
    {
      "epoch": 16.89012003693444,
      "grad_norm": 0.21170689165592194,
      "learning_rate": 2.7453000000000003e-05,
      "loss": 0.0089,
      "step": 9154
    },
    {
      "epoch": 16.89196675900277,
      "grad_norm": 0.3793889880180359,
      "learning_rate": 2.7456000000000003e-05,
      "loss": 0.0131,
      "step": 9155
    },
    {
      "epoch": 16.8938134810711,
      "grad_norm": 0.40929901599884033,
      "learning_rate": 2.7459e-05,
      "loss": 0.0116,
      "step": 9156
    },
    {
      "epoch": 16.89566020313943,
      "grad_norm": 0.8098201751708984,
      "learning_rate": 2.7462e-05,
      "loss": 0.015,
      "step": 9157
    },
    {
      "epoch": 16.897506925207757,
      "grad_norm": 0.2994246482849121,
      "learning_rate": 2.7465e-05,
      "loss": 0.0117,
      "step": 9158
    },
    {
      "epoch": 16.899353647276087,
      "grad_norm": 0.8491688370704651,
      "learning_rate": 2.7468e-05,
      "loss": 0.0389,
      "step": 9159
    },
    {
      "epoch": 16.901200369344412,
      "grad_norm": 0.29761558771133423,
      "learning_rate": 2.7471e-05,
      "loss": 0.0126,
      "step": 9160
    },
    {
      "epoch": 16.90304709141274,
      "grad_norm": 0.2771746814250946,
      "learning_rate": 2.7473999999999998e-05,
      "loss": 0.0154,
      "step": 9161
    },
    {
      "epoch": 16.90489381348107,
      "grad_norm": 0.19669193029403687,
      "learning_rate": 2.7477e-05,
      "loss": 0.0131,
      "step": 9162
    },
    {
      "epoch": 16.9067405355494,
      "grad_norm": 0.42978933453559875,
      "learning_rate": 2.748e-05,
      "loss": 0.0165,
      "step": 9163
    },
    {
      "epoch": 16.90858725761773,
      "grad_norm": 0.48822492361068726,
      "learning_rate": 2.7483e-05,
      "loss": 0.0127,
      "step": 9164
    },
    {
      "epoch": 16.910433979686058,
      "grad_norm": 0.3552548587322235,
      "learning_rate": 2.7486e-05,
      "loss": 0.0119,
      "step": 9165
    },
    {
      "epoch": 16.912280701754387,
      "grad_norm": 0.48361822962760925,
      "learning_rate": 2.7489e-05,
      "loss": 0.0135,
      "step": 9166
    },
    {
      "epoch": 16.914127423822716,
      "grad_norm": 0.2711721956729889,
      "learning_rate": 2.7492e-05,
      "loss": 0.008,
      "step": 9167
    },
    {
      "epoch": 16.91597414589104,
      "grad_norm": 0.41434890031814575,
      "learning_rate": 2.7495e-05,
      "loss": 0.0142,
      "step": 9168
    },
    {
      "epoch": 16.91782086795937,
      "grad_norm": 0.3995231091976166,
      "learning_rate": 2.7498e-05,
      "loss": 0.0163,
      "step": 9169
    },
    {
      "epoch": 16.9196675900277,
      "grad_norm": 0.6402641534805298,
      "learning_rate": 2.7501e-05,
      "loss": 0.0163,
      "step": 9170
    },
    {
      "epoch": 16.92151431209603,
      "grad_norm": 0.46368321776390076,
      "learning_rate": 2.7504e-05,
      "loss": 0.0152,
      "step": 9171
    },
    {
      "epoch": 16.92336103416436,
      "grad_norm": 0.36163800954818726,
      "learning_rate": 2.7507000000000003e-05,
      "loss": 0.0247,
      "step": 9172
    },
    {
      "epoch": 16.925207756232687,
      "grad_norm": 0.6181609034538269,
      "learning_rate": 2.7510000000000003e-05,
      "loss": 0.1281,
      "step": 9173
    },
    {
      "epoch": 16.927054478301017,
      "grad_norm": 0.7446320056915283,
      "learning_rate": 2.7513000000000002e-05,
      "loss": 0.14,
      "step": 9174
    },
    {
      "epoch": 16.928901200369346,
      "grad_norm": 0.6812530755996704,
      "learning_rate": 2.7516000000000002e-05,
      "loss": 0.1675,
      "step": 9175
    },
    {
      "epoch": 16.930747922437675,
      "grad_norm": 0.6202914714813232,
      "learning_rate": 2.7519000000000002e-05,
      "loss": 0.1464,
      "step": 9176
    },
    {
      "epoch": 16.932594644506,
      "grad_norm": 0.4810855984687805,
      "learning_rate": 2.7522000000000002e-05,
      "loss": 0.0624,
      "step": 9177
    },
    {
      "epoch": 16.93444136657433,
      "grad_norm": 0.6213238835334778,
      "learning_rate": 2.7525e-05,
      "loss": 0.0846,
      "step": 9178
    },
    {
      "epoch": 16.93628808864266,
      "grad_norm": 1.0476669073104858,
      "learning_rate": 2.7528e-05,
      "loss": 0.1087,
      "step": 9179
    },
    {
      "epoch": 16.938134810710988,
      "grad_norm": 0.5438078045845032,
      "learning_rate": 2.7531e-05,
      "loss": 0.0478,
      "step": 9180
    },
    {
      "epoch": 16.939981532779317,
      "grad_norm": 0.6464812159538269,
      "learning_rate": 2.7533999999999998e-05,
      "loss": 0.0393,
      "step": 9181
    },
    {
      "epoch": 16.941828254847646,
      "grad_norm": 0.8130463361740112,
      "learning_rate": 2.7537e-05,
      "loss": 0.0671,
      "step": 9182
    },
    {
      "epoch": 16.943674976915975,
      "grad_norm": 0.4148968756198883,
      "learning_rate": 2.754e-05,
      "loss": 0.071,
      "step": 9183
    },
    {
      "epoch": 16.945521698984304,
      "grad_norm": 0.4553473889827728,
      "learning_rate": 2.7543e-05,
      "loss": 0.0552,
      "step": 9184
    },
    {
      "epoch": 16.94736842105263,
      "grad_norm": 0.4178435802459717,
      "learning_rate": 2.7546e-05,
      "loss": 0.0193,
      "step": 9185
    },
    {
      "epoch": 16.94921514312096,
      "grad_norm": 0.4120926856994629,
      "learning_rate": 2.7549e-05,
      "loss": 0.0186,
      "step": 9186
    },
    {
      "epoch": 16.95106186518929,
      "grad_norm": 0.1929618865251541,
      "learning_rate": 2.7552e-05,
      "loss": 0.0131,
      "step": 9187
    },
    {
      "epoch": 16.952908587257618,
      "grad_norm": 0.387055367231369,
      "learning_rate": 2.7555e-05,
      "loss": 0.0126,
      "step": 9188
    },
    {
      "epoch": 16.954755309325947,
      "grad_norm": 0.17407526075839996,
      "learning_rate": 2.7558e-05,
      "loss": 0.0131,
      "step": 9189
    },
    {
      "epoch": 16.956602031394276,
      "grad_norm": 0.24599270522594452,
      "learning_rate": 2.7561e-05,
      "loss": 0.0164,
      "step": 9190
    },
    {
      "epoch": 16.958448753462605,
      "grad_norm": 0.19677557051181793,
      "learning_rate": 2.7564e-05,
      "loss": 0.0093,
      "step": 9191
    },
    {
      "epoch": 16.960295475530934,
      "grad_norm": 0.465720534324646,
      "learning_rate": 2.7567000000000002e-05,
      "loss": 0.0147,
      "step": 9192
    },
    {
      "epoch": 16.96214219759926,
      "grad_norm": 0.8160534501075745,
      "learning_rate": 2.7570000000000002e-05,
      "loss": 0.0268,
      "step": 9193
    },
    {
      "epoch": 16.96398891966759,
      "grad_norm": 0.19757269322872162,
      "learning_rate": 2.7573000000000002e-05,
      "loss": 0.008,
      "step": 9194
    },
    {
      "epoch": 16.965835641735918,
      "grad_norm": 0.3500001132488251,
      "learning_rate": 2.7576e-05,
      "loss": 0.0165,
      "step": 9195
    },
    {
      "epoch": 16.967682363804247,
      "grad_norm": 0.5525098443031311,
      "learning_rate": 2.7579e-05,
      "loss": 0.0607,
      "step": 9196
    },
    {
      "epoch": 16.969529085872576,
      "grad_norm": 0.6156070828437805,
      "learning_rate": 2.7582e-05,
      "loss": 0.0248,
      "step": 9197
    },
    {
      "epoch": 16.971375807940905,
      "grad_norm": 0.4163903594017029,
      "learning_rate": 2.7585e-05,
      "loss": 0.0157,
      "step": 9198
    },
    {
      "epoch": 16.973222530009235,
      "grad_norm": 0.3857148289680481,
      "learning_rate": 2.7588e-05,
      "loss": 0.0099,
      "step": 9199
    },
    {
      "epoch": 16.975069252077564,
      "grad_norm": 0.27057746052742004,
      "learning_rate": 2.7591e-05,
      "loss": 0.0098,
      "step": 9200
    },
    {
      "epoch": 16.97691597414589,
      "grad_norm": 0.4757899045944214,
      "learning_rate": 2.7594e-05,
      "loss": 0.0134,
      "step": 9201
    },
    {
      "epoch": 16.97876269621422,
      "grad_norm": 0.4091153144836426,
      "learning_rate": 2.7597000000000004e-05,
      "loss": 0.0117,
      "step": 9202
    },
    {
      "epoch": 16.980609418282548,
      "grad_norm": 0.16787925362586975,
      "learning_rate": 2.7600000000000003e-05,
      "loss": 0.0082,
      "step": 9203
    },
    {
      "epoch": 16.982456140350877,
      "grad_norm": 0.2270852029323578,
      "learning_rate": 2.7603000000000003e-05,
      "loss": 0.0098,
      "step": 9204
    },
    {
      "epoch": 16.984302862419206,
      "grad_norm": 0.2602526843547821,
      "learning_rate": 2.7606e-05,
      "loss": 0.0173,
      "step": 9205
    },
    {
      "epoch": 16.986149584487535,
      "grad_norm": 0.3468780219554901,
      "learning_rate": 2.7609e-05,
      "loss": 0.0138,
      "step": 9206
    },
    {
      "epoch": 16.987996306555864,
      "grad_norm": 0.1793440729379654,
      "learning_rate": 2.7612e-05,
      "loss": 0.0096,
      "step": 9207
    },
    {
      "epoch": 16.989843028624193,
      "grad_norm": 0.35675808787345886,
      "learning_rate": 2.7615e-05,
      "loss": 0.0139,
      "step": 9208
    },
    {
      "epoch": 16.991689750692522,
      "grad_norm": 0.7460472583770752,
      "learning_rate": 2.7618e-05,
      "loss": 0.0157,
      "step": 9209
    },
    {
      "epoch": 16.993536472760848,
      "grad_norm": 0.8922962546348572,
      "learning_rate": 2.7621e-05,
      "loss": 0.0165,
      "step": 9210
    },
    {
      "epoch": 16.995383194829177,
      "grad_norm": 0.4702712893486023,
      "learning_rate": 2.7624e-05,
      "loss": 0.0157,
      "step": 9211
    },
    {
      "epoch": 16.997229916897506,
      "grad_norm": 0.8927752375602722,
      "learning_rate": 2.7627e-05,
      "loss": 0.0223,
      "step": 9212
    },
    {
      "epoch": 16.999076638965835,
      "grad_norm": 0.5239241123199463,
      "learning_rate": 2.763e-05,
      "loss": 0.0123,
      "step": 9213
    },
    {
      "epoch": 17.0,
      "grad_norm": 1.5342310667037964,
      "learning_rate": 2.7633e-05,
      "loss": 0.0427,
      "step": 9214
    },
    {
      "epoch": 17.00184672206833,
      "grad_norm": 1.8125861883163452,
      "learning_rate": 2.7636e-05,
      "loss": 0.2042,
      "step": 9215
    },
    {
      "epoch": 17.00369344413666,
      "grad_norm": 0.9332644939422607,
      "learning_rate": 2.7639e-05,
      "loss": 0.1342,
      "step": 9216
    },
    {
      "epoch": 17.005540166204987,
      "grad_norm": 1.0487055778503418,
      "learning_rate": 2.7642e-05,
      "loss": 0.2083,
      "step": 9217
    },
    {
      "epoch": 17.007386888273317,
      "grad_norm": 0.5406859517097473,
      "learning_rate": 2.7645e-05,
      "loss": 0.1001,
      "step": 9218
    },
    {
      "epoch": 17.009233610341642,
      "grad_norm": 0.8032557964324951,
      "learning_rate": 2.7648e-05,
      "loss": 0.1462,
      "step": 9219
    },
    {
      "epoch": 17.01108033240997,
      "grad_norm": 0.7021965980529785,
      "learning_rate": 2.7651e-05,
      "loss": 0.0841,
      "step": 9220
    },
    {
      "epoch": 17.0129270544783,
      "grad_norm": 0.5755544900894165,
      "learning_rate": 2.7654e-05,
      "loss": 0.0954,
      "step": 9221
    },
    {
      "epoch": 17.01477377654663,
      "grad_norm": 0.6425787210464478,
      "learning_rate": 2.7657000000000003e-05,
      "loss": 0.0837,
      "step": 9222
    },
    {
      "epoch": 17.01662049861496,
      "grad_norm": 0.41704753041267395,
      "learning_rate": 2.7660000000000003e-05,
      "loss": 0.0527,
      "step": 9223
    },
    {
      "epoch": 17.018467220683288,
      "grad_norm": 0.5317704677581787,
      "learning_rate": 2.7663000000000003e-05,
      "loss": 0.0558,
      "step": 9224
    },
    {
      "epoch": 17.020313942751617,
      "grad_norm": 0.7003104090690613,
      "learning_rate": 2.7666000000000002e-05,
      "loss": 0.1217,
      "step": 9225
    },
    {
      "epoch": 17.022160664819946,
      "grad_norm": 0.38377368450164795,
      "learning_rate": 2.7669000000000002e-05,
      "loss": 0.0473,
      "step": 9226
    },
    {
      "epoch": 17.02400738688827,
      "grad_norm": 0.8063497543334961,
      "learning_rate": 2.7672000000000002e-05,
      "loss": 0.0967,
      "step": 9227
    },
    {
      "epoch": 17.0258541089566,
      "grad_norm": 0.7322953343391418,
      "learning_rate": 2.7675000000000002e-05,
      "loss": 0.1003,
      "step": 9228
    },
    {
      "epoch": 17.02770083102493,
      "grad_norm": 2.8443334102630615,
      "learning_rate": 2.7678e-05,
      "loss": 0.0507,
      "step": 9229
    },
    {
      "epoch": 17.02954755309326,
      "grad_norm": 0.2949460744857788,
      "learning_rate": 2.7680999999999998e-05,
      "loss": 0.0211,
      "step": 9230
    },
    {
      "epoch": 17.03139427516159,
      "grad_norm": 0.22077718377113342,
      "learning_rate": 2.7683999999999998e-05,
      "loss": 0.0123,
      "step": 9231
    },
    {
      "epoch": 17.033240997229917,
      "grad_norm": 0.21740186214447021,
      "learning_rate": 2.7687e-05,
      "loss": 0.0154,
      "step": 9232
    },
    {
      "epoch": 17.035087719298247,
      "grad_norm": 0.33870649337768555,
      "learning_rate": 2.769e-05,
      "loss": 0.0352,
      "step": 9233
    },
    {
      "epoch": 17.036934441366576,
      "grad_norm": 0.28785428404808044,
      "learning_rate": 2.7693e-05,
      "loss": 0.0124,
      "step": 9234
    },
    {
      "epoch": 17.0387811634349,
      "grad_norm": 0.25172266364097595,
      "learning_rate": 2.7696e-05,
      "loss": 0.0099,
      "step": 9235
    },
    {
      "epoch": 17.04062788550323,
      "grad_norm": 0.29542961716651917,
      "learning_rate": 2.7699e-05,
      "loss": 0.0116,
      "step": 9236
    },
    {
      "epoch": 17.04247460757156,
      "grad_norm": 0.3282698690891266,
      "learning_rate": 2.7702e-05,
      "loss": 0.0155,
      "step": 9237
    },
    {
      "epoch": 17.04432132963989,
      "grad_norm": 0.4257015883922577,
      "learning_rate": 2.7705e-05,
      "loss": 0.0243,
      "step": 9238
    },
    {
      "epoch": 17.046168051708218,
      "grad_norm": 0.22363309562206268,
      "learning_rate": 2.7708e-05,
      "loss": 0.0111,
      "step": 9239
    },
    {
      "epoch": 17.048014773776547,
      "grad_norm": 0.37432435154914856,
      "learning_rate": 2.7711e-05,
      "loss": 0.0154,
      "step": 9240
    },
    {
      "epoch": 17.049861495844876,
      "grad_norm": 0.27944135665893555,
      "learning_rate": 2.7714e-05,
      "loss": 0.0076,
      "step": 9241
    },
    {
      "epoch": 17.051708217913205,
      "grad_norm": 0.3036108911037445,
      "learning_rate": 2.7717000000000002e-05,
      "loss": 0.0108,
      "step": 9242
    },
    {
      "epoch": 17.053554939981534,
      "grad_norm": 0.34930723905563354,
      "learning_rate": 2.7720000000000002e-05,
      "loss": 0.013,
      "step": 9243
    },
    {
      "epoch": 17.05540166204986,
      "grad_norm": 0.3175663948059082,
      "learning_rate": 2.7723000000000002e-05,
      "loss": 0.0098,
      "step": 9244
    },
    {
      "epoch": 17.05724838411819,
      "grad_norm": 0.36189937591552734,
      "learning_rate": 2.7726000000000002e-05,
      "loss": 0.0155,
      "step": 9245
    },
    {
      "epoch": 17.05909510618652,
      "grad_norm": 0.2590970993041992,
      "learning_rate": 2.7729e-05,
      "loss": 0.0073,
      "step": 9246
    },
    {
      "epoch": 17.060941828254848,
      "grad_norm": 0.16251273453235626,
      "learning_rate": 2.7732e-05,
      "loss": 0.0042,
      "step": 9247
    },
    {
      "epoch": 17.062788550323177,
      "grad_norm": 0.28405195474624634,
      "learning_rate": 2.7735e-05,
      "loss": 0.009,
      "step": 9248
    },
    {
      "epoch": 17.064635272391506,
      "grad_norm": 0.3556298613548279,
      "learning_rate": 2.7738e-05,
      "loss": 0.0104,
      "step": 9249
    },
    {
      "epoch": 17.066481994459835,
      "grad_norm": 0.2897495627403259,
      "learning_rate": 2.7741e-05,
      "loss": 0.0141,
      "step": 9250
    },
    {
      "epoch": 17.068328716528164,
      "grad_norm": 0.5381990075111389,
      "learning_rate": 2.7744e-05,
      "loss": 0.0099,
      "step": 9251
    },
    {
      "epoch": 17.07017543859649,
      "grad_norm": 0.2446378916501999,
      "learning_rate": 2.7747000000000004e-05,
      "loss": 0.0105,
      "step": 9252
    },
    {
      "epoch": 17.07202216066482,
      "grad_norm": 0.4810533821582794,
      "learning_rate": 2.7750000000000004e-05,
      "loss": 0.0113,
      "step": 9253
    },
    {
      "epoch": 17.073868882733148,
      "grad_norm": 0.3621048629283905,
      "learning_rate": 2.7753e-05,
      "loss": 0.0106,
      "step": 9254
    },
    {
      "epoch": 17.075715604801477,
      "grad_norm": 0.2906154990196228,
      "learning_rate": 2.7756e-05,
      "loss": 0.0132,
      "step": 9255
    },
    {
      "epoch": 17.077562326869806,
      "grad_norm": 0.240259051322937,
      "learning_rate": 2.7759e-05,
      "loss": 0.006,
      "step": 9256
    },
    {
      "epoch": 17.079409048938135,
      "grad_norm": 0.35923489928245544,
      "learning_rate": 2.7762e-05,
      "loss": 0.0107,
      "step": 9257
    },
    {
      "epoch": 17.081255771006465,
      "grad_norm": 0.30543577671051025,
      "learning_rate": 2.7765e-05,
      "loss": 0.0044,
      "step": 9258
    },
    {
      "epoch": 17.083102493074794,
      "grad_norm": 0.3810657858848572,
      "learning_rate": 2.7768e-05,
      "loss": 0.0119,
      "step": 9259
    },
    {
      "epoch": 17.08494921514312,
      "grad_norm": 0.2955728769302368,
      "learning_rate": 2.7771e-05,
      "loss": 0.0096,
      "step": 9260
    },
    {
      "epoch": 17.08679593721145,
      "grad_norm": 0.9007036089897156,
      "learning_rate": 2.7774e-05,
      "loss": 0.0328,
      "step": 9261
    },
    {
      "epoch": 17.088642659279778,
      "grad_norm": 0.3021596074104309,
      "learning_rate": 2.7777e-05,
      "loss": 0.009,
      "step": 9262
    },
    {
      "epoch": 17.090489381348107,
      "grad_norm": 0.39645281434059143,
      "learning_rate": 2.778e-05,
      "loss": 0.0103,
      "step": 9263
    },
    {
      "epoch": 17.092336103416436,
      "grad_norm": 0.45379069447517395,
      "learning_rate": 2.7783e-05,
      "loss": 0.0064,
      "step": 9264
    },
    {
      "epoch": 17.094182825484765,
      "grad_norm": 0.7573075294494629,
      "learning_rate": 2.7786e-05,
      "loss": 0.1849,
      "step": 9265
    },
    {
      "epoch": 17.096029547553094,
      "grad_norm": 0.6216201186180115,
      "learning_rate": 2.7789e-05,
      "loss": 0.1987,
      "step": 9266
    },
    {
      "epoch": 17.097876269621423,
      "grad_norm": 0.5163629651069641,
      "learning_rate": 2.7792e-05,
      "loss": 0.1449,
      "step": 9267
    },
    {
      "epoch": 17.099722991689752,
      "grad_norm": 0.5870918035507202,
      "learning_rate": 2.7795e-05,
      "loss": 0.1147,
      "step": 9268
    },
    {
      "epoch": 17.101569713758078,
      "grad_norm": 0.5727010369300842,
      "learning_rate": 2.7798e-05,
      "loss": 0.1186,
      "step": 9269
    },
    {
      "epoch": 17.103416435826407,
      "grad_norm": 0.4444744884967804,
      "learning_rate": 2.7801e-05,
      "loss": 0.0816,
      "step": 9270
    },
    {
      "epoch": 17.105263157894736,
      "grad_norm": 0.5740081071853638,
      "learning_rate": 2.7804e-05,
      "loss": 0.0998,
      "step": 9271
    },
    {
      "epoch": 17.107109879963065,
      "grad_norm": 0.3582746684551239,
      "learning_rate": 2.7807e-05,
      "loss": 0.0796,
      "step": 9272
    },
    {
      "epoch": 17.108956602031395,
      "grad_norm": 0.5588990449905396,
      "learning_rate": 2.7810000000000003e-05,
      "loss": 0.0924,
      "step": 9273
    },
    {
      "epoch": 17.110803324099724,
      "grad_norm": 0.31783145666122437,
      "learning_rate": 2.7813000000000003e-05,
      "loss": 0.0601,
      "step": 9274
    },
    {
      "epoch": 17.112650046168053,
      "grad_norm": 0.553686797618866,
      "learning_rate": 2.7816000000000003e-05,
      "loss": 0.0395,
      "step": 9275
    },
    {
      "epoch": 17.114496768236382,
      "grad_norm": 0.28794318437576294,
      "learning_rate": 2.7819000000000002e-05,
      "loss": 0.0467,
      "step": 9276
    },
    {
      "epoch": 17.116343490304708,
      "grad_norm": 1.769435167312622,
      "learning_rate": 2.7822000000000002e-05,
      "loss": 0.0312,
      "step": 9277
    },
    {
      "epoch": 17.118190212373037,
      "grad_norm": 0.4049617350101471,
      "learning_rate": 2.7825000000000002e-05,
      "loss": 0.0172,
      "step": 9278
    },
    {
      "epoch": 17.120036934441366,
      "grad_norm": 0.41664329171180725,
      "learning_rate": 2.7828e-05,
      "loss": 0.0386,
      "step": 9279
    },
    {
      "epoch": 17.121883656509695,
      "grad_norm": 0.31309565901756287,
      "learning_rate": 2.7831e-05,
      "loss": 0.0355,
      "step": 9280
    },
    {
      "epoch": 17.123730378578024,
      "grad_norm": 0.26691946387290955,
      "learning_rate": 2.7833999999999998e-05,
      "loss": 0.0302,
      "step": 9281
    },
    {
      "epoch": 17.125577100646353,
      "grad_norm": 0.3606624901294708,
      "learning_rate": 2.7836999999999998e-05,
      "loss": 0.0172,
      "step": 9282
    },
    {
      "epoch": 17.127423822714682,
      "grad_norm": 0.29028576612472534,
      "learning_rate": 2.784e-05,
      "loss": 0.016,
      "step": 9283
    },
    {
      "epoch": 17.12927054478301,
      "grad_norm": 0.32634156942367554,
      "learning_rate": 2.7843e-05,
      "loss": 0.0746,
      "step": 9284
    },
    {
      "epoch": 17.131117266851337,
      "grad_norm": 0.297480970621109,
      "learning_rate": 2.7846e-05,
      "loss": 0.0164,
      "step": 9285
    },
    {
      "epoch": 17.132963988919666,
      "grad_norm": 0.3533071279525757,
      "learning_rate": 2.7849e-05,
      "loss": 0.0116,
      "step": 9286
    },
    {
      "epoch": 17.134810710987995,
      "grad_norm": 0.33252015709877014,
      "learning_rate": 2.7852e-05,
      "loss": 0.0139,
      "step": 9287
    },
    {
      "epoch": 17.136657433056325,
      "grad_norm": 0.5829310417175293,
      "learning_rate": 2.7855e-05,
      "loss": 0.0167,
      "step": 9288
    },
    {
      "epoch": 17.138504155124654,
      "grad_norm": 0.5431162118911743,
      "learning_rate": 2.7858e-05,
      "loss": 0.0258,
      "step": 9289
    },
    {
      "epoch": 17.140350877192983,
      "grad_norm": 0.279328852891922,
      "learning_rate": 2.7861e-05,
      "loss": 0.0136,
      "step": 9290
    },
    {
      "epoch": 17.142197599261312,
      "grad_norm": 0.31380873918533325,
      "learning_rate": 2.7864e-05,
      "loss": 0.0082,
      "step": 9291
    },
    {
      "epoch": 17.14404432132964,
      "grad_norm": 0.4092232286930084,
      "learning_rate": 2.7867e-05,
      "loss": 0.0074,
      "step": 9292
    },
    {
      "epoch": 17.14589104339797,
      "grad_norm": 0.29413503408432007,
      "learning_rate": 2.7870000000000003e-05,
      "loss": 0.0088,
      "step": 9293
    },
    {
      "epoch": 17.147737765466296,
      "grad_norm": 0.23219628632068634,
      "learning_rate": 2.7873000000000002e-05,
      "loss": 0.0103,
      "step": 9294
    },
    {
      "epoch": 17.149584487534625,
      "grad_norm": 1.2620007991790771,
      "learning_rate": 2.7876000000000002e-05,
      "loss": 0.0213,
      "step": 9295
    },
    {
      "epoch": 17.151431209602954,
      "grad_norm": 0.24481666088104248,
      "learning_rate": 2.7879000000000002e-05,
      "loss": 0.0053,
      "step": 9296
    },
    {
      "epoch": 17.153277931671283,
      "grad_norm": 0.30119237303733826,
      "learning_rate": 2.7882000000000002e-05,
      "loss": 0.0088,
      "step": 9297
    },
    {
      "epoch": 17.155124653739612,
      "grad_norm": 0.6697653532028198,
      "learning_rate": 2.7885e-05,
      "loss": 0.0179,
      "step": 9298
    },
    {
      "epoch": 17.15697137580794,
      "grad_norm": 0.3480161726474762,
      "learning_rate": 2.7888e-05,
      "loss": 0.0159,
      "step": 9299
    },
    {
      "epoch": 17.15881809787627,
      "grad_norm": 0.22501026093959808,
      "learning_rate": 2.7891e-05,
      "loss": 0.0094,
      "step": 9300
    },
    {
      "epoch": 17.1606648199446,
      "grad_norm": 0.37073078751564026,
      "learning_rate": 2.7894e-05,
      "loss": 0.0166,
      "step": 9301
    },
    {
      "epoch": 17.162511542012926,
      "grad_norm": 0.34021317958831787,
      "learning_rate": 2.7897e-05,
      "loss": 0.0103,
      "step": 9302
    },
    {
      "epoch": 17.164358264081255,
      "grad_norm": 0.7827439308166504,
      "learning_rate": 2.79e-05,
      "loss": 0.0143,
      "step": 9303
    },
    {
      "epoch": 17.166204986149584,
      "grad_norm": 0.39998817443847656,
      "learning_rate": 2.7903e-05,
      "loss": 0.0173,
      "step": 9304
    },
    {
      "epoch": 17.168051708217913,
      "grad_norm": 0.3274659812450409,
      "learning_rate": 2.7906e-05,
      "loss": 0.0089,
      "step": 9305
    },
    {
      "epoch": 17.169898430286242,
      "grad_norm": 0.2546210289001465,
      "learning_rate": 2.7909e-05,
      "loss": 0.0118,
      "step": 9306
    },
    {
      "epoch": 17.17174515235457,
      "grad_norm": 0.26053136587142944,
      "learning_rate": 2.7912e-05,
      "loss": 0.0078,
      "step": 9307
    },
    {
      "epoch": 17.1735918744229,
      "grad_norm": 0.4792042374610901,
      "learning_rate": 2.7915e-05,
      "loss": 0.0164,
      "step": 9308
    },
    {
      "epoch": 17.17543859649123,
      "grad_norm": 0.540781557559967,
      "learning_rate": 2.7918e-05,
      "loss": 0.0245,
      "step": 9309
    },
    {
      "epoch": 17.177285318559555,
      "grad_norm": 0.5518727898597717,
      "learning_rate": 2.7921e-05,
      "loss": 0.0173,
      "step": 9310
    },
    {
      "epoch": 17.179132040627884,
      "grad_norm": 0.5841360092163086,
      "learning_rate": 2.7924e-05,
      "loss": 0.0098,
      "step": 9311
    },
    {
      "epoch": 17.180978762696213,
      "grad_norm": 0.9632762670516968,
      "learning_rate": 2.7927e-05,
      "loss": 0.0153,
      "step": 9312
    },
    {
      "epoch": 17.182825484764543,
      "grad_norm": 0.4417622983455658,
      "learning_rate": 2.7930000000000002e-05,
      "loss": 0.0097,
      "step": 9313
    },
    {
      "epoch": 17.18467220683287,
      "grad_norm": 0.6527366042137146,
      "learning_rate": 2.7933000000000002e-05,
      "loss": 0.0261,
      "step": 9314
    },
    {
      "epoch": 17.1865189289012,
      "grad_norm": 0.8529950976371765,
      "learning_rate": 2.7936e-05,
      "loss": 0.3008,
      "step": 9315
    },
    {
      "epoch": 17.18836565096953,
      "grad_norm": 0.5253120064735413,
      "learning_rate": 2.7939e-05,
      "loss": 0.1189,
      "step": 9316
    },
    {
      "epoch": 17.19021237303786,
      "grad_norm": 0.4528539776802063,
      "learning_rate": 2.7942e-05,
      "loss": 0.1442,
      "step": 9317
    },
    {
      "epoch": 17.19205909510619,
      "grad_norm": 0.7505183219909668,
      "learning_rate": 2.7945e-05,
      "loss": 0.153,
      "step": 9318
    },
    {
      "epoch": 17.193905817174514,
      "grad_norm": 0.831365168094635,
      "learning_rate": 2.7948e-05,
      "loss": 0.1751,
      "step": 9319
    },
    {
      "epoch": 17.195752539242843,
      "grad_norm": 0.8180812001228333,
      "learning_rate": 2.7951e-05,
      "loss": 0.1076,
      "step": 9320
    },
    {
      "epoch": 17.197599261311172,
      "grad_norm": 0.5121163725852966,
      "learning_rate": 2.7954e-05,
      "loss": 0.0991,
      "step": 9321
    },
    {
      "epoch": 17.1994459833795,
      "grad_norm": 0.3973347544670105,
      "learning_rate": 2.7957e-05,
      "loss": 0.0564,
      "step": 9322
    },
    {
      "epoch": 17.20129270544783,
      "grad_norm": 0.5886101126670837,
      "learning_rate": 2.7960000000000003e-05,
      "loss": 0.0696,
      "step": 9323
    },
    {
      "epoch": 17.20313942751616,
      "grad_norm": 0.38349026441574097,
      "learning_rate": 2.7963000000000003e-05,
      "loss": 0.0792,
      "step": 9324
    },
    {
      "epoch": 17.20498614958449,
      "grad_norm": 0.4047221541404724,
      "learning_rate": 2.7966000000000003e-05,
      "loss": 0.0519,
      "step": 9325
    },
    {
      "epoch": 17.206832871652818,
      "grad_norm": 0.3595011532306671,
      "learning_rate": 2.7969000000000003e-05,
      "loss": 0.0392,
      "step": 9326
    },
    {
      "epoch": 17.208679593721143,
      "grad_norm": 0.294626384973526,
      "learning_rate": 2.7972000000000003e-05,
      "loss": 0.0484,
      "step": 9327
    },
    {
      "epoch": 17.210526315789473,
      "grad_norm": 0.3707464933395386,
      "learning_rate": 2.7975e-05,
      "loss": 0.0366,
      "step": 9328
    },
    {
      "epoch": 17.2123730378578,
      "grad_norm": 0.3111836314201355,
      "learning_rate": 2.7978e-05,
      "loss": 0.0371,
      "step": 9329
    },
    {
      "epoch": 17.21421975992613,
      "grad_norm": 0.19050897657871246,
      "learning_rate": 2.7981e-05,
      "loss": 0.0243,
      "step": 9330
    },
    {
      "epoch": 17.21606648199446,
      "grad_norm": 0.2788408100605011,
      "learning_rate": 2.7984e-05,
      "loss": 0.0133,
      "step": 9331
    },
    {
      "epoch": 17.21791320406279,
      "grad_norm": 0.3897608518600464,
      "learning_rate": 2.7986999999999998e-05,
      "loss": 0.0149,
      "step": 9332
    },
    {
      "epoch": 17.21975992613112,
      "grad_norm": 0.24043641984462738,
      "learning_rate": 2.799e-05,
      "loss": 0.0417,
      "step": 9333
    },
    {
      "epoch": 17.221606648199447,
      "grad_norm": 0.3331644833087921,
      "learning_rate": 2.7993e-05,
      "loss": 0.0124,
      "step": 9334
    },
    {
      "epoch": 17.223453370267773,
      "grad_norm": 0.2830711901187897,
      "learning_rate": 2.7996e-05,
      "loss": 0.012,
      "step": 9335
    },
    {
      "epoch": 17.225300092336102,
      "grad_norm": 0.21424072980880737,
      "learning_rate": 2.7999e-05,
      "loss": 0.0065,
      "step": 9336
    },
    {
      "epoch": 17.22714681440443,
      "grad_norm": 0.6466320753097534,
      "learning_rate": 2.8002e-05,
      "loss": 0.0103,
      "step": 9337
    },
    {
      "epoch": 17.22899353647276,
      "grad_norm": 0.2295599728822708,
      "learning_rate": 2.8005e-05,
      "loss": 0.01,
      "step": 9338
    },
    {
      "epoch": 17.23084025854109,
      "grad_norm": 0.2938486635684967,
      "learning_rate": 2.8008e-05,
      "loss": 0.0096,
      "step": 9339
    },
    {
      "epoch": 17.23268698060942,
      "grad_norm": 0.17863905429840088,
      "learning_rate": 2.8011e-05,
      "loss": 0.0069,
      "step": 9340
    },
    {
      "epoch": 17.234533702677748,
      "grad_norm": 0.1935294270515442,
      "learning_rate": 2.8014e-05,
      "loss": 0.0125,
      "step": 9341
    },
    {
      "epoch": 17.236380424746077,
      "grad_norm": 0.9676811099052429,
      "learning_rate": 2.8017e-05,
      "loss": 0.0269,
      "step": 9342
    },
    {
      "epoch": 17.238227146814406,
      "grad_norm": 0.45683619379997253,
      "learning_rate": 2.8020000000000003e-05,
      "loss": 0.0171,
      "step": 9343
    },
    {
      "epoch": 17.24007386888273,
      "grad_norm": 0.40154099464416504,
      "learning_rate": 2.8023000000000003e-05,
      "loss": 0.0114,
      "step": 9344
    },
    {
      "epoch": 17.24192059095106,
      "grad_norm": 0.4316021203994751,
      "learning_rate": 2.8026000000000002e-05,
      "loss": 0.0365,
      "step": 9345
    },
    {
      "epoch": 17.24376731301939,
      "grad_norm": 0.3077472448348999,
      "learning_rate": 2.8029000000000002e-05,
      "loss": 0.0118,
      "step": 9346
    },
    {
      "epoch": 17.24561403508772,
      "grad_norm": 0.49241816997528076,
      "learning_rate": 2.8032000000000002e-05,
      "loss": 0.0107,
      "step": 9347
    },
    {
      "epoch": 17.24746075715605,
      "grad_norm": 0.3258560299873352,
      "learning_rate": 2.8035000000000002e-05,
      "loss": 0.0114,
      "step": 9348
    },
    {
      "epoch": 17.249307479224377,
      "grad_norm": 0.33918431401252747,
      "learning_rate": 2.8038e-05,
      "loss": 0.0141,
      "step": 9349
    },
    {
      "epoch": 17.251154201292707,
      "grad_norm": 0.29143860936164856,
      "learning_rate": 2.8041e-05,
      "loss": 0.0134,
      "step": 9350
    },
    {
      "epoch": 17.253000923361036,
      "grad_norm": 0.39543747901916504,
      "learning_rate": 2.8044e-05,
      "loss": 0.0145,
      "step": 9351
    },
    {
      "epoch": 17.25484764542936,
      "grad_norm": 0.2542322874069214,
      "learning_rate": 2.8047e-05,
      "loss": 0.0113,
      "step": 9352
    },
    {
      "epoch": 17.25669436749769,
      "grad_norm": 0.28918883204460144,
      "learning_rate": 2.805e-05,
      "loss": 0.0063,
      "step": 9353
    },
    {
      "epoch": 17.25854108956602,
      "grad_norm": 0.5130068063735962,
      "learning_rate": 2.8053e-05,
      "loss": 0.0175,
      "step": 9354
    },
    {
      "epoch": 17.26038781163435,
      "grad_norm": 0.6604469418525696,
      "learning_rate": 2.8056e-05,
      "loss": 0.0121,
      "step": 9355
    },
    {
      "epoch": 17.262234533702678,
      "grad_norm": 0.21387112140655518,
      "learning_rate": 2.8059e-05,
      "loss": 0.0078,
      "step": 9356
    },
    {
      "epoch": 17.264081255771007,
      "grad_norm": 0.7465587854385376,
      "learning_rate": 2.8062e-05,
      "loss": 0.022,
      "step": 9357
    },
    {
      "epoch": 17.265927977839336,
      "grad_norm": 0.9279031157493591,
      "learning_rate": 2.8065e-05,
      "loss": 0.0261,
      "step": 9358
    },
    {
      "epoch": 17.267774699907665,
      "grad_norm": 0.23070333898067474,
      "learning_rate": 2.8068e-05,
      "loss": 0.0071,
      "step": 9359
    },
    {
      "epoch": 17.26962142197599,
      "grad_norm": 0.3000924587249756,
      "learning_rate": 2.8071e-05,
      "loss": 0.0152,
      "step": 9360
    },
    {
      "epoch": 17.27146814404432,
      "grad_norm": 0.36132705211639404,
      "learning_rate": 2.8074e-05,
      "loss": 0.0099,
      "step": 9361
    },
    {
      "epoch": 17.27331486611265,
      "grad_norm": 0.3235369920730591,
      "learning_rate": 2.8077e-05,
      "loss": 0.0123,
      "step": 9362
    },
    {
      "epoch": 17.27516158818098,
      "grad_norm": 3.1168806552886963,
      "learning_rate": 2.8080000000000002e-05,
      "loss": 0.0307,
      "step": 9363
    },
    {
      "epoch": 17.277008310249307,
      "grad_norm": 0.44572263956069946,
      "learning_rate": 2.8083000000000002e-05,
      "loss": 0.0152,
      "step": 9364
    },
    {
      "epoch": 17.278855032317637,
      "grad_norm": 1.7113220691680908,
      "learning_rate": 2.8086000000000002e-05,
      "loss": 0.1906,
      "step": 9365
    },
    {
      "epoch": 17.280701754385966,
      "grad_norm": 0.5876774191856384,
      "learning_rate": 2.8089e-05,
      "loss": 0.1274,
      "step": 9366
    },
    {
      "epoch": 17.282548476454295,
      "grad_norm": 0.8600068688392639,
      "learning_rate": 2.8092e-05,
      "loss": 0.1518,
      "step": 9367
    },
    {
      "epoch": 17.284395198522624,
      "grad_norm": 0.47339197993278503,
      "learning_rate": 2.8095e-05,
      "loss": 0.0893,
      "step": 9368
    },
    {
      "epoch": 17.28624192059095,
      "grad_norm": 0.4132726192474365,
      "learning_rate": 2.8098e-05,
      "loss": 0.0815,
      "step": 9369
    },
    {
      "epoch": 17.28808864265928,
      "grad_norm": 0.4227193295955658,
      "learning_rate": 2.8101e-05,
      "loss": 0.0712,
      "step": 9370
    },
    {
      "epoch": 17.289935364727608,
      "grad_norm": 0.45899859070777893,
      "learning_rate": 2.8104e-05,
      "loss": 0.0947,
      "step": 9371
    },
    {
      "epoch": 17.291782086795937,
      "grad_norm": 0.4160376489162445,
      "learning_rate": 2.8107e-05,
      "loss": 0.0745,
      "step": 9372
    },
    {
      "epoch": 17.293628808864266,
      "grad_norm": 0.3612839877605438,
      "learning_rate": 2.8110000000000004e-05,
      "loss": 0.0645,
      "step": 9373
    },
    {
      "epoch": 17.295475530932595,
      "grad_norm": 0.34148138761520386,
      "learning_rate": 2.8113000000000003e-05,
      "loss": 0.0461,
      "step": 9374
    },
    {
      "epoch": 17.297322253000925,
      "grad_norm": 0.4269258975982666,
      "learning_rate": 2.8116000000000003e-05,
      "loss": 0.1095,
      "step": 9375
    },
    {
      "epoch": 17.299168975069254,
      "grad_norm": 0.3814285695552826,
      "learning_rate": 2.8119000000000003e-05,
      "loss": 0.0368,
      "step": 9376
    },
    {
      "epoch": 17.30101569713758,
      "grad_norm": 0.7113535404205322,
      "learning_rate": 2.8122e-05,
      "loss": 0.0425,
      "step": 9377
    },
    {
      "epoch": 17.30286241920591,
      "grad_norm": 0.40248632431030273,
      "learning_rate": 2.8125e-05,
      "loss": 0.0601,
      "step": 9378
    },
    {
      "epoch": 17.304709141274238,
      "grad_norm": 0.3561994433403015,
      "learning_rate": 2.8128e-05,
      "loss": 0.0527,
      "step": 9379
    },
    {
      "epoch": 17.306555863342567,
      "grad_norm": 0.21789389848709106,
      "learning_rate": 2.8131e-05,
      "loss": 0.0346,
      "step": 9380
    },
    {
      "epoch": 17.308402585410896,
      "grad_norm": 0.4023694396018982,
      "learning_rate": 2.8134e-05,
      "loss": 0.0165,
      "step": 9381
    },
    {
      "epoch": 17.310249307479225,
      "grad_norm": 0.5231220722198486,
      "learning_rate": 2.8137e-05,
      "loss": 0.0317,
      "step": 9382
    },
    {
      "epoch": 17.312096029547554,
      "grad_norm": 0.3214775025844574,
      "learning_rate": 2.8139999999999998e-05,
      "loss": 0.0127,
      "step": 9383
    },
    {
      "epoch": 17.313942751615883,
      "grad_norm": 0.2625340223312378,
      "learning_rate": 2.8143e-05,
      "loss": 0.0223,
      "step": 9384
    },
    {
      "epoch": 17.31578947368421,
      "grad_norm": 0.41229403018951416,
      "learning_rate": 2.8146e-05,
      "loss": 0.0202,
      "step": 9385
    },
    {
      "epoch": 17.317636195752538,
      "grad_norm": 0.37261727452278137,
      "learning_rate": 2.8149e-05,
      "loss": 0.0201,
      "step": 9386
    },
    {
      "epoch": 17.319482917820867,
      "grad_norm": 0.16855792701244354,
      "learning_rate": 2.8152e-05,
      "loss": 0.0102,
      "step": 9387
    },
    {
      "epoch": 17.321329639889196,
      "grad_norm": 0.3535830080509186,
      "learning_rate": 2.8155e-05,
      "loss": 0.0155,
      "step": 9388
    },
    {
      "epoch": 17.323176361957525,
      "grad_norm": 0.3080908954143524,
      "learning_rate": 2.8158e-05,
      "loss": 0.02,
      "step": 9389
    },
    {
      "epoch": 17.325023084025855,
      "grad_norm": 0.32726743817329407,
      "learning_rate": 2.8161e-05,
      "loss": 0.0166,
      "step": 9390
    },
    {
      "epoch": 17.326869806094184,
      "grad_norm": 1.4310134649276733,
      "learning_rate": 2.8164e-05,
      "loss": 0.0222,
      "step": 9391
    },
    {
      "epoch": 17.328716528162513,
      "grad_norm": 0.4071342349052429,
      "learning_rate": 2.8167e-05,
      "loss": 0.0298,
      "step": 9392
    },
    {
      "epoch": 17.330563250230842,
      "grad_norm": 0.2208002209663391,
      "learning_rate": 2.817e-05,
      "loss": 0.0117,
      "step": 9393
    },
    {
      "epoch": 17.332409972299168,
      "grad_norm": 0.3327980637550354,
      "learning_rate": 2.8173000000000003e-05,
      "loss": 0.0147,
      "step": 9394
    },
    {
      "epoch": 17.334256694367497,
      "grad_norm": 0.7948895692825317,
      "learning_rate": 2.8176000000000003e-05,
      "loss": 0.0258,
      "step": 9395
    },
    {
      "epoch": 17.336103416435826,
      "grad_norm": 0.2921746075153351,
      "learning_rate": 2.8179000000000002e-05,
      "loss": 0.0074,
      "step": 9396
    },
    {
      "epoch": 17.337950138504155,
      "grad_norm": 0.6576547622680664,
      "learning_rate": 2.8182000000000002e-05,
      "loss": 0.0144,
      "step": 9397
    },
    {
      "epoch": 17.339796860572484,
      "grad_norm": 0.41292092204093933,
      "learning_rate": 2.8185000000000002e-05,
      "loss": 0.0158,
      "step": 9398
    },
    {
      "epoch": 17.341643582640813,
      "grad_norm": 0.2634788453578949,
      "learning_rate": 2.8188000000000002e-05,
      "loss": 0.0077,
      "step": 9399
    },
    {
      "epoch": 17.343490304709142,
      "grad_norm": 0.42348232865333557,
      "learning_rate": 2.8191e-05,
      "loss": 0.0092,
      "step": 9400
    },
    {
      "epoch": 17.34533702677747,
      "grad_norm": 0.7069520354270935,
      "learning_rate": 2.8194e-05,
      "loss": 0.0156,
      "step": 9401
    },
    {
      "epoch": 17.347183748845797,
      "grad_norm": 0.6102639436721802,
      "learning_rate": 2.8196999999999998e-05,
      "loss": 0.0219,
      "step": 9402
    },
    {
      "epoch": 17.349030470914126,
      "grad_norm": 0.5276309251785278,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 0.0192,
      "step": 9403
    },
    {
      "epoch": 17.350877192982455,
      "grad_norm": 0.24854065477848053,
      "learning_rate": 2.8203e-05,
      "loss": 0.0119,
      "step": 9404
    },
    {
      "epoch": 17.352723915050785,
      "grad_norm": 0.7557331919670105,
      "learning_rate": 2.8206e-05,
      "loss": 0.0199,
      "step": 9405
    },
    {
      "epoch": 17.354570637119114,
      "grad_norm": 0.5607580542564392,
      "learning_rate": 2.8209e-05,
      "loss": 0.0141,
      "step": 9406
    },
    {
      "epoch": 17.356417359187443,
      "grad_norm": 0.6083971858024597,
      "learning_rate": 2.8212e-05,
      "loss": 0.0179,
      "step": 9407
    },
    {
      "epoch": 17.358264081255772,
      "grad_norm": 0.517368733882904,
      "learning_rate": 2.8215e-05,
      "loss": 0.0138,
      "step": 9408
    },
    {
      "epoch": 17.3601108033241,
      "grad_norm": 0.20876462757587433,
      "learning_rate": 2.8218e-05,
      "loss": 0.0079,
      "step": 9409
    },
    {
      "epoch": 17.361957525392427,
      "grad_norm": 0.28480789065361023,
      "learning_rate": 2.8221e-05,
      "loss": 0.0075,
      "step": 9410
    },
    {
      "epoch": 17.363804247460756,
      "grad_norm": 0.3195708692073822,
      "learning_rate": 2.8224e-05,
      "loss": 0.0121,
      "step": 9411
    },
    {
      "epoch": 17.365650969529085,
      "grad_norm": 0.6342963576316833,
      "learning_rate": 2.8227e-05,
      "loss": 0.0279,
      "step": 9412
    },
    {
      "epoch": 17.367497691597414,
      "grad_norm": 0.44565221667289734,
      "learning_rate": 2.823e-05,
      "loss": 0.0132,
      "step": 9413
    },
    {
      "epoch": 17.369344413665743,
      "grad_norm": 1.350311517715454,
      "learning_rate": 2.8233000000000002e-05,
      "loss": 0.031,
      "step": 9414
    },
    {
      "epoch": 17.371191135734072,
      "grad_norm": 0.8127965331077576,
      "learning_rate": 2.8236000000000002e-05,
      "loss": 0.2402,
      "step": 9415
    },
    {
      "epoch": 17.3730378578024,
      "grad_norm": 0.6130790710449219,
      "learning_rate": 2.8239000000000002e-05,
      "loss": 0.2571,
      "step": 9416
    },
    {
      "epoch": 17.37488457987073,
      "grad_norm": 0.7279358506202698,
      "learning_rate": 2.8242e-05,
      "loss": 0.1271,
      "step": 9417
    },
    {
      "epoch": 17.37673130193906,
      "grad_norm": 0.730381429195404,
      "learning_rate": 2.8245e-05,
      "loss": 0.112,
      "step": 9418
    },
    {
      "epoch": 17.378578024007385,
      "grad_norm": 0.5133957266807556,
      "learning_rate": 2.8248e-05,
      "loss": 0.092,
      "step": 9419
    },
    {
      "epoch": 17.380424746075715,
      "grad_norm": 0.6243531107902527,
      "learning_rate": 2.8251e-05,
      "loss": 0.1239,
      "step": 9420
    },
    {
      "epoch": 17.382271468144044,
      "grad_norm": 0.536197304725647,
      "learning_rate": 2.8254e-05,
      "loss": 0.0683,
      "step": 9421
    },
    {
      "epoch": 17.384118190212373,
      "grad_norm": 0.5808553099632263,
      "learning_rate": 2.8257e-05,
      "loss": 0.061,
      "step": 9422
    },
    {
      "epoch": 17.385964912280702,
      "grad_norm": 0.5259320139884949,
      "learning_rate": 2.826e-05,
      "loss": 0.087,
      "step": 9423
    },
    {
      "epoch": 17.38781163434903,
      "grad_norm": 0.3729979991912842,
      "learning_rate": 2.8263000000000004e-05,
      "loss": 0.049,
      "step": 9424
    },
    {
      "epoch": 17.38965835641736,
      "grad_norm": 0.4005459249019623,
      "learning_rate": 2.8266000000000003e-05,
      "loss": 0.0563,
      "step": 9425
    },
    {
      "epoch": 17.39150507848569,
      "grad_norm": 0.8215917348861694,
      "learning_rate": 2.8269e-05,
      "loss": 0.0378,
      "step": 9426
    },
    {
      "epoch": 17.393351800554015,
      "grad_norm": 1.047788381576538,
      "learning_rate": 2.8272e-05,
      "loss": 0.1099,
      "step": 9427
    },
    {
      "epoch": 17.395198522622344,
      "grad_norm": 0.41913044452667236,
      "learning_rate": 2.8275e-05,
      "loss": 0.0354,
      "step": 9428
    },
    {
      "epoch": 17.397045244690673,
      "grad_norm": 0.29535073041915894,
      "learning_rate": 2.8278e-05,
      "loss": 0.0308,
      "step": 9429
    },
    {
      "epoch": 17.398891966759003,
      "grad_norm": 0.43361786007881165,
      "learning_rate": 2.8281e-05,
      "loss": 0.0184,
      "step": 9430
    },
    {
      "epoch": 17.40073868882733,
      "grad_norm": 0.2104760706424713,
      "learning_rate": 2.8284e-05,
      "loss": 0.0098,
      "step": 9431
    },
    {
      "epoch": 17.40258541089566,
      "grad_norm": 0.640812337398529,
      "learning_rate": 2.8287e-05,
      "loss": 0.0366,
      "step": 9432
    },
    {
      "epoch": 17.40443213296399,
      "grad_norm": 0.6072062253952026,
      "learning_rate": 2.829e-05,
      "loss": 0.0237,
      "step": 9433
    },
    {
      "epoch": 17.40627885503232,
      "grad_norm": 0.22999133169651031,
      "learning_rate": 2.8293e-05,
      "loss": 0.0106,
      "step": 9434
    },
    {
      "epoch": 17.408125577100645,
      "grad_norm": 0.360470712184906,
      "learning_rate": 2.8296e-05,
      "loss": 0.0168,
      "step": 9435
    },
    {
      "epoch": 17.409972299168974,
      "grad_norm": 0.22122471034526825,
      "learning_rate": 2.8299e-05,
      "loss": 0.0184,
      "step": 9436
    },
    {
      "epoch": 17.411819021237303,
      "grad_norm": 0.23513934016227722,
      "learning_rate": 2.8302e-05,
      "loss": 0.0137,
      "step": 9437
    },
    {
      "epoch": 17.413665743305632,
      "grad_norm": 0.4968354403972626,
      "learning_rate": 2.8305e-05,
      "loss": 0.0186,
      "step": 9438
    },
    {
      "epoch": 17.41551246537396,
      "grad_norm": 0.4751613438129425,
      "learning_rate": 2.8308e-05,
      "loss": 0.0163,
      "step": 9439
    },
    {
      "epoch": 17.41735918744229,
      "grad_norm": 0.3601779043674469,
      "learning_rate": 2.8311e-05,
      "loss": 0.0131,
      "step": 9440
    },
    {
      "epoch": 17.41920590951062,
      "grad_norm": 1.2434005737304688,
      "learning_rate": 2.8314e-05,
      "loss": 0.0195,
      "step": 9441
    },
    {
      "epoch": 17.42105263157895,
      "grad_norm": 0.6260395646095276,
      "learning_rate": 2.8317e-05,
      "loss": 0.0097,
      "step": 9442
    },
    {
      "epoch": 17.422899353647278,
      "grad_norm": 0.3804498612880707,
      "learning_rate": 2.832e-05,
      "loss": 0.0171,
      "step": 9443
    },
    {
      "epoch": 17.424746075715603,
      "grad_norm": 0.6318892240524292,
      "learning_rate": 2.8323000000000003e-05,
      "loss": 0.0204,
      "step": 9444
    },
    {
      "epoch": 17.426592797783933,
      "grad_norm": 0.48028627038002014,
      "learning_rate": 2.8326000000000003e-05,
      "loss": 0.0147,
      "step": 9445
    },
    {
      "epoch": 17.42843951985226,
      "grad_norm": 0.19748181104660034,
      "learning_rate": 2.8329000000000003e-05,
      "loss": 0.0088,
      "step": 9446
    },
    {
      "epoch": 17.43028624192059,
      "grad_norm": 0.23807014524936676,
      "learning_rate": 2.8332000000000002e-05,
      "loss": 0.0116,
      "step": 9447
    },
    {
      "epoch": 17.43213296398892,
      "grad_norm": 0.4963562786579132,
      "learning_rate": 2.8335000000000002e-05,
      "loss": 0.0096,
      "step": 9448
    },
    {
      "epoch": 17.43397968605725,
      "grad_norm": 0.5319734811782837,
      "learning_rate": 2.8338000000000002e-05,
      "loss": 0.0139,
      "step": 9449
    },
    {
      "epoch": 17.43582640812558,
      "grad_norm": 0.3593072295188904,
      "learning_rate": 2.8341000000000002e-05,
      "loss": 0.0092,
      "step": 9450
    },
    {
      "epoch": 17.437673130193907,
      "grad_norm": 0.40505778789520264,
      "learning_rate": 2.8344e-05,
      "loss": 0.0127,
      "step": 9451
    },
    {
      "epoch": 17.439519852262233,
      "grad_norm": 0.321172297000885,
      "learning_rate": 2.8346999999999998e-05,
      "loss": 0.0099,
      "step": 9452
    },
    {
      "epoch": 17.441366574330562,
      "grad_norm": 0.26844117045402527,
      "learning_rate": 2.8349999999999998e-05,
      "loss": 0.0113,
      "step": 9453
    },
    {
      "epoch": 17.44321329639889,
      "grad_norm": 0.40463778376579285,
      "learning_rate": 2.8353e-05,
      "loss": 0.0162,
      "step": 9454
    },
    {
      "epoch": 17.44506001846722,
      "grad_norm": 0.5917307734489441,
      "learning_rate": 2.8356e-05,
      "loss": 0.0275,
      "step": 9455
    },
    {
      "epoch": 17.44690674053555,
      "grad_norm": 0.5759962201118469,
      "learning_rate": 2.8359e-05,
      "loss": 0.0157,
      "step": 9456
    },
    {
      "epoch": 17.44875346260388,
      "grad_norm": 0.252387672662735,
      "learning_rate": 2.8362e-05,
      "loss": 0.0129,
      "step": 9457
    },
    {
      "epoch": 17.450600184672208,
      "grad_norm": 0.5068285465240479,
      "learning_rate": 2.8365e-05,
      "loss": 0.0172,
      "step": 9458
    },
    {
      "epoch": 17.452446906740537,
      "grad_norm": 0.2979760468006134,
      "learning_rate": 2.8368e-05,
      "loss": 0.016,
      "step": 9459
    },
    {
      "epoch": 17.454293628808863,
      "grad_norm": 0.557605504989624,
      "learning_rate": 2.8371e-05,
      "loss": 0.0218,
      "step": 9460
    },
    {
      "epoch": 17.45614035087719,
      "grad_norm": 0.6038251519203186,
      "learning_rate": 2.8374e-05,
      "loss": 0.0126,
      "step": 9461
    },
    {
      "epoch": 17.45798707294552,
      "grad_norm": 1.0946751832962036,
      "learning_rate": 2.8377e-05,
      "loss": 0.0425,
      "step": 9462
    },
    {
      "epoch": 17.45983379501385,
      "grad_norm": 0.39458906650543213,
      "learning_rate": 2.838e-05,
      "loss": 0.0178,
      "step": 9463
    },
    {
      "epoch": 17.46168051708218,
      "grad_norm": 0.388188898563385,
      "learning_rate": 2.8383000000000003e-05,
      "loss": 0.0131,
      "step": 9464
    },
    {
      "epoch": 17.46352723915051,
      "grad_norm": 1.3474907875061035,
      "learning_rate": 2.8386000000000002e-05,
      "loss": 0.2015,
      "step": 9465
    },
    {
      "epoch": 17.465373961218837,
      "grad_norm": 0.6771953105926514,
      "learning_rate": 2.8389000000000002e-05,
      "loss": 0.1534,
      "step": 9466
    },
    {
      "epoch": 17.467220683287167,
      "grad_norm": 0.586620032787323,
      "learning_rate": 2.8392000000000002e-05,
      "loss": 0.144,
      "step": 9467
    },
    {
      "epoch": 17.469067405355496,
      "grad_norm": 0.5955852270126343,
      "learning_rate": 2.8395000000000002e-05,
      "loss": 0.1074,
      "step": 9468
    },
    {
      "epoch": 17.47091412742382,
      "grad_norm": 0.7881993055343628,
      "learning_rate": 2.8398e-05,
      "loss": 0.0898,
      "step": 9469
    },
    {
      "epoch": 17.47276084949215,
      "grad_norm": 0.502046525478363,
      "learning_rate": 2.8401e-05,
      "loss": 0.0669,
      "step": 9470
    },
    {
      "epoch": 17.47460757156048,
      "grad_norm": 0.42128366231918335,
      "learning_rate": 2.8404e-05,
      "loss": 0.079,
      "step": 9471
    },
    {
      "epoch": 17.47645429362881,
      "grad_norm": 0.5567536354064941,
      "learning_rate": 2.8407e-05,
      "loss": 0.0523,
      "step": 9472
    },
    {
      "epoch": 17.478301015697138,
      "grad_norm": 0.5962188839912415,
      "learning_rate": 2.841e-05,
      "loss": 0.0686,
      "step": 9473
    },
    {
      "epoch": 17.480147737765467,
      "grad_norm": 0.5559502840042114,
      "learning_rate": 2.8413000000000004e-05,
      "loss": 0.08,
      "step": 9474
    },
    {
      "epoch": 17.481994459833796,
      "grad_norm": 0.5045327544212341,
      "learning_rate": 2.8416e-05,
      "loss": 0.0555,
      "step": 9475
    },
    {
      "epoch": 17.483841181902125,
      "grad_norm": 0.9575005173683167,
      "learning_rate": 2.8419e-05,
      "loss": 0.0283,
      "step": 9476
    },
    {
      "epoch": 17.48568790397045,
      "grad_norm": 0.6482095718383789,
      "learning_rate": 2.8422e-05,
      "loss": 0.0452,
      "step": 9477
    },
    {
      "epoch": 17.48753462603878,
      "grad_norm": 0.3243868350982666,
      "learning_rate": 2.8425e-05,
      "loss": 0.0209,
      "step": 9478
    },
    {
      "epoch": 17.48938134810711,
      "grad_norm": 0.32968437671661377,
      "learning_rate": 2.8428e-05,
      "loss": 0.0162,
      "step": 9479
    },
    {
      "epoch": 17.49122807017544,
      "grad_norm": 0.4083661437034607,
      "learning_rate": 2.8431e-05,
      "loss": 0.0213,
      "step": 9480
    },
    {
      "epoch": 17.493074792243767,
      "grad_norm": 0.3611833453178406,
      "learning_rate": 2.8434e-05,
      "loss": 0.0141,
      "step": 9481
    },
    {
      "epoch": 17.494921514312097,
      "grad_norm": 0.3172003924846649,
      "learning_rate": 2.8437e-05,
      "loss": 0.018,
      "step": 9482
    },
    {
      "epoch": 17.496768236380426,
      "grad_norm": 0.25099191069602966,
      "learning_rate": 2.844e-05,
      "loss": 0.0171,
      "step": 9483
    },
    {
      "epoch": 17.498614958448755,
      "grad_norm": 0.417826384305954,
      "learning_rate": 2.8443000000000002e-05,
      "loss": 0.042,
      "step": 9484
    },
    {
      "epoch": 17.50046168051708,
      "grad_norm": 0.5361323356628418,
      "learning_rate": 2.8446000000000002e-05,
      "loss": 0.0366,
      "step": 9485
    },
    {
      "epoch": 17.50230840258541,
      "grad_norm": 0.2949165999889374,
      "learning_rate": 2.8449e-05,
      "loss": 0.0089,
      "step": 9486
    },
    {
      "epoch": 17.50415512465374,
      "grad_norm": 0.434059739112854,
      "learning_rate": 2.8452e-05,
      "loss": 0.0176,
      "step": 9487
    },
    {
      "epoch": 17.506001846722068,
      "grad_norm": 0.3568911850452423,
      "learning_rate": 2.8455e-05,
      "loss": 0.0235,
      "step": 9488
    },
    {
      "epoch": 17.507848568790397,
      "grad_norm": 0.4586827754974365,
      "learning_rate": 2.8458e-05,
      "loss": 0.0091,
      "step": 9489
    },
    {
      "epoch": 17.509695290858726,
      "grad_norm": 0.38927754759788513,
      "learning_rate": 2.8461e-05,
      "loss": 0.03,
      "step": 9490
    },
    {
      "epoch": 17.511542012927055,
      "grad_norm": 0.27216121554374695,
      "learning_rate": 2.8464e-05,
      "loss": 0.0111,
      "step": 9491
    },
    {
      "epoch": 17.513388734995385,
      "grad_norm": 0.3073829710483551,
      "learning_rate": 2.8467e-05,
      "loss": 0.0134,
      "step": 9492
    },
    {
      "epoch": 17.51523545706371,
      "grad_norm": 0.3827543556690216,
      "learning_rate": 2.847e-05,
      "loss": 0.0128,
      "step": 9493
    },
    {
      "epoch": 17.51708217913204,
      "grad_norm": 0.3475456237792969,
      "learning_rate": 2.8473000000000003e-05,
      "loss": 0.012,
      "step": 9494
    },
    {
      "epoch": 17.51892890120037,
      "grad_norm": 0.21921490132808685,
      "learning_rate": 2.8476000000000003e-05,
      "loss": 0.0063,
      "step": 9495
    },
    {
      "epoch": 17.520775623268698,
      "grad_norm": 0.5031759738922119,
      "learning_rate": 2.8479000000000003e-05,
      "loss": 0.0162,
      "step": 9496
    },
    {
      "epoch": 17.522622345337027,
      "grad_norm": 0.3823947310447693,
      "learning_rate": 2.8482000000000003e-05,
      "loss": 0.0096,
      "step": 9497
    },
    {
      "epoch": 17.524469067405356,
      "grad_norm": 0.3628939688205719,
      "learning_rate": 2.8485000000000003e-05,
      "loss": 0.0133,
      "step": 9498
    },
    {
      "epoch": 17.526315789473685,
      "grad_norm": 0.5429162383079529,
      "learning_rate": 2.8488000000000002e-05,
      "loss": 0.0157,
      "step": 9499
    },
    {
      "epoch": 17.528162511542014,
      "grad_norm": 0.27803874015808105,
      "learning_rate": 2.8491e-05,
      "loss": 0.0076,
      "step": 9500
    },
    {
      "epoch": 17.530009233610343,
      "grad_norm": 0.4344504773616791,
      "learning_rate": 2.8494e-05,
      "loss": 0.0097,
      "step": 9501
    },
    {
      "epoch": 17.53185595567867,
      "grad_norm": 0.2885424494743347,
      "learning_rate": 2.8497e-05,
      "loss": 0.0138,
      "step": 9502
    },
    {
      "epoch": 17.533702677746998,
      "grad_norm": 0.6529332399368286,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 0.0264,
      "step": 9503
    },
    {
      "epoch": 17.535549399815327,
      "grad_norm": 1.1576621532440186,
      "learning_rate": 2.8502999999999998e-05,
      "loss": 0.014,
      "step": 9504
    },
    {
      "epoch": 17.537396121883656,
      "grad_norm": 0.26482293009757996,
      "learning_rate": 2.8506e-05,
      "loss": 0.0119,
      "step": 9505
    },
    {
      "epoch": 17.539242843951985,
      "grad_norm": 0.21488991379737854,
      "learning_rate": 2.8509e-05,
      "loss": 0.0078,
      "step": 9506
    },
    {
      "epoch": 17.541089566020315,
      "grad_norm": 0.6642892360687256,
      "learning_rate": 2.8512e-05,
      "loss": 0.0124,
      "step": 9507
    },
    {
      "epoch": 17.542936288088644,
      "grad_norm": 0.669625461101532,
      "learning_rate": 2.8515e-05,
      "loss": 0.0181,
      "step": 9508
    },
    {
      "epoch": 17.544783010156973,
      "grad_norm": 0.5279157161712646,
      "learning_rate": 2.8518e-05,
      "loss": 0.0132,
      "step": 9509
    },
    {
      "epoch": 17.5466297322253,
      "grad_norm": 0.320911705493927,
      "learning_rate": 2.8521e-05,
      "loss": 0.0156,
      "step": 9510
    },
    {
      "epoch": 17.548476454293628,
      "grad_norm": 0.6547080278396606,
      "learning_rate": 2.8524e-05,
      "loss": 0.0176,
      "step": 9511
    },
    {
      "epoch": 17.550323176361957,
      "grad_norm": 0.2697703242301941,
      "learning_rate": 2.8527e-05,
      "loss": 0.0144,
      "step": 9512
    },
    {
      "epoch": 17.552169898430286,
      "grad_norm": 0.33988258242607117,
      "learning_rate": 2.853e-05,
      "loss": 0.0069,
      "step": 9513
    },
    {
      "epoch": 17.554016620498615,
      "grad_norm": 0.9230872392654419,
      "learning_rate": 2.8533e-05,
      "loss": 0.0201,
      "step": 9514
    },
    {
      "epoch": 17.555863342566944,
      "grad_norm": 0.6196572780609131,
      "learning_rate": 2.8536000000000003e-05,
      "loss": 0.1457,
      "step": 9515
    },
    {
      "epoch": 17.557710064635273,
      "grad_norm": 0.86606365442276,
      "learning_rate": 2.8539000000000002e-05,
      "loss": 0.1609,
      "step": 9516
    },
    {
      "epoch": 17.559556786703602,
      "grad_norm": 0.7880556583404541,
      "learning_rate": 2.8542000000000002e-05,
      "loss": 0.1303,
      "step": 9517
    },
    {
      "epoch": 17.56140350877193,
      "grad_norm": 0.5721180438995361,
      "learning_rate": 2.8545000000000002e-05,
      "loss": 0.0967,
      "step": 9518
    },
    {
      "epoch": 17.563250230840257,
      "grad_norm": 0.5649641752243042,
      "learning_rate": 2.8548000000000002e-05,
      "loss": 0.1203,
      "step": 9519
    },
    {
      "epoch": 17.565096952908586,
      "grad_norm": 0.49527832865715027,
      "learning_rate": 2.8551e-05,
      "loss": 0.0973,
      "step": 9520
    },
    {
      "epoch": 17.566943674976915,
      "grad_norm": 0.4757789373397827,
      "learning_rate": 2.8554e-05,
      "loss": 0.0727,
      "step": 9521
    },
    {
      "epoch": 17.568790397045245,
      "grad_norm": 0.4874994456768036,
      "learning_rate": 2.8557e-05,
      "loss": 0.106,
      "step": 9522
    },
    {
      "epoch": 17.570637119113574,
      "grad_norm": 0.76370769739151,
      "learning_rate": 2.856e-05,
      "loss": 0.0565,
      "step": 9523
    },
    {
      "epoch": 17.572483841181903,
      "grad_norm": 0.6513908505439758,
      "learning_rate": 2.8563e-05,
      "loss": 0.0506,
      "step": 9524
    },
    {
      "epoch": 17.574330563250232,
      "grad_norm": 0.4472300410270691,
      "learning_rate": 2.8566e-05,
      "loss": 0.0472,
      "step": 9525
    },
    {
      "epoch": 17.57617728531856,
      "grad_norm": 0.5632028579711914,
      "learning_rate": 2.8569e-05,
      "loss": 0.0625,
      "step": 9526
    },
    {
      "epoch": 17.578024007386887,
      "grad_norm": 0.5988933444023132,
      "learning_rate": 2.8572e-05,
      "loss": 0.0989,
      "step": 9527
    },
    {
      "epoch": 17.579870729455216,
      "grad_norm": 0.8819764256477356,
      "learning_rate": 2.8575e-05,
      "loss": 0.0573,
      "step": 9528
    },
    {
      "epoch": 17.581717451523545,
      "grad_norm": 0.9006199836730957,
      "learning_rate": 2.8578e-05,
      "loss": 0.038,
      "step": 9529
    },
    {
      "epoch": 17.583564173591874,
      "grad_norm": 0.674299955368042,
      "learning_rate": 2.8581e-05,
      "loss": 0.0131,
      "step": 9530
    },
    {
      "epoch": 17.585410895660203,
      "grad_norm": 0.38229405879974365,
      "learning_rate": 2.8584e-05,
      "loss": 0.02,
      "step": 9531
    },
    {
      "epoch": 17.587257617728532,
      "grad_norm": 1.4004647731781006,
      "learning_rate": 2.8587e-05,
      "loss": 0.0299,
      "step": 9532
    },
    {
      "epoch": 17.58910433979686,
      "grad_norm": 0.3553142547607422,
      "learning_rate": 2.859e-05,
      "loss": 0.0347,
      "step": 9533
    },
    {
      "epoch": 17.59095106186519,
      "grad_norm": 0.2724413573741913,
      "learning_rate": 2.8593e-05,
      "loss": 0.0148,
      "step": 9534
    },
    {
      "epoch": 17.592797783933516,
      "grad_norm": 0.7119773626327515,
      "learning_rate": 2.8596000000000002e-05,
      "loss": 0.0416,
      "step": 9535
    },
    {
      "epoch": 17.594644506001845,
      "grad_norm": 0.44424229860305786,
      "learning_rate": 2.8599000000000002e-05,
      "loss": 0.0197,
      "step": 9536
    },
    {
      "epoch": 17.596491228070175,
      "grad_norm": 0.2880791425704956,
      "learning_rate": 2.8602e-05,
      "loss": 0.009,
      "step": 9537
    },
    {
      "epoch": 17.598337950138504,
      "grad_norm": 0.3062593936920166,
      "learning_rate": 2.8605e-05,
      "loss": 0.0145,
      "step": 9538
    },
    {
      "epoch": 17.600184672206833,
      "grad_norm": 0.31058281660079956,
      "learning_rate": 2.8608e-05,
      "loss": 0.0128,
      "step": 9539
    },
    {
      "epoch": 17.602031394275162,
      "grad_norm": 0.4950855076313019,
      "learning_rate": 2.8611e-05,
      "loss": 0.0419,
      "step": 9540
    },
    {
      "epoch": 17.60387811634349,
      "grad_norm": 0.16567905247211456,
      "learning_rate": 2.8614e-05,
      "loss": 0.0425,
      "step": 9541
    },
    {
      "epoch": 17.60572483841182,
      "grad_norm": 0.3481304943561554,
      "learning_rate": 2.8617e-05,
      "loss": 0.0199,
      "step": 9542
    },
    {
      "epoch": 17.607571560480146,
      "grad_norm": 0.20209775865077972,
      "learning_rate": 2.862e-05,
      "loss": 0.0083,
      "step": 9543
    },
    {
      "epoch": 17.609418282548475,
      "grad_norm": 0.29109519720077515,
      "learning_rate": 2.8623e-05,
      "loss": 0.0081,
      "step": 9544
    },
    {
      "epoch": 17.611265004616804,
      "grad_norm": 0.47210419178009033,
      "learning_rate": 2.8626000000000003e-05,
      "loss": 0.0224,
      "step": 9545
    },
    {
      "epoch": 17.613111726685133,
      "grad_norm": 0.3226254880428314,
      "learning_rate": 2.8629000000000003e-05,
      "loss": 0.015,
      "step": 9546
    },
    {
      "epoch": 17.614958448753463,
      "grad_norm": 0.3622691035270691,
      "learning_rate": 2.8632000000000003e-05,
      "loss": 0.0182,
      "step": 9547
    },
    {
      "epoch": 17.61680517082179,
      "grad_norm": 0.4353501498699188,
      "learning_rate": 2.8635000000000003e-05,
      "loss": 0.0168,
      "step": 9548
    },
    {
      "epoch": 17.61865189289012,
      "grad_norm": 0.2856418192386627,
      "learning_rate": 2.8638e-05,
      "loss": 0.0134,
      "step": 9549
    },
    {
      "epoch": 17.62049861495845,
      "grad_norm": 0.3683633506298065,
      "learning_rate": 2.8641e-05,
      "loss": 0.0168,
      "step": 9550
    },
    {
      "epoch": 17.62234533702678,
      "grad_norm": 0.36736348271369934,
      "learning_rate": 2.8644e-05,
      "loss": 0.0121,
      "step": 9551
    },
    {
      "epoch": 17.624192059095105,
      "grad_norm": 0.5249360203742981,
      "learning_rate": 2.8647e-05,
      "loss": 0.0149,
      "step": 9552
    },
    {
      "epoch": 17.626038781163434,
      "grad_norm": 0.4789264500141144,
      "learning_rate": 2.865e-05,
      "loss": 0.0119,
      "step": 9553
    },
    {
      "epoch": 17.627885503231763,
      "grad_norm": 0.6176832914352417,
      "learning_rate": 2.8652999999999998e-05,
      "loss": 0.0246,
      "step": 9554
    },
    {
      "epoch": 17.629732225300092,
      "grad_norm": 0.562190592288971,
      "learning_rate": 2.8656e-05,
      "loss": 0.0178,
      "step": 9555
    },
    {
      "epoch": 17.63157894736842,
      "grad_norm": 0.6893006563186646,
      "learning_rate": 2.8659e-05,
      "loss": 0.0157,
      "step": 9556
    },
    {
      "epoch": 17.63342566943675,
      "grad_norm": 0.7275405526161194,
      "learning_rate": 2.8662e-05,
      "loss": 0.0269,
      "step": 9557
    },
    {
      "epoch": 17.63527239150508,
      "grad_norm": 0.494080126285553,
      "learning_rate": 2.8665e-05,
      "loss": 0.012,
      "step": 9558
    },
    {
      "epoch": 17.63711911357341,
      "grad_norm": 0.32813942432403564,
      "learning_rate": 2.8668e-05,
      "loss": 0.0185,
      "step": 9559
    },
    {
      "epoch": 17.638965835641734,
      "grad_norm": 0.3075019121170044,
      "learning_rate": 2.8671e-05,
      "loss": 0.0114,
      "step": 9560
    },
    {
      "epoch": 17.640812557710063,
      "grad_norm": 0.9370800852775574,
      "learning_rate": 2.8674e-05,
      "loss": 0.0175,
      "step": 9561
    },
    {
      "epoch": 17.642659279778393,
      "grad_norm": 0.16680116951465607,
      "learning_rate": 2.8677e-05,
      "loss": 0.0043,
      "step": 9562
    },
    {
      "epoch": 17.64450600184672,
      "grad_norm": 0.4168938994407654,
      "learning_rate": 2.868e-05,
      "loss": 0.009,
      "step": 9563
    },
    {
      "epoch": 17.64635272391505,
      "grad_norm": 0.3269334137439728,
      "learning_rate": 2.8683e-05,
      "loss": 0.0097,
      "step": 9564
    },
    {
      "epoch": 17.64819944598338,
      "grad_norm": 1.061153531074524,
      "learning_rate": 2.8686000000000003e-05,
      "loss": 0.2211,
      "step": 9565
    },
    {
      "epoch": 17.65004616805171,
      "grad_norm": 0.823059618473053,
      "learning_rate": 2.8689000000000003e-05,
      "loss": 0.1503,
      "step": 9566
    },
    {
      "epoch": 17.65189289012004,
      "grad_norm": 0.634117066860199,
      "learning_rate": 2.8692000000000002e-05,
      "loss": 0.0934,
      "step": 9567
    },
    {
      "epoch": 17.653739612188367,
      "grad_norm": 0.6172256469726562,
      "learning_rate": 2.8695000000000002e-05,
      "loss": 0.1134,
      "step": 9568
    },
    {
      "epoch": 17.655586334256693,
      "grad_norm": 0.7339869141578674,
      "learning_rate": 2.8698000000000002e-05,
      "loss": 0.1427,
      "step": 9569
    },
    {
      "epoch": 17.657433056325022,
      "grad_norm": 0.6254228353500366,
      "learning_rate": 2.8701000000000002e-05,
      "loss": 0.1562,
      "step": 9570
    },
    {
      "epoch": 17.65927977839335,
      "grad_norm": 0.6750670671463013,
      "learning_rate": 2.8704e-05,
      "loss": 0.0843,
      "step": 9571
    },
    {
      "epoch": 17.66112650046168,
      "grad_norm": 0.6463918089866638,
      "learning_rate": 2.8707e-05,
      "loss": 0.0835,
      "step": 9572
    },
    {
      "epoch": 17.66297322253001,
      "grad_norm": 0.5513696074485779,
      "learning_rate": 2.871e-05,
      "loss": 0.0705,
      "step": 9573
    },
    {
      "epoch": 17.66481994459834,
      "grad_norm": 0.8210604190826416,
      "learning_rate": 2.8712999999999998e-05,
      "loss": 0.0536,
      "step": 9574
    },
    {
      "epoch": 17.666666666666668,
      "grad_norm": 0.5390028953552246,
      "learning_rate": 2.8716e-05,
      "loss": 0.0537,
      "step": 9575
    },
    {
      "epoch": 17.668513388734997,
      "grad_norm": 0.3856324255466461,
      "learning_rate": 2.8719e-05,
      "loss": 0.058,
      "step": 9576
    },
    {
      "epoch": 17.670360110803323,
      "grad_norm": 0.5079103708267212,
      "learning_rate": 2.8722e-05,
      "loss": 0.0483,
      "step": 9577
    },
    {
      "epoch": 17.67220683287165,
      "grad_norm": 0.4683472812175751,
      "learning_rate": 2.8725e-05,
      "loss": 0.0329,
      "step": 9578
    },
    {
      "epoch": 17.67405355493998,
      "grad_norm": 0.5738013982772827,
      "learning_rate": 2.8728e-05,
      "loss": 0.0681,
      "step": 9579
    },
    {
      "epoch": 17.67590027700831,
      "grad_norm": 0.5303285717964172,
      "learning_rate": 2.8731e-05,
      "loss": 0.039,
      "step": 9580
    },
    {
      "epoch": 17.67774699907664,
      "grad_norm": 0.2283354550600052,
      "learning_rate": 2.8734e-05,
      "loss": 0.0169,
      "step": 9581
    },
    {
      "epoch": 17.67959372114497,
      "grad_norm": 0.22478586435317993,
      "learning_rate": 2.8737e-05,
      "loss": 0.0121,
      "step": 9582
    },
    {
      "epoch": 17.681440443213297,
      "grad_norm": 0.22185231745243073,
      "learning_rate": 2.874e-05,
      "loss": 0.0164,
      "step": 9583
    },
    {
      "epoch": 17.683287165281627,
      "grad_norm": 0.29085099697113037,
      "learning_rate": 2.8743e-05,
      "loss": 0.0185,
      "step": 9584
    },
    {
      "epoch": 17.685133887349952,
      "grad_norm": 0.5266684293746948,
      "learning_rate": 2.8746000000000002e-05,
      "loss": 0.0191,
      "step": 9585
    },
    {
      "epoch": 17.68698060941828,
      "grad_norm": 0.20346860587596893,
      "learning_rate": 2.8749000000000002e-05,
      "loss": 0.0095,
      "step": 9586
    },
    {
      "epoch": 17.68882733148661,
      "grad_norm": 0.5799970626831055,
      "learning_rate": 2.8752000000000002e-05,
      "loss": 0.029,
      "step": 9587
    },
    {
      "epoch": 17.69067405355494,
      "grad_norm": 0.29750967025756836,
      "learning_rate": 2.8755e-05,
      "loss": 0.0166,
      "step": 9588
    },
    {
      "epoch": 17.69252077562327,
      "grad_norm": 0.2990019619464874,
      "learning_rate": 2.8758e-05,
      "loss": 0.0159,
      "step": 9589
    },
    {
      "epoch": 17.694367497691598,
      "grad_norm": 0.3774636685848236,
      "learning_rate": 2.8761e-05,
      "loss": 0.0129,
      "step": 9590
    },
    {
      "epoch": 17.696214219759927,
      "grad_norm": 0.2095613330602646,
      "learning_rate": 2.8764e-05,
      "loss": 0.0114,
      "step": 9591
    },
    {
      "epoch": 17.698060941828256,
      "grad_norm": 0.3799412250518799,
      "learning_rate": 2.8767e-05,
      "loss": 0.0167,
      "step": 9592
    },
    {
      "epoch": 17.69990766389658,
      "grad_norm": 0.29384300112724304,
      "learning_rate": 2.877e-05,
      "loss": 0.0128,
      "step": 9593
    },
    {
      "epoch": 17.70175438596491,
      "grad_norm": 0.5691142082214355,
      "learning_rate": 2.8773e-05,
      "loss": 0.0683,
      "step": 9594
    },
    {
      "epoch": 17.70360110803324,
      "grad_norm": 0.3275772035121918,
      "learning_rate": 2.8776000000000004e-05,
      "loss": 0.0142,
      "step": 9595
    },
    {
      "epoch": 17.70544783010157,
      "grad_norm": 0.27224838733673096,
      "learning_rate": 2.8779000000000003e-05,
      "loss": 0.0159,
      "step": 9596
    },
    {
      "epoch": 17.7072945521699,
      "grad_norm": 0.3701270520687103,
      "learning_rate": 2.8782000000000003e-05,
      "loss": 0.01,
      "step": 9597
    },
    {
      "epoch": 17.709141274238227,
      "grad_norm": 0.35283249616622925,
      "learning_rate": 2.8785e-05,
      "loss": 0.0122,
      "step": 9598
    },
    {
      "epoch": 17.710987996306557,
      "grad_norm": 0.28894442319869995,
      "learning_rate": 2.8788e-05,
      "loss": 0.0115,
      "step": 9599
    },
    {
      "epoch": 17.712834718374886,
      "grad_norm": 0.4092564582824707,
      "learning_rate": 2.8791e-05,
      "loss": 0.01,
      "step": 9600
    },
    {
      "epoch": 17.714681440443215,
      "grad_norm": 0.40917760133743286,
      "learning_rate": 2.8794e-05,
      "loss": 0.0087,
      "step": 9601
    },
    {
      "epoch": 17.71652816251154,
      "grad_norm": 0.26291561126708984,
      "learning_rate": 2.8797e-05,
      "loss": 0.0039,
      "step": 9602
    },
    {
      "epoch": 17.71837488457987,
      "grad_norm": 0.501798152923584,
      "learning_rate": 2.88e-05,
      "loss": 0.018,
      "step": 9603
    },
    {
      "epoch": 17.7202216066482,
      "grad_norm": 0.25262412428855896,
      "learning_rate": 2.8803e-05,
      "loss": 0.0089,
      "step": 9604
    },
    {
      "epoch": 17.722068328716528,
      "grad_norm": 0.5903002023696899,
      "learning_rate": 2.8806e-05,
      "loss": 0.0166,
      "step": 9605
    },
    {
      "epoch": 17.723915050784857,
      "grad_norm": 0.5660081505775452,
      "learning_rate": 2.8809e-05,
      "loss": 0.0163,
      "step": 9606
    },
    {
      "epoch": 17.725761772853186,
      "grad_norm": 0.5993664860725403,
      "learning_rate": 2.8812e-05,
      "loss": 0.0173,
      "step": 9607
    },
    {
      "epoch": 17.727608494921515,
      "grad_norm": 0.45771902799606323,
      "learning_rate": 2.8815e-05,
      "loss": 0.0205,
      "step": 9608
    },
    {
      "epoch": 17.729455216989845,
      "grad_norm": 0.3490065932273865,
      "learning_rate": 2.8818e-05,
      "loss": 0.0167,
      "step": 9609
    },
    {
      "epoch": 17.73130193905817,
      "grad_norm": 0.8391483426094055,
      "learning_rate": 2.8821e-05,
      "loss": 0.0231,
      "step": 9610
    },
    {
      "epoch": 17.7331486611265,
      "grad_norm": 0.3549315929412842,
      "learning_rate": 2.8824e-05,
      "loss": 0.0104,
      "step": 9611
    },
    {
      "epoch": 17.73499538319483,
      "grad_norm": 0.4805821478366852,
      "learning_rate": 2.8827e-05,
      "loss": 0.0121,
      "step": 9612
    },
    {
      "epoch": 17.736842105263158,
      "grad_norm": 0.6108695864677429,
      "learning_rate": 2.883e-05,
      "loss": 0.0255,
      "step": 9613
    },
    {
      "epoch": 17.738688827331487,
      "grad_norm": 0.6562829613685608,
      "learning_rate": 2.8833e-05,
      "loss": 0.0245,
      "step": 9614
    },
    {
      "epoch": 17.740535549399816,
      "grad_norm": 0.6468729376792908,
      "learning_rate": 2.8836000000000003e-05,
      "loss": 0.1794,
      "step": 9615
    },
    {
      "epoch": 17.742382271468145,
      "grad_norm": 0.5797168612480164,
      "learning_rate": 2.8839000000000003e-05,
      "loss": 0.1674,
      "step": 9616
    },
    {
      "epoch": 17.744228993536474,
      "grad_norm": 0.498214989900589,
      "learning_rate": 2.8842000000000003e-05,
      "loss": 0.1372,
      "step": 9617
    },
    {
      "epoch": 17.746075715604803,
      "grad_norm": 0.42887386679649353,
      "learning_rate": 2.8845000000000003e-05,
      "loss": 0.1131,
      "step": 9618
    },
    {
      "epoch": 17.74792243767313,
      "grad_norm": 0.4448026418685913,
      "learning_rate": 2.8848000000000002e-05,
      "loss": 0.0786,
      "step": 9619
    },
    {
      "epoch": 17.749769159741458,
      "grad_norm": 0.734890341758728,
      "learning_rate": 2.8851000000000002e-05,
      "loss": 0.0993,
      "step": 9620
    },
    {
      "epoch": 17.751615881809787,
      "grad_norm": 0.5033615827560425,
      "learning_rate": 2.8854000000000002e-05,
      "loss": 0.0839,
      "step": 9621
    },
    {
      "epoch": 17.753462603878116,
      "grad_norm": 0.42628228664398193,
      "learning_rate": 2.8857000000000002e-05,
      "loss": 0.054,
      "step": 9622
    },
    {
      "epoch": 17.755309325946445,
      "grad_norm": 0.46170705556869507,
      "learning_rate": 2.8859999999999998e-05,
      "loss": 0.0717,
      "step": 9623
    },
    {
      "epoch": 17.757156048014775,
      "grad_norm": 0.5042704939842224,
      "learning_rate": 2.8862999999999998e-05,
      "loss": 0.1043,
      "step": 9624
    },
    {
      "epoch": 17.759002770083104,
      "grad_norm": 0.5552589893341064,
      "learning_rate": 2.8866e-05,
      "loss": 0.0452,
      "step": 9625
    },
    {
      "epoch": 17.760849492151433,
      "grad_norm": 0.29673290252685547,
      "learning_rate": 2.8869e-05,
      "loss": 0.0418,
      "step": 9626
    },
    {
      "epoch": 17.76269621421976,
      "grad_norm": 0.5385515689849854,
      "learning_rate": 2.8872e-05,
      "loss": 0.0462,
      "step": 9627
    },
    {
      "epoch": 17.764542936288088,
      "grad_norm": 0.32758527994155884,
      "learning_rate": 2.8875e-05,
      "loss": 0.0458,
      "step": 9628
    },
    {
      "epoch": 17.766389658356417,
      "grad_norm": 0.2771989107131958,
      "learning_rate": 2.8878e-05,
      "loss": 0.0291,
      "step": 9629
    },
    {
      "epoch": 17.768236380424746,
      "grad_norm": 0.34781336784362793,
      "learning_rate": 2.8881e-05,
      "loss": 0.024,
      "step": 9630
    },
    {
      "epoch": 17.770083102493075,
      "grad_norm": 0.39459630846977234,
      "learning_rate": 2.8884e-05,
      "loss": 0.0178,
      "step": 9631
    },
    {
      "epoch": 17.771929824561404,
      "grad_norm": 0.34889426827430725,
      "learning_rate": 2.8887e-05,
      "loss": 0.027,
      "step": 9632
    },
    {
      "epoch": 17.773776546629733,
      "grad_norm": 0.21631485223770142,
      "learning_rate": 2.889e-05,
      "loss": 0.0182,
      "step": 9633
    },
    {
      "epoch": 17.775623268698062,
      "grad_norm": 0.24801243841648102,
      "learning_rate": 2.8893e-05,
      "loss": 0.034,
      "step": 9634
    },
    {
      "epoch": 17.777469990766388,
      "grad_norm": 0.270134836435318,
      "learning_rate": 2.8896e-05,
      "loss": 0.0091,
      "step": 9635
    },
    {
      "epoch": 17.779316712834717,
      "grad_norm": 0.594982922077179,
      "learning_rate": 2.8899000000000002e-05,
      "loss": 0.0151,
      "step": 9636
    },
    {
      "epoch": 17.781163434903046,
      "grad_norm": 0.6876816749572754,
      "learning_rate": 2.8902000000000002e-05,
      "loss": 0.0184,
      "step": 9637
    },
    {
      "epoch": 17.783010156971375,
      "grad_norm": 0.3381214439868927,
      "learning_rate": 2.8905000000000002e-05,
      "loss": 0.0071,
      "step": 9638
    },
    {
      "epoch": 17.784856879039705,
      "grad_norm": 0.4301603436470032,
      "learning_rate": 2.8908000000000002e-05,
      "loss": 0.0143,
      "step": 9639
    },
    {
      "epoch": 17.786703601108034,
      "grad_norm": 0.40740901231765747,
      "learning_rate": 2.8911e-05,
      "loss": 0.0147,
      "step": 9640
    },
    {
      "epoch": 17.788550323176363,
      "grad_norm": 0.5560108423233032,
      "learning_rate": 2.8914e-05,
      "loss": 0.0196,
      "step": 9641
    },
    {
      "epoch": 17.790397045244692,
      "grad_norm": 0.18544799089431763,
      "learning_rate": 2.8917e-05,
      "loss": 0.0083,
      "step": 9642
    },
    {
      "epoch": 17.792243767313018,
      "grad_norm": 0.36729657649993896,
      "learning_rate": 2.892e-05,
      "loss": 0.0136,
      "step": 9643
    },
    {
      "epoch": 17.794090489381347,
      "grad_norm": 0.24889634549617767,
      "learning_rate": 2.8923e-05,
      "loss": 0.0356,
      "step": 9644
    },
    {
      "epoch": 17.795937211449676,
      "grad_norm": 0.26268625259399414,
      "learning_rate": 2.8926e-05,
      "loss": 0.0085,
      "step": 9645
    },
    {
      "epoch": 17.797783933518005,
      "grad_norm": 0.1501934677362442,
      "learning_rate": 2.8929000000000004e-05,
      "loss": 0.0075,
      "step": 9646
    },
    {
      "epoch": 17.799630655586334,
      "grad_norm": 0.3604651391506195,
      "learning_rate": 2.8932e-05,
      "loss": 0.0234,
      "step": 9647
    },
    {
      "epoch": 17.801477377654663,
      "grad_norm": 0.25846487283706665,
      "learning_rate": 2.8935e-05,
      "loss": 0.0105,
      "step": 9648
    },
    {
      "epoch": 17.803324099722992,
      "grad_norm": 0.29089727997779846,
      "learning_rate": 2.8938e-05,
      "loss": 0.0067,
      "step": 9649
    },
    {
      "epoch": 17.80517082179132,
      "grad_norm": 0.32083141803741455,
      "learning_rate": 2.8941e-05,
      "loss": 0.0068,
      "step": 9650
    },
    {
      "epoch": 17.80701754385965,
      "grad_norm": 0.5071612000465393,
      "learning_rate": 2.8944e-05,
      "loss": 0.0216,
      "step": 9651
    },
    {
      "epoch": 17.808864265927976,
      "grad_norm": 0.4766952097415924,
      "learning_rate": 2.8947e-05,
      "loss": 0.0446,
      "step": 9652
    },
    {
      "epoch": 17.810710987996305,
      "grad_norm": 0.32992035150527954,
      "learning_rate": 2.895e-05,
      "loss": 0.0128,
      "step": 9653
    },
    {
      "epoch": 17.812557710064635,
      "grad_norm": 0.2831116020679474,
      "learning_rate": 2.8953e-05,
      "loss": 0.0113,
      "step": 9654
    },
    {
      "epoch": 17.814404432132964,
      "grad_norm": 0.45260798931121826,
      "learning_rate": 2.8956e-05,
      "loss": 0.0155,
      "step": 9655
    },
    {
      "epoch": 17.816251154201293,
      "grad_norm": 0.46074360609054565,
      "learning_rate": 2.8959000000000002e-05,
      "loss": 0.0217,
      "step": 9656
    },
    {
      "epoch": 17.818097876269622,
      "grad_norm": 1.0912634134292603,
      "learning_rate": 2.8962e-05,
      "loss": 0.015,
      "step": 9657
    },
    {
      "epoch": 17.81994459833795,
      "grad_norm": 0.34561482071876526,
      "learning_rate": 2.8965e-05,
      "loss": 0.0103,
      "step": 9658
    },
    {
      "epoch": 17.82179132040628,
      "grad_norm": 0.4716539978981018,
      "learning_rate": 2.8968e-05,
      "loss": 0.0138,
      "step": 9659
    },
    {
      "epoch": 17.823638042474606,
      "grad_norm": 0.5009711980819702,
      "learning_rate": 2.8971e-05,
      "loss": 0.0166,
      "step": 9660
    },
    {
      "epoch": 17.825484764542935,
      "grad_norm": 0.3603162169456482,
      "learning_rate": 2.8974e-05,
      "loss": 0.0167,
      "step": 9661
    },
    {
      "epoch": 17.827331486611264,
      "grad_norm": 0.4746507406234741,
      "learning_rate": 2.8977e-05,
      "loss": 0.0149,
      "step": 9662
    },
    {
      "epoch": 17.829178208679593,
      "grad_norm": 0.6108738780021667,
      "learning_rate": 2.898e-05,
      "loss": 0.0243,
      "step": 9663
    },
    {
      "epoch": 17.831024930747922,
      "grad_norm": 0.5035948753356934,
      "learning_rate": 2.8983e-05,
      "loss": 0.0122,
      "step": 9664
    },
    {
      "epoch": 17.83287165281625,
      "grad_norm": 0.535803496837616,
      "learning_rate": 2.8986e-05,
      "loss": 0.1685,
      "step": 9665
    },
    {
      "epoch": 17.83471837488458,
      "grad_norm": 0.7792430520057678,
      "learning_rate": 2.8989000000000003e-05,
      "loss": 0.1759,
      "step": 9666
    },
    {
      "epoch": 17.83656509695291,
      "grad_norm": 0.44851070642471313,
      "learning_rate": 2.8992000000000003e-05,
      "loss": 0.1387,
      "step": 9667
    },
    {
      "epoch": 17.83841181902124,
      "grad_norm": 0.531370222568512,
      "learning_rate": 2.8995000000000003e-05,
      "loss": 0.1089,
      "step": 9668
    },
    {
      "epoch": 17.840258541089565,
      "grad_norm": 0.4909587800502777,
      "learning_rate": 2.8998000000000003e-05,
      "loss": 0.1063,
      "step": 9669
    },
    {
      "epoch": 17.842105263157894,
      "grad_norm": 0.4452713429927826,
      "learning_rate": 2.9001000000000002e-05,
      "loss": 0.088,
      "step": 9670
    },
    {
      "epoch": 17.843951985226223,
      "grad_norm": 1.0999737977981567,
      "learning_rate": 2.9004000000000002e-05,
      "loss": 0.1129,
      "step": 9671
    },
    {
      "epoch": 17.845798707294552,
      "grad_norm": 0.7121695280075073,
      "learning_rate": 2.9007e-05,
      "loss": 0.1001,
      "step": 9672
    },
    {
      "epoch": 17.84764542936288,
      "grad_norm": 0.39976316690444946,
      "learning_rate": 2.901e-05,
      "loss": 0.0658,
      "step": 9673
    },
    {
      "epoch": 17.84949215143121,
      "grad_norm": 0.4128870964050293,
      "learning_rate": 2.9012999999999998e-05,
      "loss": 0.0482,
      "step": 9674
    },
    {
      "epoch": 17.85133887349954,
      "grad_norm": 0.9304009079933167,
      "learning_rate": 2.9015999999999998e-05,
      "loss": 0.0511,
      "step": 9675
    },
    {
      "epoch": 17.85318559556787,
      "grad_norm": 1.323746919631958,
      "learning_rate": 2.9019e-05,
      "loss": 0.0375,
      "step": 9676
    },
    {
      "epoch": 17.855032317636194,
      "grad_norm": 0.467035174369812,
      "learning_rate": 2.9022e-05,
      "loss": 0.039,
      "step": 9677
    },
    {
      "epoch": 17.856879039704523,
      "grad_norm": 0.37897157669067383,
      "learning_rate": 2.9025e-05,
      "loss": 0.0258,
      "step": 9678
    },
    {
      "epoch": 17.858725761772853,
      "grad_norm": 0.4991145730018616,
      "learning_rate": 2.9028e-05,
      "loss": 0.033,
      "step": 9679
    },
    {
      "epoch": 17.86057248384118,
      "grad_norm": 0.44403761625289917,
      "learning_rate": 2.9031e-05,
      "loss": 0.0419,
      "step": 9680
    },
    {
      "epoch": 17.86241920590951,
      "grad_norm": 0.37267547845840454,
      "learning_rate": 2.9034e-05,
      "loss": 0.0222,
      "step": 9681
    },
    {
      "epoch": 17.86426592797784,
      "grad_norm": 0.31096506118774414,
      "learning_rate": 2.9037e-05,
      "loss": 0.0163,
      "step": 9682
    },
    {
      "epoch": 17.86611265004617,
      "grad_norm": 0.3130188584327698,
      "learning_rate": 2.904e-05,
      "loss": 0.0174,
      "step": 9683
    },
    {
      "epoch": 17.8679593721145,
      "grad_norm": 0.22604769468307495,
      "learning_rate": 2.9043e-05,
      "loss": 0.0079,
      "step": 9684
    },
    {
      "epoch": 17.869806094182824,
      "grad_norm": 0.35768184065818787,
      "learning_rate": 2.9046e-05,
      "loss": 0.0237,
      "step": 9685
    },
    {
      "epoch": 17.871652816251153,
      "grad_norm": 0.2845885753631592,
      "learning_rate": 2.9049000000000003e-05,
      "loss": 0.0148,
      "step": 9686
    },
    {
      "epoch": 17.873499538319482,
      "grad_norm": 0.5197261571884155,
      "learning_rate": 2.9052000000000002e-05,
      "loss": 0.0163,
      "step": 9687
    },
    {
      "epoch": 17.87534626038781,
      "grad_norm": 0.3724684715270996,
      "learning_rate": 2.9055000000000002e-05,
      "loss": 0.0166,
      "step": 9688
    },
    {
      "epoch": 17.87719298245614,
      "grad_norm": 0.3093060851097107,
      "learning_rate": 2.9058000000000002e-05,
      "loss": 0.0115,
      "step": 9689
    },
    {
      "epoch": 17.87903970452447,
      "grad_norm": 0.5066264867782593,
      "learning_rate": 2.9061000000000002e-05,
      "loss": 0.023,
      "step": 9690
    },
    {
      "epoch": 17.8808864265928,
      "grad_norm": 0.32835444808006287,
      "learning_rate": 2.9064e-05,
      "loss": 0.0149,
      "step": 9691
    },
    {
      "epoch": 17.882733148661128,
      "grad_norm": 0.23925748467445374,
      "learning_rate": 2.9067e-05,
      "loss": 0.008,
      "step": 9692
    },
    {
      "epoch": 17.884579870729453,
      "grad_norm": 0.21648789942264557,
      "learning_rate": 2.907e-05,
      "loss": 0.0075,
      "step": 9693
    },
    {
      "epoch": 17.886426592797783,
      "grad_norm": 0.3470613360404968,
      "learning_rate": 2.9073e-05,
      "loss": 0.015,
      "step": 9694
    },
    {
      "epoch": 17.88827331486611,
      "grad_norm": 0.2070864886045456,
      "learning_rate": 2.9076e-05,
      "loss": 0.0048,
      "step": 9695
    },
    {
      "epoch": 17.89012003693444,
      "grad_norm": 0.48686373233795166,
      "learning_rate": 2.9079e-05,
      "loss": 0.0105,
      "step": 9696
    },
    {
      "epoch": 17.89196675900277,
      "grad_norm": 0.29262134432792664,
      "learning_rate": 2.9082e-05,
      "loss": 0.0092,
      "step": 9697
    },
    {
      "epoch": 17.8938134810711,
      "grad_norm": 0.3679444193840027,
      "learning_rate": 2.9085e-05,
      "loss": 0.0098,
      "step": 9698
    },
    {
      "epoch": 17.89566020313943,
      "grad_norm": 0.224870502948761,
      "learning_rate": 2.9088e-05,
      "loss": 0.0052,
      "step": 9699
    },
    {
      "epoch": 17.897506925207757,
      "grad_norm": 0.35356229543685913,
      "learning_rate": 2.9091e-05,
      "loss": 0.0115,
      "step": 9700
    },
    {
      "epoch": 17.899353647276087,
      "grad_norm": 0.3993721902370453,
      "learning_rate": 2.9094e-05,
      "loss": 0.0201,
      "step": 9701
    },
    {
      "epoch": 17.901200369344412,
      "grad_norm": 0.2684133052825928,
      "learning_rate": 2.9097e-05,
      "loss": 0.0096,
      "step": 9702
    },
    {
      "epoch": 17.90304709141274,
      "grad_norm": 0.3427678346633911,
      "learning_rate": 2.91e-05,
      "loss": 0.0139,
      "step": 9703
    },
    {
      "epoch": 17.90489381348107,
      "grad_norm": 0.36482563614845276,
      "learning_rate": 2.9103e-05,
      "loss": 0.0096,
      "step": 9704
    },
    {
      "epoch": 17.9067405355494,
      "grad_norm": 0.40752437710762024,
      "learning_rate": 2.9106e-05,
      "loss": 0.014,
      "step": 9705
    },
    {
      "epoch": 17.90858725761773,
      "grad_norm": 1.0441960096359253,
      "learning_rate": 2.9109000000000002e-05,
      "loss": 0.0208,
      "step": 9706
    },
    {
      "epoch": 17.910433979686058,
      "grad_norm": 0.2949042320251465,
      "learning_rate": 2.9112000000000002e-05,
      "loss": 0.0083,
      "step": 9707
    },
    {
      "epoch": 17.912280701754387,
      "grad_norm": 0.2150924652814865,
      "learning_rate": 2.9115e-05,
      "loss": 0.0062,
      "step": 9708
    },
    {
      "epoch": 17.914127423822716,
      "grad_norm": 0.8675482869148254,
      "learning_rate": 2.9118e-05,
      "loss": 0.018,
      "step": 9709
    },
    {
      "epoch": 17.91597414589104,
      "grad_norm": 0.7008863091468811,
      "learning_rate": 2.9121e-05,
      "loss": 0.0155,
      "step": 9710
    },
    {
      "epoch": 17.91782086795937,
      "grad_norm": 0.4291934370994568,
      "learning_rate": 2.9124e-05,
      "loss": 0.0142,
      "step": 9711
    },
    {
      "epoch": 17.9196675900277,
      "grad_norm": 0.8048538565635681,
      "learning_rate": 2.9127e-05,
      "loss": 0.0225,
      "step": 9712
    },
    {
      "epoch": 17.92151431209603,
      "grad_norm": 0.35469120740890503,
      "learning_rate": 2.913e-05,
      "loss": 0.0074,
      "step": 9713
    },
    {
      "epoch": 17.92336103416436,
      "grad_norm": 0.8907756209373474,
      "learning_rate": 2.9133e-05,
      "loss": 0.0212,
      "step": 9714
    },
    {
      "epoch": 17.925207756232687,
      "grad_norm": 1.0798656940460205,
      "learning_rate": 2.9136e-05,
      "loss": 0.1914,
      "step": 9715
    },
    {
      "epoch": 17.927054478301017,
      "grad_norm": 0.8650581240653992,
      "learning_rate": 2.9139000000000003e-05,
      "loss": 0.1628,
      "step": 9716
    },
    {
      "epoch": 17.928901200369346,
      "grad_norm": 0.8818985819816589,
      "learning_rate": 2.9142000000000003e-05,
      "loss": 0.1243,
      "step": 9717
    },
    {
      "epoch": 17.930747922437675,
      "grad_norm": 0.9086804986000061,
      "learning_rate": 2.9145000000000003e-05,
      "loss": 0.1201,
      "step": 9718
    },
    {
      "epoch": 17.932594644506,
      "grad_norm": 0.5983270406723022,
      "learning_rate": 2.9148000000000003e-05,
      "loss": 0.0855,
      "step": 9719
    },
    {
      "epoch": 17.93444136657433,
      "grad_norm": 0.6448835730552673,
      "learning_rate": 2.9151000000000003e-05,
      "loss": 0.0799,
      "step": 9720
    },
    {
      "epoch": 17.93628808864266,
      "grad_norm": 0.9719138145446777,
      "learning_rate": 2.9154e-05,
      "loss": 0.1022,
      "step": 9721
    },
    {
      "epoch": 17.938134810710988,
      "grad_norm": 0.4985361397266388,
      "learning_rate": 2.9157e-05,
      "loss": 0.0639,
      "step": 9722
    },
    {
      "epoch": 17.939981532779317,
      "grad_norm": 0.5949772000312805,
      "learning_rate": 2.916e-05,
      "loss": 0.0661,
      "step": 9723
    },
    {
      "epoch": 17.941828254847646,
      "grad_norm": 0.4565816819667816,
      "learning_rate": 2.9163e-05,
      "loss": 0.0676,
      "step": 9724
    },
    {
      "epoch": 17.943674976915975,
      "grad_norm": 0.7509490847587585,
      "learning_rate": 2.9165999999999998e-05,
      "loss": 0.0749,
      "step": 9725
    },
    {
      "epoch": 17.945521698984304,
      "grad_norm": 0.3464672565460205,
      "learning_rate": 2.9169e-05,
      "loss": 0.0221,
      "step": 9726
    },
    {
      "epoch": 17.94736842105263,
      "grad_norm": 0.39118891954421997,
      "learning_rate": 2.9172e-05,
      "loss": 0.0294,
      "step": 9727
    },
    {
      "epoch": 17.94921514312096,
      "grad_norm": 0.7514564394950867,
      "learning_rate": 2.9175e-05,
      "loss": 0.0638,
      "step": 9728
    },
    {
      "epoch": 17.95106186518929,
      "grad_norm": 0.35066482424736023,
      "learning_rate": 2.9178e-05,
      "loss": 0.025,
      "step": 9729
    },
    {
      "epoch": 17.952908587257618,
      "grad_norm": 0.4681917130947113,
      "learning_rate": 2.9181e-05,
      "loss": 0.0182,
      "step": 9730
    },
    {
      "epoch": 17.954755309325947,
      "grad_norm": 0.44393619894981384,
      "learning_rate": 2.9184e-05,
      "loss": 0.0193,
      "step": 9731
    },
    {
      "epoch": 17.956602031394276,
      "grad_norm": 0.31914278864860535,
      "learning_rate": 2.9187e-05,
      "loss": 0.0113,
      "step": 9732
    },
    {
      "epoch": 17.958448753462605,
      "grad_norm": 0.23968352377414703,
      "learning_rate": 2.919e-05,
      "loss": 0.0123,
      "step": 9733
    },
    {
      "epoch": 17.960295475530934,
      "grad_norm": 0.5183699131011963,
      "learning_rate": 2.9193e-05,
      "loss": 0.0124,
      "step": 9734
    },
    {
      "epoch": 17.96214219759926,
      "grad_norm": 0.19911415874958038,
      "learning_rate": 2.9196e-05,
      "loss": 0.008,
      "step": 9735
    },
    {
      "epoch": 17.96398891966759,
      "grad_norm": 0.5218525528907776,
      "learning_rate": 2.9199000000000003e-05,
      "loss": 0.0245,
      "step": 9736
    },
    {
      "epoch": 17.965835641735918,
      "grad_norm": 0.25067374110221863,
      "learning_rate": 2.9202000000000003e-05,
      "loss": 0.0136,
      "step": 9737
    },
    {
      "epoch": 17.967682363804247,
      "grad_norm": 0.3637305796146393,
      "learning_rate": 2.9205000000000002e-05,
      "loss": 0.0185,
      "step": 9738
    },
    {
      "epoch": 17.969529085872576,
      "grad_norm": 0.6495672464370728,
      "learning_rate": 2.9208000000000002e-05,
      "loss": 0.0206,
      "step": 9739
    },
    {
      "epoch": 17.971375807940905,
      "grad_norm": 0.589525043964386,
      "learning_rate": 2.9211000000000002e-05,
      "loss": 0.0179,
      "step": 9740
    },
    {
      "epoch": 17.973222530009235,
      "grad_norm": 0.46126195788383484,
      "learning_rate": 2.9214000000000002e-05,
      "loss": 0.0257,
      "step": 9741
    },
    {
      "epoch": 17.975069252077564,
      "grad_norm": 0.24324364960193634,
      "learning_rate": 2.9217e-05,
      "loss": 0.0114,
      "step": 9742
    },
    {
      "epoch": 17.97691597414589,
      "grad_norm": 0.44025933742523193,
      "learning_rate": 2.922e-05,
      "loss": 0.0161,
      "step": 9743
    },
    {
      "epoch": 17.97876269621422,
      "grad_norm": 0.3250148892402649,
      "learning_rate": 2.9223e-05,
      "loss": 0.0143,
      "step": 9744
    },
    {
      "epoch": 17.980609418282548,
      "grad_norm": 0.5272476673126221,
      "learning_rate": 2.9226e-05,
      "loss": 0.0188,
      "step": 9745
    },
    {
      "epoch": 17.982456140350877,
      "grad_norm": 2.31430983543396,
      "learning_rate": 2.9229e-05,
      "loss": 0.0309,
      "step": 9746
    },
    {
      "epoch": 17.984302862419206,
      "grad_norm": 0.20313288271427155,
      "learning_rate": 2.9232e-05,
      "loss": 0.0054,
      "step": 9747
    },
    {
      "epoch": 17.986149584487535,
      "grad_norm": 0.17994390428066254,
      "learning_rate": 2.9235e-05,
      "loss": 0.0069,
      "step": 9748
    },
    {
      "epoch": 17.987996306555864,
      "grad_norm": 0.675997257232666,
      "learning_rate": 2.9238e-05,
      "loss": 0.0066,
      "step": 9749
    },
    {
      "epoch": 17.989843028624193,
      "grad_norm": 0.5051863193511963,
      "learning_rate": 2.9241e-05,
      "loss": 0.0153,
      "step": 9750
    },
    {
      "epoch": 17.991689750692522,
      "grad_norm": 0.46311643719673157,
      "learning_rate": 2.9244e-05,
      "loss": 0.0179,
      "step": 9751
    },
    {
      "epoch": 17.993536472760848,
      "grad_norm": 0.38243719935417175,
      "learning_rate": 2.9247e-05,
      "loss": 0.0125,
      "step": 9752
    },
    {
      "epoch": 17.995383194829177,
      "grad_norm": 0.46555525064468384,
      "learning_rate": 2.925e-05,
      "loss": 0.034,
      "step": 9753
    },
    {
      "epoch": 17.997229916897506,
      "grad_norm": 0.41126081347465515,
      "learning_rate": 2.9253e-05,
      "loss": 0.0154,
      "step": 9754
    },
    {
      "epoch": 17.999076638965835,
      "grad_norm": 0.7898223996162415,
      "learning_rate": 2.9256e-05,
      "loss": 0.0344,
      "step": 9755
    },
    {
      "epoch": 18.0,
      "grad_norm": 0.8481200933456421,
      "learning_rate": 2.9259e-05,
      "loss": 0.0256,
      "step": 9756
    },
    {
      "epoch": 18.00184672206833,
      "grad_norm": 0.9342116713523865,
      "learning_rate": 2.9262000000000002e-05,
      "loss": 0.2025,
      "step": 9757
    },
    {
      "epoch": 18.00369344413666,
      "grad_norm": 0.6150652170181274,
      "learning_rate": 2.9265000000000002e-05,
      "loss": 0.1379,
      "step": 9758
    },
    {
      "epoch": 18.005540166204987,
      "grad_norm": 0.7522546052932739,
      "learning_rate": 2.9268e-05,
      "loss": 0.2386,
      "step": 9759
    },
    {
      "epoch": 18.007386888273317,
      "grad_norm": 1.3952076435089111,
      "learning_rate": 2.9271e-05,
      "loss": 0.1439,
      "step": 9760
    },
    {
      "epoch": 18.009233610341642,
      "grad_norm": 1.392382025718689,
      "learning_rate": 2.9274e-05,
      "loss": 0.1604,
      "step": 9761
    },
    {
      "epoch": 18.01108033240997,
      "grad_norm": 0.8652476072311401,
      "learning_rate": 2.9277e-05,
      "loss": 0.1213,
      "step": 9762
    },
    {
      "epoch": 18.0129270544783,
      "grad_norm": 0.510383665561676,
      "learning_rate": 2.928e-05,
      "loss": 0.0917,
      "step": 9763
    },
    {
      "epoch": 18.01477377654663,
      "grad_norm": 0.7107333540916443,
      "learning_rate": 2.9283e-05,
      "loss": 0.1355,
      "step": 9764
    },
    {
      "epoch": 18.01662049861496,
      "grad_norm": 0.5975661277770996,
      "learning_rate": 2.9286e-05,
      "loss": 0.0964,
      "step": 9765
    },
    {
      "epoch": 18.018467220683288,
      "grad_norm": 0.6261776089668274,
      "learning_rate": 2.9289e-05,
      "loss": 0.0617,
      "step": 9766
    },
    {
      "epoch": 18.020313942751617,
      "grad_norm": 0.3411554992198944,
      "learning_rate": 2.9292000000000003e-05,
      "loss": 0.0557,
      "step": 9767
    },
    {
      "epoch": 18.022160664819946,
      "grad_norm": 0.7305089235305786,
      "learning_rate": 2.9295000000000003e-05,
      "loss": 0.049,
      "step": 9768
    },
    {
      "epoch": 18.02400738688827,
      "grad_norm": 0.506694495677948,
      "learning_rate": 2.9298000000000003e-05,
      "loss": 0.0273,
      "step": 9769
    },
    {
      "epoch": 18.0258541089566,
      "grad_norm": 0.5924853682518005,
      "learning_rate": 2.9301e-05,
      "loss": 0.1094,
      "step": 9770
    },
    {
      "epoch": 18.02770083102493,
      "grad_norm": 0.35882166028022766,
      "learning_rate": 2.9304e-05,
      "loss": 0.0271,
      "step": 9771
    },
    {
      "epoch": 18.02954755309326,
      "grad_norm": 0.44903764128685,
      "learning_rate": 2.9307e-05,
      "loss": 0.0165,
      "step": 9772
    },
    {
      "epoch": 18.03139427516159,
      "grad_norm": 0.36786243319511414,
      "learning_rate": 2.931e-05,
      "loss": 0.0156,
      "step": 9773
    },
    {
      "epoch": 18.033240997229917,
      "grad_norm": 0.1827472448348999,
      "learning_rate": 2.9313e-05,
      "loss": 0.0079,
      "step": 9774
    },
    {
      "epoch": 18.035087719298247,
      "grad_norm": 0.36526259779930115,
      "learning_rate": 2.9316e-05,
      "loss": 0.0158,
      "step": 9775
    },
    {
      "epoch": 18.036934441366576,
      "grad_norm": 0.23194356262683868,
      "learning_rate": 2.9318999999999998e-05,
      "loss": 0.0098,
      "step": 9776
    },
    {
      "epoch": 18.0387811634349,
      "grad_norm": 0.6627047657966614,
      "learning_rate": 2.9322e-05,
      "loss": 0.0213,
      "step": 9777
    },
    {
      "epoch": 18.04062788550323,
      "grad_norm": 0.22040517628192902,
      "learning_rate": 2.9325e-05,
      "loss": 0.0105,
      "step": 9778
    },
    {
      "epoch": 18.04247460757156,
      "grad_norm": 0.223483607172966,
      "learning_rate": 2.9328e-05,
      "loss": 0.0093,
      "step": 9779
    },
    {
      "epoch": 18.04432132963989,
      "grad_norm": 0.6126001477241516,
      "learning_rate": 2.9331e-05,
      "loss": 0.0142,
      "step": 9780
    },
    {
      "epoch": 18.046168051708218,
      "grad_norm": 0.2615165114402771,
      "learning_rate": 2.9334e-05,
      "loss": 0.0172,
      "step": 9781
    },
    {
      "epoch": 18.048014773776547,
      "grad_norm": 0.3288552761077881,
      "learning_rate": 2.9337e-05,
      "loss": 0.0118,
      "step": 9782
    },
    {
      "epoch": 18.049861495844876,
      "grad_norm": 0.1888730674982071,
      "learning_rate": 2.934e-05,
      "loss": 0.009,
      "step": 9783
    },
    {
      "epoch": 18.051708217913205,
      "grad_norm": 0.3395959138870239,
      "learning_rate": 2.9343e-05,
      "loss": 0.0089,
      "step": 9784
    },
    {
      "epoch": 18.053554939981534,
      "grad_norm": 0.1359689086675644,
      "learning_rate": 2.9346e-05,
      "loss": 0.0043,
      "step": 9785
    },
    {
      "epoch": 18.05540166204986,
      "grad_norm": 0.3583788573741913,
      "learning_rate": 2.9349e-05,
      "loss": 0.0145,
      "step": 9786
    },
    {
      "epoch": 18.05724838411819,
      "grad_norm": 0.34380096197128296,
      "learning_rate": 2.9352000000000003e-05,
      "loss": 0.0089,
      "step": 9787
    },
    {
      "epoch": 18.05909510618652,
      "grad_norm": 0.2520500123500824,
      "learning_rate": 2.9355000000000003e-05,
      "loss": 0.0098,
      "step": 9788
    },
    {
      "epoch": 18.060941828254848,
      "grad_norm": 0.6030365824699402,
      "learning_rate": 2.9358000000000003e-05,
      "loss": 0.0146,
      "step": 9789
    },
    {
      "epoch": 18.062788550323177,
      "grad_norm": 0.29205048084259033,
      "learning_rate": 2.9361000000000002e-05,
      "loss": 0.004,
      "step": 9790
    },
    {
      "epoch": 18.064635272391506,
      "grad_norm": 0.3522581458091736,
      "learning_rate": 2.9364000000000002e-05,
      "loss": 0.0091,
      "step": 9791
    },
    {
      "epoch": 18.066481994459835,
      "grad_norm": 0.25486263632774353,
      "learning_rate": 2.9367000000000002e-05,
      "loss": 0.0066,
      "step": 9792
    },
    {
      "epoch": 18.068328716528164,
      "grad_norm": 0.2006990760564804,
      "learning_rate": 2.9370000000000002e-05,
      "loss": 0.012,
      "step": 9793
    },
    {
      "epoch": 18.07017543859649,
      "grad_norm": 0.19068430364131927,
      "learning_rate": 2.9373e-05,
      "loss": 0.0101,
      "step": 9794
    },
    {
      "epoch": 18.07202216066482,
      "grad_norm": 0.2882688045501709,
      "learning_rate": 2.9375999999999998e-05,
      "loss": 0.0086,
      "step": 9795
    },
    {
      "epoch": 18.073868882733148,
      "grad_norm": 0.27749103307724,
      "learning_rate": 2.9378999999999998e-05,
      "loss": 0.0172,
      "step": 9796
    },
    {
      "epoch": 18.075715604801477,
      "grad_norm": 0.29392507672309875,
      "learning_rate": 2.9382e-05,
      "loss": 0.0081,
      "step": 9797
    },
    {
      "epoch": 18.077562326869806,
      "grad_norm": 0.7844383120536804,
      "learning_rate": 2.9385e-05,
      "loss": 0.0278,
      "step": 9798
    },
    {
      "epoch": 18.079409048938135,
      "grad_norm": 0.18613365292549133,
      "learning_rate": 2.9388e-05,
      "loss": 0.0053,
      "step": 9799
    },
    {
      "epoch": 18.081255771006465,
      "grad_norm": 0.3608344793319702,
      "learning_rate": 2.9391e-05,
      "loss": 0.0109,
      "step": 9800
    },
    {
      "epoch": 18.083102493074794,
      "grad_norm": 0.32925644516944885,
      "learning_rate": 2.9394e-05,
      "loss": 0.008,
      "step": 9801
    },
    {
      "epoch": 18.08494921514312,
      "grad_norm": 0.21824046969413757,
      "learning_rate": 2.9397e-05,
      "loss": 0.0091,
      "step": 9802
    },
    {
      "epoch": 18.08679593721145,
      "grad_norm": 0.36572808027267456,
      "learning_rate": 2.94e-05,
      "loss": 0.0125,
      "step": 9803
    },
    {
      "epoch": 18.088642659279778,
      "grad_norm": 0.40578538179397583,
      "learning_rate": 2.9403e-05,
      "loss": 0.0148,
      "step": 9804
    },
    {
      "epoch": 18.090489381348107,
      "grad_norm": 0.4316546320915222,
      "learning_rate": 2.9406e-05,
      "loss": 0.0146,
      "step": 9805
    },
    {
      "epoch": 18.092336103416436,
      "grad_norm": 0.383820503950119,
      "learning_rate": 2.9409e-05,
      "loss": 0.007,
      "step": 9806
    },
    {
      "epoch": 18.094182825484765,
      "grad_norm": 0.9452648162841797,
      "learning_rate": 2.9412000000000002e-05,
      "loss": 0.2356,
      "step": 9807
    },
    {
      "epoch": 18.096029547553094,
      "grad_norm": 0.7241860628128052,
      "learning_rate": 2.9415000000000002e-05,
      "loss": 0.1892,
      "step": 9808
    },
    {
      "epoch": 18.097876269621423,
      "grad_norm": 1.6240252256393433,
      "learning_rate": 2.9418000000000002e-05,
      "loss": 0.0974,
      "step": 9809
    },
    {
      "epoch": 18.099722991689752,
      "grad_norm": 0.9428126811981201,
      "learning_rate": 2.9421000000000002e-05,
      "loss": 0.133,
      "step": 9810
    },
    {
      "epoch": 18.101569713758078,
      "grad_norm": 0.7279912233352661,
      "learning_rate": 2.9424e-05,
      "loss": 0.1401,
      "step": 9811
    },
    {
      "epoch": 18.103416435826407,
      "grad_norm": 0.5566812753677368,
      "learning_rate": 2.9427e-05,
      "loss": 0.0827,
      "step": 9812
    },
    {
      "epoch": 18.105263157894736,
      "grad_norm": 0.34377196431159973,
      "learning_rate": 2.943e-05,
      "loss": 0.0868,
      "step": 9813
    },
    {
      "epoch": 18.107109879963065,
      "grad_norm": 0.39494451880455017,
      "learning_rate": 2.9433e-05,
      "loss": 0.0655,
      "step": 9814
    },
    {
      "epoch": 18.108956602031395,
      "grad_norm": 0.5374976396560669,
      "learning_rate": 2.9436e-05,
      "loss": 0.0541,
      "step": 9815
    },
    {
      "epoch": 18.110803324099724,
      "grad_norm": 0.5303835272789001,
      "learning_rate": 2.9439e-05,
      "loss": 0.0617,
      "step": 9816
    },
    {
      "epoch": 18.112650046168053,
      "grad_norm": 0.4022885859012604,
      "learning_rate": 2.9442000000000004e-05,
      "loss": 0.0652,
      "step": 9817
    },
    {
      "epoch": 18.114496768236382,
      "grad_norm": 0.3138822019100189,
      "learning_rate": 2.9445000000000004e-05,
      "loss": 0.0368,
      "step": 9818
    },
    {
      "epoch": 18.116343490304708,
      "grad_norm": 0.4218757450580597,
      "learning_rate": 2.9448e-05,
      "loss": 0.0383,
      "step": 9819
    },
    {
      "epoch": 18.118190212373037,
      "grad_norm": 1.328559398651123,
      "learning_rate": 2.9451e-05,
      "loss": 0.0613,
      "step": 9820
    },
    {
      "epoch": 18.120036934441366,
      "grad_norm": 0.26926133036613464,
      "learning_rate": 2.9454e-05,
      "loss": 0.0326,
      "step": 9821
    },
    {
      "epoch": 18.121883656509695,
      "grad_norm": 0.36706602573394775,
      "learning_rate": 2.9457e-05,
      "loss": 0.0215,
      "step": 9822
    },
    {
      "epoch": 18.123730378578024,
      "grad_norm": 0.5583202838897705,
      "learning_rate": 2.946e-05,
      "loss": 0.0217,
      "step": 9823
    },
    {
      "epoch": 18.125577100646353,
      "grad_norm": 0.38190707564353943,
      "learning_rate": 2.9463e-05,
      "loss": 0.0223,
      "step": 9824
    },
    {
      "epoch": 18.127423822714682,
      "grad_norm": 0.4724488854408264,
      "learning_rate": 2.9466e-05,
      "loss": 0.0485,
      "step": 9825
    },
    {
      "epoch": 18.12927054478301,
      "grad_norm": 0.3241349458694458,
      "learning_rate": 2.9469e-05,
      "loss": 0.0307,
      "step": 9826
    },
    {
      "epoch": 18.131117266851337,
      "grad_norm": 0.12482421100139618,
      "learning_rate": 2.9472000000000002e-05,
      "loss": 0.0056,
      "step": 9827
    },
    {
      "epoch": 18.132963988919666,
      "grad_norm": 0.2672232985496521,
      "learning_rate": 2.9475e-05,
      "loss": 0.0254,
      "step": 9828
    },
    {
      "epoch": 18.134810710987995,
      "grad_norm": 0.27123939990997314,
      "learning_rate": 2.9478e-05,
      "loss": 0.0135,
      "step": 9829
    },
    {
      "epoch": 18.136657433056325,
      "grad_norm": 0.25711241364479065,
      "learning_rate": 2.9481e-05,
      "loss": 0.0118,
      "step": 9830
    },
    {
      "epoch": 18.138504155124654,
      "grad_norm": 0.540631115436554,
      "learning_rate": 2.9484e-05,
      "loss": 0.0175,
      "step": 9831
    },
    {
      "epoch": 18.140350877192983,
      "grad_norm": 0.517338216304779,
      "learning_rate": 2.9487e-05,
      "loss": 0.0123,
      "step": 9832
    },
    {
      "epoch": 18.142197599261312,
      "grad_norm": 0.19550421833992004,
      "learning_rate": 2.949e-05,
      "loss": 0.0072,
      "step": 9833
    },
    {
      "epoch": 18.14404432132964,
      "grad_norm": 0.24754598736763,
      "learning_rate": 2.9493e-05,
      "loss": 0.008,
      "step": 9834
    },
    {
      "epoch": 18.14589104339797,
      "grad_norm": 0.18133080005645752,
      "learning_rate": 2.9496e-05,
      "loss": 0.0108,
      "step": 9835
    },
    {
      "epoch": 18.147737765466296,
      "grad_norm": 0.27041253447532654,
      "learning_rate": 2.9499e-05,
      "loss": 0.0143,
      "step": 9836
    },
    {
      "epoch": 18.149584487534625,
      "grad_norm": 0.22081269323825836,
      "learning_rate": 2.9502000000000003e-05,
      "loss": 0.0075,
      "step": 9837
    },
    {
      "epoch": 18.151431209602954,
      "grad_norm": 0.37559419870376587,
      "learning_rate": 2.9505000000000003e-05,
      "loss": 0.0077,
      "step": 9838
    },
    {
      "epoch": 18.153277931671283,
      "grad_norm": 0.22016127407550812,
      "learning_rate": 2.9508000000000003e-05,
      "loss": 0.0108,
      "step": 9839
    },
    {
      "epoch": 18.155124653739612,
      "grad_norm": 0.26403334736824036,
      "learning_rate": 2.9511000000000003e-05,
      "loss": 0.0069,
      "step": 9840
    },
    {
      "epoch": 18.15697137580794,
      "grad_norm": 0.6012740731239319,
      "learning_rate": 2.9514000000000002e-05,
      "loss": 0.0201,
      "step": 9841
    },
    {
      "epoch": 18.15881809787627,
      "grad_norm": 0.3800670802593231,
      "learning_rate": 2.9517000000000002e-05,
      "loss": 0.0085,
      "step": 9842
    },
    {
      "epoch": 18.1606648199446,
      "grad_norm": 0.3758285343647003,
      "learning_rate": 2.9520000000000002e-05,
      "loss": 0.0101,
      "step": 9843
    },
    {
      "epoch": 18.162511542012926,
      "grad_norm": 0.2834491729736328,
      "learning_rate": 2.9523e-05,
      "loss": 0.0115,
      "step": 9844
    },
    {
      "epoch": 18.164358264081255,
      "grad_norm": 0.3003970682621002,
      "learning_rate": 2.9525999999999998e-05,
      "loss": 0.0082,
      "step": 9845
    },
    {
      "epoch": 18.166204986149584,
      "grad_norm": 0.4620857238769531,
      "learning_rate": 2.9528999999999998e-05,
      "loss": 0.0191,
      "step": 9846
    },
    {
      "epoch": 18.168051708217913,
      "grad_norm": 0.27000996470451355,
      "learning_rate": 2.9532e-05,
      "loss": 0.0079,
      "step": 9847
    },
    {
      "epoch": 18.169898430286242,
      "grad_norm": 0.33361855149269104,
      "learning_rate": 2.9535e-05,
      "loss": 0.007,
      "step": 9848
    },
    {
      "epoch": 18.17174515235457,
      "grad_norm": 0.3398788571357727,
      "learning_rate": 2.9538e-05,
      "loss": 0.0095,
      "step": 9849
    },
    {
      "epoch": 18.1735918744229,
      "grad_norm": 0.17393405735492706,
      "learning_rate": 2.9541e-05,
      "loss": 0.0032,
      "step": 9850
    },
    {
      "epoch": 18.17543859649123,
      "grad_norm": 0.3088524341583252,
      "learning_rate": 2.9544e-05,
      "loss": 0.0094,
      "step": 9851
    },
    {
      "epoch": 18.177285318559555,
      "grad_norm": 1.077933430671692,
      "learning_rate": 2.9547e-05,
      "loss": 0.0119,
      "step": 9852
    },
    {
      "epoch": 18.179132040627884,
      "grad_norm": 0.48251262307167053,
      "learning_rate": 2.955e-05,
      "loss": 0.0057,
      "step": 9853
    },
    {
      "epoch": 18.180978762696213,
      "grad_norm": 0.5218617916107178,
      "learning_rate": 2.9553e-05,
      "loss": 0.0137,
      "step": 9854
    },
    {
      "epoch": 18.182825484764543,
      "grad_norm": 0.6712053418159485,
      "learning_rate": 2.9556e-05,
      "loss": 0.0169,
      "step": 9855
    },
    {
      "epoch": 18.18467220683287,
      "grad_norm": 1.1995861530303955,
      "learning_rate": 2.9559e-05,
      "loss": 0.0415,
      "step": 9856
    },
    {
      "epoch": 18.1865189289012,
      "grad_norm": 0.5802194476127625,
      "learning_rate": 2.9562000000000003e-05,
      "loss": 0.1692,
      "step": 9857
    },
    {
      "epoch": 18.18836565096953,
      "grad_norm": 0.6861435174942017,
      "learning_rate": 2.9565000000000002e-05,
      "loss": 0.142,
      "step": 9858
    },
    {
      "epoch": 18.19021237303786,
      "grad_norm": 0.7514278888702393,
      "learning_rate": 2.9568000000000002e-05,
      "loss": 0.1626,
      "step": 9859
    },
    {
      "epoch": 18.19205909510619,
      "grad_norm": 0.5463578104972839,
      "learning_rate": 2.9571000000000002e-05,
      "loss": 0.0996,
      "step": 9860
    },
    {
      "epoch": 18.193905817174514,
      "grad_norm": 0.6229665875434875,
      "learning_rate": 2.9574000000000002e-05,
      "loss": 0.075,
      "step": 9861
    },
    {
      "epoch": 18.195752539242843,
      "grad_norm": 0.8171252012252808,
      "learning_rate": 2.9577e-05,
      "loss": 0.1555,
      "step": 9862
    },
    {
      "epoch": 18.197599261311172,
      "grad_norm": 0.4324992001056671,
      "learning_rate": 2.958e-05,
      "loss": 0.0936,
      "step": 9863
    },
    {
      "epoch": 18.1994459833795,
      "grad_norm": 0.7613241672515869,
      "learning_rate": 2.9583e-05,
      "loss": 0.077,
      "step": 9864
    },
    {
      "epoch": 18.20129270544783,
      "grad_norm": 0.3602001368999481,
      "learning_rate": 2.9586e-05,
      "loss": 0.0599,
      "step": 9865
    },
    {
      "epoch": 18.20313942751616,
      "grad_norm": 0.4885939359664917,
      "learning_rate": 2.9589e-05,
      "loss": 0.0588,
      "step": 9866
    },
    {
      "epoch": 18.20498614958449,
      "grad_norm": 0.5051184892654419,
      "learning_rate": 2.9592000000000004e-05,
      "loss": 0.061,
      "step": 9867
    },
    {
      "epoch": 18.206832871652818,
      "grad_norm": 0.34616658091545105,
      "learning_rate": 2.9595e-05,
      "loss": 0.0352,
      "step": 9868
    },
    {
      "epoch": 18.208679593721143,
      "grad_norm": 0.5483611226081848,
      "learning_rate": 2.9598e-05,
      "loss": 0.0965,
      "step": 9869
    },
    {
      "epoch": 18.210526315789473,
      "grad_norm": 0.45261475443840027,
      "learning_rate": 2.9601e-05,
      "loss": 0.0315,
      "step": 9870
    },
    {
      "epoch": 18.2123730378578,
      "grad_norm": 0.3057132363319397,
      "learning_rate": 2.9604e-05,
      "loss": 0.0235,
      "step": 9871
    },
    {
      "epoch": 18.21421975992613,
      "grad_norm": 0.20985980331897736,
      "learning_rate": 2.9607e-05,
      "loss": 0.0176,
      "step": 9872
    },
    {
      "epoch": 18.21606648199446,
      "grad_norm": 0.3141608238220215,
      "learning_rate": 2.961e-05,
      "loss": 0.0464,
      "step": 9873
    },
    {
      "epoch": 18.21791320406279,
      "grad_norm": 0.5170801281929016,
      "learning_rate": 2.9613e-05,
      "loss": 0.0201,
      "step": 9874
    },
    {
      "epoch": 18.21975992613112,
      "grad_norm": 0.8139154314994812,
      "learning_rate": 2.9616e-05,
      "loss": 0.0407,
      "step": 9875
    },
    {
      "epoch": 18.221606648199447,
      "grad_norm": 0.22715677320957184,
      "learning_rate": 2.9619e-05,
      "loss": 0.0155,
      "step": 9876
    },
    {
      "epoch": 18.223453370267773,
      "grad_norm": 0.33089739084243774,
      "learning_rate": 2.9622000000000002e-05,
      "loss": 0.0189,
      "step": 9877
    },
    {
      "epoch": 18.225300092336102,
      "grad_norm": 0.8802370429039001,
      "learning_rate": 2.9625000000000002e-05,
      "loss": 0.0154,
      "step": 9878
    },
    {
      "epoch": 18.22714681440443,
      "grad_norm": 0.2362622320652008,
      "learning_rate": 2.9628e-05,
      "loss": 0.0113,
      "step": 9879
    },
    {
      "epoch": 18.22899353647276,
      "grad_norm": 0.21023108065128326,
      "learning_rate": 2.9631e-05,
      "loss": 0.0074,
      "step": 9880
    },
    {
      "epoch": 18.23084025854109,
      "grad_norm": 0.39017900824546814,
      "learning_rate": 2.9634e-05,
      "loss": 0.011,
      "step": 9881
    },
    {
      "epoch": 18.23268698060942,
      "grad_norm": 0.31846851110458374,
      "learning_rate": 2.9637e-05,
      "loss": 0.0085,
      "step": 9882
    },
    {
      "epoch": 18.234533702677748,
      "grad_norm": 0.32058441638946533,
      "learning_rate": 2.964e-05,
      "loss": 0.0197,
      "step": 9883
    },
    {
      "epoch": 18.236380424746077,
      "grad_norm": 0.27096280455589294,
      "learning_rate": 2.9643e-05,
      "loss": 0.0448,
      "step": 9884
    },
    {
      "epoch": 18.238227146814406,
      "grad_norm": 0.28840863704681396,
      "learning_rate": 2.9646e-05,
      "loss": 0.012,
      "step": 9885
    },
    {
      "epoch": 18.24007386888273,
      "grad_norm": 0.2708560824394226,
      "learning_rate": 2.9649e-05,
      "loss": 0.0074,
      "step": 9886
    },
    {
      "epoch": 18.24192059095106,
      "grad_norm": 0.3992142677307129,
      "learning_rate": 2.9652e-05,
      "loss": 0.0113,
      "step": 9887
    },
    {
      "epoch": 18.24376731301939,
      "grad_norm": 0.2929539978504181,
      "learning_rate": 2.9655000000000003e-05,
      "loss": 0.0167,
      "step": 9888
    },
    {
      "epoch": 18.24561403508772,
      "grad_norm": 0.4625506103038788,
      "learning_rate": 2.9658000000000003e-05,
      "loss": 0.0103,
      "step": 9889
    },
    {
      "epoch": 18.24746075715605,
      "grad_norm": 0.2780413329601288,
      "learning_rate": 2.9661000000000003e-05,
      "loss": 0.0091,
      "step": 9890
    },
    {
      "epoch": 18.249307479224377,
      "grad_norm": 0.26614031195640564,
      "learning_rate": 2.9664000000000003e-05,
      "loss": 0.006,
      "step": 9891
    },
    {
      "epoch": 18.251154201292707,
      "grad_norm": 0.3119448125362396,
      "learning_rate": 2.9667000000000002e-05,
      "loss": 0.0069,
      "step": 9892
    },
    {
      "epoch": 18.253000923361036,
      "grad_norm": 0.4528767466545105,
      "learning_rate": 2.967e-05,
      "loss": 0.0091,
      "step": 9893
    },
    {
      "epoch": 18.25484764542936,
      "grad_norm": 0.24236981570720673,
      "learning_rate": 2.9673e-05,
      "loss": 0.0085,
      "step": 9894
    },
    {
      "epoch": 18.25669436749769,
      "grad_norm": 0.31961968541145325,
      "learning_rate": 2.9676e-05,
      "loss": 0.01,
      "step": 9895
    },
    {
      "epoch": 18.25854108956602,
      "grad_norm": 0.5327759385108948,
      "learning_rate": 2.9678999999999998e-05,
      "loss": 0.018,
      "step": 9896
    },
    {
      "epoch": 18.26038781163435,
      "grad_norm": 0.2756796181201935,
      "learning_rate": 2.9681999999999998e-05,
      "loss": 0.0066,
      "step": 9897
    },
    {
      "epoch": 18.262234533702678,
      "grad_norm": 0.2694704234600067,
      "learning_rate": 2.9685e-05,
      "loss": 0.0067,
      "step": 9898
    },
    {
      "epoch": 18.264081255771007,
      "grad_norm": 0.877261757850647,
      "learning_rate": 2.9688e-05,
      "loss": 0.0263,
      "step": 9899
    },
    {
      "epoch": 18.265927977839336,
      "grad_norm": 0.18148718774318695,
      "learning_rate": 2.9691e-05,
      "loss": 0.0065,
      "step": 9900
    },
    {
      "epoch": 18.267774699907665,
      "grad_norm": 0.17909610271453857,
      "learning_rate": 2.9694e-05,
      "loss": 0.0051,
      "step": 9901
    },
    {
      "epoch": 18.26962142197599,
      "grad_norm": 0.2902460992336273,
      "learning_rate": 2.9697e-05,
      "loss": 0.0084,
      "step": 9902
    },
    {
      "epoch": 18.27146814404432,
      "grad_norm": 0.45404717326164246,
      "learning_rate": 2.97e-05,
      "loss": 0.0147,
      "step": 9903
    },
    {
      "epoch": 18.27331486611265,
      "grad_norm": 0.3529004454612732,
      "learning_rate": 2.9703e-05,
      "loss": 0.0103,
      "step": 9904
    },
    {
      "epoch": 18.27516158818098,
      "grad_norm": 0.30562788248062134,
      "learning_rate": 2.9706e-05,
      "loss": 0.0074,
      "step": 9905
    },
    {
      "epoch": 18.277008310249307,
      "grad_norm": 0.3384891748428345,
      "learning_rate": 2.9709e-05,
      "loss": 0.0108,
      "step": 9906
    },
    {
      "epoch": 18.278855032317637,
      "grad_norm": 0.6173661351203918,
      "learning_rate": 2.9712e-05,
      "loss": 0.1869,
      "step": 9907
    },
    {
      "epoch": 18.280701754385966,
      "grad_norm": 0.5092833042144775,
      "learning_rate": 2.9715000000000003e-05,
      "loss": 0.1286,
      "step": 9908
    },
    {
      "epoch": 18.282548476454295,
      "grad_norm": 0.58086758852005,
      "learning_rate": 2.9718000000000002e-05,
      "loss": 0.1219,
      "step": 9909
    },
    {
      "epoch": 18.284395198522624,
      "grad_norm": 3.074801206588745,
      "learning_rate": 2.9721000000000002e-05,
      "loss": 0.1325,
      "step": 9910
    },
    {
      "epoch": 18.28624192059095,
      "grad_norm": 0.6098874807357788,
      "learning_rate": 2.9724000000000002e-05,
      "loss": 0.1242,
      "step": 9911
    },
    {
      "epoch": 18.28808864265928,
      "grad_norm": 0.6750804781913757,
      "learning_rate": 2.9727000000000002e-05,
      "loss": 0.049,
      "step": 9912
    },
    {
      "epoch": 18.289935364727608,
      "grad_norm": 0.7178696990013123,
      "learning_rate": 2.973e-05,
      "loss": 0.0865,
      "step": 9913
    },
    {
      "epoch": 18.291782086795937,
      "grad_norm": 0.43659573793411255,
      "learning_rate": 2.9733e-05,
      "loss": 0.0552,
      "step": 9914
    },
    {
      "epoch": 18.293628808864266,
      "grad_norm": 0.4207955002784729,
      "learning_rate": 2.9736e-05,
      "loss": 0.0653,
      "step": 9915
    },
    {
      "epoch": 18.295475530932595,
      "grad_norm": 0.3190794885158539,
      "learning_rate": 2.9739e-05,
      "loss": 0.0325,
      "step": 9916
    },
    {
      "epoch": 18.297322253000925,
      "grad_norm": 0.27810654044151306,
      "learning_rate": 2.9742e-05,
      "loss": 0.0386,
      "step": 9917
    },
    {
      "epoch": 18.299168975069254,
      "grad_norm": 0.6035851240158081,
      "learning_rate": 2.9745e-05,
      "loss": 0.0791,
      "step": 9918
    },
    {
      "epoch": 18.30101569713758,
      "grad_norm": 0.36111941933631897,
      "learning_rate": 2.9748e-05,
      "loss": 0.0429,
      "step": 9919
    },
    {
      "epoch": 18.30286241920591,
      "grad_norm": 0.2906637191772461,
      "learning_rate": 2.9751e-05,
      "loss": 0.0245,
      "step": 9920
    },
    {
      "epoch": 18.304709141274238,
      "grad_norm": 0.3560005724430084,
      "learning_rate": 2.9754e-05,
      "loss": 0.0273,
      "step": 9921
    },
    {
      "epoch": 18.306555863342567,
      "grad_norm": 0.3637906312942505,
      "learning_rate": 2.9757e-05,
      "loss": 0.0217,
      "step": 9922
    },
    {
      "epoch": 18.308402585410896,
      "grad_norm": 0.2707814574241638,
      "learning_rate": 2.976e-05,
      "loss": 0.0131,
      "step": 9923
    },
    {
      "epoch": 18.310249307479225,
      "grad_norm": 0.4414331018924713,
      "learning_rate": 2.9763e-05,
      "loss": 0.0129,
      "step": 9924
    },
    {
      "epoch": 18.312096029547554,
      "grad_norm": 0.3390336334705353,
      "learning_rate": 2.9766e-05,
      "loss": 0.014,
      "step": 9925
    },
    {
      "epoch": 18.313942751615883,
      "grad_norm": 0.35256242752075195,
      "learning_rate": 2.9769e-05,
      "loss": 0.0123,
      "step": 9926
    },
    {
      "epoch": 18.31578947368421,
      "grad_norm": 0.2255878895521164,
      "learning_rate": 2.9772e-05,
      "loss": 0.0143,
      "step": 9927
    },
    {
      "epoch": 18.317636195752538,
      "grad_norm": 0.19617587327957153,
      "learning_rate": 2.9775000000000002e-05,
      "loss": 0.0119,
      "step": 9928
    },
    {
      "epoch": 18.319482917820867,
      "grad_norm": 0.5293916463851929,
      "learning_rate": 2.9778000000000002e-05,
      "loss": 0.0222,
      "step": 9929
    },
    {
      "epoch": 18.321329639889196,
      "grad_norm": 0.3983894884586334,
      "learning_rate": 2.9781e-05,
      "loss": 0.0119,
      "step": 9930
    },
    {
      "epoch": 18.323176361957525,
      "grad_norm": 0.1999145895242691,
      "learning_rate": 2.9784e-05,
      "loss": 0.008,
      "step": 9931
    },
    {
      "epoch": 18.325023084025855,
      "grad_norm": 0.21079927682876587,
      "learning_rate": 2.9787e-05,
      "loss": 0.0083,
      "step": 9932
    },
    {
      "epoch": 18.326869806094184,
      "grad_norm": 0.18804602324962616,
      "learning_rate": 2.979e-05,
      "loss": 0.0102,
      "step": 9933
    },
    {
      "epoch": 18.328716528162513,
      "grad_norm": 0.2636594772338867,
      "learning_rate": 2.9793e-05,
      "loss": 0.0094,
      "step": 9934
    },
    {
      "epoch": 18.330563250230842,
      "grad_norm": 0.22981365025043488,
      "learning_rate": 2.9796e-05,
      "loss": 0.0278,
      "step": 9935
    },
    {
      "epoch": 18.332409972299168,
      "grad_norm": 0.2265322506427765,
      "learning_rate": 2.9799e-05,
      "loss": 0.0121,
      "step": 9936
    },
    {
      "epoch": 18.334256694367497,
      "grad_norm": 0.2998930513858795,
      "learning_rate": 2.9802e-05,
      "loss": 0.0338,
      "step": 9937
    },
    {
      "epoch": 18.336103416435826,
      "grad_norm": 0.44089028239250183,
      "learning_rate": 2.9805000000000003e-05,
      "loss": 0.0103,
      "step": 9938
    },
    {
      "epoch": 18.337950138504155,
      "grad_norm": 0.36817842721939087,
      "learning_rate": 2.9808000000000003e-05,
      "loss": 0.0167,
      "step": 9939
    },
    {
      "epoch": 18.339796860572484,
      "grad_norm": 0.17441830039024353,
      "learning_rate": 2.9811000000000003e-05,
      "loss": 0.0089,
      "step": 9940
    },
    {
      "epoch": 18.341643582640813,
      "grad_norm": 0.5821911096572876,
      "learning_rate": 2.9814000000000003e-05,
      "loss": 0.0133,
      "step": 9941
    },
    {
      "epoch": 18.343490304709142,
      "grad_norm": 0.21145932376384735,
      "learning_rate": 2.9817e-05,
      "loss": 0.0084,
      "step": 9942
    },
    {
      "epoch": 18.34533702677747,
      "grad_norm": 0.38200899958610535,
      "learning_rate": 2.982e-05,
      "loss": 0.0093,
      "step": 9943
    },
    {
      "epoch": 18.347183748845797,
      "grad_norm": 0.6706154942512512,
      "learning_rate": 2.9823e-05,
      "loss": 0.0113,
      "step": 9944
    },
    {
      "epoch": 18.349030470914126,
      "grad_norm": 0.31891512870788574,
      "learning_rate": 2.9826e-05,
      "loss": 0.0078,
      "step": 9945
    },
    {
      "epoch": 18.350877192982455,
      "grad_norm": 0.47149425745010376,
      "learning_rate": 2.9829e-05,
      "loss": 0.0139,
      "step": 9946
    },
    {
      "epoch": 18.352723915050785,
      "grad_norm": 0.30881768465042114,
      "learning_rate": 2.9831999999999998e-05,
      "loss": 0.0081,
      "step": 9947
    },
    {
      "epoch": 18.354570637119114,
      "grad_norm": 0.6706964373588562,
      "learning_rate": 2.9835e-05,
      "loss": 0.0267,
      "step": 9948
    },
    {
      "epoch": 18.356417359187443,
      "grad_norm": 0.5232123136520386,
      "learning_rate": 2.9838e-05,
      "loss": 0.0109,
      "step": 9949
    },
    {
      "epoch": 18.358264081255772,
      "grad_norm": 0.3877739906311035,
      "learning_rate": 2.9841e-05,
      "loss": 0.0165,
      "step": 9950
    },
    {
      "epoch": 18.3601108033241,
      "grad_norm": 0.4155578017234802,
      "learning_rate": 2.9844e-05,
      "loss": 0.0135,
      "step": 9951
    },
    {
      "epoch": 18.361957525392427,
      "grad_norm": 0.5564242005348206,
      "learning_rate": 2.9847e-05,
      "loss": 0.0188,
      "step": 9952
    },
    {
      "epoch": 18.363804247460756,
      "grad_norm": 0.4501713812351227,
      "learning_rate": 2.985e-05,
      "loss": 0.0102,
      "step": 9953
    },
    {
      "epoch": 18.365650969529085,
      "grad_norm": 0.15342417359352112,
      "learning_rate": 2.9853e-05,
      "loss": 0.0051,
      "step": 9954
    },
    {
      "epoch": 18.367497691597414,
      "grad_norm": 0.46972164511680603,
      "learning_rate": 2.9856e-05,
      "loss": 0.0154,
      "step": 9955
    },
    {
      "epoch": 18.369344413665743,
      "grad_norm": 0.4050143361091614,
      "learning_rate": 2.9859e-05,
      "loss": 0.0167,
      "step": 9956
    },
    {
      "epoch": 18.371191135734072,
      "grad_norm": 0.7018055319786072,
      "learning_rate": 2.9862e-05,
      "loss": 0.1757,
      "step": 9957
    },
    {
      "epoch": 18.3730378578024,
      "grad_norm": 0.46269240975379944,
      "learning_rate": 2.9865000000000003e-05,
      "loss": 0.1259,
      "step": 9958
    },
    {
      "epoch": 18.37488457987073,
      "grad_norm": 0.3974771201610565,
      "learning_rate": 2.9868000000000003e-05,
      "loss": 0.0835,
      "step": 9959
    },
    {
      "epoch": 18.37673130193906,
      "grad_norm": 0.558796226978302,
      "learning_rate": 2.9871000000000003e-05,
      "loss": 0.094,
      "step": 9960
    },
    {
      "epoch": 18.378578024007385,
      "grad_norm": 0.4177483320236206,
      "learning_rate": 2.9874000000000002e-05,
      "loss": 0.0545,
      "step": 9961
    },
    {
      "epoch": 18.380424746075715,
      "grad_norm": 0.4701012969017029,
      "learning_rate": 2.9877000000000002e-05,
      "loss": 0.0512,
      "step": 9962
    },
    {
      "epoch": 18.382271468144044,
      "grad_norm": 0.5373863577842712,
      "learning_rate": 2.9880000000000002e-05,
      "loss": 0.0975,
      "step": 9963
    },
    {
      "epoch": 18.384118190212373,
      "grad_norm": 0.5024900436401367,
      "learning_rate": 2.9883000000000002e-05,
      "loss": 0.0734,
      "step": 9964
    },
    {
      "epoch": 18.385964912280702,
      "grad_norm": 0.4085754454135895,
      "learning_rate": 2.9886e-05,
      "loss": 0.0475,
      "step": 9965
    },
    {
      "epoch": 18.38781163434903,
      "grad_norm": 0.4630957245826721,
      "learning_rate": 2.9889e-05,
      "loss": 0.0739,
      "step": 9966
    },
    {
      "epoch": 18.38965835641736,
      "grad_norm": 0.7340787649154663,
      "learning_rate": 2.9891999999999998e-05,
      "loss": 0.0373,
      "step": 9967
    },
    {
      "epoch": 18.39150507848569,
      "grad_norm": 0.3731462061405182,
      "learning_rate": 2.9895e-05,
      "loss": 0.0265,
      "step": 9968
    },
    {
      "epoch": 18.393351800554015,
      "grad_norm": 0.7383576035499573,
      "learning_rate": 2.9898e-05,
      "loss": 0.0926,
      "step": 9969
    },
    {
      "epoch": 18.395198522622344,
      "grad_norm": 0.5255272388458252,
      "learning_rate": 2.9901e-05,
      "loss": 0.0295,
      "step": 9970
    },
    {
      "epoch": 18.397045244690673,
      "grad_norm": 0.45948368310928345,
      "learning_rate": 2.9904e-05,
      "loss": 0.0407,
      "step": 9971
    },
    {
      "epoch": 18.398891966759003,
      "grad_norm": 0.8766978979110718,
      "learning_rate": 2.9907e-05,
      "loss": 0.0314,
      "step": 9972
    },
    {
      "epoch": 18.40073868882733,
      "grad_norm": 24.089262008666992,
      "learning_rate": 2.991e-05,
      "loss": 0.8891,
      "step": 9973
    },
    {
      "epoch": 18.40258541089566,
      "grad_norm": 4.03217887878418,
      "learning_rate": 2.9913e-05,
      "loss": 0.1203,
      "step": 9974
    },
    {
      "epoch": 18.40443213296399,
      "grad_norm": 0.5048349499702454,
      "learning_rate": 2.9916e-05,
      "loss": 0.0239,
      "step": 9975
    },
    {
      "epoch": 18.40627885503232,
      "grad_norm": 0.3953116536140442,
      "learning_rate": 2.9919e-05,
      "loss": 0.0257,
      "step": 9976
    },
    {
      "epoch": 18.408125577100645,
      "grad_norm": 0.3035966753959656,
      "learning_rate": 2.9922e-05,
      "loss": 0.0245,
      "step": 9977
    },
    {
      "epoch": 18.409972299168974,
      "grad_norm": 0.22685669362545013,
      "learning_rate": 2.9925000000000002e-05,
      "loss": 0.0121,
      "step": 9978
    },
    {
      "epoch": 18.411819021237303,
      "grad_norm": 0.34173470735549927,
      "learning_rate": 2.9928000000000002e-05,
      "loss": 0.0345,
      "step": 9979
    },
    {
      "epoch": 18.413665743305632,
      "grad_norm": 0.32935476303100586,
      "learning_rate": 2.9931000000000002e-05,
      "loss": 0.0144,
      "step": 9980
    },
    {
      "epoch": 18.41551246537396,
      "grad_norm": 0.27030572295188904,
      "learning_rate": 2.9934000000000002e-05,
      "loss": 0.0093,
      "step": 9981
    },
    {
      "epoch": 18.41735918744229,
      "grad_norm": 0.32887694239616394,
      "learning_rate": 2.9937e-05,
      "loss": 0.0151,
      "step": 9982
    },
    {
      "epoch": 18.41920590951062,
      "grad_norm": 0.4195692241191864,
      "learning_rate": 2.994e-05,
      "loss": 0.0142,
      "step": 9983
    },
    {
      "epoch": 18.42105263157895,
      "grad_norm": 0.21854715049266815,
      "learning_rate": 2.9943e-05,
      "loss": 0.015,
      "step": 9984
    },
    {
      "epoch": 18.422899353647278,
      "grad_norm": 0.2712455093860626,
      "learning_rate": 2.9946e-05,
      "loss": 0.0078,
      "step": 9985
    },
    {
      "epoch": 18.424746075715603,
      "grad_norm": 0.2517973780632019,
      "learning_rate": 2.9949e-05,
      "loss": 0.0099,
      "step": 9986
    },
    {
      "epoch": 18.426592797783933,
      "grad_norm": 0.44632387161254883,
      "learning_rate": 2.9952e-05,
      "loss": 0.0156,
      "step": 9987
    },
    {
      "epoch": 18.42843951985226,
      "grad_norm": 0.7028042674064636,
      "learning_rate": 2.9955000000000004e-05,
      "loss": 0.0141,
      "step": 9988
    },
    {
      "epoch": 18.43028624192059,
      "grad_norm": 0.26331591606140137,
      "learning_rate": 2.9958000000000004e-05,
      "loss": 0.0103,
      "step": 9989
    },
    {
      "epoch": 18.43213296398892,
      "grad_norm": 0.1908412128686905,
      "learning_rate": 2.9961000000000003e-05,
      "loss": 0.0057,
      "step": 9990
    },
    {
      "epoch": 18.43397968605725,
      "grad_norm": 0.28224021196365356,
      "learning_rate": 2.9964e-05,
      "loss": 0.0078,
      "step": 9991
    },
    {
      "epoch": 18.43582640812558,
      "grad_norm": 0.369598925113678,
      "learning_rate": 2.9967e-05,
      "loss": 0.0101,
      "step": 9992
    },
    {
      "epoch": 18.437673130193907,
      "grad_norm": 0.4752858281135559,
      "learning_rate": 2.997e-05,
      "loss": 0.0109,
      "step": 9993
    },
    {
      "epoch": 18.439519852262233,
      "grad_norm": 0.35530319809913635,
      "learning_rate": 2.9973e-05,
      "loss": 0.0107,
      "step": 9994
    },
    {
      "epoch": 18.441366574330562,
      "grad_norm": 3.2367780208587646,
      "learning_rate": 2.9976e-05,
      "loss": 0.0246,
      "step": 9995
    },
    {
      "epoch": 18.44321329639889,
      "grad_norm": 0.46068835258483887,
      "learning_rate": 2.9979e-05,
      "loss": 0.0106,
      "step": 9996
    },
    {
      "epoch": 18.44506001846722,
      "grad_norm": 0.462532639503479,
      "learning_rate": 2.9982e-05,
      "loss": 0.0207,
      "step": 9997
    },
    {
      "epoch": 18.44690674053555,
      "grad_norm": 0.29335060715675354,
      "learning_rate": 2.9985000000000002e-05,
      "loss": 0.012,
      "step": 9998
    },
    {
      "epoch": 18.44875346260388,
      "grad_norm": 0.34464287757873535,
      "learning_rate": 2.9988e-05,
      "loss": 0.0168,
      "step": 9999
    },
    {
      "epoch": 18.450600184672208,
      "grad_norm": 0.5311354398727417,
      "learning_rate": 2.9991e-05,
      "loss": 0.0198,
      "step": 10000
    },
    {
      "epoch": 18.450600184672208,
      "eval_cer": 0.11346138891049071,
      "eval_loss": 0.400764524936676,
      "eval_runtime": 16.6034,
      "eval_samples_per_second": 18.31,
      "eval_steps_per_second": 0.602,
      "eval_wer": 0.4002302379125096,
      "step": 10000
    },
    {
      "epoch": 18.452446906740537,
      "grad_norm": 1.1134467124938965,
      "learning_rate": 2.9994e-05,
      "loss": 0.0213,
      "step": 10001
    },
    {
      "epoch": 18.454293628808863,
      "grad_norm": 0.3000761568546295,
      "learning_rate": 2.9997e-05,
      "loss": 0.0153,
      "step": 10002
    },
    {
      "epoch": 18.45614035087719,
      "grad_norm": 0.6289483308792114,
      "learning_rate": 3e-05,
      "loss": 0.0259,
      "step": 10003
    },
    {
      "epoch": 18.45798707294552,
      "grad_norm": 0.45202159881591797,
      "learning_rate": 2.999966666666667e-05,
      "loss": 0.0163,
      "step": 10004
    },
    {
      "epoch": 18.45983379501385,
      "grad_norm": 0.435899555683136,
      "learning_rate": 2.9999333333333333e-05,
      "loss": 0.0245,
      "step": 10005
    },
    {
      "epoch": 18.46168051708218,
      "grad_norm": 0.5257439017295837,
      "learning_rate": 2.9999000000000002e-05,
      "loss": 0.0327,
      "step": 10006
    },
    {
      "epoch": 18.46352723915051,
      "grad_norm": 1.0339168310165405,
      "learning_rate": 2.9998666666666668e-05,
      "loss": 0.2506,
      "step": 10007
    },
    {
      "epoch": 18.465373961218837,
      "grad_norm": 0.5932863354682922,
      "learning_rate": 2.9998333333333334e-05,
      "loss": 0.1746,
      "step": 10008
    },
    {
      "epoch": 18.467220683287167,
      "grad_norm": 0.6096854209899902,
      "learning_rate": 2.9998e-05,
      "loss": 0.1864,
      "step": 10009
    },
    {
      "epoch": 18.469067405355496,
      "grad_norm": 0.49512627720832825,
      "learning_rate": 2.999766666666667e-05,
      "loss": 0.0995,
      "step": 10010
    },
    {
      "epoch": 18.47091412742382,
      "grad_norm": 0.4499821960926056,
      "learning_rate": 2.999733333333333e-05,
      "loss": 0.0955,
      "step": 10011
    },
    {
      "epoch": 18.47276084949215,
      "grad_norm": 0.5281078219413757,
      "learning_rate": 2.9997e-05,
      "loss": 0.0588,
      "step": 10012
    },
    {
      "epoch": 18.47460757156048,
      "grad_norm": 0.5754829049110413,
      "learning_rate": 2.999666666666667e-05,
      "loss": 0.0842,
      "step": 10013
    },
    {
      "epoch": 18.47645429362881,
      "grad_norm": 0.4375404417514801,
      "learning_rate": 2.9996333333333333e-05,
      "loss": 0.0733,
      "step": 10014
    },
    {
      "epoch": 18.478301015697138,
      "grad_norm": 0.4932893216609955,
      "learning_rate": 2.9996000000000002e-05,
      "loss": 0.0756,
      "step": 10015
    },
    {
      "epoch": 18.480147737765467,
      "grad_norm": 0.6486257314682007,
      "learning_rate": 2.9995666666666668e-05,
      "loss": 0.1231,
      "step": 10016
    },
    {
      "epoch": 18.481994459833796,
      "grad_norm": 0.3846319317817688,
      "learning_rate": 2.9995333333333334e-05,
      "loss": 0.0681,
      "step": 10017
    },
    {
      "epoch": 18.483841181902125,
      "grad_norm": 1.1660290956497192,
      "learning_rate": 2.9995e-05,
      "loss": 0.0722,
      "step": 10018
    },
    {
      "epoch": 18.48568790397045,
      "grad_norm": 0.4112194776535034,
      "learning_rate": 2.9994666666666666e-05,
      "loss": 0.0335,
      "step": 10019
    },
    {
      "epoch": 18.48753462603878,
      "grad_norm": 0.29463788866996765,
      "learning_rate": 2.9994333333333335e-05,
      "loss": 0.0234,
      "step": 10020
    },
    {
      "epoch": 18.48938134810711,
      "grad_norm": 0.37340521812438965,
      "learning_rate": 2.9994e-05,
      "loss": 0.039,
      "step": 10021
    },
    {
      "epoch": 18.49122807017544,
      "grad_norm": 0.7228174209594727,
      "learning_rate": 2.9993666666666667e-05,
      "loss": 0.0459,
      "step": 10022
    },
    {
      "epoch": 18.493074792243767,
      "grad_norm": 0.4925774335861206,
      "learning_rate": 2.9993333333333333e-05,
      "loss": 0.0171,
      "step": 10023
    },
    {
      "epoch": 18.494921514312097,
      "grad_norm": 0.5816299319267273,
      "learning_rate": 2.9993000000000002e-05,
      "loss": 0.0207,
      "step": 10024
    },
    {
      "epoch": 18.496768236380426,
      "grad_norm": 0.442291796207428,
      "learning_rate": 2.9992666666666665e-05,
      "loss": 0.0354,
      "step": 10025
    },
    {
      "epoch": 18.498614958448755,
      "grad_norm": 0.2628393769264221,
      "learning_rate": 2.9992333333333334e-05,
      "loss": 0.0115,
      "step": 10026
    },
    {
      "epoch": 18.50046168051708,
      "grad_norm": 0.33153483271598816,
      "learning_rate": 2.9992e-05,
      "loss": 0.0152,
      "step": 10027
    },
    {
      "epoch": 18.50230840258541,
      "grad_norm": 0.3420373797416687,
      "learning_rate": 2.9991666666666666e-05,
      "loss": 0.0238,
      "step": 10028
    },
    {
      "epoch": 18.50415512465374,
      "grad_norm": 0.6902183890342712,
      "learning_rate": 2.9991333333333335e-05,
      "loss": 0.0103,
      "step": 10029
    },
    {
      "epoch": 18.506001846722068,
      "grad_norm": 0.3361265957355499,
      "learning_rate": 2.9991e-05,
      "loss": 0.0204,
      "step": 10030
    },
    {
      "epoch": 18.507848568790397,
      "grad_norm": 0.32917317748069763,
      "learning_rate": 2.9990666666666667e-05,
      "loss": 0.0101,
      "step": 10031
    },
    {
      "epoch": 18.509695290858726,
      "grad_norm": 1.2217377424240112,
      "learning_rate": 2.9990333333333333e-05,
      "loss": 0.0163,
      "step": 10032
    },
    {
      "epoch": 18.511542012927055,
      "grad_norm": 0.3950839638710022,
      "learning_rate": 2.9990000000000003e-05,
      "loss": 0.0213,
      "step": 10033
    },
    {
      "epoch": 18.513388734995385,
      "grad_norm": 0.24514494836330414,
      "learning_rate": 2.9989666666666665e-05,
      "loss": 0.0075,
      "step": 10034
    },
    {
      "epoch": 18.51523545706371,
      "grad_norm": 0.4704078733921051,
      "learning_rate": 2.9989333333333334e-05,
      "loss": 0.015,
      "step": 10035
    },
    {
      "epoch": 18.51708217913204,
      "grad_norm": 0.522798478603363,
      "learning_rate": 2.9989e-05,
      "loss": 0.0175,
      "step": 10036
    },
    {
      "epoch": 18.51892890120037,
      "grad_norm": 0.1550319790840149,
      "learning_rate": 2.9988666666666666e-05,
      "loss": 0.0086,
      "step": 10037
    },
    {
      "epoch": 18.520775623268698,
      "grad_norm": 0.20868565142154694,
      "learning_rate": 2.9988333333333336e-05,
      "loss": 0.0088,
      "step": 10038
    },
    {
      "epoch": 18.522622345337027,
      "grad_norm": 0.6053646802902222,
      "learning_rate": 2.9988e-05,
      "loss": 0.0214,
      "step": 10039
    },
    {
      "epoch": 18.524469067405356,
      "grad_norm": 0.7348169088363647,
      "learning_rate": 2.9987666666666667e-05,
      "loss": 0.0219,
      "step": 10040
    },
    {
      "epoch": 18.526315789473685,
      "grad_norm": 0.3019208014011383,
      "learning_rate": 2.9987333333333333e-05,
      "loss": 0.0156,
      "step": 10041
    },
    {
      "epoch": 18.528162511542014,
      "grad_norm": 0.9538294672966003,
      "learning_rate": 2.9987000000000003e-05,
      "loss": 0.0154,
      "step": 10042
    },
    {
      "epoch": 18.530009233610343,
      "grad_norm": 0.9015957117080688,
      "learning_rate": 2.9986666666666665e-05,
      "loss": 0.0223,
      "step": 10043
    },
    {
      "epoch": 18.53185595567867,
      "grad_norm": 0.31876128911972046,
      "learning_rate": 2.9986333333333335e-05,
      "loss": 0.0102,
      "step": 10044
    },
    {
      "epoch": 18.533702677746998,
      "grad_norm": 0.22382700443267822,
      "learning_rate": 2.9986000000000004e-05,
      "loss": 0.0146,
      "step": 10045
    },
    {
      "epoch": 18.535549399815327,
      "grad_norm": 0.3416086733341217,
      "learning_rate": 2.9985666666666666e-05,
      "loss": 0.0124,
      "step": 10046
    },
    {
      "epoch": 18.537396121883656,
      "grad_norm": 0.42624083161354065,
      "learning_rate": 2.9985333333333336e-05,
      "loss": 0.0128,
      "step": 10047
    },
    {
      "epoch": 18.539242843951985,
      "grad_norm": 0.8317400217056274,
      "learning_rate": 2.9985000000000002e-05,
      "loss": 0.0254,
      "step": 10048
    },
    {
      "epoch": 18.541089566020315,
      "grad_norm": 0.7831063866615295,
      "learning_rate": 2.9984666666666668e-05,
      "loss": 0.0205,
      "step": 10049
    },
    {
      "epoch": 18.542936288088644,
      "grad_norm": 0.3047102391719818,
      "learning_rate": 2.9984333333333334e-05,
      "loss": 0.0147,
      "step": 10050
    },
    {
      "epoch": 18.544783010156973,
      "grad_norm": 0.3798275291919708,
      "learning_rate": 2.9984e-05,
      "loss": 0.0135,
      "step": 10051
    },
    {
      "epoch": 18.5466297322253,
      "grad_norm": 0.5144087076187134,
      "learning_rate": 2.9983666666666665e-05,
      "loss": 0.0144,
      "step": 10052
    },
    {
      "epoch": 18.548476454293628,
      "grad_norm": 0.3479164242744446,
      "learning_rate": 2.9983333333333335e-05,
      "loss": 0.0189,
      "step": 10053
    },
    {
      "epoch": 18.550323176361957,
      "grad_norm": 0.29653459787368774,
      "learning_rate": 2.9983e-05,
      "loss": 0.0286,
      "step": 10054
    },
    {
      "epoch": 18.552169898430286,
      "grad_norm": 0.5336138606071472,
      "learning_rate": 2.9982666666666667e-05,
      "loss": 0.0248,
      "step": 10055
    },
    {
      "epoch": 18.554016620498615,
      "grad_norm": 0.43095484375953674,
      "learning_rate": 2.9982333333333336e-05,
      "loss": 0.0118,
      "step": 10056
    },
    {
      "epoch": 18.555863342566944,
      "grad_norm": 0.8118236660957336,
      "learning_rate": 2.9982e-05,
      "loss": 0.1899,
      "step": 10057
    },
    {
      "epoch": 18.557710064635273,
      "grad_norm": 1.0012409687042236,
      "learning_rate": 2.9981666666666668e-05,
      "loss": 0.2917,
      "step": 10058
    },
    {
      "epoch": 18.559556786703602,
      "grad_norm": 0.6601340174674988,
      "learning_rate": 2.9981333333333334e-05,
      "loss": 0.1182,
      "step": 10059
    },
    {
      "epoch": 18.56140350877193,
      "grad_norm": 0.6776450872421265,
      "learning_rate": 2.9981e-05,
      "loss": 0.1185,
      "step": 10060
    },
    {
      "epoch": 18.563250230840257,
      "grad_norm": 0.588324248790741,
      "learning_rate": 2.9980666666666666e-05,
      "loss": 0.1353,
      "step": 10061
    },
    {
      "epoch": 18.565096952908586,
      "grad_norm": 0.5905701518058777,
      "learning_rate": 2.9980333333333335e-05,
      "loss": 0.1602,
      "step": 10062
    },
    {
      "epoch": 18.566943674976915,
      "grad_norm": 0.5700792074203491,
      "learning_rate": 2.998e-05,
      "loss": 0.1245,
      "step": 10063
    },
    {
      "epoch": 18.568790397045245,
      "grad_norm": 0.5920236706733704,
      "learning_rate": 2.9979666666666667e-05,
      "loss": 0.0669,
      "step": 10064
    },
    {
      "epoch": 18.570637119113574,
      "grad_norm": 0.7023180723190308,
      "learning_rate": 2.9979333333333336e-05,
      "loss": 0.1148,
      "step": 10065
    },
    {
      "epoch": 18.572483841181903,
      "grad_norm": 0.3596070110797882,
      "learning_rate": 2.9979e-05,
      "loss": 0.0552,
      "step": 10066
    },
    {
      "epoch": 18.574330563250232,
      "grad_norm": 0.765246570110321,
      "learning_rate": 2.9978666666666668e-05,
      "loss": 0.0738,
      "step": 10067
    },
    {
      "epoch": 18.57617728531856,
      "grad_norm": 0.5413464307785034,
      "learning_rate": 2.9978333333333334e-05,
      "loss": 0.07,
      "step": 10068
    },
    {
      "epoch": 18.578024007386887,
      "grad_norm": 0.5046939253807068,
      "learning_rate": 2.9978e-05,
      "loss": 0.051,
      "step": 10069
    },
    {
      "epoch": 18.579870729455216,
      "grad_norm": 0.3605642020702362,
      "learning_rate": 2.997766666666667e-05,
      "loss": 0.037,
      "step": 10070
    },
    {
      "epoch": 18.581717451523545,
      "grad_norm": 0.3019397556781769,
      "learning_rate": 2.9977333333333335e-05,
      "loss": 0.0261,
      "step": 10071
    },
    {
      "epoch": 18.583564173591874,
      "grad_norm": 0.468206524848938,
      "learning_rate": 2.9977e-05,
      "loss": 0.0353,
      "step": 10072
    },
    {
      "epoch": 18.585410895660203,
      "grad_norm": 0.2793387770652771,
      "learning_rate": 2.9976666666666667e-05,
      "loss": 0.0193,
      "step": 10073
    },
    {
      "epoch": 18.587257617728532,
      "grad_norm": 0.33137214183807373,
      "learning_rate": 2.9976333333333336e-05,
      "loss": 0.0214,
      "step": 10074
    },
    {
      "epoch": 18.58910433979686,
      "grad_norm": 0.8799513578414917,
      "learning_rate": 2.9976e-05,
      "loss": 0.0167,
      "step": 10075
    },
    {
      "epoch": 18.59095106186519,
      "grad_norm": 0.1753733605146408,
      "learning_rate": 2.9975666666666668e-05,
      "loss": 0.0092,
      "step": 10076
    },
    {
      "epoch": 18.592797783933516,
      "grad_norm": 0.3424614667892456,
      "learning_rate": 2.9975333333333334e-05,
      "loss": 0.0161,
      "step": 10077
    },
    {
      "epoch": 18.594644506001845,
      "grad_norm": 0.1594821959733963,
      "learning_rate": 2.9975e-05,
      "loss": 0.0104,
      "step": 10078
    },
    {
      "epoch": 18.596491228070175,
      "grad_norm": 0.401669979095459,
      "learning_rate": 2.997466666666667e-05,
      "loss": 0.0175,
      "step": 10079
    },
    {
      "epoch": 18.598337950138504,
      "grad_norm": 0.5343904495239258,
      "learning_rate": 2.9974333333333332e-05,
      "loss": 0.0169,
      "step": 10080
    },
    {
      "epoch": 18.600184672206833,
      "grad_norm": 0.639664351940155,
      "learning_rate": 2.9974e-05,
      "loss": 0.0211,
      "step": 10081
    },
    {
      "epoch": 18.602031394275162,
      "grad_norm": 0.34157347679138184,
      "learning_rate": 2.9973666666666667e-05,
      "loss": 0.0319,
      "step": 10082
    },
    {
      "epoch": 18.60387811634349,
      "grad_norm": 0.8372710943222046,
      "learning_rate": 2.9973333333333333e-05,
      "loss": 0.0505,
      "step": 10083
    },
    {
      "epoch": 18.60572483841182,
      "grad_norm": 0.23561322689056396,
      "learning_rate": 2.9973e-05,
      "loss": 0.0081,
      "step": 10084
    },
    {
      "epoch": 18.607571560480146,
      "grad_norm": 0.3529277741909027,
      "learning_rate": 2.997266666666667e-05,
      "loss": 0.0089,
      "step": 10085
    },
    {
      "epoch": 18.609418282548475,
      "grad_norm": 0.2528364658355713,
      "learning_rate": 2.997233333333333e-05,
      "loss": 0.0144,
      "step": 10086
    },
    {
      "epoch": 18.611265004616804,
      "grad_norm": 0.513348400592804,
      "learning_rate": 2.9972e-05,
      "loss": 0.0071,
      "step": 10087
    },
    {
      "epoch": 18.613111726685133,
      "grad_norm": 0.332783579826355,
      "learning_rate": 2.997166666666667e-05,
      "loss": 0.0113,
      "step": 10088
    },
    {
      "epoch": 18.614958448753463,
      "grad_norm": 0.2563062310218811,
      "learning_rate": 2.9971333333333332e-05,
      "loss": 0.0111,
      "step": 10089
    },
    {
      "epoch": 18.61680517082179,
      "grad_norm": 0.39945706725120544,
      "learning_rate": 2.9971e-05,
      "loss": 0.0078,
      "step": 10090
    },
    {
      "epoch": 18.61865189289012,
      "grad_norm": 0.30604419112205505,
      "learning_rate": 2.9970666666666667e-05,
      "loss": 0.0099,
      "step": 10091
    },
    {
      "epoch": 18.62049861495845,
      "grad_norm": 0.3137086033821106,
      "learning_rate": 2.9970333333333333e-05,
      "loss": 0.0108,
      "step": 10092
    },
    {
      "epoch": 18.62234533702678,
      "grad_norm": 0.3253783881664276,
      "learning_rate": 2.997e-05,
      "loss": 0.0171,
      "step": 10093
    },
    {
      "epoch": 18.624192059095105,
      "grad_norm": 1.2647370100021362,
      "learning_rate": 2.996966666666667e-05,
      "loss": 0.1085,
      "step": 10094
    },
    {
      "epoch": 18.626038781163434,
      "grad_norm": 0.2971855103969574,
      "learning_rate": 2.9969333333333335e-05,
      "loss": 0.0139,
      "step": 10095
    },
    {
      "epoch": 18.627885503231763,
      "grad_norm": 0.29171839356422424,
      "learning_rate": 2.9969e-05,
      "loss": 0.0123,
      "step": 10096
    },
    {
      "epoch": 18.629732225300092,
      "grad_norm": 0.3409866392612457,
      "learning_rate": 2.996866666666667e-05,
      "loss": 0.0153,
      "step": 10097
    },
    {
      "epoch": 18.63157894736842,
      "grad_norm": 0.3979312479496002,
      "learning_rate": 2.9968333333333332e-05,
      "loss": 0.0191,
      "step": 10098
    },
    {
      "epoch": 18.63342566943675,
      "grad_norm": 0.4440046548843384,
      "learning_rate": 2.9968000000000002e-05,
      "loss": 0.0141,
      "step": 10099
    },
    {
      "epoch": 18.63527239150508,
      "grad_norm": 0.3148280382156372,
      "learning_rate": 2.9967666666666668e-05,
      "loss": 0.0157,
      "step": 10100
    },
    {
      "epoch": 18.63711911357341,
      "grad_norm": 0.7379143834114075,
      "learning_rate": 2.9967333333333334e-05,
      "loss": 0.0142,
      "step": 10101
    },
    {
      "epoch": 18.638965835641734,
      "grad_norm": 0.4341832399368286,
      "learning_rate": 2.9967e-05,
      "loss": 0.0222,
      "step": 10102
    },
    {
      "epoch": 18.640812557710063,
      "grad_norm": 0.25606244802474976,
      "learning_rate": 2.996666666666667e-05,
      "loss": 0.0088,
      "step": 10103
    },
    {
      "epoch": 18.642659279778393,
      "grad_norm": 0.32717734575271606,
      "learning_rate": 2.9966333333333335e-05,
      "loss": 0.0195,
      "step": 10104
    },
    {
      "epoch": 18.64450600184672,
      "grad_norm": 0.5025584697723389,
      "learning_rate": 2.9966e-05,
      "loss": 0.0113,
      "step": 10105
    },
    {
      "epoch": 18.64635272391505,
      "grad_norm": 0.30470889806747437,
      "learning_rate": 2.996566666666667e-05,
      "loss": 0.0127,
      "step": 10106
    },
    {
      "epoch": 18.64819944598338,
      "grad_norm": 3.4498097896575928,
      "learning_rate": 2.9965333333333333e-05,
      "loss": 0.2126,
      "step": 10107
    },
    {
      "epoch": 18.65004616805171,
      "grad_norm": 0.6399456262588501,
      "learning_rate": 2.9965000000000002e-05,
      "loss": 0.1395,
      "step": 10108
    },
    {
      "epoch": 18.65189289012004,
      "grad_norm": 0.5964101552963257,
      "learning_rate": 2.9964666666666664e-05,
      "loss": 0.1193,
      "step": 10109
    },
    {
      "epoch": 18.653739612188367,
      "grad_norm": 1.0935943126678467,
      "learning_rate": 2.9964333333333334e-05,
      "loss": 0.1248,
      "step": 10110
    },
    {
      "epoch": 18.655586334256693,
      "grad_norm": 1.1805049180984497,
      "learning_rate": 2.9964e-05,
      "loss": 0.1114,
      "step": 10111
    },
    {
      "epoch": 18.657433056325022,
      "grad_norm": 0.5595899224281311,
      "learning_rate": 2.9963666666666666e-05,
      "loss": 0.0861,
      "step": 10112
    },
    {
      "epoch": 18.65927977839335,
      "grad_norm": 0.4162147343158722,
      "learning_rate": 2.9963333333333335e-05,
      "loss": 0.056,
      "step": 10113
    },
    {
      "epoch": 18.66112650046168,
      "grad_norm": 0.8674030303955078,
      "learning_rate": 2.9963e-05,
      "loss": 0.1408,
      "step": 10114
    },
    {
      "epoch": 18.66297322253001,
      "grad_norm": 0.37030184268951416,
      "learning_rate": 2.9962666666666667e-05,
      "loss": 0.0534,
      "step": 10115
    },
    {
      "epoch": 18.66481994459834,
      "grad_norm": 0.4186866283416748,
      "learning_rate": 2.9962333333333333e-05,
      "loss": 0.0514,
      "step": 10116
    },
    {
      "epoch": 18.666666666666668,
      "grad_norm": 0.776822566986084,
      "learning_rate": 2.9962000000000002e-05,
      "loss": 0.0497,
      "step": 10117
    },
    {
      "epoch": 18.668513388734997,
      "grad_norm": 0.41464748978614807,
      "learning_rate": 2.9961666666666665e-05,
      "loss": 0.0481,
      "step": 10118
    },
    {
      "epoch": 18.670360110803323,
      "grad_norm": 1.1878679990768433,
      "learning_rate": 2.9961333333333334e-05,
      "loss": 0.0557,
      "step": 10119
    },
    {
      "epoch": 18.67220683287165,
      "grad_norm": 0.563939094543457,
      "learning_rate": 2.9961000000000003e-05,
      "loss": 0.0647,
      "step": 10120
    },
    {
      "epoch": 18.67405355493998,
      "grad_norm": 0.8240351676940918,
      "learning_rate": 2.9960666666666666e-05,
      "loss": 0.0339,
      "step": 10121
    },
    {
      "epoch": 18.67590027700831,
      "grad_norm": 0.19536948204040527,
      "learning_rate": 2.9960333333333335e-05,
      "loss": 0.0106,
      "step": 10122
    },
    {
      "epoch": 18.67774699907664,
      "grad_norm": 1.515914797782898,
      "learning_rate": 2.996e-05,
      "loss": 0.0125,
      "step": 10123
    },
    {
      "epoch": 18.67959372114497,
      "grad_norm": 0.20640091598033905,
      "learning_rate": 2.9959666666666667e-05,
      "loss": 0.0189,
      "step": 10124
    },
    {
      "epoch": 18.681440443213297,
      "grad_norm": 0.44630637764930725,
      "learning_rate": 2.9959333333333333e-05,
      "loss": 0.0666,
      "step": 10125
    },
    {
      "epoch": 18.683287165281627,
      "grad_norm": 0.34298625588417053,
      "learning_rate": 2.9959000000000002e-05,
      "loss": 0.044,
      "step": 10126
    },
    {
      "epoch": 18.685133887349952,
      "grad_norm": 0.4227137267589569,
      "learning_rate": 2.9958666666666665e-05,
      "loss": 0.0153,
      "step": 10127
    },
    {
      "epoch": 18.68698060941828,
      "grad_norm": 0.21673138439655304,
      "learning_rate": 2.9958333333333334e-05,
      "loss": 0.01,
      "step": 10128
    },
    {
      "epoch": 18.68882733148661,
      "grad_norm": 0.1855924427509308,
      "learning_rate": 2.9958000000000004e-05,
      "loss": 0.0117,
      "step": 10129
    },
    {
      "epoch": 18.69067405355494,
      "grad_norm": 0.32578045129776,
      "learning_rate": 2.9957666666666666e-05,
      "loss": 0.0059,
      "step": 10130
    },
    {
      "epoch": 18.69252077562327,
      "grad_norm": 0.38381683826446533,
      "learning_rate": 2.9957333333333335e-05,
      "loss": 0.0166,
      "step": 10131
    },
    {
      "epoch": 18.694367497691598,
      "grad_norm": 0.4086329936981201,
      "learning_rate": 2.9957e-05,
      "loss": 0.0116,
      "step": 10132
    },
    {
      "epoch": 18.696214219759927,
      "grad_norm": 0.5231785178184509,
      "learning_rate": 2.9956666666666667e-05,
      "loss": 0.0156,
      "step": 10133
    },
    {
      "epoch": 18.698060941828256,
      "grad_norm": 0.22241975367069244,
      "learning_rate": 2.9956333333333333e-05,
      "loss": 0.0072,
      "step": 10134
    },
    {
      "epoch": 18.69990766389658,
      "grad_norm": 0.3974694013595581,
      "learning_rate": 2.9956000000000003e-05,
      "loss": 0.019,
      "step": 10135
    },
    {
      "epoch": 18.70175438596491,
      "grad_norm": 0.32537904381752014,
      "learning_rate": 2.9955666666666665e-05,
      "loss": 0.0131,
      "step": 10136
    },
    {
      "epoch": 18.70360110803324,
      "grad_norm": 0.6127800941467285,
      "learning_rate": 2.9955333333333334e-05,
      "loss": 0.0242,
      "step": 10137
    },
    {
      "epoch": 18.70544783010157,
      "grad_norm": 0.29393771290779114,
      "learning_rate": 2.9955000000000004e-05,
      "loss": 0.0121,
      "step": 10138
    },
    {
      "epoch": 18.7072945521699,
      "grad_norm": 0.4710220992565155,
      "learning_rate": 2.9954666666666666e-05,
      "loss": 0.0169,
      "step": 10139
    },
    {
      "epoch": 18.709141274238227,
      "grad_norm": 0.3239732086658478,
      "learning_rate": 2.9954333333333336e-05,
      "loss": 0.0096,
      "step": 10140
    },
    {
      "epoch": 18.710987996306557,
      "grad_norm": 0.3971695601940155,
      "learning_rate": 2.9953999999999998e-05,
      "loss": 0.0108,
      "step": 10141
    },
    {
      "epoch": 18.712834718374886,
      "grad_norm": 0.37625211477279663,
      "learning_rate": 2.9953666666666667e-05,
      "loss": 0.0158,
      "step": 10142
    },
    {
      "epoch": 18.714681440443215,
      "grad_norm": 0.6671343445777893,
      "learning_rate": 2.9953333333333333e-05,
      "loss": 0.0184,
      "step": 10143
    },
    {
      "epoch": 18.71652816251154,
      "grad_norm": 0.4562927186489105,
      "learning_rate": 2.9953e-05,
      "loss": 0.0165,
      "step": 10144
    },
    {
      "epoch": 18.71837488457987,
      "grad_norm": 0.270826518535614,
      "learning_rate": 2.995266666666667e-05,
      "loss": 0.0107,
      "step": 10145
    },
    {
      "epoch": 18.7202216066482,
      "grad_norm": 0.2679862678050995,
      "learning_rate": 2.9952333333333335e-05,
      "loss": 0.0163,
      "step": 10146
    },
    {
      "epoch": 18.722068328716528,
      "grad_norm": 0.42214062809944153,
      "learning_rate": 2.9952e-05,
      "loss": 0.013,
      "step": 10147
    },
    {
      "epoch": 18.723915050784857,
      "grad_norm": 0.3402858376502991,
      "learning_rate": 2.9951666666666666e-05,
      "loss": 0.0124,
      "step": 10148
    },
    {
      "epoch": 18.725761772853186,
      "grad_norm": 0.38823774456977844,
      "learning_rate": 2.9951333333333336e-05,
      "loss": 0.017,
      "step": 10149
    },
    {
      "epoch": 18.727608494921515,
      "grad_norm": 0.6362157464027405,
      "learning_rate": 2.9951e-05,
      "loss": 0.0208,
      "step": 10150
    },
    {
      "epoch": 18.729455216989845,
      "grad_norm": 0.21173816919326782,
      "learning_rate": 2.9950666666666668e-05,
      "loss": 0.0083,
      "step": 10151
    },
    {
      "epoch": 18.73130193905817,
      "grad_norm": 0.5997298955917358,
      "learning_rate": 2.9950333333333334e-05,
      "loss": 0.0177,
      "step": 10152
    },
    {
      "epoch": 18.7331486611265,
      "grad_norm": 0.670458972454071,
      "learning_rate": 2.995e-05,
      "loss": 0.0171,
      "step": 10153
    },
    {
      "epoch": 18.73499538319483,
      "grad_norm": 0.4021938741207123,
      "learning_rate": 2.994966666666667e-05,
      "loss": 0.0112,
      "step": 10154
    },
    {
      "epoch": 18.736842105263158,
      "grad_norm": 0.7811278104782104,
      "learning_rate": 2.9949333333333335e-05,
      "loss": 0.0159,
      "step": 10155
    },
    {
      "epoch": 18.738688827331487,
      "grad_norm": 0.6017845273017883,
      "learning_rate": 2.9949e-05,
      "loss": 0.016,
      "step": 10156
    },
    {
      "epoch": 18.740535549399816,
      "grad_norm": 1.6666468381881714,
      "learning_rate": 2.9948666666666667e-05,
      "loss": 0.3443,
      "step": 10157
    },
    {
      "epoch": 18.742382271468145,
      "grad_norm": 0.7472243309020996,
      "learning_rate": 2.9948333333333336e-05,
      "loss": 0.1577,
      "step": 10158
    },
    {
      "epoch": 18.744228993536474,
      "grad_norm": 0.7569506168365479,
      "learning_rate": 2.9948e-05,
      "loss": 0.1193,
      "step": 10159
    },
    {
      "epoch": 18.746075715604803,
      "grad_norm": 0.7633649110794067,
      "learning_rate": 2.9947666666666668e-05,
      "loss": 0.113,
      "step": 10160
    },
    {
      "epoch": 18.74792243767313,
      "grad_norm": 0.6256909370422363,
      "learning_rate": 2.9947333333333334e-05,
      "loss": 0.0973,
      "step": 10161
    },
    {
      "epoch": 18.749769159741458,
      "grad_norm": 1.0657498836517334,
      "learning_rate": 2.9947e-05,
      "loss": 0.118,
      "step": 10162
    },
    {
      "epoch": 18.751615881809787,
      "grad_norm": 0.7749153971672058,
      "learning_rate": 2.994666666666667e-05,
      "loss": 0.0741,
      "step": 10163
    },
    {
      "epoch": 18.753462603878116,
      "grad_norm": 0.4885929822921753,
      "learning_rate": 2.9946333333333335e-05,
      "loss": 0.0539,
      "step": 10164
    },
    {
      "epoch": 18.755309325946445,
      "grad_norm": 0.765285313129425,
      "learning_rate": 2.9946e-05,
      "loss": 0.0759,
      "step": 10165
    },
    {
      "epoch": 18.757156048014775,
      "grad_norm": 0.6439147591590881,
      "learning_rate": 2.9945666666666667e-05,
      "loss": 0.0714,
      "step": 10166
    },
    {
      "epoch": 18.759002770083104,
      "grad_norm": 0.5060761570930481,
      "learning_rate": 2.9945333333333336e-05,
      "loss": 0.0639,
      "step": 10167
    },
    {
      "epoch": 18.760849492151433,
      "grad_norm": 0.34558436274528503,
      "learning_rate": 2.9945e-05,
      "loss": 0.0341,
      "step": 10168
    },
    {
      "epoch": 18.76269621421976,
      "grad_norm": 0.6426031589508057,
      "learning_rate": 2.9944666666666668e-05,
      "loss": 0.0938,
      "step": 10169
    },
    {
      "epoch": 18.764542936288088,
      "grad_norm": 0.6749737858772278,
      "learning_rate": 2.9944333333333334e-05,
      "loss": 0.0304,
      "step": 10170
    },
    {
      "epoch": 18.766389658356417,
      "grad_norm": 0.3478706479072571,
      "learning_rate": 2.9944e-05,
      "loss": 0.0617,
      "step": 10171
    },
    {
      "epoch": 18.768236380424746,
      "grad_norm": 0.24540524184703827,
      "learning_rate": 2.994366666666667e-05,
      "loss": 0.0209,
      "step": 10172
    },
    {
      "epoch": 18.770083102493075,
      "grad_norm": 0.5011123418807983,
      "learning_rate": 2.9943333333333332e-05,
      "loss": 0.03,
      "step": 10173
    },
    {
      "epoch": 18.771929824561404,
      "grad_norm": 0.27377021312713623,
      "learning_rate": 2.9943e-05,
      "loss": 0.0177,
      "step": 10174
    },
    {
      "epoch": 18.773776546629733,
      "grad_norm": 0.1973041296005249,
      "learning_rate": 2.9942666666666667e-05,
      "loss": 0.0161,
      "step": 10175
    },
    {
      "epoch": 18.775623268698062,
      "grad_norm": 0.2237047255039215,
      "learning_rate": 2.9942333333333333e-05,
      "loss": 0.012,
      "step": 10176
    },
    {
      "epoch": 18.777469990766388,
      "grad_norm": 0.34094229340553284,
      "learning_rate": 2.9942e-05,
      "loss": 0.0444,
      "step": 10177
    },
    {
      "epoch": 18.779316712834717,
      "grad_norm": 0.4286072850227356,
      "learning_rate": 2.9941666666666668e-05,
      "loss": 0.0139,
      "step": 10178
    },
    {
      "epoch": 18.781163434903046,
      "grad_norm": 1.1234757900238037,
      "learning_rate": 2.9941333333333334e-05,
      "loss": 0.0222,
      "step": 10179
    },
    {
      "epoch": 18.783010156971375,
      "grad_norm": 0.4694407880306244,
      "learning_rate": 2.9941e-05,
      "loss": 0.0124,
      "step": 10180
    },
    {
      "epoch": 18.784856879039705,
      "grad_norm": 0.2294221818447113,
      "learning_rate": 2.994066666666667e-05,
      "loss": 0.0114,
      "step": 10181
    },
    {
      "epoch": 18.786703601108034,
      "grad_norm": 0.2664346396923065,
      "learning_rate": 2.9940333333333332e-05,
      "loss": 0.0152,
      "step": 10182
    },
    {
      "epoch": 18.788550323176363,
      "grad_norm": 0.2968711853027344,
      "learning_rate": 2.994e-05,
      "loss": 0.0098,
      "step": 10183
    },
    {
      "epoch": 18.790397045244692,
      "grad_norm": 0.39534589648246765,
      "learning_rate": 2.9939666666666667e-05,
      "loss": 0.017,
      "step": 10184
    },
    {
      "epoch": 18.792243767313018,
      "grad_norm": 0.19877228140830994,
      "learning_rate": 2.9939333333333333e-05,
      "loss": 0.0086,
      "step": 10185
    },
    {
      "epoch": 18.794090489381347,
      "grad_norm": 0.178855299949646,
      "learning_rate": 2.9939e-05,
      "loss": 0.0083,
      "step": 10186
    },
    {
      "epoch": 18.795937211449676,
      "grad_norm": 0.38069117069244385,
      "learning_rate": 2.993866666666667e-05,
      "loss": 0.0153,
      "step": 10187
    },
    {
      "epoch": 18.797783933518005,
      "grad_norm": 0.7297185659408569,
      "learning_rate": 2.9938333333333334e-05,
      "loss": 0.0485,
      "step": 10188
    },
    {
      "epoch": 18.799630655586334,
      "grad_norm": 0.32633522152900696,
      "learning_rate": 2.9938e-05,
      "loss": 0.0135,
      "step": 10189
    },
    {
      "epoch": 18.801477377654663,
      "grad_norm": 0.3482316732406616,
      "learning_rate": 2.993766666666667e-05,
      "loss": 0.0131,
      "step": 10190
    },
    {
      "epoch": 18.803324099722992,
      "grad_norm": 0.5848433375358582,
      "learning_rate": 2.9937333333333332e-05,
      "loss": 0.0304,
      "step": 10191
    },
    {
      "epoch": 18.80517082179132,
      "grad_norm": 0.46898210048675537,
      "learning_rate": 2.9937e-05,
      "loss": 0.0138,
      "step": 10192
    },
    {
      "epoch": 18.80701754385965,
      "grad_norm": 0.5427939295768738,
      "learning_rate": 2.9936666666666667e-05,
      "loss": 0.012,
      "step": 10193
    },
    {
      "epoch": 18.808864265927976,
      "grad_norm": 0.2933759391307831,
      "learning_rate": 2.9936333333333333e-05,
      "loss": 0.0093,
      "step": 10194
    },
    {
      "epoch": 18.810710987996305,
      "grad_norm": 0.4972042739391327,
      "learning_rate": 2.9936000000000003e-05,
      "loss": 0.0144,
      "step": 10195
    },
    {
      "epoch": 18.812557710064635,
      "grad_norm": 0.4755079448223114,
      "learning_rate": 2.993566666666667e-05,
      "loss": 0.0387,
      "step": 10196
    },
    {
      "epoch": 18.814404432132964,
      "grad_norm": 0.414993017911911,
      "learning_rate": 2.9935333333333335e-05,
      "loss": 0.0176,
      "step": 10197
    },
    {
      "epoch": 18.816251154201293,
      "grad_norm": 0.30449292063713074,
      "learning_rate": 2.9935e-05,
      "loss": 0.0084,
      "step": 10198
    },
    {
      "epoch": 18.818097876269622,
      "grad_norm": 0.595119059085846,
      "learning_rate": 2.993466666666667e-05,
      "loss": 0.0251,
      "step": 10199
    },
    {
      "epoch": 18.81994459833795,
      "grad_norm": 0.5565841794013977,
      "learning_rate": 2.9934333333333332e-05,
      "loss": 0.0178,
      "step": 10200
    },
    {
      "epoch": 18.82179132040628,
      "grad_norm": 0.17402859032154083,
      "learning_rate": 2.9934000000000002e-05,
      "loss": 0.0046,
      "step": 10201
    },
    {
      "epoch": 18.823638042474606,
      "grad_norm": 0.3507221043109894,
      "learning_rate": 2.9933666666666664e-05,
      "loss": 0.0089,
      "step": 10202
    },
    {
      "epoch": 18.825484764542935,
      "grad_norm": 0.5356447696685791,
      "learning_rate": 2.9933333333333334e-05,
      "loss": 0.037,
      "step": 10203
    },
    {
      "epoch": 18.827331486611264,
      "grad_norm": 0.41256019473075867,
      "learning_rate": 2.9933000000000003e-05,
      "loss": 0.0147,
      "step": 10204
    },
    {
      "epoch": 18.829178208679593,
      "grad_norm": 0.3081376254558563,
      "learning_rate": 2.9932666666666665e-05,
      "loss": 0.0118,
      "step": 10205
    },
    {
      "epoch": 18.831024930747922,
      "grad_norm": 0.45098060369491577,
      "learning_rate": 2.9932333333333335e-05,
      "loss": 0.0128,
      "step": 10206
    },
    {
      "epoch": 18.83287165281625,
      "grad_norm": 0.7322559356689453,
      "learning_rate": 2.9932e-05,
      "loss": 0.1536,
      "step": 10207
    },
    {
      "epoch": 18.83471837488458,
      "grad_norm": 0.6320260167121887,
      "learning_rate": 2.9931666666666667e-05,
      "loss": 0.1585,
      "step": 10208
    },
    {
      "epoch": 18.83656509695291,
      "grad_norm": 0.5659246444702148,
      "learning_rate": 2.9931333333333333e-05,
      "loss": 0.1664,
      "step": 10209
    },
    {
      "epoch": 18.83841181902124,
      "grad_norm": 0.38136017322540283,
      "learning_rate": 2.9931000000000002e-05,
      "loss": 0.0818,
      "step": 10210
    },
    {
      "epoch": 18.840258541089565,
      "grad_norm": 0.4973321557044983,
      "learning_rate": 2.9930666666666668e-05,
      "loss": 0.0997,
      "step": 10211
    },
    {
      "epoch": 18.842105263157894,
      "grad_norm": 0.4908638596534729,
      "learning_rate": 2.9930333333333334e-05,
      "loss": 0.1203,
      "step": 10212
    },
    {
      "epoch": 18.843951985226223,
      "grad_norm": 0.48943161964416504,
      "learning_rate": 2.9930000000000003e-05,
      "loss": 0.0635,
      "step": 10213
    },
    {
      "epoch": 18.845798707294552,
      "grad_norm": 0.7102023363113403,
      "learning_rate": 2.9929666666666666e-05,
      "loss": 0.0842,
      "step": 10214
    },
    {
      "epoch": 18.84764542936288,
      "grad_norm": 0.4362596571445465,
      "learning_rate": 2.9929333333333335e-05,
      "loss": 0.0639,
      "step": 10215
    },
    {
      "epoch": 18.84949215143121,
      "grad_norm": 0.4935445785522461,
      "learning_rate": 2.9929e-05,
      "loss": 0.0571,
      "step": 10216
    },
    {
      "epoch": 18.85133887349954,
      "grad_norm": 0.5463587641716003,
      "learning_rate": 2.9928666666666667e-05,
      "loss": 0.094,
      "step": 10217
    },
    {
      "epoch": 18.85318559556787,
      "grad_norm": 0.39658188819885254,
      "learning_rate": 2.9928333333333333e-05,
      "loss": 0.0355,
      "step": 10218
    },
    {
      "epoch": 18.855032317636194,
      "grad_norm": 0.4764830768108368,
      "learning_rate": 2.9928000000000002e-05,
      "loss": 0.0387,
      "step": 10219
    },
    {
      "epoch": 18.856879039704523,
      "grad_norm": 0.28915467858314514,
      "learning_rate": 2.9927666666666668e-05,
      "loss": 0.0298,
      "step": 10220
    },
    {
      "epoch": 18.858725761772853,
      "grad_norm": 0.6550969481468201,
      "learning_rate": 2.9927333333333334e-05,
      "loss": 0.0526,
      "step": 10221
    },
    {
      "epoch": 18.86057248384118,
      "grad_norm": 0.22818061709403992,
      "learning_rate": 2.9927000000000003e-05,
      "loss": 0.0323,
      "step": 10222
    },
    {
      "epoch": 18.86241920590951,
      "grad_norm": 0.7346168756484985,
      "learning_rate": 2.9926666666666666e-05,
      "loss": 0.0663,
      "step": 10223
    },
    {
      "epoch": 18.86426592797784,
      "grad_norm": 0.27641716599464417,
      "learning_rate": 2.9926333333333335e-05,
      "loss": 0.0141,
      "step": 10224
    },
    {
      "epoch": 18.86611265004617,
      "grad_norm": 0.32033631205558777,
      "learning_rate": 2.9926e-05,
      "loss": 0.0237,
      "step": 10225
    },
    {
      "epoch": 18.8679593721145,
      "grad_norm": 0.38354334235191345,
      "learning_rate": 2.9925666666666667e-05,
      "loss": 0.0234,
      "step": 10226
    },
    {
      "epoch": 18.869806094182824,
      "grad_norm": 0.13038446009159088,
      "learning_rate": 2.9925333333333333e-05,
      "loss": 0.0054,
      "step": 10227
    },
    {
      "epoch": 18.871652816251153,
      "grad_norm": 0.3191926181316376,
      "learning_rate": 2.9925000000000002e-05,
      "loss": 0.0218,
      "step": 10228
    },
    {
      "epoch": 18.873499538319482,
      "grad_norm": 0.26581940054893494,
      "learning_rate": 2.9924666666666668e-05,
      "loss": 0.0141,
      "step": 10229
    },
    {
      "epoch": 18.87534626038781,
      "grad_norm": 0.26856064796447754,
      "learning_rate": 2.9924333333333334e-05,
      "loss": 0.0152,
      "step": 10230
    },
    {
      "epoch": 18.87719298245614,
      "grad_norm": 0.5691299438476562,
      "learning_rate": 2.9924e-05,
      "loss": 0.0213,
      "step": 10231
    },
    {
      "epoch": 18.87903970452447,
      "grad_norm": 0.773888111114502,
      "learning_rate": 2.9923666666666666e-05,
      "loss": 0.0162,
      "step": 10232
    },
    {
      "epoch": 18.8808864265928,
      "grad_norm": 0.4445539116859436,
      "learning_rate": 2.9923333333333335e-05,
      "loss": 0.0156,
      "step": 10233
    },
    {
      "epoch": 18.882733148661128,
      "grad_norm": 0.30708056688308716,
      "learning_rate": 2.9922999999999998e-05,
      "loss": 0.0085,
      "step": 10234
    },
    {
      "epoch": 18.884579870729453,
      "grad_norm": 0.5381516814231873,
      "learning_rate": 2.9922666666666667e-05,
      "loss": 0.0181,
      "step": 10235
    },
    {
      "epoch": 18.886426592797783,
      "grad_norm": 0.2624804675579071,
      "learning_rate": 2.9922333333333333e-05,
      "loss": 0.0084,
      "step": 10236
    },
    {
      "epoch": 18.88827331486611,
      "grad_norm": 0.3261619508266449,
      "learning_rate": 2.9922e-05,
      "loss": 0.0096,
      "step": 10237
    },
    {
      "epoch": 18.89012003693444,
      "grad_norm": 0.3803158402442932,
      "learning_rate": 2.992166666666667e-05,
      "loss": 0.0132,
      "step": 10238
    },
    {
      "epoch": 18.89196675900277,
      "grad_norm": 0.29096710681915283,
      "learning_rate": 2.9921333333333334e-05,
      "loss": 0.0137,
      "step": 10239
    },
    {
      "epoch": 18.8938134810711,
      "grad_norm": 0.6219228506088257,
      "learning_rate": 2.9921e-05,
      "loss": 0.0155,
      "step": 10240
    },
    {
      "epoch": 18.89566020313943,
      "grad_norm": 0.6573033332824707,
      "learning_rate": 2.9920666666666666e-05,
      "loss": 0.0208,
      "step": 10241
    },
    {
      "epoch": 18.897506925207757,
      "grad_norm": 0.43783968687057495,
      "learning_rate": 2.9920333333333336e-05,
      "loss": 0.0149,
      "step": 10242
    },
    {
      "epoch": 18.899353647276087,
      "grad_norm": 0.44213783740997314,
      "learning_rate": 2.9919999999999998e-05,
      "loss": 0.0154,
      "step": 10243
    },
    {
      "epoch": 18.901200369344412,
      "grad_norm": 0.5574394464492798,
      "learning_rate": 2.9919666666666667e-05,
      "loss": 0.0177,
      "step": 10244
    },
    {
      "epoch": 18.90304709141274,
      "grad_norm": 0.2930018901824951,
      "learning_rate": 2.9919333333333337e-05,
      "loss": 0.0087,
      "step": 10245
    },
    {
      "epoch": 18.90489381348107,
      "grad_norm": 0.5472742915153503,
      "learning_rate": 2.9919e-05,
      "loss": 0.0123,
      "step": 10246
    },
    {
      "epoch": 18.9067405355494,
      "grad_norm": 0.33874809741973877,
      "learning_rate": 2.991866666666667e-05,
      "loss": 0.0141,
      "step": 10247
    },
    {
      "epoch": 18.90858725761773,
      "grad_norm": 0.6086728572845459,
      "learning_rate": 2.9918333333333335e-05,
      "loss": 0.0116,
      "step": 10248
    },
    {
      "epoch": 18.910433979686058,
      "grad_norm": 0.6606793999671936,
      "learning_rate": 2.9918e-05,
      "loss": 0.0206,
      "step": 10249
    },
    {
      "epoch": 18.912280701754387,
      "grad_norm": 0.8574153780937195,
      "learning_rate": 2.9917666666666666e-05,
      "loss": 0.0226,
      "step": 10250
    },
    {
      "epoch": 18.914127423822716,
      "grad_norm": 0.23669973015785217,
      "learning_rate": 2.9917333333333336e-05,
      "loss": 0.0108,
      "step": 10251
    },
    {
      "epoch": 18.91597414589104,
      "grad_norm": 0.6156405806541443,
      "learning_rate": 2.9917e-05,
      "loss": 0.0233,
      "step": 10252
    },
    {
      "epoch": 18.91782086795937,
      "grad_norm": 0.29926902055740356,
      "learning_rate": 2.9916666666666668e-05,
      "loss": 0.0107,
      "step": 10253
    },
    {
      "epoch": 18.9196675900277,
      "grad_norm": 0.2887798845767975,
      "learning_rate": 2.9916333333333337e-05,
      "loss": 0.0153,
      "step": 10254
    },
    {
      "epoch": 18.92151431209603,
      "grad_norm": 0.5958914756774902,
      "learning_rate": 2.9916e-05,
      "loss": 0.0135,
      "step": 10255
    },
    {
      "epoch": 18.92336103416436,
      "grad_norm": 0.6727198362350464,
      "learning_rate": 2.991566666666667e-05,
      "loss": 0.0115,
      "step": 10256
    },
    {
      "epoch": 18.925207756232687,
      "grad_norm": 0.5990599393844604,
      "learning_rate": 2.9915333333333335e-05,
      "loss": 0.1305,
      "step": 10257
    },
    {
      "epoch": 18.927054478301017,
      "grad_norm": 0.7151499390602112,
      "learning_rate": 2.9915e-05,
      "loss": 0.1737,
      "step": 10258
    },
    {
      "epoch": 18.928901200369346,
      "grad_norm": 0.5996337532997131,
      "learning_rate": 2.9914666666666667e-05,
      "loss": 0.1591,
      "step": 10259
    },
    {
      "epoch": 18.930747922437675,
      "grad_norm": 0.7856311202049255,
      "learning_rate": 2.9914333333333336e-05,
      "loss": 0.1495,
      "step": 10260
    },
    {
      "epoch": 18.932594644506,
      "grad_norm": 0.5247669219970703,
      "learning_rate": 2.9914000000000002e-05,
      "loss": 0.0875,
      "step": 10261
    },
    {
      "epoch": 18.93444136657433,
      "grad_norm": 0.5936292409896851,
      "learning_rate": 2.9913666666666668e-05,
      "loss": 0.0807,
      "step": 10262
    },
    {
      "epoch": 18.93628808864266,
      "grad_norm": 0.595765233039856,
      "learning_rate": 2.9913333333333334e-05,
      "loss": 0.0844,
      "step": 10263
    },
    {
      "epoch": 18.938134810710988,
      "grad_norm": 0.5866780877113342,
      "learning_rate": 2.9913e-05,
      "loss": 0.0498,
      "step": 10264
    },
    {
      "epoch": 18.939981532779317,
      "grad_norm": 0.40006527304649353,
      "learning_rate": 2.991266666666667e-05,
      "loss": 0.0411,
      "step": 10265
    },
    {
      "epoch": 18.941828254847646,
      "grad_norm": 0.3506452441215515,
      "learning_rate": 2.991233333333333e-05,
      "loss": 0.0972,
      "step": 10266
    },
    {
      "epoch": 18.943674976915975,
      "grad_norm": 0.2703535854816437,
      "learning_rate": 2.9912e-05,
      "loss": 0.0624,
      "step": 10267
    },
    {
      "epoch": 18.945521698984304,
      "grad_norm": 0.841524600982666,
      "learning_rate": 2.9911666666666667e-05,
      "loss": 0.0254,
      "step": 10268
    },
    {
      "epoch": 18.94736842105263,
      "grad_norm": 0.3244848847389221,
      "learning_rate": 2.9911333333333333e-05,
      "loss": 0.0193,
      "step": 10269
    },
    {
      "epoch": 18.94921514312096,
      "grad_norm": 3.271413564682007,
      "learning_rate": 2.9911000000000002e-05,
      "loss": 0.0191,
      "step": 10270
    },
    {
      "epoch": 18.95106186518929,
      "grad_norm": 0.24318455159664154,
      "learning_rate": 2.9910666666666668e-05,
      "loss": 0.0213,
      "step": 10271
    },
    {
      "epoch": 18.952908587257618,
      "grad_norm": 0.617788553237915,
      "learning_rate": 2.9910333333333334e-05,
      "loss": 0.0208,
      "step": 10272
    },
    {
      "epoch": 18.954755309325947,
      "grad_norm": 0.33688652515411377,
      "learning_rate": 2.991e-05,
      "loss": 0.0092,
      "step": 10273
    },
    {
      "epoch": 18.956602031394276,
      "grad_norm": 1.3016357421875,
      "learning_rate": 2.990966666666667e-05,
      "loss": 0.0223,
      "step": 10274
    },
    {
      "epoch": 18.958448753462605,
      "grad_norm": 0.3748088479042053,
      "learning_rate": 2.9909333333333332e-05,
      "loss": 0.0134,
      "step": 10275
    },
    {
      "epoch": 18.960295475530934,
      "grad_norm": 0.5534132719039917,
      "learning_rate": 2.9909e-05,
      "loss": 0.0253,
      "step": 10276
    },
    {
      "epoch": 18.96214219759926,
      "grad_norm": 0.4077107906341553,
      "learning_rate": 2.9908666666666667e-05,
      "loss": 0.0282,
      "step": 10277
    },
    {
      "epoch": 18.96398891966759,
      "grad_norm": 0.355773389339447,
      "learning_rate": 2.9908333333333333e-05,
      "loss": 0.019,
      "step": 10278
    },
    {
      "epoch": 18.965835641735918,
      "grad_norm": 0.42264899611473083,
      "learning_rate": 2.9908000000000002e-05,
      "loss": 0.009,
      "step": 10279
    },
    {
      "epoch": 18.967682363804247,
      "grad_norm": 0.40342891216278076,
      "learning_rate": 2.9907666666666668e-05,
      "loss": 0.0136,
      "step": 10280
    },
    {
      "epoch": 18.969529085872576,
      "grad_norm": 0.5260114669799805,
      "learning_rate": 2.9907333333333334e-05,
      "loss": 0.0631,
      "step": 10281
    },
    {
      "epoch": 18.971375807940905,
      "grad_norm": 0.46058377623558044,
      "learning_rate": 2.9907e-05,
      "loss": 0.0161,
      "step": 10282
    },
    {
      "epoch": 18.973222530009235,
      "grad_norm": 0.6268635988235474,
      "learning_rate": 2.990666666666667e-05,
      "loss": 0.0119,
      "step": 10283
    },
    {
      "epoch": 18.975069252077564,
      "grad_norm": 0.4749756157398224,
      "learning_rate": 2.9906333333333332e-05,
      "loss": 0.0122,
      "step": 10284
    },
    {
      "epoch": 18.97691597414589,
      "grad_norm": 0.2996837794780731,
      "learning_rate": 2.9906e-05,
      "loss": 0.0116,
      "step": 10285
    },
    {
      "epoch": 18.97876269621422,
      "grad_norm": 0.4519551694393158,
      "learning_rate": 2.9905666666666667e-05,
      "loss": 0.0142,
      "step": 10286
    },
    {
      "epoch": 18.980609418282548,
      "grad_norm": 0.30716121196746826,
      "learning_rate": 2.9905333333333333e-05,
      "loss": 0.0125,
      "step": 10287
    },
    {
      "epoch": 18.982456140350877,
      "grad_norm": 0.34883931279182434,
      "learning_rate": 2.9905000000000003e-05,
      "loss": 0.0156,
      "step": 10288
    },
    {
      "epoch": 18.984302862419206,
      "grad_norm": 0.4810369312763214,
      "learning_rate": 2.990466666666667e-05,
      "loss": 0.0261,
      "step": 10289
    },
    {
      "epoch": 18.986149584487535,
      "grad_norm": 0.46950608491897583,
      "learning_rate": 2.9904333333333334e-05,
      "loss": 0.015,
      "step": 10290
    },
    {
      "epoch": 18.987996306555864,
      "grad_norm": 0.30146342515945435,
      "learning_rate": 2.9904e-05,
      "loss": 0.017,
      "step": 10291
    },
    {
      "epoch": 18.989843028624193,
      "grad_norm": 0.5918002724647522,
      "learning_rate": 2.9903666666666666e-05,
      "loss": 0.021,
      "step": 10292
    },
    {
      "epoch": 18.991689750692522,
      "grad_norm": 0.25285613536834717,
      "learning_rate": 2.9903333333333332e-05,
      "loss": 0.0108,
      "step": 10293
    },
    {
      "epoch": 18.993536472760848,
      "grad_norm": 0.5188554525375366,
      "learning_rate": 2.9903e-05,
      "loss": 0.0269,
      "step": 10294
    },
    {
      "epoch": 18.995383194829177,
      "grad_norm": 0.22588178515434265,
      "learning_rate": 2.9902666666666667e-05,
      "loss": 0.0087,
      "step": 10295
    },
    {
      "epoch": 18.997229916897506,
      "grad_norm": 0.33000004291534424,
      "learning_rate": 2.9902333333333333e-05,
      "loss": 0.0087,
      "step": 10296
    },
    {
      "epoch": 18.999076638965835,
      "grad_norm": 0.39947277307510376,
      "learning_rate": 2.9902000000000003e-05,
      "loss": 0.012,
      "step": 10297
    },
    {
      "epoch": 19.0,
      "grad_norm": 0.390468955039978,
      "learning_rate": 2.9901666666666665e-05,
      "loss": 0.0301,
      "step": 10298
    },
    {
      "epoch": 19.00184672206833,
      "grad_norm": 0.7757726311683655,
      "learning_rate": 2.9901333333333335e-05,
      "loss": 0.1456,
      "step": 10299
    },
    {
      "epoch": 19.00369344413666,
      "grad_norm": 0.7913090586662292,
      "learning_rate": 2.9901e-05,
      "loss": 0.1691,
      "step": 10300
    },
    {
      "epoch": 19.005540166204987,
      "grad_norm": 0.677194356918335,
      "learning_rate": 2.9900666666666666e-05,
      "loss": 0.0949,
      "step": 10301
    },
    {
      "epoch": 19.007386888273317,
      "grad_norm": 0.788819432258606,
      "learning_rate": 2.9900333333333332e-05,
      "loss": 0.0753,
      "step": 10302
    },
    {
      "epoch": 19.009233610341642,
      "grad_norm": 0.5638797283172607,
      "learning_rate": 2.9900000000000002e-05,
      "loss": 0.0793,
      "step": 10303
    },
    {
      "epoch": 19.01108033240997,
      "grad_norm": 0.4615686237812042,
      "learning_rate": 2.9899666666666668e-05,
      "loss": 0.1133,
      "step": 10304
    },
    {
      "epoch": 19.0129270544783,
      "grad_norm": 0.5038096904754639,
      "learning_rate": 2.9899333333333334e-05,
      "loss": 0.0578,
      "step": 10305
    },
    {
      "epoch": 19.01477377654663,
      "grad_norm": 0.767895519733429,
      "learning_rate": 2.9899000000000003e-05,
      "loss": 0.0495,
      "step": 10306
    },
    {
      "epoch": 19.01662049861496,
      "grad_norm": 0.35079145431518555,
      "learning_rate": 2.9898666666666665e-05,
      "loss": 0.0444,
      "step": 10307
    },
    {
      "epoch": 19.018467220683288,
      "grad_norm": 0.43407732248306274,
      "learning_rate": 2.9898333333333335e-05,
      "loss": 0.0537,
      "step": 10308
    },
    {
      "epoch": 19.020313942751617,
      "grad_norm": 0.26736965775489807,
      "learning_rate": 2.9898e-05,
      "loss": 0.0311,
      "step": 10309
    },
    {
      "epoch": 19.022160664819946,
      "grad_norm": 0.45536327362060547,
      "learning_rate": 2.9897666666666667e-05,
      "loss": 0.0389,
      "step": 10310
    },
    {
      "epoch": 19.02400738688827,
      "grad_norm": 0.31458866596221924,
      "learning_rate": 2.9897333333333336e-05,
      "loss": 0.0386,
      "step": 10311
    },
    {
      "epoch": 19.0258541089566,
      "grad_norm": 0.2078089565038681,
      "learning_rate": 2.9897000000000002e-05,
      "loss": 0.0229,
      "step": 10312
    },
    {
      "epoch": 19.02770083102493,
      "grad_norm": 0.42538657784461975,
      "learning_rate": 2.9896666666666668e-05,
      "loss": 0.0267,
      "step": 10313
    },
    {
      "epoch": 19.02954755309326,
      "grad_norm": 0.7776997685432434,
      "learning_rate": 2.9896333333333334e-05,
      "loss": 0.0207,
      "step": 10314
    },
    {
      "epoch": 19.03139427516159,
      "grad_norm": 0.46976688504219055,
      "learning_rate": 2.9896000000000003e-05,
      "loss": 0.0272,
      "step": 10315
    },
    {
      "epoch": 19.033240997229917,
      "grad_norm": 0.1779782623052597,
      "learning_rate": 2.9895666666666666e-05,
      "loss": 0.0086,
      "step": 10316
    },
    {
      "epoch": 19.035087719298247,
      "grad_norm": 0.4215569794178009,
      "learning_rate": 2.9895333333333335e-05,
      "loss": 0.0153,
      "step": 10317
    },
    {
      "epoch": 19.036934441366576,
      "grad_norm": 0.37013107538223267,
      "learning_rate": 2.9895e-05,
      "loss": 0.0166,
      "step": 10318
    },
    {
      "epoch": 19.0387811634349,
      "grad_norm": 0.1785709410905838,
      "learning_rate": 2.9894666666666667e-05,
      "loss": 0.0082,
      "step": 10319
    },
    {
      "epoch": 19.04062788550323,
      "grad_norm": 0.7148463129997253,
      "learning_rate": 2.9894333333333336e-05,
      "loss": 0.0334,
      "step": 10320
    },
    {
      "epoch": 19.04247460757156,
      "grad_norm": 0.35422831773757935,
      "learning_rate": 2.9894e-05,
      "loss": 0.0128,
      "step": 10321
    },
    {
      "epoch": 19.04432132963989,
      "grad_norm": 0.19529224932193756,
      "learning_rate": 2.9893666666666668e-05,
      "loss": 0.0104,
      "step": 10322
    },
    {
      "epoch": 19.046168051708218,
      "grad_norm": 0.21543395519256592,
      "learning_rate": 2.9893333333333334e-05,
      "loss": 0.012,
      "step": 10323
    },
    {
      "epoch": 19.048014773776547,
      "grad_norm": 0.25531303882598877,
      "learning_rate": 2.9893e-05,
      "loss": 0.0082,
      "step": 10324
    },
    {
      "epoch": 19.049861495844876,
      "grad_norm": 0.20938219130039215,
      "learning_rate": 2.9892666666666666e-05,
      "loss": 0.0104,
      "step": 10325
    },
    {
      "epoch": 19.051708217913205,
      "grad_norm": 0.3913450837135315,
      "learning_rate": 2.9892333333333335e-05,
      "loss": 0.0077,
      "step": 10326
    },
    {
      "epoch": 19.053554939981534,
      "grad_norm": 0.16339915990829468,
      "learning_rate": 2.9891999999999998e-05,
      "loss": 0.0056,
      "step": 10327
    },
    {
      "epoch": 19.05540166204986,
      "grad_norm": 0.2552281618118286,
      "learning_rate": 2.9891666666666667e-05,
      "loss": 0.006,
      "step": 10328
    },
    {
      "epoch": 19.05724838411819,
      "grad_norm": 0.35827621817588806,
      "learning_rate": 2.9891333333333336e-05,
      "loss": 0.0098,
      "step": 10329
    },
    {
      "epoch": 19.05909510618652,
      "grad_norm": 0.2173592448234558,
      "learning_rate": 2.9891e-05,
      "loss": 0.0077,
      "step": 10330
    },
    {
      "epoch": 19.060941828254848,
      "grad_norm": 0.2610386312007904,
      "learning_rate": 2.9890666666666668e-05,
      "loss": 0.0131,
      "step": 10331
    },
    {
      "epoch": 19.062788550323177,
      "grad_norm": 0.18036872148513794,
      "learning_rate": 2.9890333333333334e-05,
      "loss": 0.0064,
      "step": 10332
    },
    {
      "epoch": 19.064635272391506,
      "grad_norm": 0.5067037343978882,
      "learning_rate": 2.989e-05,
      "loss": 0.0105,
      "step": 10333
    },
    {
      "epoch": 19.066481994459835,
      "grad_norm": 0.7099167704582214,
      "learning_rate": 2.9889666666666666e-05,
      "loss": 0.0082,
      "step": 10334
    },
    {
      "epoch": 19.068328716528164,
      "grad_norm": 0.1525428295135498,
      "learning_rate": 2.9889333333333335e-05,
      "loss": 0.0053,
      "step": 10335
    },
    {
      "epoch": 19.07017543859649,
      "grad_norm": 0.31399956345558167,
      "learning_rate": 2.9889e-05,
      "loss": 0.0113,
      "step": 10336
    },
    {
      "epoch": 19.07202216066482,
      "grad_norm": 0.45584020018577576,
      "learning_rate": 2.9888666666666667e-05,
      "loss": 0.0089,
      "step": 10337
    },
    {
      "epoch": 19.073868882733148,
      "grad_norm": 0.6998863220214844,
      "learning_rate": 2.9888333333333337e-05,
      "loss": 0.0135,
      "step": 10338
    },
    {
      "epoch": 19.075715604801477,
      "grad_norm": 0.3413309156894684,
      "learning_rate": 2.9888e-05,
      "loss": 0.0111,
      "step": 10339
    },
    {
      "epoch": 19.077562326869806,
      "grad_norm": 0.25519177317619324,
      "learning_rate": 2.988766666666667e-05,
      "loss": 0.0064,
      "step": 10340
    },
    {
      "epoch": 19.079409048938135,
      "grad_norm": 0.5262901186943054,
      "learning_rate": 2.9887333333333334e-05,
      "loss": 0.0077,
      "step": 10341
    },
    {
      "epoch": 19.081255771006465,
      "grad_norm": 0.9745499491691589,
      "learning_rate": 2.9887e-05,
      "loss": 0.0123,
      "step": 10342
    },
    {
      "epoch": 19.083102493074794,
      "grad_norm": 0.34378862380981445,
      "learning_rate": 2.9886666666666666e-05,
      "loss": 0.0072,
      "step": 10343
    },
    {
      "epoch": 19.08494921514312,
      "grad_norm": 0.300809770822525,
      "learning_rate": 2.9886333333333336e-05,
      "loss": 0.0063,
      "step": 10344
    },
    {
      "epoch": 19.08679593721145,
      "grad_norm": 0.34393367171287537,
      "learning_rate": 2.9886e-05,
      "loss": 0.0055,
      "step": 10345
    },
    {
      "epoch": 19.088642659279778,
      "grad_norm": 0.2126002013683319,
      "learning_rate": 2.9885666666666667e-05,
      "loss": 0.0057,
      "step": 10346
    },
    {
      "epoch": 19.090489381348107,
      "grad_norm": 0.394257515668869,
      "learning_rate": 2.9885333333333337e-05,
      "loss": 0.0123,
      "step": 10347
    },
    {
      "epoch": 19.092336103416436,
      "grad_norm": 0.4051159620285034,
      "learning_rate": 2.9885e-05,
      "loss": 0.031,
      "step": 10348
    },
    {
      "epoch": 19.094182825484765,
      "grad_norm": 0.7709018588066101,
      "learning_rate": 2.988466666666667e-05,
      "loss": 0.2503,
      "step": 10349
    },
    {
      "epoch": 19.096029547553094,
      "grad_norm": 0.6019572019577026,
      "learning_rate": 2.9884333333333335e-05,
      "loss": 0.1228,
      "step": 10350
    },
    {
      "epoch": 19.097876269621423,
      "grad_norm": 0.5292881727218628,
      "learning_rate": 2.9884e-05,
      "loss": 0.1702,
      "step": 10351
    },
    {
      "epoch": 19.099722991689752,
      "grad_norm": 0.5636563301086426,
      "learning_rate": 2.9883666666666666e-05,
      "loss": 0.1217,
      "step": 10352
    },
    {
      "epoch": 19.101569713758078,
      "grad_norm": 0.45421287417411804,
      "learning_rate": 2.9883333333333332e-05,
      "loss": 0.0916,
      "step": 10353
    },
    {
      "epoch": 19.103416435826407,
      "grad_norm": 0.6303958892822266,
      "learning_rate": 2.9883000000000002e-05,
      "loss": 0.0635,
      "step": 10354
    },
    {
      "epoch": 19.105263157894736,
      "grad_norm": 0.7112407088279724,
      "learning_rate": 2.9882666666666668e-05,
      "loss": 0.0863,
      "step": 10355
    },
    {
      "epoch": 19.107109879963065,
      "grad_norm": 0.6119093894958496,
      "learning_rate": 2.9882333333333334e-05,
      "loss": 0.0879,
      "step": 10356
    },
    {
      "epoch": 19.108956602031395,
      "grad_norm": 0.42396941781044006,
      "learning_rate": 2.9882e-05,
      "loss": 0.0857,
      "step": 10357
    },
    {
      "epoch": 19.110803324099724,
      "grad_norm": 0.6566585302352905,
      "learning_rate": 2.988166666666667e-05,
      "loss": 0.0547,
      "step": 10358
    },
    {
      "epoch": 19.112650046168053,
      "grad_norm": 0.37996527552604675,
      "learning_rate": 2.988133333333333e-05,
      "loss": 0.0465,
      "step": 10359
    },
    {
      "epoch": 19.114496768236382,
      "grad_norm": 0.8353615999221802,
      "learning_rate": 2.9881e-05,
      "loss": 0.0446,
      "step": 10360
    },
    {
      "epoch": 19.116343490304708,
      "grad_norm": 0.3163006901741028,
      "learning_rate": 2.988066666666667e-05,
      "loss": 0.0311,
      "step": 10361
    },
    {
      "epoch": 19.118190212373037,
      "grad_norm": 0.4427793323993683,
      "learning_rate": 2.9880333333333333e-05,
      "loss": 0.0476,
      "step": 10362
    },
    {
      "epoch": 19.120036934441366,
      "grad_norm": 0.24254950881004333,
      "learning_rate": 2.9880000000000002e-05,
      "loss": 0.0231,
      "step": 10363
    },
    {
      "epoch": 19.121883656509695,
      "grad_norm": 0.38209423422813416,
      "learning_rate": 2.9879666666666668e-05,
      "loss": 0.0591,
      "step": 10364
    },
    {
      "epoch": 19.123730378578024,
      "grad_norm": 0.18305137753486633,
      "learning_rate": 2.9879333333333334e-05,
      "loss": 0.0107,
      "step": 10365
    },
    {
      "epoch": 19.125577100646353,
      "grad_norm": 0.42425912618637085,
      "learning_rate": 2.9879e-05,
      "loss": 0.0102,
      "step": 10366
    },
    {
      "epoch": 19.127423822714682,
      "grad_norm": 0.24442718923091888,
      "learning_rate": 2.987866666666667e-05,
      "loss": 0.0418,
      "step": 10367
    },
    {
      "epoch": 19.12927054478301,
      "grad_norm": 0.3955996036529541,
      "learning_rate": 2.987833333333333e-05,
      "loss": 0.0108,
      "step": 10368
    },
    {
      "epoch": 19.131117266851337,
      "grad_norm": 0.1432962566614151,
      "learning_rate": 2.9878e-05,
      "loss": 0.0078,
      "step": 10369
    },
    {
      "epoch": 19.132963988919666,
      "grad_norm": 0.2589716911315918,
      "learning_rate": 2.987766666666667e-05,
      "loss": 0.0146,
      "step": 10370
    },
    {
      "epoch": 19.134810710987995,
      "grad_norm": 0.22195583581924438,
      "learning_rate": 2.9877333333333333e-05,
      "loss": 0.0093,
      "step": 10371
    },
    {
      "epoch": 19.136657433056325,
      "grad_norm": 0.1908663958311081,
      "learning_rate": 2.9877000000000002e-05,
      "loss": 0.0044,
      "step": 10372
    },
    {
      "epoch": 19.138504155124654,
      "grad_norm": 0.154612734913826,
      "learning_rate": 2.9876666666666668e-05,
      "loss": 0.0076,
      "step": 10373
    },
    {
      "epoch": 19.140350877192983,
      "grad_norm": 0.4131031334400177,
      "learning_rate": 2.9876333333333334e-05,
      "loss": 0.0117,
      "step": 10374
    },
    {
      "epoch": 19.142197599261312,
      "grad_norm": 0.3654535412788391,
      "learning_rate": 2.9876e-05,
      "loss": 0.0149,
      "step": 10375
    },
    {
      "epoch": 19.14404432132964,
      "grad_norm": 0.24689264595508575,
      "learning_rate": 2.987566666666667e-05,
      "loss": 0.0095,
      "step": 10376
    },
    {
      "epoch": 19.14589104339797,
      "grad_norm": 0.24347740411758423,
      "learning_rate": 2.9875333333333332e-05,
      "loss": 0.0082,
      "step": 10377
    },
    {
      "epoch": 19.147737765466296,
      "grad_norm": 0.531166672706604,
      "learning_rate": 2.9875e-05,
      "loss": 0.0578,
      "step": 10378
    },
    {
      "epoch": 19.149584487534625,
      "grad_norm": 0.4409041404724121,
      "learning_rate": 2.987466666666667e-05,
      "loss": 0.0162,
      "step": 10379
    },
    {
      "epoch": 19.151431209602954,
      "grad_norm": 0.35764533281326294,
      "learning_rate": 2.9874333333333333e-05,
      "loss": 0.0084,
      "step": 10380
    },
    {
      "epoch": 19.153277931671283,
      "grad_norm": 0.4136661887168884,
      "learning_rate": 2.9874000000000002e-05,
      "loss": 0.0075,
      "step": 10381
    },
    {
      "epoch": 19.155124653739612,
      "grad_norm": 1.1931157112121582,
      "learning_rate": 2.9873666666666665e-05,
      "loss": 0.0154,
      "step": 10382
    },
    {
      "epoch": 19.15697137580794,
      "grad_norm": 0.1597888022661209,
      "learning_rate": 2.9873333333333334e-05,
      "loss": 0.0054,
      "step": 10383
    },
    {
      "epoch": 19.15881809787627,
      "grad_norm": 0.4281260669231415,
      "learning_rate": 2.9873e-05,
      "loss": 0.0405,
      "step": 10384
    },
    {
      "epoch": 19.1606648199446,
      "grad_norm": 0.4385799169540405,
      "learning_rate": 2.9872666666666666e-05,
      "loss": 0.0112,
      "step": 10385
    },
    {
      "epoch": 19.162511542012926,
      "grad_norm": 0.22523953020572662,
      "learning_rate": 2.9872333333333335e-05,
      "loss": 0.0085,
      "step": 10386
    },
    {
      "epoch": 19.164358264081255,
      "grad_norm": 0.36893364787101746,
      "learning_rate": 2.9872e-05,
      "loss": 0.0204,
      "step": 10387
    },
    {
      "epoch": 19.166204986149584,
      "grad_norm": 0.7368447184562683,
      "learning_rate": 2.9871666666666667e-05,
      "loss": 0.0126,
      "step": 10388
    },
    {
      "epoch": 19.168051708217913,
      "grad_norm": 0.29873305559158325,
      "learning_rate": 2.9871333333333333e-05,
      "loss": 0.0117,
      "step": 10389
    },
    {
      "epoch": 19.169898430286242,
      "grad_norm": 0.4472907483577728,
      "learning_rate": 2.9871000000000003e-05,
      "loss": 0.009,
      "step": 10390
    },
    {
      "epoch": 19.17174515235457,
      "grad_norm": 0.9163946509361267,
      "learning_rate": 2.9870666666666665e-05,
      "loss": 0.0127,
      "step": 10391
    },
    {
      "epoch": 19.1735918744229,
      "grad_norm": 0.2877427935600281,
      "learning_rate": 2.9870333333333334e-05,
      "loss": 0.0112,
      "step": 10392
    },
    {
      "epoch": 19.17543859649123,
      "grad_norm": 0.5438757538795471,
      "learning_rate": 2.987e-05,
      "loss": 0.0092,
      "step": 10393
    },
    {
      "epoch": 19.177285318559555,
      "grad_norm": 0.34730347990989685,
      "learning_rate": 2.9869666666666666e-05,
      "loss": 0.0083,
      "step": 10394
    },
    {
      "epoch": 19.179132040627884,
      "grad_norm": 0.49923327565193176,
      "learning_rate": 2.9869333333333336e-05,
      "loss": 0.0143,
      "step": 10395
    },
    {
      "epoch": 19.180978762696213,
      "grad_norm": 0.19998763501644135,
      "learning_rate": 2.9869e-05,
      "loss": 0.0047,
      "step": 10396
    },
    {
      "epoch": 19.182825484764543,
      "grad_norm": 0.4820026457309723,
      "learning_rate": 2.9868666666666667e-05,
      "loss": 0.0136,
      "step": 10397
    },
    {
      "epoch": 19.18467220683287,
      "grad_norm": 0.5864433646202087,
      "learning_rate": 2.9868333333333333e-05,
      "loss": 0.0157,
      "step": 10398
    },
    {
      "epoch": 19.1865189289012,
      "grad_norm": 0.5068093538284302,
      "learning_rate": 2.9868000000000003e-05,
      "loss": 0.1275,
      "step": 10399
    },
    {
      "epoch": 19.18836565096953,
      "grad_norm": 0.4917617440223694,
      "learning_rate": 2.9867666666666665e-05,
      "loss": 0.1254,
      "step": 10400
    },
    {
      "epoch": 19.19021237303786,
      "grad_norm": 0.8632218837738037,
      "learning_rate": 2.9867333333333335e-05,
      "loss": 0.1655,
      "step": 10401
    },
    {
      "epoch": 19.19205909510619,
      "grad_norm": 0.4661218225955963,
      "learning_rate": 2.9867e-05,
      "loss": 0.0876,
      "step": 10402
    },
    {
      "epoch": 19.193905817174514,
      "grad_norm": 0.6275524497032166,
      "learning_rate": 2.9866666666666666e-05,
      "loss": 0.0873,
      "step": 10403
    },
    {
      "epoch": 19.195752539242843,
      "grad_norm": 0.4922047555446625,
      "learning_rate": 2.9866333333333336e-05,
      "loss": 0.0897,
      "step": 10404
    },
    {
      "epoch": 19.197599261311172,
      "grad_norm": 0.43131202459335327,
      "learning_rate": 2.9866000000000002e-05,
      "loss": 0.0591,
      "step": 10405
    },
    {
      "epoch": 19.1994459833795,
      "grad_norm": 0.536503255367279,
      "learning_rate": 2.9865666666666668e-05,
      "loss": 0.0911,
      "step": 10406
    },
    {
      "epoch": 19.20129270544783,
      "grad_norm": 0.3472905457019806,
      "learning_rate": 2.9865333333333334e-05,
      "loss": 0.0381,
      "step": 10407
    },
    {
      "epoch": 19.20313942751616,
      "grad_norm": 0.49963095784187317,
      "learning_rate": 2.9865000000000003e-05,
      "loss": 0.0509,
      "step": 10408
    },
    {
      "epoch": 19.20498614958449,
      "grad_norm": 0.48727062344551086,
      "learning_rate": 2.9864666666666665e-05,
      "loss": 0.0497,
      "step": 10409
    },
    {
      "epoch": 19.206832871652818,
      "grad_norm": 0.33055925369262695,
      "learning_rate": 2.9864333333333335e-05,
      "loss": 0.0492,
      "step": 10410
    },
    {
      "epoch": 19.208679593721143,
      "grad_norm": 0.4543503224849701,
      "learning_rate": 2.9864000000000004e-05,
      "loss": 0.0365,
      "step": 10411
    },
    {
      "epoch": 19.210526315789473,
      "grad_norm": 0.316006600856781,
      "learning_rate": 2.9863666666666667e-05,
      "loss": 0.0228,
      "step": 10412
    },
    {
      "epoch": 19.2123730378578,
      "grad_norm": 0.39846211671829224,
      "learning_rate": 2.9863333333333336e-05,
      "loss": 0.0144,
      "step": 10413
    },
    {
      "epoch": 19.21421975992613,
      "grad_norm": 0.32504454255104065,
      "learning_rate": 2.9863e-05,
      "loss": 0.0176,
      "step": 10414
    },
    {
      "epoch": 19.21606648199446,
      "grad_norm": 0.3825388252735138,
      "learning_rate": 2.9862666666666668e-05,
      "loss": 0.0137,
      "step": 10415
    },
    {
      "epoch": 19.21791320406279,
      "grad_norm": 0.276236355304718,
      "learning_rate": 2.9862333333333334e-05,
      "loss": 0.0174,
      "step": 10416
    },
    {
      "epoch": 19.21975992613112,
      "grad_norm": 0.31894567608833313,
      "learning_rate": 2.9862e-05,
      "loss": 0.0318,
      "step": 10417
    },
    {
      "epoch": 19.221606648199447,
      "grad_norm": 0.39254486560821533,
      "learning_rate": 2.9861666666666666e-05,
      "loss": 0.0322,
      "step": 10418
    },
    {
      "epoch": 19.223453370267773,
      "grad_norm": 0.3326973021030426,
      "learning_rate": 2.9861333333333335e-05,
      "loss": 0.0087,
      "step": 10419
    },
    {
      "epoch": 19.225300092336102,
      "grad_norm": 0.29619577527046204,
      "learning_rate": 2.9861e-05,
      "loss": 0.0089,
      "step": 10420
    },
    {
      "epoch": 19.22714681440443,
      "grad_norm": 0.3656005561351776,
      "learning_rate": 2.9860666666666667e-05,
      "loss": 0.0185,
      "step": 10421
    },
    {
      "epoch": 19.22899353647276,
      "grad_norm": 0.26014333963394165,
      "learning_rate": 2.9860333333333336e-05,
      "loss": 0.008,
      "step": 10422
    },
    {
      "epoch": 19.23084025854109,
      "grad_norm": 0.5052050948143005,
      "learning_rate": 2.986e-05,
      "loss": 0.0136,
      "step": 10423
    },
    {
      "epoch": 19.23268698060942,
      "grad_norm": 0.3282339572906494,
      "learning_rate": 2.9859666666666668e-05,
      "loss": 0.0141,
      "step": 10424
    },
    {
      "epoch": 19.234533702677748,
      "grad_norm": 0.33655408024787903,
      "learning_rate": 2.9859333333333334e-05,
      "loss": 0.0134,
      "step": 10425
    },
    {
      "epoch": 19.236380424746077,
      "grad_norm": 0.2944200336933136,
      "learning_rate": 2.9859e-05,
      "loss": 0.0103,
      "step": 10426
    },
    {
      "epoch": 19.238227146814406,
      "grad_norm": 0.49892765283584595,
      "learning_rate": 2.9858666666666666e-05,
      "loss": 0.0275,
      "step": 10427
    },
    {
      "epoch": 19.24007386888273,
      "grad_norm": 0.4131115972995758,
      "learning_rate": 2.9858333333333335e-05,
      "loss": 0.0096,
      "step": 10428
    },
    {
      "epoch": 19.24192059095106,
      "grad_norm": 0.2645561397075653,
      "learning_rate": 2.9858e-05,
      "loss": 0.0065,
      "step": 10429
    },
    {
      "epoch": 19.24376731301939,
      "grad_norm": 0.40666913986206055,
      "learning_rate": 2.9857666666666667e-05,
      "loss": 0.0111,
      "step": 10430
    },
    {
      "epoch": 19.24561403508772,
      "grad_norm": 0.22525639832019806,
      "learning_rate": 2.9857333333333336e-05,
      "loss": 0.0064,
      "step": 10431
    },
    {
      "epoch": 19.24746075715605,
      "grad_norm": 0.400501549243927,
      "learning_rate": 2.9857e-05,
      "loss": 0.0097,
      "step": 10432
    },
    {
      "epoch": 19.249307479224377,
      "grad_norm": 0.28048983216285706,
      "learning_rate": 2.9856666666666668e-05,
      "loss": 0.0093,
      "step": 10433
    },
    {
      "epoch": 19.251154201292707,
      "grad_norm": 0.21725383400917053,
      "learning_rate": 2.9856333333333334e-05,
      "loss": 0.0069,
      "step": 10434
    },
    {
      "epoch": 19.253000923361036,
      "grad_norm": 0.294262558221817,
      "learning_rate": 2.9856e-05,
      "loss": 0.0098,
      "step": 10435
    },
    {
      "epoch": 19.25484764542936,
      "grad_norm": 0.43420305848121643,
      "learning_rate": 2.985566666666667e-05,
      "loss": 0.0189,
      "step": 10436
    },
    {
      "epoch": 19.25669436749769,
      "grad_norm": 0.19349953532218933,
      "learning_rate": 2.9855333333333335e-05,
      "loss": 0.0043,
      "step": 10437
    },
    {
      "epoch": 19.25854108956602,
      "grad_norm": 0.26871469616889954,
      "learning_rate": 2.9855e-05,
      "loss": 0.0068,
      "step": 10438
    },
    {
      "epoch": 19.26038781163435,
      "grad_norm": 0.18552646040916443,
      "learning_rate": 2.9854666666666667e-05,
      "loss": 0.005,
      "step": 10439
    },
    {
      "epoch": 19.262234533702678,
      "grad_norm": 0.2697330117225647,
      "learning_rate": 2.9854333333333337e-05,
      "loss": 0.0094,
      "step": 10440
    },
    {
      "epoch": 19.264081255771007,
      "grad_norm": 0.47541746497154236,
      "learning_rate": 2.9854e-05,
      "loss": 0.0094,
      "step": 10441
    },
    {
      "epoch": 19.265927977839336,
      "grad_norm": 0.2967643141746521,
      "learning_rate": 2.985366666666667e-05,
      "loss": 0.0071,
      "step": 10442
    },
    {
      "epoch": 19.267774699907665,
      "grad_norm": 0.3433935046195984,
      "learning_rate": 2.985333333333333e-05,
      "loss": 0.0118,
      "step": 10443
    },
    {
      "epoch": 19.26962142197599,
      "grad_norm": 0.2878125011920929,
      "learning_rate": 2.9853e-05,
      "loss": 0.0088,
      "step": 10444
    },
    {
      "epoch": 19.27146814404432,
      "grad_norm": 0.8284724354743958,
      "learning_rate": 2.985266666666667e-05,
      "loss": 0.0102,
      "step": 10445
    },
    {
      "epoch": 19.27331486611265,
      "grad_norm": 0.244842529296875,
      "learning_rate": 2.9852333333333332e-05,
      "loss": 0.0118,
      "step": 10446
    },
    {
      "epoch": 19.27516158818098,
      "grad_norm": 0.3617515563964844,
      "learning_rate": 2.9852e-05,
      "loss": 0.0093,
      "step": 10447
    },
    {
      "epoch": 19.277008310249307,
      "grad_norm": 0.1818387806415558,
      "learning_rate": 2.9851666666666667e-05,
      "loss": 0.0045,
      "step": 10448
    },
    {
      "epoch": 19.278855032317637,
      "grad_norm": 0.6636005640029907,
      "learning_rate": 2.9851333333333333e-05,
      "loss": 0.1463,
      "step": 10449
    },
    {
      "epoch": 19.280701754385966,
      "grad_norm": 0.501599133014679,
      "learning_rate": 2.9851e-05,
      "loss": 0.1471,
      "step": 10450
    },
    {
      "epoch": 19.282548476454295,
      "grad_norm": 0.46136799454689026,
      "learning_rate": 2.985066666666667e-05,
      "loss": 0.1049,
      "step": 10451
    },
    {
      "epoch": 19.284395198522624,
      "grad_norm": 0.5088426470756531,
      "learning_rate": 2.985033333333333e-05,
      "loss": 0.1087,
      "step": 10452
    },
    {
      "epoch": 19.28624192059095,
      "grad_norm": 0.5903878808021545,
      "learning_rate": 2.985e-05,
      "loss": 0.1165,
      "step": 10453
    },
    {
      "epoch": 19.28808864265928,
      "grad_norm": 0.45213332772254944,
      "learning_rate": 2.984966666666667e-05,
      "loss": 0.066,
      "step": 10454
    },
    {
      "epoch": 19.289935364727608,
      "grad_norm": 0.4082927703857422,
      "learning_rate": 2.9849333333333332e-05,
      "loss": 0.0713,
      "step": 10455
    },
    {
      "epoch": 19.291782086795937,
      "grad_norm": 0.4666750729084015,
      "learning_rate": 2.9849000000000002e-05,
      "loss": 0.06,
      "step": 10456
    },
    {
      "epoch": 19.293628808864266,
      "grad_norm": 0.6132091283798218,
      "learning_rate": 2.9848666666666668e-05,
      "loss": 0.0729,
      "step": 10457
    },
    {
      "epoch": 19.295475530932595,
      "grad_norm": 0.37567561864852905,
      "learning_rate": 2.9848333333333334e-05,
      "loss": 0.0659,
      "step": 10458
    },
    {
      "epoch": 19.297322253000925,
      "grad_norm": 0.30930113792419434,
      "learning_rate": 2.9848e-05,
      "loss": 0.0263,
      "step": 10459
    },
    {
      "epoch": 19.299168975069254,
      "grad_norm": 0.3227839171886444,
      "learning_rate": 2.984766666666667e-05,
      "loss": 0.0456,
      "step": 10460
    },
    {
      "epoch": 19.30101569713758,
      "grad_norm": 0.451276034116745,
      "learning_rate": 2.9847333333333335e-05,
      "loss": 0.0269,
      "step": 10461
    },
    {
      "epoch": 19.30286241920591,
      "grad_norm": 0.754874050617218,
      "learning_rate": 2.9847e-05,
      "loss": 0.0179,
      "step": 10462
    },
    {
      "epoch": 19.304709141274238,
      "grad_norm": 0.20438246428966522,
      "learning_rate": 2.984666666666667e-05,
      "loss": 0.0115,
      "step": 10463
    },
    {
      "epoch": 19.306555863342567,
      "grad_norm": 0.2762463390827179,
      "learning_rate": 2.9846333333333333e-05,
      "loss": 0.0283,
      "step": 10464
    },
    {
      "epoch": 19.308402585410896,
      "grad_norm": 0.34477874636650085,
      "learning_rate": 2.9846000000000002e-05,
      "loss": 0.0225,
      "step": 10465
    },
    {
      "epoch": 19.310249307479225,
      "grad_norm": 0.5555924773216248,
      "learning_rate": 2.9845666666666668e-05,
      "loss": 0.0574,
      "step": 10466
    },
    {
      "epoch": 19.312096029547554,
      "grad_norm": 0.2817211449146271,
      "learning_rate": 2.9845333333333334e-05,
      "loss": 0.039,
      "step": 10467
    },
    {
      "epoch": 19.313942751615883,
      "grad_norm": 0.3302725553512573,
      "learning_rate": 2.9845e-05,
      "loss": 0.0107,
      "step": 10468
    },
    {
      "epoch": 19.31578947368421,
      "grad_norm": 0.29507848620414734,
      "learning_rate": 2.984466666666667e-05,
      "loss": 0.0093,
      "step": 10469
    },
    {
      "epoch": 19.317636195752538,
      "grad_norm": 0.2905113399028778,
      "learning_rate": 2.9844333333333335e-05,
      "loss": 0.0118,
      "step": 10470
    },
    {
      "epoch": 19.319482917820867,
      "grad_norm": 0.7147642970085144,
      "learning_rate": 2.9844e-05,
      "loss": 0.0197,
      "step": 10471
    },
    {
      "epoch": 19.321329639889196,
      "grad_norm": 0.14920946955680847,
      "learning_rate": 2.9843666666666667e-05,
      "loss": 0.0055,
      "step": 10472
    },
    {
      "epoch": 19.323176361957525,
      "grad_norm": 0.30008870363235474,
      "learning_rate": 2.9843333333333333e-05,
      "loss": 0.011,
      "step": 10473
    },
    {
      "epoch": 19.325023084025855,
      "grad_norm": 0.27200573682785034,
      "learning_rate": 2.9843000000000002e-05,
      "loss": 0.0085,
      "step": 10474
    },
    {
      "epoch": 19.326869806094184,
      "grad_norm": 0.2959764301776886,
      "learning_rate": 2.9842666666666665e-05,
      "loss": 0.0062,
      "step": 10475
    },
    {
      "epoch": 19.328716528162513,
      "grad_norm": 0.5562206506729126,
      "learning_rate": 2.9842333333333334e-05,
      "loss": 0.0083,
      "step": 10476
    },
    {
      "epoch": 19.330563250230842,
      "grad_norm": 0.42132067680358887,
      "learning_rate": 2.9842e-05,
      "loss": 0.0168,
      "step": 10477
    },
    {
      "epoch": 19.332409972299168,
      "grad_norm": 0.7031576037406921,
      "learning_rate": 2.9841666666666666e-05,
      "loss": 0.0189,
      "step": 10478
    },
    {
      "epoch": 19.334256694367497,
      "grad_norm": 0.23381592333316803,
      "learning_rate": 2.9841333333333335e-05,
      "loss": 0.0073,
      "step": 10479
    },
    {
      "epoch": 19.336103416435826,
      "grad_norm": 0.50837641954422,
      "learning_rate": 2.9841e-05,
      "loss": 0.0149,
      "step": 10480
    },
    {
      "epoch": 19.337950138504155,
      "grad_norm": 0.47945070266723633,
      "learning_rate": 2.9840666666666667e-05,
      "loss": 0.0093,
      "step": 10481
    },
    {
      "epoch": 19.339796860572484,
      "grad_norm": 0.1993795931339264,
      "learning_rate": 2.9840333333333333e-05,
      "loss": 0.0091,
      "step": 10482
    },
    {
      "epoch": 19.341643582640813,
      "grad_norm": 0.25505030155181885,
      "learning_rate": 2.9840000000000002e-05,
      "loss": 0.0047,
      "step": 10483
    },
    {
      "epoch": 19.343490304709142,
      "grad_norm": 0.2103731483221054,
      "learning_rate": 2.9839666666666665e-05,
      "loss": 0.0061,
      "step": 10484
    },
    {
      "epoch": 19.34533702677747,
      "grad_norm": 0.3151203989982605,
      "learning_rate": 2.9839333333333334e-05,
      "loss": 0.0113,
      "step": 10485
    },
    {
      "epoch": 19.347183748845797,
      "grad_norm": 0.365931898355484,
      "learning_rate": 2.9839000000000003e-05,
      "loss": 0.0097,
      "step": 10486
    },
    {
      "epoch": 19.349030470914126,
      "grad_norm": 0.25502514839172363,
      "learning_rate": 2.9838666666666666e-05,
      "loss": 0.0106,
      "step": 10487
    },
    {
      "epoch": 19.350877192982455,
      "grad_norm": 0.25653916597366333,
      "learning_rate": 2.9838333333333335e-05,
      "loss": 0.007,
      "step": 10488
    },
    {
      "epoch": 19.352723915050785,
      "grad_norm": 0.441602498292923,
      "learning_rate": 2.9838e-05,
      "loss": 0.0204,
      "step": 10489
    },
    {
      "epoch": 19.354570637119114,
      "grad_norm": 0.12745323777198792,
      "learning_rate": 2.9837666666666667e-05,
      "loss": 0.0051,
      "step": 10490
    },
    {
      "epoch": 19.356417359187443,
      "grad_norm": 0.3819082975387573,
      "learning_rate": 2.9837333333333333e-05,
      "loss": 0.0084,
      "step": 10491
    },
    {
      "epoch": 19.358264081255772,
      "grad_norm": 0.4130721688270569,
      "learning_rate": 2.9837000000000002e-05,
      "loss": 0.0104,
      "step": 10492
    },
    {
      "epoch": 19.3601108033241,
      "grad_norm": 0.2628061771392822,
      "learning_rate": 2.9836666666666665e-05,
      "loss": 0.0093,
      "step": 10493
    },
    {
      "epoch": 19.361957525392427,
      "grad_norm": 0.23770292103290558,
      "learning_rate": 2.9836333333333334e-05,
      "loss": 0.0052,
      "step": 10494
    },
    {
      "epoch": 19.363804247460756,
      "grad_norm": 0.4211592376232147,
      "learning_rate": 2.9836000000000004e-05,
      "loss": 0.0252,
      "step": 10495
    },
    {
      "epoch": 19.365650969529085,
      "grad_norm": 0.4491584002971649,
      "learning_rate": 2.9835666666666666e-05,
      "loss": 0.0161,
      "step": 10496
    },
    {
      "epoch": 19.367497691597414,
      "grad_norm": 0.38713496923446655,
      "learning_rate": 2.9835333333333336e-05,
      "loss": 0.0126,
      "step": 10497
    },
    {
      "epoch": 19.369344413665743,
      "grad_norm": 0.36408424377441406,
      "learning_rate": 2.9835e-05,
      "loss": 0.0093,
      "step": 10498
    },
    {
      "epoch": 19.371191135734072,
      "grad_norm": 0.767509937286377,
      "learning_rate": 2.9834666666666667e-05,
      "loss": 0.1899,
      "step": 10499
    },
    {
      "epoch": 19.3730378578024,
      "grad_norm": 0.5073058009147644,
      "learning_rate": 2.9834333333333333e-05,
      "loss": 0.1397,
      "step": 10500
    },
    {
      "epoch": 19.37488457987073,
      "grad_norm": 0.5196423530578613,
      "learning_rate": 2.9834000000000003e-05,
      "loss": 0.0866,
      "step": 10501
    },
    {
      "epoch": 19.37673130193906,
      "grad_norm": 0.4803215265274048,
      "learning_rate": 2.9833666666666665e-05,
      "loss": 0.0727,
      "step": 10502
    },
    {
      "epoch": 19.378578024007385,
      "grad_norm": 0.9240314364433289,
      "learning_rate": 2.9833333333333335e-05,
      "loss": 0.1469,
      "step": 10503
    },
    {
      "epoch": 19.380424746075715,
      "grad_norm": 0.3998948633670807,
      "learning_rate": 2.9833e-05,
      "loss": 0.0637,
      "step": 10504
    },
    {
      "epoch": 19.382271468144044,
      "grad_norm": 0.5939068794250488,
      "learning_rate": 2.9832666666666666e-05,
      "loss": 0.0628,
      "step": 10505
    },
    {
      "epoch": 19.384118190212373,
      "grad_norm": 0.38086938858032227,
      "learning_rate": 2.9832333333333336e-05,
      "loss": 0.066,
      "step": 10506
    },
    {
      "epoch": 19.385964912280702,
      "grad_norm": 0.41734108328819275,
      "learning_rate": 2.9831999999999998e-05,
      "loss": 0.0603,
      "step": 10507
    },
    {
      "epoch": 19.38781163434903,
      "grad_norm": 0.49323397874832153,
      "learning_rate": 2.9831666666666668e-05,
      "loss": 0.0567,
      "step": 10508
    },
    {
      "epoch": 19.38965835641736,
      "grad_norm": 0.3245604932308197,
      "learning_rate": 2.9831333333333334e-05,
      "loss": 0.0372,
      "step": 10509
    },
    {
      "epoch": 19.39150507848569,
      "grad_norm": 0.24436530470848083,
      "learning_rate": 2.9831e-05,
      "loss": 0.0352,
      "step": 10510
    },
    {
      "epoch": 19.393351800554015,
      "grad_norm": 0.35334378480911255,
      "learning_rate": 2.983066666666667e-05,
      "loss": 0.0754,
      "step": 10511
    },
    {
      "epoch": 19.395198522622344,
      "grad_norm": 0.38580289483070374,
      "learning_rate": 2.9830333333333335e-05,
      "loss": 0.0679,
      "step": 10512
    },
    {
      "epoch": 19.397045244690673,
      "grad_norm": 0.31692564487457275,
      "learning_rate": 2.983e-05,
      "loss": 0.0186,
      "step": 10513
    },
    {
      "epoch": 19.398891966759003,
      "grad_norm": 0.31757763028144836,
      "learning_rate": 2.9829666666666667e-05,
      "loss": 0.0256,
      "step": 10514
    },
    {
      "epoch": 19.40073868882733,
      "grad_norm": 0.16572915017604828,
      "learning_rate": 2.9829333333333336e-05,
      "loss": 0.011,
      "step": 10515
    },
    {
      "epoch": 19.40258541089566,
      "grad_norm": 0.4005686342716217,
      "learning_rate": 2.9829e-05,
      "loss": 0.017,
      "step": 10516
    },
    {
      "epoch": 19.40443213296399,
      "grad_norm": 0.33007410168647766,
      "learning_rate": 2.9828666666666668e-05,
      "loss": 0.0127,
      "step": 10517
    },
    {
      "epoch": 19.40627885503232,
      "grad_norm": 0.09764605760574341,
      "learning_rate": 2.9828333333333334e-05,
      "loss": 0.0049,
      "step": 10518
    },
    {
      "epoch": 19.408125577100645,
      "grad_norm": 0.29652926325798035,
      "learning_rate": 2.9828e-05,
      "loss": 0.0177,
      "step": 10519
    },
    {
      "epoch": 19.409972299168974,
      "grad_norm": 0.27024373412132263,
      "learning_rate": 2.982766666666667e-05,
      "loss": 0.0134,
      "step": 10520
    },
    {
      "epoch": 19.411819021237303,
      "grad_norm": 0.18943774700164795,
      "learning_rate": 2.9827333333333335e-05,
      "loss": 0.0157,
      "step": 10521
    },
    {
      "epoch": 19.413665743305632,
      "grad_norm": 0.11868040263652802,
      "learning_rate": 2.9827e-05,
      "loss": 0.0042,
      "step": 10522
    },
    {
      "epoch": 19.41551246537396,
      "grad_norm": 0.611986517906189,
      "learning_rate": 2.9826666666666667e-05,
      "loss": 0.0121,
      "step": 10523
    },
    {
      "epoch": 19.41735918744229,
      "grad_norm": 0.3093113899230957,
      "learning_rate": 2.9826333333333336e-05,
      "loss": 0.0242,
      "step": 10524
    },
    {
      "epoch": 19.41920590951062,
      "grad_norm": 0.7205303311347961,
      "learning_rate": 2.9826e-05,
      "loss": 0.0176,
      "step": 10525
    },
    {
      "epoch": 19.42105263157895,
      "grad_norm": 0.36200499534606934,
      "learning_rate": 2.9825666666666668e-05,
      "loss": 0.0125,
      "step": 10526
    },
    {
      "epoch": 19.422899353647278,
      "grad_norm": 0.2035263627767563,
      "learning_rate": 2.9825333333333334e-05,
      "loss": 0.0099,
      "step": 10527
    },
    {
      "epoch": 19.424746075715603,
      "grad_norm": 0.27171552181243896,
      "learning_rate": 2.9825e-05,
      "loss": 0.0079,
      "step": 10528
    },
    {
      "epoch": 19.426592797783933,
      "grad_norm": 0.3976079523563385,
      "learning_rate": 2.982466666666667e-05,
      "loss": 0.0458,
      "step": 10529
    },
    {
      "epoch": 19.42843951985226,
      "grad_norm": 0.19442453980445862,
      "learning_rate": 2.9824333333333335e-05,
      "loss": 0.0085,
      "step": 10530
    },
    {
      "epoch": 19.43028624192059,
      "grad_norm": 0.30339083075523376,
      "learning_rate": 2.9824e-05,
      "loss": 0.0115,
      "step": 10531
    },
    {
      "epoch": 19.43213296398892,
      "grad_norm": 1.2128410339355469,
      "learning_rate": 2.9823666666666667e-05,
      "loss": 0.0042,
      "step": 10532
    },
    {
      "epoch": 19.43397968605725,
      "grad_norm": 0.8668030500411987,
      "learning_rate": 2.9823333333333333e-05,
      "loss": 0.0131,
      "step": 10533
    },
    {
      "epoch": 19.43582640812558,
      "grad_norm": 0.1962207853794098,
      "learning_rate": 2.9823e-05,
      "loss": 0.0043,
      "step": 10534
    },
    {
      "epoch": 19.437673130193907,
      "grad_norm": 0.21301503479480743,
      "learning_rate": 2.9822666666666668e-05,
      "loss": 0.0068,
      "step": 10535
    },
    {
      "epoch": 19.439519852262233,
      "grad_norm": 0.7212603092193604,
      "learning_rate": 2.9822333333333334e-05,
      "loss": 0.0152,
      "step": 10536
    },
    {
      "epoch": 19.441366574330562,
      "grad_norm": 0.4302648901939392,
      "learning_rate": 2.9822e-05,
      "loss": 0.0095,
      "step": 10537
    },
    {
      "epoch": 19.44321329639889,
      "grad_norm": 0.2859456539154053,
      "learning_rate": 2.982166666666667e-05,
      "loss": 0.0051,
      "step": 10538
    },
    {
      "epoch": 19.44506001846722,
      "grad_norm": 0.2743951976299286,
      "learning_rate": 2.9821333333333332e-05,
      "loss": 0.0076,
      "step": 10539
    },
    {
      "epoch": 19.44690674053555,
      "grad_norm": 0.7187720537185669,
      "learning_rate": 2.9821e-05,
      "loss": 0.0136,
      "step": 10540
    },
    {
      "epoch": 19.44875346260388,
      "grad_norm": 0.4372936487197876,
      "learning_rate": 2.9820666666666667e-05,
      "loss": 0.0122,
      "step": 10541
    },
    {
      "epoch": 19.450600184672208,
      "grad_norm": 0.5092134475708008,
      "learning_rate": 2.9820333333333333e-05,
      "loss": 0.0155,
      "step": 10542
    },
    {
      "epoch": 19.452446906740537,
      "grad_norm": 0.41750380396842957,
      "learning_rate": 2.982e-05,
      "loss": 0.0096,
      "step": 10543
    },
    {
      "epoch": 19.454293628808863,
      "grad_norm": 0.5092931985855103,
      "learning_rate": 2.981966666666667e-05,
      "loss": 0.0092,
      "step": 10544
    },
    {
      "epoch": 19.45614035087719,
      "grad_norm": 0.25309839844703674,
      "learning_rate": 2.9819333333333334e-05,
      "loss": 0.0116,
      "step": 10545
    },
    {
      "epoch": 19.45798707294552,
      "grad_norm": 0.6120051145553589,
      "learning_rate": 2.9819e-05,
      "loss": 0.0154,
      "step": 10546
    },
    {
      "epoch": 19.45983379501385,
      "grad_norm": 0.2491397112607956,
      "learning_rate": 2.981866666666667e-05,
      "loss": 0.0083,
      "step": 10547
    },
    {
      "epoch": 19.46168051708218,
      "grad_norm": 0.3732081353664398,
      "learning_rate": 2.9818333333333332e-05,
      "loss": 0.0104,
      "step": 10548
    },
    {
      "epoch": 19.46352723915051,
      "grad_norm": 0.5665434002876282,
      "learning_rate": 2.9818e-05,
      "loss": 0.1143,
      "step": 10549
    },
    {
      "epoch": 19.465373961218837,
      "grad_norm": 0.5090430974960327,
      "learning_rate": 2.9817666666666667e-05,
      "loss": 0.1119,
      "step": 10550
    },
    {
      "epoch": 19.467220683287167,
      "grad_norm": 0.7848197817802429,
      "learning_rate": 2.9817333333333333e-05,
      "loss": 0.164,
      "step": 10551
    },
    {
      "epoch": 19.469067405355496,
      "grad_norm": 0.6643295288085938,
      "learning_rate": 2.9817e-05,
      "loss": 0.1209,
      "step": 10552
    },
    {
      "epoch": 19.47091412742382,
      "grad_norm": 0.7260629534721375,
      "learning_rate": 2.981666666666667e-05,
      "loss": 0.1687,
      "step": 10553
    },
    {
      "epoch": 19.47276084949215,
      "grad_norm": 0.575410783290863,
      "learning_rate": 2.9816333333333335e-05,
      "loss": 0.0625,
      "step": 10554
    },
    {
      "epoch": 19.47460757156048,
      "grad_norm": 0.5349397659301758,
      "learning_rate": 2.9816e-05,
      "loss": 0.0714,
      "step": 10555
    },
    {
      "epoch": 19.47645429362881,
      "grad_norm": 0.39579540491104126,
      "learning_rate": 2.981566666666667e-05,
      "loss": 0.0591,
      "step": 10556
    },
    {
      "epoch": 19.478301015697138,
      "grad_norm": 0.47131359577178955,
      "learning_rate": 2.9815333333333332e-05,
      "loss": 0.0645,
      "step": 10557
    },
    {
      "epoch": 19.480147737765467,
      "grad_norm": 0.42419734597206116,
      "learning_rate": 2.9815e-05,
      "loss": 0.0878,
      "step": 10558
    },
    {
      "epoch": 19.481994459833796,
      "grad_norm": 0.5721330642700195,
      "learning_rate": 2.9814666666666668e-05,
      "loss": 0.0719,
      "step": 10559
    },
    {
      "epoch": 19.483841181902125,
      "grad_norm": 0.3947785794734955,
      "learning_rate": 2.9814333333333334e-05,
      "loss": 0.0369,
      "step": 10560
    },
    {
      "epoch": 19.48568790397045,
      "grad_norm": 0.4918561577796936,
      "learning_rate": 2.9814000000000003e-05,
      "loss": 0.0688,
      "step": 10561
    },
    {
      "epoch": 19.48753462603878,
      "grad_norm": 0.9153860807418823,
      "learning_rate": 2.981366666666667e-05,
      "loss": 0.0445,
      "step": 10562
    },
    {
      "epoch": 19.48938134810711,
      "grad_norm": 0.5050480961799622,
      "learning_rate": 2.9813333333333335e-05,
      "loss": 0.0213,
      "step": 10563
    },
    {
      "epoch": 19.49122807017544,
      "grad_norm": 0.28121042251586914,
      "learning_rate": 2.9813e-05,
      "loss": 0.0113,
      "step": 10564
    },
    {
      "epoch": 19.493074792243767,
      "grad_norm": 0.5272451043128967,
      "learning_rate": 2.9812666666666667e-05,
      "loss": 0.0162,
      "step": 10565
    },
    {
      "epoch": 19.494921514312097,
      "grad_norm": 0.29827800393104553,
      "learning_rate": 2.9812333333333333e-05,
      "loss": 0.0086,
      "step": 10566
    },
    {
      "epoch": 19.496768236380426,
      "grad_norm": 0.19304965436458588,
      "learning_rate": 2.9812000000000002e-05,
      "loss": 0.0139,
      "step": 10567
    },
    {
      "epoch": 19.498614958448755,
      "grad_norm": 0.440072625875473,
      "learning_rate": 2.9811666666666664e-05,
      "loss": 0.0164,
      "step": 10568
    },
    {
      "epoch": 19.50046168051708,
      "grad_norm": 0.16277796030044556,
      "learning_rate": 2.9811333333333334e-05,
      "loss": 0.0073,
      "step": 10569
    },
    {
      "epoch": 19.50230840258541,
      "grad_norm": 0.22500938177108765,
      "learning_rate": 2.9811000000000003e-05,
      "loss": 0.0063,
      "step": 10570
    },
    {
      "epoch": 19.50415512465374,
      "grad_norm": 0.20329682528972626,
      "learning_rate": 2.9810666666666666e-05,
      "loss": 0.0084,
      "step": 10571
    },
    {
      "epoch": 19.506001846722068,
      "grad_norm": 0.5552082657814026,
      "learning_rate": 2.9810333333333335e-05,
      "loss": 0.0118,
      "step": 10572
    },
    {
      "epoch": 19.507848568790397,
      "grad_norm": 0.4801299273967743,
      "learning_rate": 2.981e-05,
      "loss": 0.0174,
      "step": 10573
    },
    {
      "epoch": 19.509695290858726,
      "grad_norm": 0.6265919208526611,
      "learning_rate": 2.9809666666666667e-05,
      "loss": 0.0325,
      "step": 10574
    },
    {
      "epoch": 19.511542012927055,
      "grad_norm": 0.24671681225299835,
      "learning_rate": 2.9809333333333333e-05,
      "loss": 0.0067,
      "step": 10575
    },
    {
      "epoch": 19.513388734995385,
      "grad_norm": 0.28970077633857727,
      "learning_rate": 2.9809000000000002e-05,
      "loss": 0.0059,
      "step": 10576
    },
    {
      "epoch": 19.51523545706371,
      "grad_norm": 0.24169009923934937,
      "learning_rate": 2.9808666666666665e-05,
      "loss": 0.0088,
      "step": 10577
    },
    {
      "epoch": 19.51708217913204,
      "grad_norm": 0.2239508479833603,
      "learning_rate": 2.9808333333333334e-05,
      "loss": 0.0058,
      "step": 10578
    },
    {
      "epoch": 19.51892890120037,
      "grad_norm": 0.3484097719192505,
      "learning_rate": 2.9808000000000003e-05,
      "loss": 0.0085,
      "step": 10579
    },
    {
      "epoch": 19.520775623268698,
      "grad_norm": 0.35717564821243286,
      "learning_rate": 2.9807666666666666e-05,
      "loss": 0.0077,
      "step": 10580
    },
    {
      "epoch": 19.522622345337027,
      "grad_norm": 0.30840328335762024,
      "learning_rate": 2.9807333333333335e-05,
      "loss": 0.0074,
      "step": 10581
    },
    {
      "epoch": 19.524469067405356,
      "grad_norm": 0.3130101263523102,
      "learning_rate": 2.9807e-05,
      "loss": 0.0137,
      "step": 10582
    },
    {
      "epoch": 19.526315789473685,
      "grad_norm": 0.3575838506221771,
      "learning_rate": 2.9806666666666667e-05,
      "loss": 0.0114,
      "step": 10583
    },
    {
      "epoch": 19.528162511542014,
      "grad_norm": 0.2482454925775528,
      "learning_rate": 2.9806333333333333e-05,
      "loss": 0.0052,
      "step": 10584
    },
    {
      "epoch": 19.530009233610343,
      "grad_norm": 0.4640428125858307,
      "learning_rate": 2.9806000000000002e-05,
      "loss": 0.0395,
      "step": 10585
    },
    {
      "epoch": 19.53185595567867,
      "grad_norm": 0.5307456851005554,
      "learning_rate": 2.9805666666666668e-05,
      "loss": 0.0148,
      "step": 10586
    },
    {
      "epoch": 19.533702677746998,
      "grad_norm": 0.583564281463623,
      "learning_rate": 2.9805333333333334e-05,
      "loss": 0.0129,
      "step": 10587
    },
    {
      "epoch": 19.535549399815327,
      "grad_norm": 0.4926765561103821,
      "learning_rate": 2.9805000000000003e-05,
      "loss": 0.0093,
      "step": 10588
    },
    {
      "epoch": 19.537396121883656,
      "grad_norm": 0.5021626949310303,
      "learning_rate": 2.9804666666666666e-05,
      "loss": 0.0101,
      "step": 10589
    },
    {
      "epoch": 19.539242843951985,
      "grad_norm": 0.36017680168151855,
      "learning_rate": 2.9804333333333335e-05,
      "loss": 0.0085,
      "step": 10590
    },
    {
      "epoch": 19.541089566020315,
      "grad_norm": 0.6824921369552612,
      "learning_rate": 2.9804e-05,
      "loss": 0.0115,
      "step": 10591
    },
    {
      "epoch": 19.542936288088644,
      "grad_norm": 0.5039044618606567,
      "learning_rate": 2.9803666666666667e-05,
      "loss": 0.014,
      "step": 10592
    },
    {
      "epoch": 19.544783010156973,
      "grad_norm": 0.6053094267845154,
      "learning_rate": 2.9803333333333333e-05,
      "loss": 0.0114,
      "step": 10593
    },
    {
      "epoch": 19.5466297322253,
      "grad_norm": 0.48566383123397827,
      "learning_rate": 2.9803e-05,
      "loss": 0.0129,
      "step": 10594
    },
    {
      "epoch": 19.548476454293628,
      "grad_norm": 0.609473705291748,
      "learning_rate": 2.980266666666667e-05,
      "loss": 0.0178,
      "step": 10595
    },
    {
      "epoch": 19.550323176361957,
      "grad_norm": 0.6189708113670349,
      "learning_rate": 2.9802333333333334e-05,
      "loss": 0.0345,
      "step": 10596
    },
    {
      "epoch": 19.552169898430286,
      "grad_norm": 0.3618892729282379,
      "learning_rate": 2.9802e-05,
      "loss": 0.0121,
      "step": 10597
    },
    {
      "epoch": 19.554016620498615,
      "grad_norm": 0.42415696382522583,
      "learning_rate": 2.9801666666666666e-05,
      "loss": 0.0184,
      "step": 10598
    },
    {
      "epoch": 19.555863342566944,
      "grad_norm": 0.5935088992118835,
      "learning_rate": 2.9801333333333336e-05,
      "loss": 0.1555,
      "step": 10599
    },
    {
      "epoch": 19.557710064635273,
      "grad_norm": 0.6377758979797363,
      "learning_rate": 2.9800999999999998e-05,
      "loss": 0.1726,
      "step": 10600
    },
    {
      "epoch": 19.559556786703602,
      "grad_norm": 0.6348208785057068,
      "learning_rate": 2.9800666666666667e-05,
      "loss": 0.1333,
      "step": 10601
    },
    {
      "epoch": 19.56140350877193,
      "grad_norm": 0.4387953579425812,
      "learning_rate": 2.9800333333333333e-05,
      "loss": 0.0984,
      "step": 10602
    },
    {
      "epoch": 19.563250230840257,
      "grad_norm": 0.3931863307952881,
      "learning_rate": 2.98e-05,
      "loss": 0.078,
      "step": 10603
    },
    {
      "epoch": 19.565096952908586,
      "grad_norm": 0.4532827138900757,
      "learning_rate": 2.979966666666667e-05,
      "loss": 0.0909,
      "step": 10604
    },
    {
      "epoch": 19.566943674976915,
      "grad_norm": 0.4221654236316681,
      "learning_rate": 2.9799333333333335e-05,
      "loss": 0.0949,
      "step": 10605
    },
    {
      "epoch": 19.568790397045245,
      "grad_norm": 0.4770489037036896,
      "learning_rate": 2.9799e-05,
      "loss": 0.0448,
      "step": 10606
    },
    {
      "epoch": 19.570637119113574,
      "grad_norm": 0.4773814380168915,
      "learning_rate": 2.9798666666666666e-05,
      "loss": 0.0767,
      "step": 10607
    },
    {
      "epoch": 19.572483841181903,
      "grad_norm": 0.504423975944519,
      "learning_rate": 2.9798333333333336e-05,
      "loss": 0.0507,
      "step": 10608
    },
    {
      "epoch": 19.574330563250232,
      "grad_norm": 0.6374596953392029,
      "learning_rate": 2.9797999999999998e-05,
      "loss": 0.0608,
      "step": 10609
    },
    {
      "epoch": 19.57617728531856,
      "grad_norm": 0.4656340479850769,
      "learning_rate": 2.9797666666666668e-05,
      "loss": 0.0768,
      "step": 10610
    },
    {
      "epoch": 19.578024007386887,
      "grad_norm": 0.8195959329605103,
      "learning_rate": 2.9797333333333337e-05,
      "loss": 0.0287,
      "step": 10611
    },
    {
      "epoch": 19.579870729455216,
      "grad_norm": 0.2056460976600647,
      "learning_rate": 2.9797e-05,
      "loss": 0.0167,
      "step": 10612
    },
    {
      "epoch": 19.581717451523545,
      "grad_norm": 0.33398884534835815,
      "learning_rate": 2.979666666666667e-05,
      "loss": 0.0303,
      "step": 10613
    },
    {
      "epoch": 19.583564173591874,
      "grad_norm": 0.37782248854637146,
      "learning_rate": 2.9796333333333335e-05,
      "loss": 0.0335,
      "step": 10614
    },
    {
      "epoch": 19.585410895660203,
      "grad_norm": 0.2887199819087982,
      "learning_rate": 2.9796e-05,
      "loss": 0.0159,
      "step": 10615
    },
    {
      "epoch": 19.587257617728532,
      "grad_norm": 0.3071005046367645,
      "learning_rate": 2.9795666666666667e-05,
      "loss": 0.0184,
      "step": 10616
    },
    {
      "epoch": 19.58910433979686,
      "grad_norm": 0.3234884738922119,
      "learning_rate": 2.9795333333333336e-05,
      "loss": 0.0149,
      "step": 10617
    },
    {
      "epoch": 19.59095106186519,
      "grad_norm": 0.12488967925310135,
      "learning_rate": 2.9795e-05,
      "loss": 0.0044,
      "step": 10618
    },
    {
      "epoch": 19.592797783933516,
      "grad_norm": 0.2693093717098236,
      "learning_rate": 2.9794666666666668e-05,
      "loss": 0.0404,
      "step": 10619
    },
    {
      "epoch": 19.594644506001845,
      "grad_norm": 0.3381086587905884,
      "learning_rate": 2.9794333333333337e-05,
      "loss": 0.0104,
      "step": 10620
    },
    {
      "epoch": 19.596491228070175,
      "grad_norm": 0.5766419172286987,
      "learning_rate": 2.9794e-05,
      "loss": 0.0143,
      "step": 10621
    },
    {
      "epoch": 19.598337950138504,
      "grad_norm": 0.15342627465724945,
      "learning_rate": 2.979366666666667e-05,
      "loss": 0.0057,
      "step": 10622
    },
    {
      "epoch": 19.600184672206833,
      "grad_norm": 0.5277054309844971,
      "learning_rate": 2.979333333333333e-05,
      "loss": 0.0134,
      "step": 10623
    },
    {
      "epoch": 19.602031394275162,
      "grad_norm": 0.19860132038593292,
      "learning_rate": 2.9793e-05,
      "loss": 0.0079,
      "step": 10624
    },
    {
      "epoch": 19.60387811634349,
      "grad_norm": 0.4835110008716583,
      "learning_rate": 2.9792666666666667e-05,
      "loss": 0.0112,
      "step": 10625
    },
    {
      "epoch": 19.60572483841182,
      "grad_norm": 0.26355403661727905,
      "learning_rate": 2.9792333333333333e-05,
      "loss": 0.0431,
      "step": 10626
    },
    {
      "epoch": 19.607571560480146,
      "grad_norm": 0.4434733986854553,
      "learning_rate": 2.9792e-05,
      "loss": 0.0167,
      "step": 10627
    },
    {
      "epoch": 19.609418282548475,
      "grad_norm": 0.5944486260414124,
      "learning_rate": 2.9791666666666668e-05,
      "loss": 0.02,
      "step": 10628
    },
    {
      "epoch": 19.611265004616804,
      "grad_norm": 0.49193495512008667,
      "learning_rate": 2.9791333333333334e-05,
      "loss": 0.0188,
      "step": 10629
    },
    {
      "epoch": 19.613111726685133,
      "grad_norm": 0.5033032298088074,
      "learning_rate": 2.9791e-05,
      "loss": 0.0112,
      "step": 10630
    },
    {
      "epoch": 19.614958448753463,
      "grad_norm": 0.4277038872241974,
      "learning_rate": 2.979066666666667e-05,
      "loss": 0.0142,
      "step": 10631
    },
    {
      "epoch": 19.61680517082179,
      "grad_norm": 0.20894268155097961,
      "learning_rate": 2.9790333333333332e-05,
      "loss": 0.0051,
      "step": 10632
    },
    {
      "epoch": 19.61865189289012,
      "grad_norm": 0.22185315191745758,
      "learning_rate": 2.979e-05,
      "loss": 0.0087,
      "step": 10633
    },
    {
      "epoch": 19.62049861495845,
      "grad_norm": 0.26102548837661743,
      "learning_rate": 2.9789666666666667e-05,
      "loss": 0.0075,
      "step": 10634
    },
    {
      "epoch": 19.62234533702678,
      "grad_norm": 0.5750803351402283,
      "learning_rate": 2.9789333333333333e-05,
      "loss": 0.0062,
      "step": 10635
    },
    {
      "epoch": 19.624192059095105,
      "grad_norm": 0.19431185722351074,
      "learning_rate": 2.9789000000000002e-05,
      "loss": 0.0078,
      "step": 10636
    },
    {
      "epoch": 19.626038781163434,
      "grad_norm": 1.0597251653671265,
      "learning_rate": 2.9788666666666668e-05,
      "loss": 0.0333,
      "step": 10637
    },
    {
      "epoch": 19.627885503231763,
      "grad_norm": 0.5158454775810242,
      "learning_rate": 2.9788333333333334e-05,
      "loss": 0.0161,
      "step": 10638
    },
    {
      "epoch": 19.629732225300092,
      "grad_norm": 0.33750757575035095,
      "learning_rate": 2.9788e-05,
      "loss": 0.0082,
      "step": 10639
    },
    {
      "epoch": 19.63157894736842,
      "grad_norm": 0.5329627990722656,
      "learning_rate": 2.978766666666667e-05,
      "loss": 0.0166,
      "step": 10640
    },
    {
      "epoch": 19.63342566943675,
      "grad_norm": 0.7116667628288269,
      "learning_rate": 2.9787333333333332e-05,
      "loss": 0.0176,
      "step": 10641
    },
    {
      "epoch": 19.63527239150508,
      "grad_norm": 0.2800839841365814,
      "learning_rate": 2.9787e-05,
      "loss": 0.0074,
      "step": 10642
    },
    {
      "epoch": 19.63711911357341,
      "grad_norm": 0.556308925151825,
      "learning_rate": 2.9786666666666667e-05,
      "loss": 0.0205,
      "step": 10643
    },
    {
      "epoch": 19.638965835641734,
      "grad_norm": 0.8026602864265442,
      "learning_rate": 2.9786333333333333e-05,
      "loss": 0.0147,
      "step": 10644
    },
    {
      "epoch": 19.640812557710063,
      "grad_norm": 0.4995995759963989,
      "learning_rate": 2.9786000000000002e-05,
      "loss": 0.0106,
      "step": 10645
    },
    {
      "epoch": 19.642659279778393,
      "grad_norm": 0.556323230266571,
      "learning_rate": 2.978566666666667e-05,
      "loss": 0.0162,
      "step": 10646
    },
    {
      "epoch": 19.64450600184672,
      "grad_norm": 0.5479616522789001,
      "learning_rate": 2.9785333333333334e-05,
      "loss": 0.0109,
      "step": 10647
    },
    {
      "epoch": 19.64635272391505,
      "grad_norm": 0.9250813126564026,
      "learning_rate": 2.9785e-05,
      "loss": 0.0098,
      "step": 10648
    },
    {
      "epoch": 19.64819944598338,
      "grad_norm": 1.283340334892273,
      "learning_rate": 2.978466666666667e-05,
      "loss": 0.2004,
      "step": 10649
    },
    {
      "epoch": 19.65004616805171,
      "grad_norm": 0.6739901900291443,
      "learning_rate": 2.9784333333333332e-05,
      "loss": 0.217,
      "step": 10650
    },
    {
      "epoch": 19.65189289012004,
      "grad_norm": 0.5376523733139038,
      "learning_rate": 2.9784e-05,
      "loss": 0.1042,
      "step": 10651
    },
    {
      "epoch": 19.653739612188367,
      "grad_norm": 0.7716366648674011,
      "learning_rate": 2.9783666666666667e-05,
      "loss": 0.1006,
      "step": 10652
    },
    {
      "epoch": 19.655586334256693,
      "grad_norm": 0.4473935663700104,
      "learning_rate": 2.9783333333333333e-05,
      "loss": 0.0795,
      "step": 10653
    },
    {
      "epoch": 19.657433056325022,
      "grad_norm": 0.47197431325912476,
      "learning_rate": 2.9783000000000003e-05,
      "loss": 0.0708,
      "step": 10654
    },
    {
      "epoch": 19.65927977839335,
      "grad_norm": 0.530225396156311,
      "learning_rate": 2.9782666666666665e-05,
      "loss": 0.0684,
      "step": 10655
    },
    {
      "epoch": 19.66112650046168,
      "grad_norm": 0.3782411813735962,
      "learning_rate": 2.9782333333333335e-05,
      "loss": 0.0747,
      "step": 10656
    },
    {
      "epoch": 19.66297322253001,
      "grad_norm": 0.5400944352149963,
      "learning_rate": 2.9782e-05,
      "loss": 0.0515,
      "step": 10657
    },
    {
      "epoch": 19.66481994459834,
      "grad_norm": 0.6011316776275635,
      "learning_rate": 2.9781666666666666e-05,
      "loss": 0.0601,
      "step": 10658
    },
    {
      "epoch": 19.666666666666668,
      "grad_norm": 0.38033759593963623,
      "learning_rate": 2.9781333333333332e-05,
      "loss": 0.037,
      "step": 10659
    },
    {
      "epoch": 19.668513388734997,
      "grad_norm": 0.3871672451496124,
      "learning_rate": 2.9781e-05,
      "loss": 0.0573,
      "step": 10660
    },
    {
      "epoch": 19.670360110803323,
      "grad_norm": 0.35777178406715393,
      "learning_rate": 2.9780666666666668e-05,
      "loss": 0.048,
      "step": 10661
    },
    {
      "epoch": 19.67220683287165,
      "grad_norm": 0.4704509675502777,
      "learning_rate": 2.9780333333333334e-05,
      "loss": 0.0589,
      "step": 10662
    },
    {
      "epoch": 19.67405355493998,
      "grad_norm": 2.055455207824707,
      "learning_rate": 2.9780000000000003e-05,
      "loss": 0.078,
      "step": 10663
    },
    {
      "epoch": 19.67590027700831,
      "grad_norm": 0.25962790846824646,
      "learning_rate": 2.9779666666666665e-05,
      "loss": 0.0214,
      "step": 10664
    },
    {
      "epoch": 19.67774699907664,
      "grad_norm": 0.2785275876522064,
      "learning_rate": 2.9779333333333335e-05,
      "loss": 0.0127,
      "step": 10665
    },
    {
      "epoch": 19.67959372114497,
      "grad_norm": 0.37023863196372986,
      "learning_rate": 2.9779e-05,
      "loss": 0.0133,
      "step": 10666
    },
    {
      "epoch": 19.681440443213297,
      "grad_norm": 0.40001484751701355,
      "learning_rate": 2.9778666666666667e-05,
      "loss": 0.0324,
      "step": 10667
    },
    {
      "epoch": 19.683287165281627,
      "grad_norm": 0.26877978444099426,
      "learning_rate": 2.9778333333333333e-05,
      "loss": 0.0163,
      "step": 10668
    },
    {
      "epoch": 19.685133887349952,
      "grad_norm": 0.33660438656806946,
      "learning_rate": 2.9778000000000002e-05,
      "loss": 0.0141,
      "step": 10669
    },
    {
      "epoch": 19.68698060941828,
      "grad_norm": 0.3006255328655243,
      "learning_rate": 2.9777666666666668e-05,
      "loss": 0.0151,
      "step": 10670
    },
    {
      "epoch": 19.68882733148661,
      "grad_norm": 0.6988603472709656,
      "learning_rate": 2.9777333333333334e-05,
      "loss": 0.0268,
      "step": 10671
    },
    {
      "epoch": 19.69067405355494,
      "grad_norm": 0.4372357428073883,
      "learning_rate": 2.9777000000000003e-05,
      "loss": 0.009,
      "step": 10672
    },
    {
      "epoch": 19.69252077562327,
      "grad_norm": 0.9028867483139038,
      "learning_rate": 2.9776666666666666e-05,
      "loss": 0.0234,
      "step": 10673
    },
    {
      "epoch": 19.694367497691598,
      "grad_norm": 0.6994834542274475,
      "learning_rate": 2.9776333333333335e-05,
      "loss": 0.0181,
      "step": 10674
    },
    {
      "epoch": 19.696214219759927,
      "grad_norm": 1.2151726484298706,
      "learning_rate": 2.9776e-05,
      "loss": 0.0125,
      "step": 10675
    },
    {
      "epoch": 19.698060941828256,
      "grad_norm": 0.18534334003925323,
      "learning_rate": 2.9775666666666667e-05,
      "loss": 0.0095,
      "step": 10676
    },
    {
      "epoch": 19.69990766389658,
      "grad_norm": 0.21873116493225098,
      "learning_rate": 2.9775333333333333e-05,
      "loss": 0.0119,
      "step": 10677
    },
    {
      "epoch": 19.70175438596491,
      "grad_norm": 0.40969711542129517,
      "learning_rate": 2.9775000000000002e-05,
      "loss": 0.0103,
      "step": 10678
    },
    {
      "epoch": 19.70360110803324,
      "grad_norm": 0.5559830069541931,
      "learning_rate": 2.9774666666666668e-05,
      "loss": 0.0238,
      "step": 10679
    },
    {
      "epoch": 19.70544783010157,
      "grad_norm": 0.5412967205047607,
      "learning_rate": 2.9774333333333334e-05,
      "loss": 0.0146,
      "step": 10680
    },
    {
      "epoch": 19.7072945521699,
      "grad_norm": 0.12473435699939728,
      "learning_rate": 2.9774000000000003e-05,
      "loss": 0.0044,
      "step": 10681
    },
    {
      "epoch": 19.709141274238227,
      "grad_norm": 0.586225688457489,
      "learning_rate": 2.9773666666666666e-05,
      "loss": 0.0134,
      "step": 10682
    },
    {
      "epoch": 19.710987996306557,
      "grad_norm": 0.44378283619880676,
      "learning_rate": 2.9773333333333335e-05,
      "loss": 0.0173,
      "step": 10683
    },
    {
      "epoch": 19.712834718374886,
      "grad_norm": 0.5063177943229675,
      "learning_rate": 2.9772999999999998e-05,
      "loss": 0.0213,
      "step": 10684
    },
    {
      "epoch": 19.714681440443215,
      "grad_norm": 0.5654175281524658,
      "learning_rate": 2.9772666666666667e-05,
      "loss": 0.0101,
      "step": 10685
    },
    {
      "epoch": 19.71652816251154,
      "grad_norm": 0.222188800573349,
      "learning_rate": 2.9772333333333336e-05,
      "loss": 0.007,
      "step": 10686
    },
    {
      "epoch": 19.71837488457987,
      "grad_norm": 0.448893666267395,
      "learning_rate": 2.9772e-05,
      "loss": 0.0122,
      "step": 10687
    },
    {
      "epoch": 19.7202216066482,
      "grad_norm": 1.0693613290786743,
      "learning_rate": 2.9771666666666668e-05,
      "loss": 0.0169,
      "step": 10688
    },
    {
      "epoch": 19.722068328716528,
      "grad_norm": 0.35767197608947754,
      "learning_rate": 2.9771333333333334e-05,
      "loss": 0.0113,
      "step": 10689
    },
    {
      "epoch": 19.723915050784857,
      "grad_norm": 0.8080740571022034,
      "learning_rate": 2.9771e-05,
      "loss": 0.023,
      "step": 10690
    },
    {
      "epoch": 19.725761772853186,
      "grad_norm": 0.4295150339603424,
      "learning_rate": 2.9770666666666666e-05,
      "loss": 0.0161,
      "step": 10691
    },
    {
      "epoch": 19.727608494921515,
      "grad_norm": 0.3795710802078247,
      "learning_rate": 2.9770333333333335e-05,
      "loss": 0.0109,
      "step": 10692
    },
    {
      "epoch": 19.729455216989845,
      "grad_norm": 0.45034053921699524,
      "learning_rate": 2.9769999999999998e-05,
      "loss": 0.0193,
      "step": 10693
    },
    {
      "epoch": 19.73130193905817,
      "grad_norm": 0.4286927878856659,
      "learning_rate": 2.9769666666666667e-05,
      "loss": 0.0115,
      "step": 10694
    },
    {
      "epoch": 19.7331486611265,
      "grad_norm": 0.556318461894989,
      "learning_rate": 2.9769333333333337e-05,
      "loss": 0.0162,
      "step": 10695
    },
    {
      "epoch": 19.73499538319483,
      "grad_norm": 0.20577329397201538,
      "learning_rate": 2.9769e-05,
      "loss": 0.0077,
      "step": 10696
    },
    {
      "epoch": 19.736842105263158,
      "grad_norm": 0.16284756362438202,
      "learning_rate": 2.976866666666667e-05,
      "loss": 0.0067,
      "step": 10697
    },
    {
      "epoch": 19.738688827331487,
      "grad_norm": 1.3887354135513306,
      "learning_rate": 2.9768333333333334e-05,
      "loss": 0.0343,
      "step": 10698
    },
    {
      "epoch": 19.740535549399816,
      "grad_norm": 0.5854371786117554,
      "learning_rate": 2.9768e-05,
      "loss": 0.1647,
      "step": 10699
    },
    {
      "epoch": 19.742382271468145,
      "grad_norm": 0.5120023488998413,
      "learning_rate": 2.9767666666666666e-05,
      "loss": 0.1378,
      "step": 10700
    },
    {
      "epoch": 19.744228993536474,
      "grad_norm": 0.46874797344207764,
      "learning_rate": 2.9767333333333336e-05,
      "loss": 0.1202,
      "step": 10701
    },
    {
      "epoch": 19.746075715604803,
      "grad_norm": 0.5347771644592285,
      "learning_rate": 2.9766999999999998e-05,
      "loss": 0.1216,
      "step": 10702
    },
    {
      "epoch": 19.74792243767313,
      "grad_norm": 0.5150458216667175,
      "learning_rate": 2.9766666666666667e-05,
      "loss": 0.0987,
      "step": 10703
    },
    {
      "epoch": 19.749769159741458,
      "grad_norm": 0.5133240818977356,
      "learning_rate": 2.9766333333333337e-05,
      "loss": 0.1145,
      "step": 10704
    },
    {
      "epoch": 19.751615881809787,
      "grad_norm": 0.4530353844165802,
      "learning_rate": 2.9766e-05,
      "loss": 0.068,
      "step": 10705
    },
    {
      "epoch": 19.753462603878116,
      "grad_norm": 0.3664388656616211,
      "learning_rate": 2.976566666666667e-05,
      "loss": 0.0582,
      "step": 10706
    },
    {
      "epoch": 19.755309325946445,
      "grad_norm": 0.6541823148727417,
      "learning_rate": 2.9765333333333335e-05,
      "loss": 0.0758,
      "step": 10707
    },
    {
      "epoch": 19.757156048014775,
      "grad_norm": 0.3894199728965759,
      "learning_rate": 2.9765e-05,
      "loss": 0.0526,
      "step": 10708
    },
    {
      "epoch": 19.759002770083104,
      "grad_norm": 0.6326406002044678,
      "learning_rate": 2.9764666666666666e-05,
      "loss": 0.0544,
      "step": 10709
    },
    {
      "epoch": 19.760849492151433,
      "grad_norm": 0.5390200614929199,
      "learning_rate": 2.9764333333333336e-05,
      "loss": 0.04,
      "step": 10710
    },
    {
      "epoch": 19.76269621421976,
      "grad_norm": 0.2868020534515381,
      "learning_rate": 2.9764e-05,
      "loss": 0.0276,
      "step": 10711
    },
    {
      "epoch": 19.764542936288088,
      "grad_norm": 0.364552766084671,
      "learning_rate": 2.9763666666666668e-05,
      "loss": 0.0221,
      "step": 10712
    },
    {
      "epoch": 19.766389658356417,
      "grad_norm": 0.4122895300388336,
      "learning_rate": 2.9763333333333337e-05,
      "loss": 0.0272,
      "step": 10713
    },
    {
      "epoch": 19.768236380424746,
      "grad_norm": 0.3642270863056183,
      "learning_rate": 2.9763e-05,
      "loss": 0.0507,
      "step": 10714
    },
    {
      "epoch": 19.770083102493075,
      "grad_norm": 0.30101460218429565,
      "learning_rate": 2.976266666666667e-05,
      "loss": 0.0406,
      "step": 10715
    },
    {
      "epoch": 19.771929824561404,
      "grad_norm": 0.27228695154190063,
      "learning_rate": 2.976233333333333e-05,
      "loss": 0.0136,
      "step": 10716
    },
    {
      "epoch": 19.773776546629733,
      "grad_norm": 0.3003411293029785,
      "learning_rate": 2.9762e-05,
      "loss": 0.0117,
      "step": 10717
    },
    {
      "epoch": 19.775623268698062,
      "grad_norm": 0.4910516142845154,
      "learning_rate": 2.9761666666666667e-05,
      "loss": 0.0123,
      "step": 10718
    },
    {
      "epoch": 19.777469990766388,
      "grad_norm": 0.27473193407058716,
      "learning_rate": 2.9761333333333333e-05,
      "loss": 0.0144,
      "step": 10719
    },
    {
      "epoch": 19.779316712834717,
      "grad_norm": 0.20299680531024933,
      "learning_rate": 2.9761000000000002e-05,
      "loss": 0.012,
      "step": 10720
    },
    {
      "epoch": 19.781163434903046,
      "grad_norm": 0.16587220132350922,
      "learning_rate": 2.9760666666666668e-05,
      "loss": 0.0071,
      "step": 10721
    },
    {
      "epoch": 19.783010156971375,
      "grad_norm": 0.6967406868934631,
      "learning_rate": 2.9760333333333334e-05,
      "loss": 0.0166,
      "step": 10722
    },
    {
      "epoch": 19.784856879039705,
      "grad_norm": 0.2570898234844208,
      "learning_rate": 2.976e-05,
      "loss": 0.0105,
      "step": 10723
    },
    {
      "epoch": 19.786703601108034,
      "grad_norm": 0.3881112337112427,
      "learning_rate": 2.975966666666667e-05,
      "loss": 0.0115,
      "step": 10724
    },
    {
      "epoch": 19.788550323176363,
      "grad_norm": 0.4173627495765686,
      "learning_rate": 2.975933333333333e-05,
      "loss": 0.0127,
      "step": 10725
    },
    {
      "epoch": 19.790397045244692,
      "grad_norm": 0.4523608684539795,
      "learning_rate": 2.9759e-05,
      "loss": 0.0142,
      "step": 10726
    },
    {
      "epoch": 19.792243767313018,
      "grad_norm": 0.1756921112537384,
      "learning_rate": 2.9758666666666667e-05,
      "loss": 0.0044,
      "step": 10727
    },
    {
      "epoch": 19.794090489381347,
      "grad_norm": 0.2844432592391968,
      "learning_rate": 2.9758333333333333e-05,
      "loss": 0.0079,
      "step": 10728
    },
    {
      "epoch": 19.795937211449676,
      "grad_norm": 0.2023969292640686,
      "learning_rate": 2.9758000000000002e-05,
      "loss": 0.0102,
      "step": 10729
    },
    {
      "epoch": 19.797783933518005,
      "grad_norm": 0.2869013249874115,
      "learning_rate": 2.9757666666666668e-05,
      "loss": 0.0088,
      "step": 10730
    },
    {
      "epoch": 19.799630655586334,
      "grad_norm": 0.2981497347354889,
      "learning_rate": 2.9757333333333334e-05,
      "loss": 0.0106,
      "step": 10731
    },
    {
      "epoch": 19.801477377654663,
      "grad_norm": 0.6983348727226257,
      "learning_rate": 2.9757e-05,
      "loss": 0.0142,
      "step": 10732
    },
    {
      "epoch": 19.803324099722992,
      "grad_norm": 0.28833532333374023,
      "learning_rate": 2.975666666666667e-05,
      "loss": 0.0117,
      "step": 10733
    },
    {
      "epoch": 19.80517082179132,
      "grad_norm": 0.5392321944236755,
      "learning_rate": 2.9756333333333332e-05,
      "loss": 0.0118,
      "step": 10734
    },
    {
      "epoch": 19.80701754385965,
      "grad_norm": 0.29473716020584106,
      "learning_rate": 2.9756e-05,
      "loss": 0.0104,
      "step": 10735
    },
    {
      "epoch": 19.808864265927976,
      "grad_norm": 0.24462805688381195,
      "learning_rate": 2.975566666666667e-05,
      "loss": 0.0084,
      "step": 10736
    },
    {
      "epoch": 19.810710987996305,
      "grad_norm": 0.20876191556453705,
      "learning_rate": 2.9755333333333333e-05,
      "loss": 0.0104,
      "step": 10737
    },
    {
      "epoch": 19.812557710064635,
      "grad_norm": 1.0548560619354248,
      "learning_rate": 2.9755000000000002e-05,
      "loss": 0.0134,
      "step": 10738
    },
    {
      "epoch": 19.814404432132964,
      "grad_norm": 0.18749235570430756,
      "learning_rate": 2.9754666666666668e-05,
      "loss": 0.0052,
      "step": 10739
    },
    {
      "epoch": 19.816251154201293,
      "grad_norm": 0.5859012603759766,
      "learning_rate": 2.9754333333333334e-05,
      "loss": 0.0073,
      "step": 10740
    },
    {
      "epoch": 19.818097876269622,
      "grad_norm": 0.3927551209926605,
      "learning_rate": 2.9754e-05,
      "loss": 0.0147,
      "step": 10741
    },
    {
      "epoch": 19.81994459833795,
      "grad_norm": 0.2841923236846924,
      "learning_rate": 2.975366666666667e-05,
      "loss": 0.0089,
      "step": 10742
    },
    {
      "epoch": 19.82179132040628,
      "grad_norm": 0.26941174268722534,
      "learning_rate": 2.9753333333333332e-05,
      "loss": 0.0128,
      "step": 10743
    },
    {
      "epoch": 19.823638042474606,
      "grad_norm": 0.32388725876808167,
      "learning_rate": 2.9753e-05,
      "loss": 0.0132,
      "step": 10744
    },
    {
      "epoch": 19.825484764542935,
      "grad_norm": 0.9663119316101074,
      "learning_rate": 2.9752666666666667e-05,
      "loss": 0.0159,
      "step": 10745
    },
    {
      "epoch": 19.827331486611264,
      "grad_norm": 0.2247830033302307,
      "learning_rate": 2.9752333333333333e-05,
      "loss": 0.0092,
      "step": 10746
    },
    {
      "epoch": 19.829178208679593,
      "grad_norm": 0.169455423951149,
      "learning_rate": 2.9752000000000002e-05,
      "loss": 0.0068,
      "step": 10747
    },
    {
      "epoch": 19.831024930747922,
      "grad_norm": 0.6221871972084045,
      "learning_rate": 2.9751666666666665e-05,
      "loss": 0.0156,
      "step": 10748
    },
    {
      "epoch": 19.83287165281625,
      "grad_norm": 0.65969318151474,
      "learning_rate": 2.9751333333333334e-05,
      "loss": 0.186,
      "step": 10749
    },
    {
      "epoch": 19.83471837488458,
      "grad_norm": 0.587780237197876,
      "learning_rate": 2.9751e-05,
      "loss": 0.1271,
      "step": 10750
    },
    {
      "epoch": 19.83656509695291,
      "grad_norm": 0.6122603416442871,
      "learning_rate": 2.9750666666666666e-05,
      "loss": 0.1059,
      "step": 10751
    },
    {
      "epoch": 19.83841181902124,
      "grad_norm": 0.9975597262382507,
      "learning_rate": 2.9750333333333332e-05,
      "loss": 0.1889,
      "step": 10752
    },
    {
      "epoch": 19.840258541089565,
      "grad_norm": 0.4894188940525055,
      "learning_rate": 2.975e-05,
      "loss": 0.0773,
      "step": 10753
    },
    {
      "epoch": 19.842105263157894,
      "grad_norm": 0.5466663837432861,
      "learning_rate": 2.9749666666666667e-05,
      "loss": 0.0746,
      "step": 10754
    },
    {
      "epoch": 19.843951985226223,
      "grad_norm": 0.44391313195228577,
      "learning_rate": 2.9749333333333333e-05,
      "loss": 0.0762,
      "step": 10755
    },
    {
      "epoch": 19.845798707294552,
      "grad_norm": 0.7194757461547852,
      "learning_rate": 2.9749000000000003e-05,
      "loss": 0.0869,
      "step": 10756
    },
    {
      "epoch": 19.84764542936288,
      "grad_norm": 0.42723938822746277,
      "learning_rate": 2.9748666666666665e-05,
      "loss": 0.0938,
      "step": 10757
    },
    {
      "epoch": 19.84949215143121,
      "grad_norm": 0.4974992871284485,
      "learning_rate": 2.9748333333333335e-05,
      "loss": 0.0541,
      "step": 10758
    },
    {
      "epoch": 19.85133887349954,
      "grad_norm": 0.27553194761276245,
      "learning_rate": 2.9748e-05,
      "loss": 0.0281,
      "step": 10759
    },
    {
      "epoch": 19.85318559556787,
      "grad_norm": 0.5271324515342712,
      "learning_rate": 2.9747666666666666e-05,
      "loss": 0.0481,
      "step": 10760
    },
    {
      "epoch": 19.855032317636194,
      "grad_norm": 0.704689085483551,
      "learning_rate": 2.9747333333333336e-05,
      "loss": 0.0303,
      "step": 10761
    },
    {
      "epoch": 19.856879039704523,
      "grad_norm": 0.41468116641044617,
      "learning_rate": 2.9747e-05,
      "loss": 0.0265,
      "step": 10762
    },
    {
      "epoch": 19.858725761772853,
      "grad_norm": 0.49711140990257263,
      "learning_rate": 2.9746666666666668e-05,
      "loss": 0.0488,
      "step": 10763
    },
    {
      "epoch": 19.86057248384118,
      "grad_norm": 0.3405868709087372,
      "learning_rate": 2.9746333333333334e-05,
      "loss": 0.0256,
      "step": 10764
    },
    {
      "epoch": 19.86241920590951,
      "grad_norm": 0.41757965087890625,
      "learning_rate": 2.9746000000000003e-05,
      "loss": 0.0204,
      "step": 10765
    },
    {
      "epoch": 19.86426592797784,
      "grad_norm": 0.36645424365997314,
      "learning_rate": 2.9745666666666665e-05,
      "loss": 0.0261,
      "step": 10766
    },
    {
      "epoch": 19.86611265004617,
      "grad_norm": 0.2549731433391571,
      "learning_rate": 2.9745333333333335e-05,
      "loss": 0.014,
      "step": 10767
    },
    {
      "epoch": 19.8679593721145,
      "grad_norm": 0.4444524645805359,
      "learning_rate": 2.9745e-05,
      "loss": 0.0123,
      "step": 10768
    },
    {
      "epoch": 19.869806094182824,
      "grad_norm": 0.49080604314804077,
      "learning_rate": 2.9744666666666667e-05,
      "loss": 0.0274,
      "step": 10769
    },
    {
      "epoch": 19.871652816251153,
      "grad_norm": 0.3654726445674896,
      "learning_rate": 2.9744333333333336e-05,
      "loss": 0.0119,
      "step": 10770
    },
    {
      "epoch": 19.873499538319482,
      "grad_norm": 0.3014700412750244,
      "learning_rate": 2.9744000000000002e-05,
      "loss": 0.0097,
      "step": 10771
    },
    {
      "epoch": 19.87534626038781,
      "grad_norm": 0.2731376588344574,
      "learning_rate": 2.9743666666666668e-05,
      "loss": 0.0108,
      "step": 10772
    },
    {
      "epoch": 19.87719298245614,
      "grad_norm": 0.21904848515987396,
      "learning_rate": 2.9743333333333334e-05,
      "loss": 0.0114,
      "step": 10773
    },
    {
      "epoch": 19.87903970452447,
      "grad_norm": 0.41635486483573914,
      "learning_rate": 2.9743000000000003e-05,
      "loss": 0.0156,
      "step": 10774
    },
    {
      "epoch": 19.8808864265928,
      "grad_norm": 0.3734789490699768,
      "learning_rate": 2.9742666666666666e-05,
      "loss": 0.0113,
      "step": 10775
    },
    {
      "epoch": 19.882733148661128,
      "grad_norm": 0.44643911719322205,
      "learning_rate": 2.9742333333333335e-05,
      "loss": 0.0119,
      "step": 10776
    },
    {
      "epoch": 19.884579870729453,
      "grad_norm": 0.2010699212551117,
      "learning_rate": 2.9742e-05,
      "loss": 0.0094,
      "step": 10777
    },
    {
      "epoch": 19.886426592797783,
      "grad_norm": 0.3544909656047821,
      "learning_rate": 2.9741666666666667e-05,
      "loss": 0.011,
      "step": 10778
    },
    {
      "epoch": 19.88827331486611,
      "grad_norm": 0.3866400122642517,
      "learning_rate": 2.9741333333333336e-05,
      "loss": 0.0073,
      "step": 10779
    },
    {
      "epoch": 19.89012003693444,
      "grad_norm": 0.34151676297187805,
      "learning_rate": 2.9741e-05,
      "loss": 0.0169,
      "step": 10780
    },
    {
      "epoch": 19.89196675900277,
      "grad_norm": 0.29591402411460876,
      "learning_rate": 2.9740666666666668e-05,
      "loss": 0.008,
      "step": 10781
    },
    {
      "epoch": 19.8938134810711,
      "grad_norm": 0.30311036109924316,
      "learning_rate": 2.9740333333333334e-05,
      "loss": 0.0125,
      "step": 10782
    },
    {
      "epoch": 19.89566020313943,
      "grad_norm": 0.19861190021038055,
      "learning_rate": 2.974e-05,
      "loss": 0.0068,
      "step": 10783
    },
    {
      "epoch": 19.897506925207757,
      "grad_norm": 0.22376330196857452,
      "learning_rate": 2.9739666666666666e-05,
      "loss": 0.0081,
      "step": 10784
    },
    {
      "epoch": 19.899353647276087,
      "grad_norm": 0.4291958808898926,
      "learning_rate": 2.9739333333333335e-05,
      "loss": 0.0147,
      "step": 10785
    },
    {
      "epoch": 19.901200369344412,
      "grad_norm": 0.3459694981575012,
      "learning_rate": 2.9739e-05,
      "loss": 0.0287,
      "step": 10786
    },
    {
      "epoch": 19.90304709141274,
      "grad_norm": 0.3165818154811859,
      "learning_rate": 2.9738666666666667e-05,
      "loss": 0.0101,
      "step": 10787
    },
    {
      "epoch": 19.90489381348107,
      "grad_norm": 0.16458719968795776,
      "learning_rate": 2.9738333333333336e-05,
      "loss": 0.0063,
      "step": 10788
    },
    {
      "epoch": 19.9067405355494,
      "grad_norm": 0.26842695474624634,
      "learning_rate": 2.9738e-05,
      "loss": 0.0077,
      "step": 10789
    },
    {
      "epoch": 19.90858725761773,
      "grad_norm": 0.6238064765930176,
      "learning_rate": 2.9737666666666668e-05,
      "loss": 0.0105,
      "step": 10790
    },
    {
      "epoch": 19.910433979686058,
      "grad_norm": 0.45984452962875366,
      "learning_rate": 2.9737333333333334e-05,
      "loss": 0.0136,
      "step": 10791
    },
    {
      "epoch": 19.912280701754387,
      "grad_norm": 0.22845137119293213,
      "learning_rate": 2.9737e-05,
      "loss": 0.0095,
      "step": 10792
    },
    {
      "epoch": 19.914127423822716,
      "grad_norm": 0.6064977645874023,
      "learning_rate": 2.9736666666666666e-05,
      "loss": 0.0135,
      "step": 10793
    },
    {
      "epoch": 19.91597414589104,
      "grad_norm": 0.6048700213432312,
      "learning_rate": 2.9736333333333335e-05,
      "loss": 0.0103,
      "step": 10794
    },
    {
      "epoch": 19.91782086795937,
      "grad_norm": 0.6151441931724548,
      "learning_rate": 2.9736e-05,
      "loss": 0.0128,
      "step": 10795
    },
    {
      "epoch": 19.9196675900277,
      "grad_norm": 0.722559928894043,
      "learning_rate": 2.9735666666666667e-05,
      "loss": 0.0122,
      "step": 10796
    },
    {
      "epoch": 19.92151431209603,
      "grad_norm": 0.9557310342788696,
      "learning_rate": 2.9735333333333337e-05,
      "loss": 0.016,
      "step": 10797
    },
    {
      "epoch": 19.92336103416436,
      "grad_norm": 0.5047065615653992,
      "learning_rate": 2.9735e-05,
      "loss": 0.0143,
      "step": 10798
    },
    {
      "epoch": 19.925207756232687,
      "grad_norm": 0.9226062297821045,
      "learning_rate": 2.973466666666667e-05,
      "loss": 0.1909,
      "step": 10799
    },
    {
      "epoch": 19.927054478301017,
      "grad_norm": 0.7328792214393616,
      "learning_rate": 2.9734333333333334e-05,
      "loss": 0.1344,
      "step": 10800
    },
    {
      "epoch": 19.928901200369346,
      "grad_norm": 0.9491218328475952,
      "learning_rate": 2.9734e-05,
      "loss": 0.1537,
      "step": 10801
    },
    {
      "epoch": 19.930747922437675,
      "grad_norm": 0.7314680218696594,
      "learning_rate": 2.9733666666666666e-05,
      "loss": 0.1021,
      "step": 10802
    },
    {
      "epoch": 19.932594644506,
      "grad_norm": 0.3304039537906647,
      "learning_rate": 2.9733333333333336e-05,
      "loss": 0.07,
      "step": 10803
    },
    {
      "epoch": 19.93444136657433,
      "grad_norm": 0.7055652737617493,
      "learning_rate": 2.9733e-05,
      "loss": 0.1535,
      "step": 10804
    },
    {
      "epoch": 19.93628808864266,
      "grad_norm": 0.4285171627998352,
      "learning_rate": 2.9732666666666667e-05,
      "loss": 0.0605,
      "step": 10805
    },
    {
      "epoch": 19.938134810710988,
      "grad_norm": 0.5491939783096313,
      "learning_rate": 2.9732333333333333e-05,
      "loss": 0.0579,
      "step": 10806
    },
    {
      "epoch": 19.939981532779317,
      "grad_norm": 0.6970460414886475,
      "learning_rate": 2.9732e-05,
      "loss": 0.0504,
      "step": 10807
    },
    {
      "epoch": 19.941828254847646,
      "grad_norm": 0.36753565073013306,
      "learning_rate": 2.973166666666667e-05,
      "loss": 0.0547,
      "step": 10808
    },
    {
      "epoch": 19.943674976915975,
      "grad_norm": 0.4938269555568695,
      "learning_rate": 2.973133333333333e-05,
      "loss": 0.0613,
      "step": 10809
    },
    {
      "epoch": 19.945521698984304,
      "grad_norm": 0.2081955224275589,
      "learning_rate": 2.9731e-05,
      "loss": 0.0236,
      "step": 10810
    },
    {
      "epoch": 19.94736842105263,
      "grad_norm": 0.3069004714488983,
      "learning_rate": 2.973066666666667e-05,
      "loss": 0.0327,
      "step": 10811
    },
    {
      "epoch": 19.94921514312096,
      "grad_norm": 0.5289403796195984,
      "learning_rate": 2.9730333333333332e-05,
      "loss": 0.0276,
      "step": 10812
    },
    {
      "epoch": 19.95106186518929,
      "grad_norm": 0.26222914457321167,
      "learning_rate": 2.973e-05,
      "loss": 0.0095,
      "step": 10813
    },
    {
      "epoch": 19.952908587257618,
      "grad_norm": 0.4635355770587921,
      "learning_rate": 2.9729666666666668e-05,
      "loss": 0.029,
      "step": 10814
    },
    {
      "epoch": 19.954755309325947,
      "grad_norm": 0.43034881353378296,
      "learning_rate": 2.9729333333333334e-05,
      "loss": 0.0311,
      "step": 10815
    },
    {
      "epoch": 19.956602031394276,
      "grad_norm": 0.2856021821498871,
      "learning_rate": 2.9729e-05,
      "loss": 0.0167,
      "step": 10816
    },
    {
      "epoch": 19.958448753462605,
      "grad_norm": 0.21279774606227875,
      "learning_rate": 2.972866666666667e-05,
      "loss": 0.0122,
      "step": 10817
    },
    {
      "epoch": 19.960295475530934,
      "grad_norm": 0.18603555858135223,
      "learning_rate": 2.972833333333333e-05,
      "loss": 0.0056,
      "step": 10818
    },
    {
      "epoch": 19.96214219759926,
      "grad_norm": 0.25018876791000366,
      "learning_rate": 2.9728e-05,
      "loss": 0.0151,
      "step": 10819
    },
    {
      "epoch": 19.96398891966759,
      "grad_norm": 0.23212258517742157,
      "learning_rate": 2.972766666666667e-05,
      "loss": 0.0104,
      "step": 10820
    },
    {
      "epoch": 19.965835641735918,
      "grad_norm": 0.5990542769432068,
      "learning_rate": 2.9727333333333333e-05,
      "loss": 0.0096,
      "step": 10821
    },
    {
      "epoch": 19.967682363804247,
      "grad_norm": 0.29488661885261536,
      "learning_rate": 2.9727000000000002e-05,
      "loss": 0.0179,
      "step": 10822
    },
    {
      "epoch": 19.969529085872576,
      "grad_norm": 0.2727646827697754,
      "learning_rate": 2.9726666666666668e-05,
      "loss": 0.0326,
      "step": 10823
    },
    {
      "epoch": 19.971375807940905,
      "grad_norm": 0.21818789839744568,
      "learning_rate": 2.9726333333333334e-05,
      "loss": 0.0071,
      "step": 10824
    },
    {
      "epoch": 19.973222530009235,
      "grad_norm": 0.3400208055973053,
      "learning_rate": 2.9726e-05,
      "loss": 0.0087,
      "step": 10825
    },
    {
      "epoch": 19.975069252077564,
      "grad_norm": 0.4294982850551605,
      "learning_rate": 2.972566666666667e-05,
      "loss": 0.0164,
      "step": 10826
    },
    {
      "epoch": 19.97691597414589,
      "grad_norm": 0.20422349870204926,
      "learning_rate": 2.9725333333333335e-05,
      "loss": 0.0102,
      "step": 10827
    },
    {
      "epoch": 19.97876269621422,
      "grad_norm": 0.3575103282928467,
      "learning_rate": 2.9725e-05,
      "loss": 0.0097,
      "step": 10828
    },
    {
      "epoch": 19.980609418282548,
      "grad_norm": 0.43506139516830444,
      "learning_rate": 2.972466666666667e-05,
      "loss": 0.0115,
      "step": 10829
    },
    {
      "epoch": 19.982456140350877,
      "grad_norm": 0.39745500683784485,
      "learning_rate": 2.9724333333333333e-05,
      "loss": 0.0112,
      "step": 10830
    },
    {
      "epoch": 19.984302862419206,
      "grad_norm": 0.32903939485549927,
      "learning_rate": 2.9724000000000002e-05,
      "loss": 0.0168,
      "step": 10831
    },
    {
      "epoch": 19.986149584487535,
      "grad_norm": 0.5643303990364075,
      "learning_rate": 2.9723666666666668e-05,
      "loss": 0.0216,
      "step": 10832
    },
    {
      "epoch": 19.987996306555864,
      "grad_norm": 0.27045926451683044,
      "learning_rate": 2.9723333333333334e-05,
      "loss": 0.0096,
      "step": 10833
    },
    {
      "epoch": 19.989843028624193,
      "grad_norm": 0.26068615913391113,
      "learning_rate": 2.9723e-05,
      "loss": 0.0103,
      "step": 10834
    },
    {
      "epoch": 19.991689750692522,
      "grad_norm": 0.31268391013145447,
      "learning_rate": 2.9722666666666666e-05,
      "loss": 0.0086,
      "step": 10835
    },
    {
      "epoch": 19.993536472760848,
      "grad_norm": 0.26672235131263733,
      "learning_rate": 2.9722333333333335e-05,
      "loss": 0.0077,
      "step": 10836
    },
    {
      "epoch": 19.995383194829177,
      "grad_norm": 0.36160510778427124,
      "learning_rate": 2.9722e-05,
      "loss": 0.0095,
      "step": 10837
    },
    {
      "epoch": 19.997229916897506,
      "grad_norm": 0.7074096202850342,
      "learning_rate": 2.9721666666666667e-05,
      "loss": 0.0127,
      "step": 10838
    },
    {
      "epoch": 19.999076638965835,
      "grad_norm": 0.31743553280830383,
      "learning_rate": 2.9721333333333333e-05,
      "loss": 0.0114,
      "step": 10839
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.265717089176178,
      "learning_rate": 2.9721000000000002e-05,
      "loss": 0.0028,
      "step": 10840
    },
    {
      "epoch": 20.00184672206833,
      "grad_norm": 0.49848538637161255,
      "learning_rate": 2.9720666666666665e-05,
      "loss": 0.152,
      "step": 10841
    },
    {
      "epoch": 20.00369344413666,
      "grad_norm": 0.6284631490707397,
      "learning_rate": 2.9720333333333334e-05,
      "loss": 0.166,
      "step": 10842
    },
    {
      "epoch": 20.005540166204987,
      "grad_norm": 0.4607674181461334,
      "learning_rate": 2.972e-05,
      "loss": 0.1029,
      "step": 10843
    },
    {
      "epoch": 20.007386888273317,
      "grad_norm": 0.5444645285606384,
      "learning_rate": 2.9719666666666666e-05,
      "loss": 0.0861,
      "step": 10844
    },
    {
      "epoch": 20.009233610341642,
      "grad_norm": 0.4144648313522339,
      "learning_rate": 2.9719333333333335e-05,
      "loss": 0.0896,
      "step": 10845
    },
    {
      "epoch": 20.01108033240997,
      "grad_norm": 0.5311521887779236,
      "learning_rate": 2.9719e-05,
      "loss": 0.0902,
      "step": 10846
    },
    {
      "epoch": 20.0129270544783,
      "grad_norm": 0.3980214297771454,
      "learning_rate": 2.9718666666666667e-05,
      "loss": 0.0537,
      "step": 10847
    },
    {
      "epoch": 20.01477377654663,
      "grad_norm": 0.6435825824737549,
      "learning_rate": 2.9718333333333333e-05,
      "loss": 0.1102,
      "step": 10848
    },
    {
      "epoch": 20.01662049861496,
      "grad_norm": 0.41154757142066956,
      "learning_rate": 2.9718000000000002e-05,
      "loss": 0.0489,
      "step": 10849
    },
    {
      "epoch": 20.018467220683288,
      "grad_norm": 0.4965069890022278,
      "learning_rate": 2.9717666666666665e-05,
      "loss": 0.0433,
      "step": 10850
    },
    {
      "epoch": 20.020313942751617,
      "grad_norm": 0.3039763271808624,
      "learning_rate": 2.9717333333333334e-05,
      "loss": 0.0643,
      "step": 10851
    },
    {
      "epoch": 20.022160664819946,
      "grad_norm": 0.30471768975257874,
      "learning_rate": 2.9717e-05,
      "loss": 0.0276,
      "step": 10852
    },
    {
      "epoch": 20.02400738688827,
      "grad_norm": 0.3537667393684387,
      "learning_rate": 2.9716666666666666e-05,
      "loss": 0.0572,
      "step": 10853
    },
    {
      "epoch": 20.0258541089566,
      "grad_norm": 0.3218044638633728,
      "learning_rate": 2.9716333333333336e-05,
      "loss": 0.032,
      "step": 10854
    },
    {
      "epoch": 20.02770083102493,
      "grad_norm": 0.21395094692707062,
      "learning_rate": 2.9716e-05,
      "loss": 0.0155,
      "step": 10855
    },
    {
      "epoch": 20.02954755309326,
      "grad_norm": 0.1753055602312088,
      "learning_rate": 2.9715666666666667e-05,
      "loss": 0.0098,
      "step": 10856
    },
    {
      "epoch": 20.03139427516159,
      "grad_norm": 0.1765054166316986,
      "learning_rate": 2.9715333333333333e-05,
      "loss": 0.0081,
      "step": 10857
    },
    {
      "epoch": 20.033240997229917,
      "grad_norm": 0.15867426991462708,
      "learning_rate": 2.9715000000000003e-05,
      "loss": 0.0093,
      "step": 10858
    },
    {
      "epoch": 20.035087719298247,
      "grad_norm": 0.2986152172088623,
      "learning_rate": 2.9714666666666665e-05,
      "loss": 0.0131,
      "step": 10859
    },
    {
      "epoch": 20.036934441366576,
      "grad_norm": 0.5510702729225159,
      "learning_rate": 2.9714333333333335e-05,
      "loss": 0.0383,
      "step": 10860
    },
    {
      "epoch": 20.0387811634349,
      "grad_norm": 0.21956251561641693,
      "learning_rate": 2.9714000000000004e-05,
      "loss": 0.01,
      "step": 10861
    },
    {
      "epoch": 20.04062788550323,
      "grad_norm": 0.33517369627952576,
      "learning_rate": 2.9713666666666666e-05,
      "loss": 0.0155,
      "step": 10862
    },
    {
      "epoch": 20.04247460757156,
      "grad_norm": 0.5131813287734985,
      "learning_rate": 2.9713333333333336e-05,
      "loss": 0.0175,
      "step": 10863
    },
    {
      "epoch": 20.04432132963989,
      "grad_norm": 0.18757785856723785,
      "learning_rate": 2.9713e-05,
      "loss": 0.0084,
      "step": 10864
    },
    {
      "epoch": 20.046168051708218,
      "grad_norm": 0.14929111301898956,
      "learning_rate": 2.9712666666666668e-05,
      "loss": 0.0058,
      "step": 10865
    },
    {
      "epoch": 20.048014773776547,
      "grad_norm": 0.3974211513996124,
      "learning_rate": 2.9712333333333334e-05,
      "loss": 0.0118,
      "step": 10866
    },
    {
      "epoch": 20.049861495844876,
      "grad_norm": 0.3090802729129791,
      "learning_rate": 2.9712e-05,
      "loss": 0.0074,
      "step": 10867
    },
    {
      "epoch": 20.051708217913205,
      "grad_norm": 0.249526709318161,
      "learning_rate": 2.9711666666666665e-05,
      "loss": 0.0131,
      "step": 10868
    },
    {
      "epoch": 20.053554939981534,
      "grad_norm": 0.2154981642961502,
      "learning_rate": 2.9711333333333335e-05,
      "loss": 0.0055,
      "step": 10869
    },
    {
      "epoch": 20.05540166204986,
      "grad_norm": 0.18604527413845062,
      "learning_rate": 2.9711e-05,
      "loss": 0.0062,
      "step": 10870
    },
    {
      "epoch": 20.05724838411819,
      "grad_norm": 0.26657775044441223,
      "learning_rate": 2.9710666666666667e-05,
      "loss": 0.0076,
      "step": 10871
    },
    {
      "epoch": 20.05909510618652,
      "grad_norm": 0.16824635863304138,
      "learning_rate": 2.9710333333333336e-05,
      "loss": 0.0036,
      "step": 10872
    },
    {
      "epoch": 20.060941828254848,
      "grad_norm": 0.22853249311447144,
      "learning_rate": 2.971e-05,
      "loss": 0.0062,
      "step": 10873
    },
    {
      "epoch": 20.062788550323177,
      "grad_norm": 0.1347638964653015,
      "learning_rate": 2.9709666666666668e-05,
      "loss": 0.0043,
      "step": 10874
    },
    {
      "epoch": 20.064635272391506,
      "grad_norm": 0.5696449279785156,
      "learning_rate": 2.9709333333333334e-05,
      "loss": 0.0202,
      "step": 10875
    },
    {
      "epoch": 20.066481994459835,
      "grad_norm": 0.1813577264547348,
      "learning_rate": 2.9709e-05,
      "loss": 0.006,
      "step": 10876
    },
    {
      "epoch": 20.068328716528164,
      "grad_norm": 0.24455659091472626,
      "learning_rate": 2.970866666666667e-05,
      "loss": 0.036,
      "step": 10877
    },
    {
      "epoch": 20.07017543859649,
      "grad_norm": 0.26324525475502014,
      "learning_rate": 2.9708333333333335e-05,
      "loss": 0.0101,
      "step": 10878
    },
    {
      "epoch": 20.07202216066482,
      "grad_norm": 0.3263266682624817,
      "learning_rate": 2.9708e-05,
      "loss": 0.0065,
      "step": 10879
    },
    {
      "epoch": 20.073868882733148,
      "grad_norm": 0.43081149458885193,
      "learning_rate": 2.9707666666666667e-05,
      "loss": 0.0163,
      "step": 10880
    },
    {
      "epoch": 20.075715604801477,
      "grad_norm": 0.2542450726032257,
      "learning_rate": 2.9707333333333336e-05,
      "loss": 0.007,
      "step": 10881
    },
    {
      "epoch": 20.077562326869806,
      "grad_norm": 0.5075932741165161,
      "learning_rate": 2.9707e-05,
      "loss": 0.0137,
      "step": 10882
    },
    {
      "epoch": 20.079409048938135,
      "grad_norm": 0.47679972648620605,
      "learning_rate": 2.9706666666666668e-05,
      "loss": 0.0112,
      "step": 10883
    },
    {
      "epoch": 20.081255771006465,
      "grad_norm": 0.2431761473417282,
      "learning_rate": 2.9706333333333334e-05,
      "loss": 0.0082,
      "step": 10884
    },
    {
      "epoch": 20.083102493074794,
      "grad_norm": 0.3411027789115906,
      "learning_rate": 2.9706e-05,
      "loss": 0.0096,
      "step": 10885
    },
    {
      "epoch": 20.08494921514312,
      "grad_norm": 0.4069492816925049,
      "learning_rate": 2.970566666666667e-05,
      "loss": 0.0056,
      "step": 10886
    },
    {
      "epoch": 20.08679593721145,
      "grad_norm": 0.283371239900589,
      "learning_rate": 2.9705333333333335e-05,
      "loss": 0.0101,
      "step": 10887
    },
    {
      "epoch": 20.088642659279778,
      "grad_norm": 0.1481645703315735,
      "learning_rate": 2.9705e-05,
      "loss": 0.0033,
      "step": 10888
    },
    {
      "epoch": 20.090489381348107,
      "grad_norm": 0.45300835371017456,
      "learning_rate": 2.9704666666666667e-05,
      "loss": 0.0091,
      "step": 10889
    },
    {
      "epoch": 20.092336103416436,
      "grad_norm": 0.6289262175559998,
      "learning_rate": 2.9704333333333336e-05,
      "loss": 0.0061,
      "step": 10890
    },
    {
      "epoch": 20.094182825484765,
      "grad_norm": 0.5154815316200256,
      "learning_rate": 2.9704e-05,
      "loss": 0.1376,
      "step": 10891
    },
    {
      "epoch": 20.096029547553094,
      "grad_norm": 0.49002134799957275,
      "learning_rate": 2.9703666666666668e-05,
      "loss": 0.1471,
      "step": 10892
    },
    {
      "epoch": 20.097876269621423,
      "grad_norm": 0.4354952871799469,
      "learning_rate": 2.9703333333333334e-05,
      "loss": 0.1032,
      "step": 10893
    },
    {
      "epoch": 20.099722991689752,
      "grad_norm": 0.46918243169784546,
      "learning_rate": 2.9703e-05,
      "loss": 0.0712,
      "step": 10894
    },
    {
      "epoch": 20.101569713758078,
      "grad_norm": 0.350827693939209,
      "learning_rate": 2.970266666666667e-05,
      "loss": 0.0663,
      "step": 10895
    },
    {
      "epoch": 20.103416435826407,
      "grad_norm": 0.4532705247402191,
      "learning_rate": 2.9702333333333332e-05,
      "loss": 0.0965,
      "step": 10896
    },
    {
      "epoch": 20.105263157894736,
      "grad_norm": 0.4961085915565491,
      "learning_rate": 2.9702e-05,
      "loss": 0.0687,
      "step": 10897
    },
    {
      "epoch": 20.107109879963065,
      "grad_norm": 0.41538205742836,
      "learning_rate": 2.9701666666666667e-05,
      "loss": 0.0523,
      "step": 10898
    },
    {
      "epoch": 20.108956602031395,
      "grad_norm": 0.8226404190063477,
      "learning_rate": 2.9701333333333333e-05,
      "loss": 0.0539,
      "step": 10899
    },
    {
      "epoch": 20.110803324099724,
      "grad_norm": 0.3654651343822479,
      "learning_rate": 2.9701e-05,
      "loss": 0.0602,
      "step": 10900
    },
    {
      "epoch": 20.112650046168053,
      "grad_norm": 0.3377811908721924,
      "learning_rate": 2.970066666666667e-05,
      "loss": 0.0319,
      "step": 10901
    },
    {
      "epoch": 20.114496768236382,
      "grad_norm": 0.34735211730003357,
      "learning_rate": 2.9700333333333334e-05,
      "loss": 0.0557,
      "step": 10902
    },
    {
      "epoch": 20.116343490304708,
      "grad_norm": 0.3268611431121826,
      "learning_rate": 2.97e-05,
      "loss": 0.0273,
      "step": 10903
    },
    {
      "epoch": 20.118190212373037,
      "grad_norm": 0.3331509530544281,
      "learning_rate": 2.969966666666667e-05,
      "loss": 0.0452,
      "step": 10904
    },
    {
      "epoch": 20.120036934441366,
      "grad_norm": 0.2206970751285553,
      "learning_rate": 2.9699333333333332e-05,
      "loss": 0.0233,
      "step": 10905
    },
    {
      "epoch": 20.121883656509695,
      "grad_norm": 0.30379095673561096,
      "learning_rate": 2.9699e-05,
      "loss": 0.0092,
      "step": 10906
    },
    {
      "epoch": 20.123730378578024,
      "grad_norm": 0.6868551969528198,
      "learning_rate": 2.9698666666666667e-05,
      "loss": 0.0436,
      "step": 10907
    },
    {
      "epoch": 20.125577100646353,
      "grad_norm": 0.38484877347946167,
      "learning_rate": 2.9698333333333333e-05,
      "loss": 0.0208,
      "step": 10908
    },
    {
      "epoch": 20.127423822714682,
      "grad_norm": 0.2958061099052429,
      "learning_rate": 2.9698e-05,
      "loss": 0.016,
      "step": 10909
    },
    {
      "epoch": 20.12927054478301,
      "grad_norm": 0.16953197121620178,
      "learning_rate": 2.969766666666667e-05,
      "loss": 0.0077,
      "step": 10910
    },
    {
      "epoch": 20.131117266851337,
      "grad_norm": 0.2099265456199646,
      "learning_rate": 2.9697333333333335e-05,
      "loss": 0.0121,
      "step": 10911
    },
    {
      "epoch": 20.132963988919666,
      "grad_norm": 0.2694770097732544,
      "learning_rate": 2.9697e-05,
      "loss": 0.0101,
      "step": 10912
    },
    {
      "epoch": 20.134810710987995,
      "grad_norm": 0.4015445411205292,
      "learning_rate": 2.969666666666667e-05,
      "loss": 0.0099,
      "step": 10913
    },
    {
      "epoch": 20.136657433056325,
      "grad_norm": 0.5667865872383118,
      "learning_rate": 2.9696333333333332e-05,
      "loss": 0.0103,
      "step": 10914
    },
    {
      "epoch": 20.138504155124654,
      "grad_norm": 0.3129112422466278,
      "learning_rate": 2.9696e-05,
      "loss": 0.007,
      "step": 10915
    },
    {
      "epoch": 20.140350877192983,
      "grad_norm": 0.3248129189014435,
      "learning_rate": 2.9695666666666668e-05,
      "loss": 0.0132,
      "step": 10916
    },
    {
      "epoch": 20.142197599261312,
      "grad_norm": 0.24504362046718597,
      "learning_rate": 2.9695333333333334e-05,
      "loss": 0.0073,
      "step": 10917
    },
    {
      "epoch": 20.14404432132964,
      "grad_norm": 0.3159309923648834,
      "learning_rate": 2.9695e-05,
      "loss": 0.0079,
      "step": 10918
    },
    {
      "epoch": 20.14589104339797,
      "grad_norm": 0.4856783151626587,
      "learning_rate": 2.969466666666667e-05,
      "loss": 0.0138,
      "step": 10919
    },
    {
      "epoch": 20.147737765466296,
      "grad_norm": 0.3120325803756714,
      "learning_rate": 2.9694333333333335e-05,
      "loss": 0.0057,
      "step": 10920
    },
    {
      "epoch": 20.149584487534625,
      "grad_norm": 0.25674888491630554,
      "learning_rate": 2.9694e-05,
      "loss": 0.0064,
      "step": 10921
    },
    {
      "epoch": 20.151431209602954,
      "grad_norm": 0.17604240775108337,
      "learning_rate": 2.969366666666667e-05,
      "loss": 0.004,
      "step": 10922
    },
    {
      "epoch": 20.153277931671283,
      "grad_norm": 0.3554757237434387,
      "learning_rate": 2.9693333333333333e-05,
      "loss": 0.0107,
      "step": 10923
    },
    {
      "epoch": 20.155124653739612,
      "grad_norm": 0.31348365545272827,
      "learning_rate": 2.9693000000000002e-05,
      "loss": 0.012,
      "step": 10924
    },
    {
      "epoch": 20.15697137580794,
      "grad_norm": 0.39087775349617004,
      "learning_rate": 2.9692666666666668e-05,
      "loss": 0.0082,
      "step": 10925
    },
    {
      "epoch": 20.15881809787627,
      "grad_norm": 0.24652767181396484,
      "learning_rate": 2.9692333333333334e-05,
      "loss": 0.0078,
      "step": 10926
    },
    {
      "epoch": 20.1606648199446,
      "grad_norm": 0.3166562616825104,
      "learning_rate": 2.9692000000000003e-05,
      "loss": 0.0133,
      "step": 10927
    },
    {
      "epoch": 20.162511542012926,
      "grad_norm": 0.2203536182641983,
      "learning_rate": 2.9691666666666666e-05,
      "loss": 0.0087,
      "step": 10928
    },
    {
      "epoch": 20.164358264081255,
      "grad_norm": 0.2523220181465149,
      "learning_rate": 2.9691333333333335e-05,
      "loss": 0.0117,
      "step": 10929
    },
    {
      "epoch": 20.166204986149584,
      "grad_norm": 0.6120240092277527,
      "learning_rate": 2.9691e-05,
      "loss": 0.0143,
      "step": 10930
    },
    {
      "epoch": 20.168051708217913,
      "grad_norm": 0.34441477060317993,
      "learning_rate": 2.9690666666666667e-05,
      "loss": 0.0097,
      "step": 10931
    },
    {
      "epoch": 20.169898430286242,
      "grad_norm": 0.35958561301231384,
      "learning_rate": 2.9690333333333333e-05,
      "loss": 0.0057,
      "step": 10932
    },
    {
      "epoch": 20.17174515235457,
      "grad_norm": 0.20114681124687195,
      "learning_rate": 2.9690000000000002e-05,
      "loss": 0.0064,
      "step": 10933
    },
    {
      "epoch": 20.1735918744229,
      "grad_norm": 0.40742751955986023,
      "learning_rate": 2.9689666666666665e-05,
      "loss": 0.0136,
      "step": 10934
    },
    {
      "epoch": 20.17543859649123,
      "grad_norm": 0.21357586979866028,
      "learning_rate": 2.9689333333333334e-05,
      "loss": 0.0075,
      "step": 10935
    },
    {
      "epoch": 20.177285318559555,
      "grad_norm": 0.2256045937538147,
      "learning_rate": 2.9689000000000003e-05,
      "loss": 0.0073,
      "step": 10936
    },
    {
      "epoch": 20.179132040627884,
      "grad_norm": 1.3588895797729492,
      "learning_rate": 2.9688666666666666e-05,
      "loss": 0.0132,
      "step": 10937
    },
    {
      "epoch": 20.180978762696213,
      "grad_norm": 0.20366744697093964,
      "learning_rate": 2.9688333333333335e-05,
      "loss": 0.0066,
      "step": 10938
    },
    {
      "epoch": 20.182825484764543,
      "grad_norm": 0.20821727812290192,
      "learning_rate": 2.9688e-05,
      "loss": 0.0058,
      "step": 10939
    },
    {
      "epoch": 20.18467220683287,
      "grad_norm": 0.15946674346923828,
      "learning_rate": 2.9687666666666667e-05,
      "loss": 0.0052,
      "step": 10940
    },
    {
      "epoch": 20.1865189289012,
      "grad_norm": 0.46555081009864807,
      "learning_rate": 2.9687333333333333e-05,
      "loss": 0.1196,
      "step": 10941
    },
    {
      "epoch": 20.18836565096953,
      "grad_norm": 0.5835827589035034,
      "learning_rate": 2.9687000000000002e-05,
      "loss": 0.1611,
      "step": 10942
    },
    {
      "epoch": 20.19021237303786,
      "grad_norm": 0.5876694321632385,
      "learning_rate": 2.9686666666666665e-05,
      "loss": 0.1476,
      "step": 10943
    },
    {
      "epoch": 20.19205909510619,
      "grad_norm": 0.37970054149627686,
      "learning_rate": 2.9686333333333334e-05,
      "loss": 0.0706,
      "step": 10944
    },
    {
      "epoch": 20.193905817174514,
      "grad_norm": 0.5568404793739319,
      "learning_rate": 2.9686000000000003e-05,
      "loss": 0.0907,
      "step": 10945
    },
    {
      "epoch": 20.195752539242843,
      "grad_norm": 0.3703577220439911,
      "learning_rate": 2.9685666666666666e-05,
      "loss": 0.0511,
      "step": 10946
    },
    {
      "epoch": 20.197599261311172,
      "grad_norm": 0.4699932634830475,
      "learning_rate": 2.9685333333333335e-05,
      "loss": 0.0769,
      "step": 10947
    },
    {
      "epoch": 20.1994459833795,
      "grad_norm": 0.50783371925354,
      "learning_rate": 2.9685e-05,
      "loss": 0.0718,
      "step": 10948
    },
    {
      "epoch": 20.20129270544783,
      "grad_norm": 0.47592678666114807,
      "learning_rate": 2.9684666666666667e-05,
      "loss": 0.0514,
      "step": 10949
    },
    {
      "epoch": 20.20313942751616,
      "grad_norm": 0.4134548306465149,
      "learning_rate": 2.9684333333333333e-05,
      "loss": 0.0467,
      "step": 10950
    },
    {
      "epoch": 20.20498614958449,
      "grad_norm": 0.4005981981754303,
      "learning_rate": 2.9684000000000002e-05,
      "loss": 0.0494,
      "step": 10951
    },
    {
      "epoch": 20.206832871652818,
      "grad_norm": 0.35495683550834656,
      "learning_rate": 2.968366666666667e-05,
      "loss": 0.0497,
      "step": 10952
    },
    {
      "epoch": 20.208679593721143,
      "grad_norm": 0.3859303593635559,
      "learning_rate": 2.9683333333333334e-05,
      "loss": 0.0386,
      "step": 10953
    },
    {
      "epoch": 20.210526315789473,
      "grad_norm": 0.4294539988040924,
      "learning_rate": 2.9683000000000004e-05,
      "loss": 0.0125,
      "step": 10954
    },
    {
      "epoch": 20.2123730378578,
      "grad_norm": 0.23544616997241974,
      "learning_rate": 2.9682666666666666e-05,
      "loss": 0.0209,
      "step": 10955
    },
    {
      "epoch": 20.21421975992613,
      "grad_norm": 0.2471626102924347,
      "learning_rate": 2.9682333333333335e-05,
      "loss": 0.0429,
      "step": 10956
    },
    {
      "epoch": 20.21606648199446,
      "grad_norm": 0.2147773653268814,
      "learning_rate": 2.9681999999999998e-05,
      "loss": 0.0265,
      "step": 10957
    },
    {
      "epoch": 20.21791320406279,
      "grad_norm": 0.20125578343868256,
      "learning_rate": 2.9681666666666667e-05,
      "loss": 0.0147,
      "step": 10958
    },
    {
      "epoch": 20.21975992613112,
      "grad_norm": 0.37904635071754456,
      "learning_rate": 2.9681333333333333e-05,
      "loss": 0.0075,
      "step": 10959
    },
    {
      "epoch": 20.221606648199447,
      "grad_norm": 0.4701364040374756,
      "learning_rate": 2.9681e-05,
      "loss": 0.0311,
      "step": 10960
    },
    {
      "epoch": 20.223453370267773,
      "grad_norm": 0.4250022768974304,
      "learning_rate": 2.968066666666667e-05,
      "loss": 0.0734,
      "step": 10961
    },
    {
      "epoch": 20.225300092336102,
      "grad_norm": 0.22264167666435242,
      "learning_rate": 2.9680333333333334e-05,
      "loss": 0.0054,
      "step": 10962
    },
    {
      "epoch": 20.22714681440443,
      "grad_norm": 0.08203784376382828,
      "learning_rate": 2.968e-05,
      "loss": 0.0032,
      "step": 10963
    },
    {
      "epoch": 20.22899353647276,
      "grad_norm": 0.2587185204029083,
      "learning_rate": 2.9679666666666666e-05,
      "loss": 0.0142,
      "step": 10964
    },
    {
      "epoch": 20.23084025854109,
      "grad_norm": 0.47471538186073303,
      "learning_rate": 2.9679333333333336e-05,
      "loss": 0.0133,
      "step": 10965
    },
    {
      "epoch": 20.23268698060942,
      "grad_norm": 0.612973690032959,
      "learning_rate": 2.9678999999999998e-05,
      "loss": 0.0109,
      "step": 10966
    },
    {
      "epoch": 20.234533702677748,
      "grad_norm": 0.4560556411743164,
      "learning_rate": 2.9678666666666668e-05,
      "loss": 0.0105,
      "step": 10967
    },
    {
      "epoch": 20.236380424746077,
      "grad_norm": 0.25839492678642273,
      "learning_rate": 2.9678333333333334e-05,
      "loss": 0.0099,
      "step": 10968
    },
    {
      "epoch": 20.238227146814406,
      "grad_norm": 0.16658927500247955,
      "learning_rate": 2.9678e-05,
      "loss": 0.0078,
      "step": 10969
    },
    {
      "epoch": 20.24007386888273,
      "grad_norm": 0.37365013360977173,
      "learning_rate": 2.967766666666667e-05,
      "loss": 0.0072,
      "step": 10970
    },
    {
      "epoch": 20.24192059095106,
      "grad_norm": 0.2552949786186218,
      "learning_rate": 2.9677333333333335e-05,
      "loss": 0.0802,
      "step": 10971
    },
    {
      "epoch": 20.24376731301939,
      "grad_norm": 0.37917381525039673,
      "learning_rate": 2.9677e-05,
      "loss": 0.0102,
      "step": 10972
    },
    {
      "epoch": 20.24561403508772,
      "grad_norm": 0.2963426411151886,
      "learning_rate": 2.9676666666666667e-05,
      "loss": 0.0123,
      "step": 10973
    },
    {
      "epoch": 20.24746075715605,
      "grad_norm": 0.20354008674621582,
      "learning_rate": 2.9676333333333336e-05,
      "loss": 0.0096,
      "step": 10974
    },
    {
      "epoch": 20.249307479224377,
      "grad_norm": 0.18017128109931946,
      "learning_rate": 2.9676e-05,
      "loss": 0.0049,
      "step": 10975
    },
    {
      "epoch": 20.251154201292707,
      "grad_norm": 0.5300956964492798,
      "learning_rate": 2.9675666666666668e-05,
      "loss": 0.0078,
      "step": 10976
    },
    {
      "epoch": 20.253000923361036,
      "grad_norm": 0.31972944736480713,
      "learning_rate": 2.9675333333333337e-05,
      "loss": 0.0069,
      "step": 10977
    },
    {
      "epoch": 20.25484764542936,
      "grad_norm": 0.28348520398139954,
      "learning_rate": 2.9675e-05,
      "loss": 0.0096,
      "step": 10978
    },
    {
      "epoch": 20.25669436749769,
      "grad_norm": 0.4289419949054718,
      "learning_rate": 2.967466666666667e-05,
      "loss": 0.0149,
      "step": 10979
    },
    {
      "epoch": 20.25854108956602,
      "grad_norm": 0.3754739463329315,
      "learning_rate": 2.9674333333333335e-05,
      "loss": 0.0074,
      "step": 10980
    },
    {
      "epoch": 20.26038781163435,
      "grad_norm": 0.2263016402721405,
      "learning_rate": 2.9674e-05,
      "loss": 0.0058,
      "step": 10981
    },
    {
      "epoch": 20.262234533702678,
      "grad_norm": 0.2127392590045929,
      "learning_rate": 2.9673666666666667e-05,
      "loss": 0.0095,
      "step": 10982
    },
    {
      "epoch": 20.264081255771007,
      "grad_norm": 0.2642723321914673,
      "learning_rate": 2.9673333333333336e-05,
      "loss": 0.0076,
      "step": 10983
    },
    {
      "epoch": 20.265927977839336,
      "grad_norm": 0.7602730393409729,
      "learning_rate": 2.9673e-05,
      "loss": 0.0221,
      "step": 10984
    },
    {
      "epoch": 20.267774699907665,
      "grad_norm": 0.36623716354370117,
      "learning_rate": 2.9672666666666668e-05,
      "loss": 0.0107,
      "step": 10985
    },
    {
      "epoch": 20.26962142197599,
      "grad_norm": 0.3910073935985565,
      "learning_rate": 2.9672333333333334e-05,
      "loss": 0.0084,
      "step": 10986
    },
    {
      "epoch": 20.27146814404432,
      "grad_norm": 0.412586510181427,
      "learning_rate": 2.9672e-05,
      "loss": 0.0073,
      "step": 10987
    },
    {
      "epoch": 20.27331486611265,
      "grad_norm": 0.3259273171424866,
      "learning_rate": 2.967166666666667e-05,
      "loss": 0.0054,
      "step": 10988
    },
    {
      "epoch": 20.27516158818098,
      "grad_norm": 0.15527020394802094,
      "learning_rate": 2.9671333333333332e-05,
      "loss": 0.0077,
      "step": 10989
    },
    {
      "epoch": 20.277008310249307,
      "grad_norm": 0.34989339113235474,
      "learning_rate": 2.9671e-05,
      "loss": 0.0196,
      "step": 10990
    },
    {
      "epoch": 20.278855032317637,
      "grad_norm": 0.5065251588821411,
      "learning_rate": 2.9670666666666667e-05,
      "loss": 0.1234,
      "step": 10991
    },
    {
      "epoch": 20.280701754385966,
      "grad_norm": 0.5200304388999939,
      "learning_rate": 2.9670333333333333e-05,
      "loss": 0.201,
      "step": 10992
    },
    {
      "epoch": 20.282548476454295,
      "grad_norm": 0.5644885897636414,
      "learning_rate": 2.967e-05,
      "loss": 0.1444,
      "step": 10993
    },
    {
      "epoch": 20.284395198522624,
      "grad_norm": 0.5528456568717957,
      "learning_rate": 2.9669666666666668e-05,
      "loss": 0.1003,
      "step": 10994
    },
    {
      "epoch": 20.28624192059095,
      "grad_norm": 0.5606066584587097,
      "learning_rate": 2.9669333333333334e-05,
      "loss": 0.0624,
      "step": 10995
    },
    {
      "epoch": 20.28808864265928,
      "grad_norm": 0.5903345346450806,
      "learning_rate": 2.9669e-05,
      "loss": 0.0414,
      "step": 10996
    },
    {
      "epoch": 20.289935364727608,
      "grad_norm": 0.471769779920578,
      "learning_rate": 2.966866666666667e-05,
      "loss": 0.0666,
      "step": 10997
    },
    {
      "epoch": 20.291782086795937,
      "grad_norm": 0.5256743431091309,
      "learning_rate": 2.9668333333333332e-05,
      "loss": 0.0724,
      "step": 10998
    },
    {
      "epoch": 20.293628808864266,
      "grad_norm": 0.41306188702583313,
      "learning_rate": 2.9668e-05,
      "loss": 0.0388,
      "step": 10999
    },
    {
      "epoch": 20.295475530932595,
      "grad_norm": 0.46824705600738525,
      "learning_rate": 2.9667666666666667e-05,
      "loss": 0.0516,
      "step": 11000
    },
    {
      "epoch": 20.295475530932595,
      "eval_cer": 0.1074992871400057,
      "eval_loss": 0.30857551097869873,
      "eval_runtime": 16.2156,
      "eval_samples_per_second": 18.747,
      "eval_steps_per_second": 0.617,
      "eval_wer": 0.37010744435917114,
      "step": 11000
    },
    {
      "epoch": 20.297322253000925,
      "grad_norm": 0.4235368072986603,
      "learning_rate": 2.9667333333333333e-05,
      "loss": 0.0344,
      "step": 11001
    },
    {
      "epoch": 20.299168975069254,
      "grad_norm": 0.6765577793121338,
      "learning_rate": 2.9667000000000002e-05,
      "loss": 0.0277,
      "step": 11002
    },
    {
      "epoch": 20.30101569713758,
      "grad_norm": 0.276685893535614,
      "learning_rate": 2.966666666666667e-05,
      "loss": 0.0208,
      "step": 11003
    },
    {
      "epoch": 20.30286241920591,
      "grad_norm": 0.2812119722366333,
      "learning_rate": 2.9666333333333334e-05,
      "loss": 0.024,
      "step": 11004
    },
    {
      "epoch": 20.304709141274238,
      "grad_norm": 0.33644866943359375,
      "learning_rate": 2.9666e-05,
      "loss": 0.028,
      "step": 11005
    },
    {
      "epoch": 20.306555863342567,
      "grad_norm": 0.198531374335289,
      "learning_rate": 2.966566666666667e-05,
      "loss": 0.0174,
      "step": 11006
    },
    {
      "epoch": 20.308402585410896,
      "grad_norm": 0.19303059577941895,
      "learning_rate": 2.9665333333333332e-05,
      "loss": 0.0096,
      "step": 11007
    },
    {
      "epoch": 20.310249307479225,
      "grad_norm": 0.26009589433670044,
      "learning_rate": 2.9665e-05,
      "loss": 0.0146,
      "step": 11008
    },
    {
      "epoch": 20.312096029547554,
      "grad_norm": 0.34054481983184814,
      "learning_rate": 2.9664666666666667e-05,
      "loss": 0.0158,
      "step": 11009
    },
    {
      "epoch": 20.313942751615883,
      "grad_norm": 0.1274450719356537,
      "learning_rate": 2.9664333333333333e-05,
      "loss": 0.0059,
      "step": 11010
    },
    {
      "epoch": 20.31578947368421,
      "grad_norm": 0.41350147128105164,
      "learning_rate": 2.9664000000000003e-05,
      "loss": 0.0063,
      "step": 11011
    },
    {
      "epoch": 20.317636195752538,
      "grad_norm": 0.39728933572769165,
      "learning_rate": 2.966366666666667e-05,
      "loss": 0.0108,
      "step": 11012
    },
    {
      "epoch": 20.319482917820867,
      "grad_norm": 0.3444744646549225,
      "learning_rate": 2.9663333333333334e-05,
      "loss": 0.0109,
      "step": 11013
    },
    {
      "epoch": 20.321329639889196,
      "grad_norm": 0.30115723609924316,
      "learning_rate": 2.9663e-05,
      "loss": 0.0098,
      "step": 11014
    },
    {
      "epoch": 20.323176361957525,
      "grad_norm": 0.12758363783359528,
      "learning_rate": 2.966266666666667e-05,
      "loss": 0.0034,
      "step": 11015
    },
    {
      "epoch": 20.325023084025855,
      "grad_norm": 0.32861289381980896,
      "learning_rate": 2.9662333333333332e-05,
      "loss": 0.01,
      "step": 11016
    },
    {
      "epoch": 20.326869806094184,
      "grad_norm": 0.4190202057361603,
      "learning_rate": 2.9662e-05,
      "loss": 0.0337,
      "step": 11017
    },
    {
      "epoch": 20.328716528162513,
      "grad_norm": 0.24281994998455048,
      "learning_rate": 2.9661666666666664e-05,
      "loss": 0.0061,
      "step": 11018
    },
    {
      "epoch": 20.330563250230842,
      "grad_norm": 0.2888421416282654,
      "learning_rate": 2.9661333333333333e-05,
      "loss": 0.0131,
      "step": 11019
    },
    {
      "epoch": 20.332409972299168,
      "grad_norm": 0.27400413155555725,
      "learning_rate": 2.9661000000000003e-05,
      "loss": 0.0076,
      "step": 11020
    },
    {
      "epoch": 20.334256694367497,
      "grad_norm": 0.35427457094192505,
      "learning_rate": 2.9660666666666665e-05,
      "loss": 0.0121,
      "step": 11021
    },
    {
      "epoch": 20.336103416435826,
      "grad_norm": 0.27263957262039185,
      "learning_rate": 2.9660333333333335e-05,
      "loss": 0.0069,
      "step": 11022
    },
    {
      "epoch": 20.337950138504155,
      "grad_norm": 0.15267528593540192,
      "learning_rate": 2.966e-05,
      "loss": 0.0055,
      "step": 11023
    },
    {
      "epoch": 20.339796860572484,
      "grad_norm": 0.21431012451648712,
      "learning_rate": 2.9659666666666667e-05,
      "loss": 0.0055,
      "step": 11024
    },
    {
      "epoch": 20.341643582640813,
      "grad_norm": 0.5318107604980469,
      "learning_rate": 2.9659333333333332e-05,
      "loss": 0.0098,
      "step": 11025
    },
    {
      "epoch": 20.343490304709142,
      "grad_norm": 0.07205729931592941,
      "learning_rate": 2.9659000000000002e-05,
      "loss": 0.0018,
      "step": 11026
    },
    {
      "epoch": 20.34533702677747,
      "grad_norm": 0.22027286887168884,
      "learning_rate": 2.9658666666666668e-05,
      "loss": 0.0069,
      "step": 11027
    },
    {
      "epoch": 20.347183748845797,
      "grad_norm": 0.21012485027313232,
      "learning_rate": 2.9658333333333334e-05,
      "loss": 0.006,
      "step": 11028
    },
    {
      "epoch": 20.349030470914126,
      "grad_norm": 0.15638655424118042,
      "learning_rate": 2.9658000000000003e-05,
      "loss": 0.0065,
      "step": 11029
    },
    {
      "epoch": 20.350877192982455,
      "grad_norm": 0.5567883253097534,
      "learning_rate": 2.9657666666666666e-05,
      "loss": 0.0186,
      "step": 11030
    },
    {
      "epoch": 20.352723915050785,
      "grad_norm": 0.5367173552513123,
      "learning_rate": 2.9657333333333335e-05,
      "loss": 0.007,
      "step": 11031
    },
    {
      "epoch": 20.354570637119114,
      "grad_norm": 0.8661971688270569,
      "learning_rate": 2.9657e-05,
      "loss": 0.0171,
      "step": 11032
    },
    {
      "epoch": 20.356417359187443,
      "grad_norm": 0.4001052677631378,
      "learning_rate": 2.9656666666666667e-05,
      "loss": 0.0109,
      "step": 11033
    },
    {
      "epoch": 20.358264081255772,
      "grad_norm": 0.24180084466934204,
      "learning_rate": 2.9656333333333333e-05,
      "loss": 0.0051,
      "step": 11034
    },
    {
      "epoch": 20.3601108033241,
      "grad_norm": 0.5838049650192261,
      "learning_rate": 2.9656000000000002e-05,
      "loss": 0.0061,
      "step": 11035
    },
    {
      "epoch": 20.361957525392427,
      "grad_norm": 0.35159727931022644,
      "learning_rate": 2.9655666666666668e-05,
      "loss": 0.011,
      "step": 11036
    },
    {
      "epoch": 20.363804247460756,
      "grad_norm": 0.3918987214565277,
      "learning_rate": 2.9655333333333334e-05,
      "loss": 0.0112,
      "step": 11037
    },
    {
      "epoch": 20.365650969529085,
      "grad_norm": 0.1534157693386078,
      "learning_rate": 2.9655000000000003e-05,
      "loss": 0.0059,
      "step": 11038
    },
    {
      "epoch": 20.367497691597414,
      "grad_norm": 0.6456652879714966,
      "learning_rate": 2.9654666666666666e-05,
      "loss": 0.02,
      "step": 11039
    },
    {
      "epoch": 20.369344413665743,
      "grad_norm": 0.2508772909641266,
      "learning_rate": 2.9654333333333335e-05,
      "loss": 0.0045,
      "step": 11040
    },
    {
      "epoch": 20.371191135734072,
      "grad_norm": 0.5599384307861328,
      "learning_rate": 2.9654e-05,
      "loss": 0.1466,
      "step": 11041
    },
    {
      "epoch": 20.3730378578024,
      "grad_norm": 0.5619744062423706,
      "learning_rate": 2.9653666666666667e-05,
      "loss": 0.1297,
      "step": 11042
    },
    {
      "epoch": 20.37488457987073,
      "grad_norm": 0.5407826900482178,
      "learning_rate": 2.9653333333333333e-05,
      "loss": 0.1643,
      "step": 11043
    },
    {
      "epoch": 20.37673130193906,
      "grad_norm": 0.7417288422584534,
      "learning_rate": 2.9653000000000002e-05,
      "loss": 0.0849,
      "step": 11044
    },
    {
      "epoch": 20.378578024007385,
      "grad_norm": 0.357843279838562,
      "learning_rate": 2.9652666666666668e-05,
      "loss": 0.0829,
      "step": 11045
    },
    {
      "epoch": 20.380424746075715,
      "grad_norm": 0.5893060564994812,
      "learning_rate": 2.9652333333333334e-05,
      "loss": 0.1218,
      "step": 11046
    },
    {
      "epoch": 20.382271468144044,
      "grad_norm": 0.35972005128860474,
      "learning_rate": 2.9652e-05,
      "loss": 0.042,
      "step": 11047
    },
    {
      "epoch": 20.384118190212373,
      "grad_norm": 0.4211483895778656,
      "learning_rate": 2.9651666666666666e-05,
      "loss": 0.055,
      "step": 11048
    },
    {
      "epoch": 20.385964912280702,
      "grad_norm": 0.5754496455192566,
      "learning_rate": 2.9651333333333335e-05,
      "loss": 0.0691,
      "step": 11049
    },
    {
      "epoch": 20.38781163434903,
      "grad_norm": 0.38568854331970215,
      "learning_rate": 2.9650999999999998e-05,
      "loss": 0.0362,
      "step": 11050
    },
    {
      "epoch": 20.38965835641736,
      "grad_norm": 0.2110164612531662,
      "learning_rate": 2.9650666666666667e-05,
      "loss": 0.0279,
      "step": 11051
    },
    {
      "epoch": 20.39150507848569,
      "grad_norm": 0.5566184520721436,
      "learning_rate": 2.9650333333333336e-05,
      "loss": 0.0469,
      "step": 11052
    },
    {
      "epoch": 20.393351800554015,
      "grad_norm": 0.38283446431159973,
      "learning_rate": 2.965e-05,
      "loss": 0.0326,
      "step": 11053
    },
    {
      "epoch": 20.395198522622344,
      "grad_norm": 0.2858939468860626,
      "learning_rate": 2.964966666666667e-05,
      "loss": 0.036,
      "step": 11054
    },
    {
      "epoch": 20.397045244690673,
      "grad_norm": 0.22442243993282318,
      "learning_rate": 2.9649333333333334e-05,
      "loss": 0.0156,
      "step": 11055
    },
    {
      "epoch": 20.398891966759003,
      "grad_norm": 0.2950870096683502,
      "learning_rate": 2.9649e-05,
      "loss": 0.0242,
      "step": 11056
    },
    {
      "epoch": 20.40073868882733,
      "grad_norm": 0.20051699876785278,
      "learning_rate": 2.9648666666666666e-05,
      "loss": 0.0054,
      "step": 11057
    },
    {
      "epoch": 20.40258541089566,
      "grad_norm": 0.15492866933345795,
      "learning_rate": 2.9648333333333335e-05,
      "loss": 0.0067,
      "step": 11058
    },
    {
      "epoch": 20.40443213296399,
      "grad_norm": 0.34917473793029785,
      "learning_rate": 2.9647999999999998e-05,
      "loss": 0.0182,
      "step": 11059
    },
    {
      "epoch": 20.40627885503232,
      "grad_norm": 0.18603120744228363,
      "learning_rate": 2.9647666666666667e-05,
      "loss": 0.008,
      "step": 11060
    },
    {
      "epoch": 20.408125577100645,
      "grad_norm": 0.3769189715385437,
      "learning_rate": 2.9647333333333337e-05,
      "loss": 0.0078,
      "step": 11061
    },
    {
      "epoch": 20.409972299168974,
      "grad_norm": 0.18756088614463806,
      "learning_rate": 2.9647e-05,
      "loss": 0.0203,
      "step": 11062
    },
    {
      "epoch": 20.411819021237303,
      "grad_norm": 0.25873249769210815,
      "learning_rate": 2.964666666666667e-05,
      "loss": 0.0101,
      "step": 11063
    },
    {
      "epoch": 20.413665743305632,
      "grad_norm": 0.2606721520423889,
      "learning_rate": 2.9646333333333334e-05,
      "loss": 0.0067,
      "step": 11064
    },
    {
      "epoch": 20.41551246537396,
      "grad_norm": 0.277649849653244,
      "learning_rate": 2.9646e-05,
      "loss": 0.0089,
      "step": 11065
    },
    {
      "epoch": 20.41735918744229,
      "grad_norm": 0.2441006600856781,
      "learning_rate": 2.9645666666666666e-05,
      "loss": 0.0083,
      "step": 11066
    },
    {
      "epoch": 20.41920590951062,
      "grad_norm": 0.17656227946281433,
      "learning_rate": 2.9645333333333336e-05,
      "loss": 0.0067,
      "step": 11067
    },
    {
      "epoch": 20.42105263157895,
      "grad_norm": 0.18494001030921936,
      "learning_rate": 2.9644999999999998e-05,
      "loss": 0.0085,
      "step": 11068
    },
    {
      "epoch": 20.422899353647278,
      "grad_norm": 0.27652615308761597,
      "learning_rate": 2.9644666666666668e-05,
      "loss": 0.0077,
      "step": 11069
    },
    {
      "epoch": 20.424746075715603,
      "grad_norm": 0.2581216096878052,
      "learning_rate": 2.9644333333333337e-05,
      "loss": 0.0068,
      "step": 11070
    },
    {
      "epoch": 20.426592797783933,
      "grad_norm": 0.2612696588039398,
      "learning_rate": 2.9644e-05,
      "loss": 0.0083,
      "step": 11071
    },
    {
      "epoch": 20.42843951985226,
      "grad_norm": 0.07239628583192825,
      "learning_rate": 2.964366666666667e-05,
      "loss": 0.002,
      "step": 11072
    },
    {
      "epoch": 20.43028624192059,
      "grad_norm": 0.29392409324645996,
      "learning_rate": 2.9643333333333335e-05,
      "loss": 0.0112,
      "step": 11073
    },
    {
      "epoch": 20.43213296398892,
      "grad_norm": 0.5558052062988281,
      "learning_rate": 2.9643e-05,
      "loss": 0.0073,
      "step": 11074
    },
    {
      "epoch": 20.43397968605725,
      "grad_norm": 0.23736676573753357,
      "learning_rate": 2.9642666666666667e-05,
      "loss": 0.0074,
      "step": 11075
    },
    {
      "epoch": 20.43582640812558,
      "grad_norm": 0.29687535762786865,
      "learning_rate": 2.9642333333333336e-05,
      "loss": 0.0087,
      "step": 11076
    },
    {
      "epoch": 20.437673130193907,
      "grad_norm": 0.13251857459545135,
      "learning_rate": 2.9642000000000002e-05,
      "loss": 0.0049,
      "step": 11077
    },
    {
      "epoch": 20.439519852262233,
      "grad_norm": 0.23094314336776733,
      "learning_rate": 2.9641666666666668e-05,
      "loss": 0.0053,
      "step": 11078
    },
    {
      "epoch": 20.441366574330562,
      "grad_norm": 0.283113956451416,
      "learning_rate": 2.9641333333333334e-05,
      "loss": 0.0108,
      "step": 11079
    },
    {
      "epoch": 20.44321329639889,
      "grad_norm": 0.32321351766586304,
      "learning_rate": 2.9641e-05,
      "loss": 0.0084,
      "step": 11080
    },
    {
      "epoch": 20.44506001846722,
      "grad_norm": 0.47560450434684753,
      "learning_rate": 2.964066666666667e-05,
      "loss": 0.0148,
      "step": 11081
    },
    {
      "epoch": 20.44690674053555,
      "grad_norm": 0.4224452078342438,
      "learning_rate": 2.964033333333333e-05,
      "loss": 0.0163,
      "step": 11082
    },
    {
      "epoch": 20.44875346260388,
      "grad_norm": 0.216268852353096,
      "learning_rate": 2.964e-05,
      "loss": 0.0041,
      "step": 11083
    },
    {
      "epoch": 20.450600184672208,
      "grad_norm": 0.7119795680046082,
      "learning_rate": 2.9639666666666667e-05,
      "loss": 0.0102,
      "step": 11084
    },
    {
      "epoch": 20.452446906740537,
      "grad_norm": 0.3483209013938904,
      "learning_rate": 2.9639333333333333e-05,
      "loss": 0.0078,
      "step": 11085
    },
    {
      "epoch": 20.454293628808863,
      "grad_norm": 0.3994562327861786,
      "learning_rate": 2.9639000000000002e-05,
      "loss": 0.0054,
      "step": 11086
    },
    {
      "epoch": 20.45614035087719,
      "grad_norm": 0.2146139144897461,
      "learning_rate": 2.9638666666666668e-05,
      "loss": 0.0079,
      "step": 11087
    },
    {
      "epoch": 20.45798707294552,
      "grad_norm": 0.3482670783996582,
      "learning_rate": 2.9638333333333334e-05,
      "loss": 0.007,
      "step": 11088
    },
    {
      "epoch": 20.45983379501385,
      "grad_norm": 0.20434588193893433,
      "learning_rate": 2.9638e-05,
      "loss": 0.0048,
      "step": 11089
    },
    {
      "epoch": 20.46168051708218,
      "grad_norm": 0.137790709733963,
      "learning_rate": 2.963766666666667e-05,
      "loss": 0.0045,
      "step": 11090
    },
    {
      "epoch": 20.46352723915051,
      "grad_norm": 0.6156465411186218,
      "learning_rate": 2.963733333333333e-05,
      "loss": 0.1216,
      "step": 11091
    },
    {
      "epoch": 20.465373961218837,
      "grad_norm": 0.4434069097042084,
      "learning_rate": 2.9637e-05,
      "loss": 0.0953,
      "step": 11092
    },
    {
      "epoch": 20.467220683287167,
      "grad_norm": 0.4450667202472687,
      "learning_rate": 2.9636666666666667e-05,
      "loss": 0.0877,
      "step": 11093
    },
    {
      "epoch": 20.469067405355496,
      "grad_norm": 0.4277558922767639,
      "learning_rate": 2.9636333333333333e-05,
      "loss": 0.0637,
      "step": 11094
    },
    {
      "epoch": 20.47091412742382,
      "grad_norm": 0.5416181087493896,
      "learning_rate": 2.9636000000000002e-05,
      "loss": 0.102,
      "step": 11095
    },
    {
      "epoch": 20.47276084949215,
      "grad_norm": 0.5285162329673767,
      "learning_rate": 2.9635666666666668e-05,
      "loss": 0.0814,
      "step": 11096
    },
    {
      "epoch": 20.47460757156048,
      "grad_norm": 0.4584277868270874,
      "learning_rate": 2.9635333333333334e-05,
      "loss": 0.0593,
      "step": 11097
    },
    {
      "epoch": 20.47645429362881,
      "grad_norm": 0.3677422106266022,
      "learning_rate": 2.9635e-05,
      "loss": 0.0497,
      "step": 11098
    },
    {
      "epoch": 20.478301015697138,
      "grad_norm": 0.4277951717376709,
      "learning_rate": 2.963466666666667e-05,
      "loss": 0.0475,
      "step": 11099
    },
    {
      "epoch": 20.480147737765467,
      "grad_norm": 0.31714293360710144,
      "learning_rate": 2.9634333333333332e-05,
      "loss": 0.0422,
      "step": 11100
    },
    {
      "epoch": 20.481994459833796,
      "grad_norm": 0.26784515380859375,
      "learning_rate": 2.9634e-05,
      "loss": 0.0283,
      "step": 11101
    },
    {
      "epoch": 20.483841181902125,
      "grad_norm": 0.44170352816581726,
      "learning_rate": 2.963366666666667e-05,
      "loss": 0.0497,
      "step": 11102
    },
    {
      "epoch": 20.48568790397045,
      "grad_norm": 1.118957757949829,
      "learning_rate": 2.9633333333333333e-05,
      "loss": 0.1057,
      "step": 11103
    },
    {
      "epoch": 20.48753462603878,
      "grad_norm": 0.6414685845375061,
      "learning_rate": 2.9633000000000002e-05,
      "loss": 0.0244,
      "step": 11104
    },
    {
      "epoch": 20.48938134810711,
      "grad_norm": 0.27597081661224365,
      "learning_rate": 2.963266666666667e-05,
      "loss": 0.0329,
      "step": 11105
    },
    {
      "epoch": 20.49122807017544,
      "grad_norm": 0.27471840381622314,
      "learning_rate": 2.9632333333333334e-05,
      "loss": 0.0144,
      "step": 11106
    },
    {
      "epoch": 20.493074792243767,
      "grad_norm": 0.3149142265319824,
      "learning_rate": 2.9632e-05,
      "loss": 0.0455,
      "step": 11107
    },
    {
      "epoch": 20.494921514312097,
      "grad_norm": 0.19767074286937714,
      "learning_rate": 2.9631666666666666e-05,
      "loss": 0.0082,
      "step": 11108
    },
    {
      "epoch": 20.496768236380426,
      "grad_norm": 0.21706773340702057,
      "learning_rate": 2.9631333333333332e-05,
      "loss": 0.0102,
      "step": 11109
    },
    {
      "epoch": 20.498614958448755,
      "grad_norm": 0.2238224446773529,
      "learning_rate": 2.9631e-05,
      "loss": 0.0078,
      "step": 11110
    },
    {
      "epoch": 20.50046168051708,
      "grad_norm": 0.30950242280960083,
      "learning_rate": 2.9630666666666667e-05,
      "loss": 0.0295,
      "step": 11111
    },
    {
      "epoch": 20.50230840258541,
      "grad_norm": 0.35704588890075684,
      "learning_rate": 2.9630333333333333e-05,
      "loss": 0.0102,
      "step": 11112
    },
    {
      "epoch": 20.50415512465374,
      "grad_norm": 0.16436102986335754,
      "learning_rate": 2.9630000000000003e-05,
      "loss": 0.0106,
      "step": 11113
    },
    {
      "epoch": 20.506001846722068,
      "grad_norm": 0.1160278245806694,
      "learning_rate": 2.9629666666666665e-05,
      "loss": 0.0083,
      "step": 11114
    },
    {
      "epoch": 20.507848568790397,
      "grad_norm": 0.15938252210617065,
      "learning_rate": 2.9629333333333334e-05,
      "loss": 0.0056,
      "step": 11115
    },
    {
      "epoch": 20.509695290858726,
      "grad_norm": 0.1849542111158371,
      "learning_rate": 2.9629e-05,
      "loss": 0.0063,
      "step": 11116
    },
    {
      "epoch": 20.511542012927055,
      "grad_norm": 0.20571953058242798,
      "learning_rate": 2.9628666666666666e-05,
      "loss": 0.005,
      "step": 11117
    },
    {
      "epoch": 20.513388734995385,
      "grad_norm": 0.361489474773407,
      "learning_rate": 2.9628333333333332e-05,
      "loss": 0.014,
      "step": 11118
    },
    {
      "epoch": 20.51523545706371,
      "grad_norm": 0.37601128220558167,
      "learning_rate": 2.9628e-05,
      "loss": 0.0427,
      "step": 11119
    },
    {
      "epoch": 20.51708217913204,
      "grad_norm": 0.3030451536178589,
      "learning_rate": 2.9627666666666668e-05,
      "loss": 0.0216,
      "step": 11120
    },
    {
      "epoch": 20.51892890120037,
      "grad_norm": 0.20974400639533997,
      "learning_rate": 2.9627333333333333e-05,
      "loss": 0.0072,
      "step": 11121
    },
    {
      "epoch": 20.520775623268698,
      "grad_norm": 0.47172707319259644,
      "learning_rate": 2.9627000000000003e-05,
      "loss": 0.0201,
      "step": 11122
    },
    {
      "epoch": 20.522622345337027,
      "grad_norm": 0.4215916097164154,
      "learning_rate": 2.9626666666666665e-05,
      "loss": 0.0121,
      "step": 11123
    },
    {
      "epoch": 20.524469067405356,
      "grad_norm": 0.27536407113075256,
      "learning_rate": 2.9626333333333335e-05,
      "loss": 0.0067,
      "step": 11124
    },
    {
      "epoch": 20.526315789473685,
      "grad_norm": 0.1958393007516861,
      "learning_rate": 2.9626e-05,
      "loss": 0.0049,
      "step": 11125
    },
    {
      "epoch": 20.528162511542014,
      "grad_norm": 0.35039401054382324,
      "learning_rate": 2.9625666666666667e-05,
      "loss": 0.0093,
      "step": 11126
    },
    {
      "epoch": 20.530009233610343,
      "grad_norm": 0.32164257764816284,
      "learning_rate": 2.9625333333333336e-05,
      "loss": 0.011,
      "step": 11127
    },
    {
      "epoch": 20.53185595567867,
      "grad_norm": 0.4968523681163788,
      "learning_rate": 2.9625000000000002e-05,
      "loss": 0.0134,
      "step": 11128
    },
    {
      "epoch": 20.533702677746998,
      "grad_norm": 0.21272744238376617,
      "learning_rate": 2.9624666666666668e-05,
      "loss": 0.0042,
      "step": 11129
    },
    {
      "epoch": 20.535549399815327,
      "grad_norm": 0.5690340995788574,
      "learning_rate": 2.9624333333333334e-05,
      "loss": 0.0095,
      "step": 11130
    },
    {
      "epoch": 20.537396121883656,
      "grad_norm": 0.2647596597671509,
      "learning_rate": 2.9624000000000003e-05,
      "loss": 0.0067,
      "step": 11131
    },
    {
      "epoch": 20.539242843951985,
      "grad_norm": 0.33502963185310364,
      "learning_rate": 2.9623666666666666e-05,
      "loss": 0.0091,
      "step": 11132
    },
    {
      "epoch": 20.541089566020315,
      "grad_norm": 0.4991375207901001,
      "learning_rate": 2.9623333333333335e-05,
      "loss": 0.0105,
      "step": 11133
    },
    {
      "epoch": 20.542936288088644,
      "grad_norm": 0.8269820809364319,
      "learning_rate": 2.9623e-05,
      "loss": 0.0098,
      "step": 11134
    },
    {
      "epoch": 20.544783010156973,
      "grad_norm": 0.13691353797912598,
      "learning_rate": 2.9622666666666667e-05,
      "loss": 0.0061,
      "step": 11135
    },
    {
      "epoch": 20.5466297322253,
      "grad_norm": 0.28916093707084656,
      "learning_rate": 2.9622333333333336e-05,
      "loss": 0.0084,
      "step": 11136
    },
    {
      "epoch": 20.548476454293628,
      "grad_norm": 0.5460681319236755,
      "learning_rate": 2.9622000000000002e-05,
      "loss": 0.0335,
      "step": 11137
    },
    {
      "epoch": 20.550323176361957,
      "grad_norm": 0.7758886814117432,
      "learning_rate": 2.9621666666666668e-05,
      "loss": 0.0142,
      "step": 11138
    },
    {
      "epoch": 20.552169898430286,
      "grad_norm": 0.2703908383846283,
      "learning_rate": 2.9621333333333334e-05,
      "loss": 0.0045,
      "step": 11139
    },
    {
      "epoch": 20.554016620498615,
      "grad_norm": 0.3861091434955597,
      "learning_rate": 2.9621e-05,
      "loss": 0.0199,
      "step": 11140
    },
    {
      "epoch": 20.555863342566944,
      "grad_norm": 0.5069757699966431,
      "learning_rate": 2.9620666666666666e-05,
      "loss": 0.1421,
      "step": 11141
    },
    {
      "epoch": 20.557710064635273,
      "grad_norm": 0.5670900344848633,
      "learning_rate": 2.9620333333333335e-05,
      "loss": 0.1451,
      "step": 11142
    },
    {
      "epoch": 20.559556786703602,
      "grad_norm": 0.7019637823104858,
      "learning_rate": 2.9619999999999998e-05,
      "loss": 0.1063,
      "step": 11143
    },
    {
      "epoch": 20.56140350877193,
      "grad_norm": 0.49798068404197693,
      "learning_rate": 2.9619666666666667e-05,
      "loss": 0.0769,
      "step": 11144
    },
    {
      "epoch": 20.563250230840257,
      "grad_norm": 0.5604007840156555,
      "learning_rate": 2.9619333333333336e-05,
      "loss": 0.1241,
      "step": 11145
    },
    {
      "epoch": 20.565096952908586,
      "grad_norm": 0.4972505271434784,
      "learning_rate": 2.9619e-05,
      "loss": 0.0856,
      "step": 11146
    },
    {
      "epoch": 20.566943674976915,
      "grad_norm": 0.470397412776947,
      "learning_rate": 2.9618666666666668e-05,
      "loss": 0.0645,
      "step": 11147
    },
    {
      "epoch": 20.568790397045245,
      "grad_norm": 0.5640853643417358,
      "learning_rate": 2.9618333333333334e-05,
      "loss": 0.0911,
      "step": 11148
    },
    {
      "epoch": 20.570637119113574,
      "grad_norm": 0.4170297682285309,
      "learning_rate": 2.9618e-05,
      "loss": 0.0676,
      "step": 11149
    },
    {
      "epoch": 20.572483841181903,
      "grad_norm": 0.3935358226299286,
      "learning_rate": 2.9617666666666666e-05,
      "loss": 0.0803,
      "step": 11150
    },
    {
      "epoch": 20.574330563250232,
      "grad_norm": 0.3682790696620941,
      "learning_rate": 2.9617333333333335e-05,
      "loss": 0.0415,
      "step": 11151
    },
    {
      "epoch": 20.57617728531856,
      "grad_norm": 0.28599610924720764,
      "learning_rate": 2.9617e-05,
      "loss": 0.0252,
      "step": 11152
    },
    {
      "epoch": 20.578024007386887,
      "grad_norm": 0.31602001190185547,
      "learning_rate": 2.9616666666666667e-05,
      "loss": 0.0568,
      "step": 11153
    },
    {
      "epoch": 20.579870729455216,
      "grad_norm": 0.45824578404426575,
      "learning_rate": 2.9616333333333336e-05,
      "loss": 0.0582,
      "step": 11154
    },
    {
      "epoch": 20.581717451523545,
      "grad_norm": 0.35455620288848877,
      "learning_rate": 2.9616e-05,
      "loss": 0.0232,
      "step": 11155
    },
    {
      "epoch": 20.583564173591874,
      "grad_norm": 0.247396320104599,
      "learning_rate": 2.961566666666667e-05,
      "loss": 0.0116,
      "step": 11156
    },
    {
      "epoch": 20.585410895660203,
      "grad_norm": 0.17564378678798676,
      "learning_rate": 2.9615333333333334e-05,
      "loss": 0.0103,
      "step": 11157
    },
    {
      "epoch": 20.587257617728532,
      "grad_norm": 0.5125971436500549,
      "learning_rate": 2.9615e-05,
      "loss": 0.0322,
      "step": 11158
    },
    {
      "epoch": 20.58910433979686,
      "grad_norm": 0.18899650871753693,
      "learning_rate": 2.9614666666666666e-05,
      "loss": 0.0117,
      "step": 11159
    },
    {
      "epoch": 20.59095106186519,
      "grad_norm": 0.35293349623680115,
      "learning_rate": 2.9614333333333335e-05,
      "loss": 0.0161,
      "step": 11160
    },
    {
      "epoch": 20.592797783933516,
      "grad_norm": 0.15498006343841553,
      "learning_rate": 2.9614e-05,
      "loss": 0.0053,
      "step": 11161
    },
    {
      "epoch": 20.594644506001845,
      "grad_norm": 0.24944724142551422,
      "learning_rate": 2.9613666666666667e-05,
      "loss": 0.0109,
      "step": 11162
    },
    {
      "epoch": 20.596491228070175,
      "grad_norm": 0.2556169033050537,
      "learning_rate": 2.9613333333333337e-05,
      "loss": 0.0103,
      "step": 11163
    },
    {
      "epoch": 20.598337950138504,
      "grad_norm": 0.72690749168396,
      "learning_rate": 2.9613e-05,
      "loss": 0.0188,
      "step": 11164
    },
    {
      "epoch": 20.600184672206833,
      "grad_norm": 0.2584100663661957,
      "learning_rate": 2.961266666666667e-05,
      "loss": 0.0133,
      "step": 11165
    },
    {
      "epoch": 20.602031394275162,
      "grad_norm": 0.2549022436141968,
      "learning_rate": 2.9612333333333334e-05,
      "loss": 0.0111,
      "step": 11166
    },
    {
      "epoch": 20.60387811634349,
      "grad_norm": 0.6710214018821716,
      "learning_rate": 2.9612e-05,
      "loss": 0.0151,
      "step": 11167
    },
    {
      "epoch": 20.60572483841182,
      "grad_norm": 0.43617209792137146,
      "learning_rate": 2.9611666666666666e-05,
      "loss": 0.0092,
      "step": 11168
    },
    {
      "epoch": 20.607571560480146,
      "grad_norm": 0.245264932513237,
      "learning_rate": 2.9611333333333332e-05,
      "loss": 0.0119,
      "step": 11169
    },
    {
      "epoch": 20.609418282548475,
      "grad_norm": 0.20647655427455902,
      "learning_rate": 2.9611e-05,
      "loss": 0.007,
      "step": 11170
    },
    {
      "epoch": 20.611265004616804,
      "grad_norm": 0.4248480498790741,
      "learning_rate": 2.9610666666666668e-05,
      "loss": 0.0096,
      "step": 11171
    },
    {
      "epoch": 20.613111726685133,
      "grad_norm": 0.36499524116516113,
      "learning_rate": 2.9610333333333333e-05,
      "loss": 0.0136,
      "step": 11172
    },
    {
      "epoch": 20.614958448753463,
      "grad_norm": 0.5641077756881714,
      "learning_rate": 2.961e-05,
      "loss": 0.0111,
      "step": 11173
    },
    {
      "epoch": 20.61680517082179,
      "grad_norm": 0.3565385043621063,
      "learning_rate": 2.960966666666667e-05,
      "loss": 0.0135,
      "step": 11174
    },
    {
      "epoch": 20.61865189289012,
      "grad_norm": 0.5080560445785522,
      "learning_rate": 2.960933333333333e-05,
      "loss": 0.0158,
      "step": 11175
    },
    {
      "epoch": 20.62049861495845,
      "grad_norm": 0.12717655301094055,
      "learning_rate": 2.9609e-05,
      "loss": 0.0054,
      "step": 11176
    },
    {
      "epoch": 20.62234533702678,
      "grad_norm": 0.2625749707221985,
      "learning_rate": 2.960866666666667e-05,
      "loss": 0.009,
      "step": 11177
    },
    {
      "epoch": 20.624192059095105,
      "grad_norm": 0.6564839482307434,
      "learning_rate": 2.9608333333333332e-05,
      "loss": 0.0474,
      "step": 11178
    },
    {
      "epoch": 20.626038781163434,
      "grad_norm": 0.20139797031879425,
      "learning_rate": 2.9608000000000002e-05,
      "loss": 0.0083,
      "step": 11179
    },
    {
      "epoch": 20.627885503231763,
      "grad_norm": 0.8026975989341736,
      "learning_rate": 2.9607666666666668e-05,
      "loss": 0.0114,
      "step": 11180
    },
    {
      "epoch": 20.629732225300092,
      "grad_norm": 0.24883051216602325,
      "learning_rate": 2.9607333333333334e-05,
      "loss": 0.006,
      "step": 11181
    },
    {
      "epoch": 20.63157894736842,
      "grad_norm": 0.33926114439964294,
      "learning_rate": 2.9607e-05,
      "loss": 0.0101,
      "step": 11182
    },
    {
      "epoch": 20.63342566943675,
      "grad_norm": 0.8261702656745911,
      "learning_rate": 2.960666666666667e-05,
      "loss": 0.0232,
      "step": 11183
    },
    {
      "epoch": 20.63527239150508,
      "grad_norm": 0.5510255694389343,
      "learning_rate": 2.960633333333333e-05,
      "loss": 0.0088,
      "step": 11184
    },
    {
      "epoch": 20.63711911357341,
      "grad_norm": 0.49745795130729675,
      "learning_rate": 2.9606e-05,
      "loss": 0.0144,
      "step": 11185
    },
    {
      "epoch": 20.638965835641734,
      "grad_norm": 0.30604270100593567,
      "learning_rate": 2.960566666666667e-05,
      "loss": 0.0101,
      "step": 11186
    },
    {
      "epoch": 20.640812557710063,
      "grad_norm": 0.8259574770927429,
      "learning_rate": 2.9605333333333333e-05,
      "loss": 0.0117,
      "step": 11187
    },
    {
      "epoch": 20.642659279778393,
      "grad_norm": 0.3971351683139801,
      "learning_rate": 2.9605000000000002e-05,
      "loss": 0.0272,
      "step": 11188
    },
    {
      "epoch": 20.64450600184672,
      "grad_norm": 0.34049734473228455,
      "learning_rate": 2.9604666666666668e-05,
      "loss": 0.0072,
      "step": 11189
    },
    {
      "epoch": 20.64635272391505,
      "grad_norm": 0.3927757441997528,
      "learning_rate": 2.9604333333333334e-05,
      "loss": 0.0136,
      "step": 11190
    },
    {
      "epoch": 20.64819944598338,
      "grad_norm": 0.7999992370605469,
      "learning_rate": 2.9604e-05,
      "loss": 0.1988,
      "step": 11191
    },
    {
      "epoch": 20.65004616805171,
      "grad_norm": 0.7569670677185059,
      "learning_rate": 2.960366666666667e-05,
      "loss": 0.1572,
      "step": 11192
    },
    {
      "epoch": 20.65189289012004,
      "grad_norm": 0.5596485733985901,
      "learning_rate": 2.960333333333333e-05,
      "loss": 0.0999,
      "step": 11193
    },
    {
      "epoch": 20.653739612188367,
      "grad_norm": 0.45623841881752014,
      "learning_rate": 2.9603e-05,
      "loss": 0.0895,
      "step": 11194
    },
    {
      "epoch": 20.655586334256693,
      "grad_norm": 0.40664881467819214,
      "learning_rate": 2.960266666666667e-05,
      "loss": 0.1002,
      "step": 11195
    },
    {
      "epoch": 20.657433056325022,
      "grad_norm": 0.40975096821784973,
      "learning_rate": 2.9602333333333333e-05,
      "loss": 0.0723,
      "step": 11196
    },
    {
      "epoch": 20.65927977839335,
      "grad_norm": 0.3656075894832611,
      "learning_rate": 2.9602000000000002e-05,
      "loss": 0.0678,
      "step": 11197
    },
    {
      "epoch": 20.66112650046168,
      "grad_norm": 0.35924556851387024,
      "learning_rate": 2.9601666666666665e-05,
      "loss": 0.0438,
      "step": 11198
    },
    {
      "epoch": 20.66297322253001,
      "grad_norm": 0.3122684061527252,
      "learning_rate": 2.9601333333333334e-05,
      "loss": 0.0412,
      "step": 11199
    },
    {
      "epoch": 20.66481994459834,
      "grad_norm": 0.49023181200027466,
      "learning_rate": 2.9601e-05,
      "loss": 0.0986,
      "step": 11200
    },
    {
      "epoch": 20.666666666666668,
      "grad_norm": 0.3593541979789734,
      "learning_rate": 2.9600666666666666e-05,
      "loss": 0.0508,
      "step": 11201
    },
    {
      "epoch": 20.668513388734997,
      "grad_norm": 0.450988233089447,
      "learning_rate": 2.9600333333333335e-05,
      "loss": 0.0432,
      "step": 11202
    },
    {
      "epoch": 20.670360110803323,
      "grad_norm": 0.44862470030784607,
      "learning_rate": 2.96e-05,
      "loss": 0.0232,
      "step": 11203
    },
    {
      "epoch": 20.67220683287165,
      "grad_norm": 0.22612938284873962,
      "learning_rate": 2.9599666666666667e-05,
      "loss": 0.0186,
      "step": 11204
    },
    {
      "epoch": 20.67405355493998,
      "grad_norm": 0.30688098073005676,
      "learning_rate": 2.9599333333333333e-05,
      "loss": 0.0207,
      "step": 11205
    },
    {
      "epoch": 20.67590027700831,
      "grad_norm": 0.21418403089046478,
      "learning_rate": 2.9599000000000002e-05,
      "loss": 0.0097,
      "step": 11206
    },
    {
      "epoch": 20.67774699907664,
      "grad_norm": 0.4343891143798828,
      "learning_rate": 2.9598666666666665e-05,
      "loss": 0.0291,
      "step": 11207
    },
    {
      "epoch": 20.67959372114497,
      "grad_norm": 0.7051688432693481,
      "learning_rate": 2.9598333333333334e-05,
      "loss": 0.0312,
      "step": 11208
    },
    {
      "epoch": 20.681440443213297,
      "grad_norm": 0.2645828127861023,
      "learning_rate": 2.9598e-05,
      "loss": 0.0093,
      "step": 11209
    },
    {
      "epoch": 20.683287165281627,
      "grad_norm": 0.32652685046195984,
      "learning_rate": 2.9597666666666666e-05,
      "loss": 0.012,
      "step": 11210
    },
    {
      "epoch": 20.685133887349952,
      "grad_norm": 0.18169867992401123,
      "learning_rate": 2.9597333333333335e-05,
      "loss": 0.0102,
      "step": 11211
    },
    {
      "epoch": 20.68698060941828,
      "grad_norm": 0.28606170415878296,
      "learning_rate": 2.9597e-05,
      "loss": 0.0186,
      "step": 11212
    },
    {
      "epoch": 20.68882733148661,
      "grad_norm": 0.4741019606590271,
      "learning_rate": 2.9596666666666667e-05,
      "loss": 0.0109,
      "step": 11213
    },
    {
      "epoch": 20.69067405355494,
      "grad_norm": 0.380709171295166,
      "learning_rate": 2.9596333333333333e-05,
      "loss": 0.0111,
      "step": 11214
    },
    {
      "epoch": 20.69252077562327,
      "grad_norm": 0.2992388606071472,
      "learning_rate": 2.9596000000000003e-05,
      "loss": 0.0114,
      "step": 11215
    },
    {
      "epoch": 20.694367497691598,
      "grad_norm": 0.4169728755950928,
      "learning_rate": 2.9595666666666665e-05,
      "loss": 0.011,
      "step": 11216
    },
    {
      "epoch": 20.696214219759927,
      "grad_norm": 0.8805220127105713,
      "learning_rate": 2.9595333333333334e-05,
      "loss": 0.0061,
      "step": 11217
    },
    {
      "epoch": 20.698060941828256,
      "grad_norm": 0.12588036060333252,
      "learning_rate": 2.9595e-05,
      "loss": 0.0044,
      "step": 11218
    },
    {
      "epoch": 20.69990766389658,
      "grad_norm": 0.5865361094474792,
      "learning_rate": 2.9594666666666666e-05,
      "loss": 0.0191,
      "step": 11219
    },
    {
      "epoch": 20.70175438596491,
      "grad_norm": 0.1972983181476593,
      "learning_rate": 2.9594333333333336e-05,
      "loss": 0.0075,
      "step": 11220
    },
    {
      "epoch": 20.70360110803324,
      "grad_norm": 0.15591812133789062,
      "learning_rate": 2.9594e-05,
      "loss": 0.004,
      "step": 11221
    },
    {
      "epoch": 20.70544783010157,
      "grad_norm": 0.29317209124565125,
      "learning_rate": 2.9593666666666668e-05,
      "loss": 0.0115,
      "step": 11222
    },
    {
      "epoch": 20.7072945521699,
      "grad_norm": 0.4665067195892334,
      "learning_rate": 2.9593333333333333e-05,
      "loss": 0.015,
      "step": 11223
    },
    {
      "epoch": 20.709141274238227,
      "grad_norm": 0.48972606658935547,
      "learning_rate": 2.9593000000000003e-05,
      "loss": 0.0102,
      "step": 11224
    },
    {
      "epoch": 20.710987996306557,
      "grad_norm": 0.2213616967201233,
      "learning_rate": 2.9592666666666665e-05,
      "loss": 0.0093,
      "step": 11225
    },
    {
      "epoch": 20.712834718374886,
      "grad_norm": 0.6633093357086182,
      "learning_rate": 2.9592333333333335e-05,
      "loss": 0.0152,
      "step": 11226
    },
    {
      "epoch": 20.714681440443215,
      "grad_norm": 0.8084264993667603,
      "learning_rate": 2.9592000000000004e-05,
      "loss": 0.0149,
      "step": 11227
    },
    {
      "epoch": 20.71652816251154,
      "grad_norm": 0.2868703007698059,
      "learning_rate": 2.9591666666666667e-05,
      "loss": 0.0089,
      "step": 11228
    },
    {
      "epoch": 20.71837488457987,
      "grad_norm": 0.3528291881084442,
      "learning_rate": 2.9591333333333336e-05,
      "loss": 0.0092,
      "step": 11229
    },
    {
      "epoch": 20.7202216066482,
      "grad_norm": 0.26255181431770325,
      "learning_rate": 2.9591e-05,
      "loss": 0.0113,
      "step": 11230
    },
    {
      "epoch": 20.722068328716528,
      "grad_norm": 0.14820496737957,
      "learning_rate": 2.9590666666666668e-05,
      "loss": 0.0047,
      "step": 11231
    },
    {
      "epoch": 20.723915050784857,
      "grad_norm": 0.5360812544822693,
      "learning_rate": 2.9590333333333334e-05,
      "loss": 0.012,
      "step": 11232
    },
    {
      "epoch": 20.725761772853186,
      "grad_norm": 0.18604981899261475,
      "learning_rate": 2.959e-05,
      "loss": 0.0051,
      "step": 11233
    },
    {
      "epoch": 20.727608494921515,
      "grad_norm": 0.4018900692462921,
      "learning_rate": 2.9589666666666666e-05,
      "loss": 0.0111,
      "step": 11234
    },
    {
      "epoch": 20.729455216989845,
      "grad_norm": 0.818670392036438,
      "learning_rate": 2.9589333333333335e-05,
      "loss": 0.0166,
      "step": 11235
    },
    {
      "epoch": 20.73130193905817,
      "grad_norm": 0.243108868598938,
      "learning_rate": 2.9589e-05,
      "loss": 0.0046,
      "step": 11236
    },
    {
      "epoch": 20.7331486611265,
      "grad_norm": 0.13450288772583008,
      "learning_rate": 2.9588666666666667e-05,
      "loss": 0.0051,
      "step": 11237
    },
    {
      "epoch": 20.73499538319483,
      "grad_norm": 0.40568941831588745,
      "learning_rate": 2.9588333333333336e-05,
      "loss": 0.0113,
      "step": 11238
    },
    {
      "epoch": 20.736842105263158,
      "grad_norm": 0.4287243187427521,
      "learning_rate": 2.9588e-05,
      "loss": 0.013,
      "step": 11239
    },
    {
      "epoch": 20.738688827331487,
      "grad_norm": 0.4001462459564209,
      "learning_rate": 2.9587666666666668e-05,
      "loss": 0.0114,
      "step": 11240
    },
    {
      "epoch": 20.740535549399816,
      "grad_norm": 0.7578105926513672,
      "learning_rate": 2.9587333333333334e-05,
      "loss": 0.2375,
      "step": 11241
    },
    {
      "epoch": 20.742382271468145,
      "grad_norm": 0.43286678194999695,
      "learning_rate": 2.9587e-05,
      "loss": 0.1031,
      "step": 11242
    },
    {
      "epoch": 20.744228993536474,
      "grad_norm": 0.4050258994102478,
      "learning_rate": 2.9586666666666666e-05,
      "loss": 0.075,
      "step": 11243
    },
    {
      "epoch": 20.746075715604803,
      "grad_norm": 0.42576155066490173,
      "learning_rate": 2.9586333333333335e-05,
      "loss": 0.093,
      "step": 11244
    },
    {
      "epoch": 20.74792243767313,
      "grad_norm": 0.8448416590690613,
      "learning_rate": 2.9586e-05,
      "loss": 0.1502,
      "step": 11245
    },
    {
      "epoch": 20.749769159741458,
      "grad_norm": 0.46882346272468567,
      "learning_rate": 2.9585666666666667e-05,
      "loss": 0.0749,
      "step": 11246
    },
    {
      "epoch": 20.751615881809787,
      "grad_norm": 0.4115670323371887,
      "learning_rate": 2.9585333333333336e-05,
      "loss": 0.0778,
      "step": 11247
    },
    {
      "epoch": 20.753462603878116,
      "grad_norm": 0.49006929993629456,
      "learning_rate": 2.9585e-05,
      "loss": 0.064,
      "step": 11248
    },
    {
      "epoch": 20.755309325946445,
      "grad_norm": 0.3789594769477844,
      "learning_rate": 2.9584666666666668e-05,
      "loss": 0.0396,
      "step": 11249
    },
    {
      "epoch": 20.757156048014775,
      "grad_norm": 0.5852547883987427,
      "learning_rate": 2.9584333333333334e-05,
      "loss": 0.043,
      "step": 11250
    },
    {
      "epoch": 20.759002770083104,
      "grad_norm": 0.4392510950565338,
      "learning_rate": 2.9584e-05,
      "loss": 0.0386,
      "step": 11251
    },
    {
      "epoch": 20.760849492151433,
      "grad_norm": 0.338418573141098,
      "learning_rate": 2.958366666666667e-05,
      "loss": 0.0723,
      "step": 11252
    },
    {
      "epoch": 20.76269621421976,
      "grad_norm": 0.2645232379436493,
      "learning_rate": 2.9583333333333335e-05,
      "loss": 0.0524,
      "step": 11253
    },
    {
      "epoch": 20.764542936288088,
      "grad_norm": 0.24157041311264038,
      "learning_rate": 2.9583e-05,
      "loss": 0.0259,
      "step": 11254
    },
    {
      "epoch": 20.766389658356417,
      "grad_norm": 0.38394519686698914,
      "learning_rate": 2.9582666666666667e-05,
      "loss": 0.0432,
      "step": 11255
    },
    {
      "epoch": 20.768236380424746,
      "grad_norm": 0.3139944076538086,
      "learning_rate": 2.9582333333333336e-05,
      "loss": 0.0203,
      "step": 11256
    },
    {
      "epoch": 20.770083102493075,
      "grad_norm": 0.46563613414764404,
      "learning_rate": 2.9582e-05,
      "loss": 0.031,
      "step": 11257
    },
    {
      "epoch": 20.771929824561404,
      "grad_norm": 0.30651748180389404,
      "learning_rate": 2.958166666666667e-05,
      "loss": 0.0157,
      "step": 11258
    },
    {
      "epoch": 20.773776546629733,
      "grad_norm": 0.22431708872318268,
      "learning_rate": 2.958133333333333e-05,
      "loss": 0.0104,
      "step": 11259
    },
    {
      "epoch": 20.775623268698062,
      "grad_norm": 0.5509371161460876,
      "learning_rate": 2.9581e-05,
      "loss": 0.0187,
      "step": 11260
    },
    {
      "epoch": 20.777469990766388,
      "grad_norm": 0.22594115138053894,
      "learning_rate": 2.958066666666667e-05,
      "loss": 0.0075,
      "step": 11261
    },
    {
      "epoch": 20.779316712834717,
      "grad_norm": 0.19666936993598938,
      "learning_rate": 2.9580333333333332e-05,
      "loss": 0.0088,
      "step": 11262
    },
    {
      "epoch": 20.781163434903046,
      "grad_norm": 0.22689498960971832,
      "learning_rate": 2.958e-05,
      "loss": 0.0057,
      "step": 11263
    },
    {
      "epoch": 20.783010156971375,
      "grad_norm": 0.394515722990036,
      "learning_rate": 2.9579666666666667e-05,
      "loss": 0.0163,
      "step": 11264
    },
    {
      "epoch": 20.784856879039705,
      "grad_norm": 0.16962333023548126,
      "learning_rate": 2.9579333333333333e-05,
      "loss": 0.0056,
      "step": 11265
    },
    {
      "epoch": 20.786703601108034,
      "grad_norm": 0.20703470706939697,
      "learning_rate": 2.9579e-05,
      "loss": 0.016,
      "step": 11266
    },
    {
      "epoch": 20.788550323176363,
      "grad_norm": 0.29677820205688477,
      "learning_rate": 2.957866666666667e-05,
      "loss": 0.0085,
      "step": 11267
    },
    {
      "epoch": 20.790397045244692,
      "grad_norm": 0.19197024405002594,
      "learning_rate": 2.9578333333333334e-05,
      "loss": 0.0052,
      "step": 11268
    },
    {
      "epoch": 20.792243767313018,
      "grad_norm": 0.4533921182155609,
      "learning_rate": 2.9578e-05,
      "loss": 0.0119,
      "step": 11269
    },
    {
      "epoch": 20.794090489381347,
      "grad_norm": 0.22189201414585114,
      "learning_rate": 2.957766666666667e-05,
      "loss": 0.0053,
      "step": 11270
    },
    {
      "epoch": 20.795937211449676,
      "grad_norm": 0.5494603514671326,
      "learning_rate": 2.9577333333333332e-05,
      "loss": 0.0431,
      "step": 11271
    },
    {
      "epoch": 20.797783933518005,
      "grad_norm": 0.3049589991569519,
      "learning_rate": 2.9577e-05,
      "loss": 0.0086,
      "step": 11272
    },
    {
      "epoch": 20.799630655586334,
      "grad_norm": 0.23531177639961243,
      "learning_rate": 2.9576666666666668e-05,
      "loss": 0.0098,
      "step": 11273
    },
    {
      "epoch": 20.801477377654663,
      "grad_norm": 0.24381548166275024,
      "learning_rate": 2.9576333333333333e-05,
      "loss": 0.0053,
      "step": 11274
    },
    {
      "epoch": 20.803324099722992,
      "grad_norm": 2.4563019275665283,
      "learning_rate": 2.9576e-05,
      "loss": 0.0231,
      "step": 11275
    },
    {
      "epoch": 20.80517082179132,
      "grad_norm": 0.33990874886512756,
      "learning_rate": 2.957566666666667e-05,
      "loss": 0.0053,
      "step": 11276
    },
    {
      "epoch": 20.80701754385965,
      "grad_norm": 0.35326576232910156,
      "learning_rate": 2.9575333333333335e-05,
      "loss": 0.0083,
      "step": 11277
    },
    {
      "epoch": 20.808864265927976,
      "grad_norm": 0.34734803438186646,
      "learning_rate": 2.9575e-05,
      "loss": 0.0132,
      "step": 11278
    },
    {
      "epoch": 20.810710987996305,
      "grad_norm": 0.8471444845199585,
      "learning_rate": 2.957466666666667e-05,
      "loss": 0.0208,
      "step": 11279
    },
    {
      "epoch": 20.812557710064635,
      "grad_norm": 0.4880143105983734,
      "learning_rate": 2.9574333333333332e-05,
      "loss": 0.0061,
      "step": 11280
    },
    {
      "epoch": 20.814404432132964,
      "grad_norm": 0.23279136419296265,
      "learning_rate": 2.9574000000000002e-05,
      "loss": 0.006,
      "step": 11281
    },
    {
      "epoch": 20.816251154201293,
      "grad_norm": 0.19613172113895416,
      "learning_rate": 2.9573666666666668e-05,
      "loss": 0.0063,
      "step": 11282
    },
    {
      "epoch": 20.818097876269622,
      "grad_norm": 0.20421376824378967,
      "learning_rate": 2.9573333333333334e-05,
      "loss": 0.0047,
      "step": 11283
    },
    {
      "epoch": 20.81994459833795,
      "grad_norm": 0.3376994729042053,
      "learning_rate": 2.9573e-05,
      "loss": 0.0045,
      "step": 11284
    },
    {
      "epoch": 20.82179132040628,
      "grad_norm": 0.39231762290000916,
      "learning_rate": 2.957266666666667e-05,
      "loss": 0.0188,
      "step": 11285
    },
    {
      "epoch": 20.823638042474606,
      "grad_norm": 0.10870233923196793,
      "learning_rate": 2.9572333333333335e-05,
      "loss": 0.0026,
      "step": 11286
    },
    {
      "epoch": 20.825484764542935,
      "grad_norm": 0.697699248790741,
      "learning_rate": 2.9572e-05,
      "loss": 0.0022,
      "step": 11287
    },
    {
      "epoch": 20.827331486611264,
      "grad_norm": 0.3991001844406128,
      "learning_rate": 2.957166666666667e-05,
      "loss": 0.0127,
      "step": 11288
    },
    {
      "epoch": 20.829178208679593,
      "grad_norm": 0.9220645427703857,
      "learning_rate": 2.9571333333333333e-05,
      "loss": 0.0284,
      "step": 11289
    },
    {
      "epoch": 20.831024930747922,
      "grad_norm": 0.41163426637649536,
      "learning_rate": 2.9571000000000002e-05,
      "loss": 0.0243,
      "step": 11290
    },
    {
      "epoch": 20.83287165281625,
      "grad_norm": 0.9080430865287781,
      "learning_rate": 2.9570666666666665e-05,
      "loss": 0.1517,
      "step": 11291
    },
    {
      "epoch": 20.83471837488458,
      "grad_norm": 0.43430566787719727,
      "learning_rate": 2.9570333333333334e-05,
      "loss": 0.1335,
      "step": 11292
    },
    {
      "epoch": 20.83656509695291,
      "grad_norm": 0.5547497272491455,
      "learning_rate": 2.957e-05,
      "loss": 0.0931,
      "step": 11293
    },
    {
      "epoch": 20.83841181902124,
      "grad_norm": 0.5292896032333374,
      "learning_rate": 2.9569666666666666e-05,
      "loss": 0.1158,
      "step": 11294
    },
    {
      "epoch": 20.840258541089565,
      "grad_norm": 0.4992031753063202,
      "learning_rate": 2.9569333333333335e-05,
      "loss": 0.099,
      "step": 11295
    },
    {
      "epoch": 20.842105263157894,
      "grad_norm": 0.4766106903553009,
      "learning_rate": 2.9569e-05,
      "loss": 0.0715,
      "step": 11296
    },
    {
      "epoch": 20.843951985226223,
      "grad_norm": 0.5213163495063782,
      "learning_rate": 2.9568666666666667e-05,
      "loss": 0.0692,
      "step": 11297
    },
    {
      "epoch": 20.845798707294552,
      "grad_norm": 0.5166041851043701,
      "learning_rate": 2.9568333333333333e-05,
      "loss": 0.0706,
      "step": 11298
    },
    {
      "epoch": 20.84764542936288,
      "grad_norm": 0.3912571966648102,
      "learning_rate": 2.9568000000000002e-05,
      "loss": 0.0737,
      "step": 11299
    },
    {
      "epoch": 20.84949215143121,
      "grad_norm": 0.47118571400642395,
      "learning_rate": 2.9567666666666665e-05,
      "loss": 0.0371,
      "step": 11300
    },
    {
      "epoch": 20.85133887349954,
      "grad_norm": 0.49245890974998474,
      "learning_rate": 2.9567333333333334e-05,
      "loss": 0.0704,
      "step": 11301
    },
    {
      "epoch": 20.85318559556787,
      "grad_norm": 0.33564913272857666,
      "learning_rate": 2.9567000000000003e-05,
      "loss": 0.0356,
      "step": 11302
    },
    {
      "epoch": 20.855032317636194,
      "grad_norm": 0.36780816316604614,
      "learning_rate": 2.9566666666666666e-05,
      "loss": 0.0806,
      "step": 11303
    },
    {
      "epoch": 20.856879039704523,
      "grad_norm": 0.6674977540969849,
      "learning_rate": 2.9566333333333335e-05,
      "loss": 0.0328,
      "step": 11304
    },
    {
      "epoch": 20.858725761772853,
      "grad_norm": 0.497979998588562,
      "learning_rate": 2.9566e-05,
      "loss": 0.0261,
      "step": 11305
    },
    {
      "epoch": 20.86057248384118,
      "grad_norm": 0.23606769740581512,
      "learning_rate": 2.9565666666666667e-05,
      "loss": 0.0222,
      "step": 11306
    },
    {
      "epoch": 20.86241920590951,
      "grad_norm": 0.2154080867767334,
      "learning_rate": 2.9565333333333333e-05,
      "loss": 0.0083,
      "step": 11307
    },
    {
      "epoch": 20.86426592797784,
      "grad_norm": 0.23044618964195251,
      "learning_rate": 2.9565000000000002e-05,
      "loss": 0.0141,
      "step": 11308
    },
    {
      "epoch": 20.86611265004617,
      "grad_norm": 0.7383080124855042,
      "learning_rate": 2.9564666666666665e-05,
      "loss": 0.0095,
      "step": 11309
    },
    {
      "epoch": 20.8679593721145,
      "grad_norm": 0.21569529175758362,
      "learning_rate": 2.9564333333333334e-05,
      "loss": 0.0129,
      "step": 11310
    },
    {
      "epoch": 20.869806094182824,
      "grad_norm": 0.2047816663980484,
      "learning_rate": 2.9564000000000004e-05,
      "loss": 0.0076,
      "step": 11311
    },
    {
      "epoch": 20.871652816251153,
      "grad_norm": 0.3197352886199951,
      "learning_rate": 2.9563666666666666e-05,
      "loss": 0.0184,
      "step": 11312
    },
    {
      "epoch": 20.873499538319482,
      "grad_norm": 0.3038652539253235,
      "learning_rate": 2.9563333333333335e-05,
      "loss": 0.0109,
      "step": 11313
    },
    {
      "epoch": 20.87534626038781,
      "grad_norm": 0.6552397608757019,
      "learning_rate": 2.9563e-05,
      "loss": 0.0222,
      "step": 11314
    },
    {
      "epoch": 20.87719298245614,
      "grad_norm": 0.5822405815124512,
      "learning_rate": 2.9562666666666667e-05,
      "loss": 0.0196,
      "step": 11315
    },
    {
      "epoch": 20.87903970452447,
      "grad_norm": 0.12892888486385345,
      "learning_rate": 2.9562333333333333e-05,
      "loss": 0.0051,
      "step": 11316
    },
    {
      "epoch": 20.8808864265928,
      "grad_norm": 0.1823236346244812,
      "learning_rate": 2.9562000000000003e-05,
      "loss": 0.0082,
      "step": 11317
    },
    {
      "epoch": 20.882733148661128,
      "grad_norm": 0.17489029467105865,
      "learning_rate": 2.956166666666667e-05,
      "loss": 0.0063,
      "step": 11318
    },
    {
      "epoch": 20.884579870729453,
      "grad_norm": 0.24932153522968292,
      "learning_rate": 2.9561333333333334e-05,
      "loss": 0.0132,
      "step": 11319
    },
    {
      "epoch": 20.886426592797783,
      "grad_norm": 0.1961432546377182,
      "learning_rate": 2.9561e-05,
      "loss": 0.0066,
      "step": 11320
    },
    {
      "epoch": 20.88827331486611,
      "grad_norm": 0.426115483045578,
      "learning_rate": 2.9560666666666666e-05,
      "loss": 0.0112,
      "step": 11321
    },
    {
      "epoch": 20.89012003693444,
      "grad_norm": 0.29435256123542786,
      "learning_rate": 2.9560333333333336e-05,
      "loss": 0.0102,
      "step": 11322
    },
    {
      "epoch": 20.89196675900277,
      "grad_norm": 0.31088516116142273,
      "learning_rate": 2.9559999999999998e-05,
      "loss": 0.0079,
      "step": 11323
    },
    {
      "epoch": 20.8938134810711,
      "grad_norm": 1.155611515045166,
      "learning_rate": 2.9559666666666668e-05,
      "loss": 0.0215,
      "step": 11324
    },
    {
      "epoch": 20.89566020313943,
      "grad_norm": 0.31312328577041626,
      "learning_rate": 2.9559333333333333e-05,
      "loss": 0.0121,
      "step": 11325
    },
    {
      "epoch": 20.897506925207757,
      "grad_norm": 0.27958038449287415,
      "learning_rate": 2.9559e-05,
      "loss": 0.0043,
      "step": 11326
    },
    {
      "epoch": 20.899353647276087,
      "grad_norm": 0.18234357237815857,
      "learning_rate": 2.955866666666667e-05,
      "loss": 0.0049,
      "step": 11327
    },
    {
      "epoch": 20.901200369344412,
      "grad_norm": 0.39806634187698364,
      "learning_rate": 2.9558333333333335e-05,
      "loss": 0.0139,
      "step": 11328
    },
    {
      "epoch": 20.90304709141274,
      "grad_norm": 0.5366166234016418,
      "learning_rate": 2.9558e-05,
      "loss": 0.0209,
      "step": 11329
    },
    {
      "epoch": 20.90489381348107,
      "grad_norm": 0.39424633979797363,
      "learning_rate": 2.9557666666666667e-05,
      "loss": 0.0112,
      "step": 11330
    },
    {
      "epoch": 20.9067405355494,
      "grad_norm": 0.4483903646469116,
      "learning_rate": 2.9557333333333336e-05,
      "loss": 0.0151,
      "step": 11331
    },
    {
      "epoch": 20.90858725761773,
      "grad_norm": 0.30543461441993713,
      "learning_rate": 2.9557e-05,
      "loss": 0.0119,
      "step": 11332
    },
    {
      "epoch": 20.910433979686058,
      "grad_norm": 0.24922345578670502,
      "learning_rate": 2.9556666666666668e-05,
      "loss": 0.0057,
      "step": 11333
    },
    {
      "epoch": 20.912280701754387,
      "grad_norm": 0.1164715439081192,
      "learning_rate": 2.9556333333333334e-05,
      "loss": 0.0034,
      "step": 11334
    },
    {
      "epoch": 20.914127423822716,
      "grad_norm": 0.1792861968278885,
      "learning_rate": 2.9556e-05,
      "loss": 0.004,
      "step": 11335
    },
    {
      "epoch": 20.91597414589104,
      "grad_norm": 0.5263696908950806,
      "learning_rate": 2.955566666666667e-05,
      "loss": 0.0165,
      "step": 11336
    },
    {
      "epoch": 20.91782086795937,
      "grad_norm": 0.36606866121292114,
      "learning_rate": 2.9555333333333335e-05,
      "loss": 0.0095,
      "step": 11337
    },
    {
      "epoch": 20.9196675900277,
      "grad_norm": 0.31590327620506287,
      "learning_rate": 2.9555e-05,
      "loss": 0.0078,
      "step": 11338
    },
    {
      "epoch": 20.92151431209603,
      "grad_norm": 0.5138630270957947,
      "learning_rate": 2.9554666666666667e-05,
      "loss": 0.0203,
      "step": 11339
    },
    {
      "epoch": 20.92336103416436,
      "grad_norm": 0.23742391169071198,
      "learning_rate": 2.9554333333333336e-05,
      "loss": 0.0203,
      "step": 11340
    },
    {
      "epoch": 20.925207756232687,
      "grad_norm": 0.6078528165817261,
      "learning_rate": 2.9554e-05,
      "loss": 0.1362,
      "step": 11341
    },
    {
      "epoch": 20.927054478301017,
      "grad_norm": 0.7384791970252991,
      "learning_rate": 2.9553666666666668e-05,
      "loss": 0.1074,
      "step": 11342
    },
    {
      "epoch": 20.928901200369346,
      "grad_norm": 0.5082293152809143,
      "learning_rate": 2.9553333333333334e-05,
      "loss": 0.1437,
      "step": 11343
    },
    {
      "epoch": 20.930747922437675,
      "grad_norm": 0.4633669853210449,
      "learning_rate": 2.9553e-05,
      "loss": 0.0835,
      "step": 11344
    },
    {
      "epoch": 20.932594644506,
      "grad_norm": 0.5880398750305176,
      "learning_rate": 2.955266666666667e-05,
      "loss": 0.1012,
      "step": 11345
    },
    {
      "epoch": 20.93444136657433,
      "grad_norm": 0.5600230097770691,
      "learning_rate": 2.9552333333333335e-05,
      "loss": 0.0635,
      "step": 11346
    },
    {
      "epoch": 20.93628808864266,
      "grad_norm": 0.5349959135055542,
      "learning_rate": 2.9552e-05,
      "loss": 0.078,
      "step": 11347
    },
    {
      "epoch": 20.938134810710988,
      "grad_norm": 0.6552622318267822,
      "learning_rate": 2.9551666666666667e-05,
      "loss": 0.0464,
      "step": 11348
    },
    {
      "epoch": 20.939981532779317,
      "grad_norm": 1.903660774230957,
      "learning_rate": 2.9551333333333333e-05,
      "loss": 0.0537,
      "step": 11349
    },
    {
      "epoch": 20.941828254847646,
      "grad_norm": 0.5627049803733826,
      "learning_rate": 2.9551e-05,
      "loss": 0.0254,
      "step": 11350
    },
    {
      "epoch": 20.943674976915975,
      "grad_norm": 0.3178408741950989,
      "learning_rate": 2.9550666666666668e-05,
      "loss": 0.0695,
      "step": 11351
    },
    {
      "epoch": 20.945521698984304,
      "grad_norm": 0.32476845383644104,
      "learning_rate": 2.9550333333333334e-05,
      "loss": 0.0187,
      "step": 11352
    },
    {
      "epoch": 20.94736842105263,
      "grad_norm": 0.37900370359420776,
      "learning_rate": 2.955e-05,
      "loss": 0.0195,
      "step": 11353
    },
    {
      "epoch": 20.94921514312096,
      "grad_norm": 0.38062894344329834,
      "learning_rate": 2.954966666666667e-05,
      "loss": 0.0125,
      "step": 11354
    },
    {
      "epoch": 20.95106186518929,
      "grad_norm": 0.26303040981292725,
      "learning_rate": 2.9549333333333332e-05,
      "loss": 0.0125,
      "step": 11355
    },
    {
      "epoch": 20.952908587257618,
      "grad_norm": 0.36702829599380493,
      "learning_rate": 2.9549e-05,
      "loss": 0.0298,
      "step": 11356
    },
    {
      "epoch": 20.954755309325947,
      "grad_norm": 0.2526761591434479,
      "learning_rate": 2.9548666666666667e-05,
      "loss": 0.0101,
      "step": 11357
    },
    {
      "epoch": 20.956602031394276,
      "grad_norm": 0.3408384919166565,
      "learning_rate": 2.9548333333333333e-05,
      "loss": 0.0137,
      "step": 11358
    },
    {
      "epoch": 20.958448753462605,
      "grad_norm": 0.21575458347797394,
      "learning_rate": 2.9548e-05,
      "loss": 0.0101,
      "step": 11359
    },
    {
      "epoch": 20.960295475530934,
      "grad_norm": 0.16458436846733093,
      "learning_rate": 2.954766666666667e-05,
      "loss": 0.0086,
      "step": 11360
    },
    {
      "epoch": 20.96214219759926,
      "grad_norm": 0.4075638949871063,
      "learning_rate": 2.9547333333333334e-05,
      "loss": 0.0236,
      "step": 11361
    },
    {
      "epoch": 20.96398891966759,
      "grad_norm": 0.21881143748760223,
      "learning_rate": 2.9547e-05,
      "loss": 0.0091,
      "step": 11362
    },
    {
      "epoch": 20.965835641735918,
      "grad_norm": 0.18376675248146057,
      "learning_rate": 2.954666666666667e-05,
      "loss": 0.0093,
      "step": 11363
    },
    {
      "epoch": 20.967682363804247,
      "grad_norm": 0.12338406592607498,
      "learning_rate": 2.9546333333333332e-05,
      "loss": 0.0044,
      "step": 11364
    },
    {
      "epoch": 20.969529085872576,
      "grad_norm": 0.7155386209487915,
      "learning_rate": 2.9546e-05,
      "loss": 0.0099,
      "step": 11365
    },
    {
      "epoch": 20.971375807940905,
      "grad_norm": 0.1896418035030365,
      "learning_rate": 2.9545666666666667e-05,
      "loss": 0.0063,
      "step": 11366
    },
    {
      "epoch": 20.973222530009235,
      "grad_norm": 0.6268427968025208,
      "learning_rate": 2.9545333333333333e-05,
      "loss": 0.0114,
      "step": 11367
    },
    {
      "epoch": 20.975069252077564,
      "grad_norm": 0.25467798113822937,
      "learning_rate": 2.9545000000000003e-05,
      "loss": 0.0052,
      "step": 11368
    },
    {
      "epoch": 20.97691597414589,
      "grad_norm": 0.0892992615699768,
      "learning_rate": 2.954466666666667e-05,
      "loss": 0.0033,
      "step": 11369
    },
    {
      "epoch": 20.97876269621422,
      "grad_norm": 0.5027537941932678,
      "learning_rate": 2.9544333333333334e-05,
      "loss": 0.0097,
      "step": 11370
    },
    {
      "epoch": 20.980609418282548,
      "grad_norm": 0.4184074401855469,
      "learning_rate": 2.9544e-05,
      "loss": 0.0202,
      "step": 11371
    },
    {
      "epoch": 20.982456140350877,
      "grad_norm": 0.17174819111824036,
      "learning_rate": 2.954366666666667e-05,
      "loss": 0.0042,
      "step": 11372
    },
    {
      "epoch": 20.984302862419206,
      "grad_norm": 0.30898192524909973,
      "learning_rate": 2.9543333333333332e-05,
      "loss": 0.0131,
      "step": 11373
    },
    {
      "epoch": 20.986149584487535,
      "grad_norm": 0.3742638826370239,
      "learning_rate": 2.9543e-05,
      "loss": 0.0091,
      "step": 11374
    },
    {
      "epoch": 20.987996306555864,
      "grad_norm": 0.5427600741386414,
      "learning_rate": 2.9542666666666668e-05,
      "loss": 0.0128,
      "step": 11375
    },
    {
      "epoch": 20.989843028624193,
      "grad_norm": 0.2309504598379135,
      "learning_rate": 2.9542333333333333e-05,
      "loss": 0.009,
      "step": 11376
    },
    {
      "epoch": 20.991689750692522,
      "grad_norm": 0.3140806257724762,
      "learning_rate": 2.9542000000000003e-05,
      "loss": 0.0081,
      "step": 11377
    },
    {
      "epoch": 20.993536472760848,
      "grad_norm": 0.41816213726997375,
      "learning_rate": 2.954166666666667e-05,
      "loss": 0.006,
      "step": 11378
    },
    {
      "epoch": 20.995383194829177,
      "grad_norm": 0.3930540084838867,
      "learning_rate": 2.9541333333333335e-05,
      "loss": 0.0079,
      "step": 11379
    },
    {
      "epoch": 20.997229916897506,
      "grad_norm": 0.3547838628292084,
      "learning_rate": 2.9541e-05,
      "loss": 0.0126,
      "step": 11380
    },
    {
      "epoch": 20.999076638965835,
      "grad_norm": 0.40626728534698486,
      "learning_rate": 2.9540666666666667e-05,
      "loss": 0.0176,
      "step": 11381
    },
    {
      "epoch": 21.0,
      "grad_norm": 0.15839992463588715,
      "learning_rate": 2.9540333333333332e-05,
      "loss": 0.0027,
      "step": 11382
    },
    {
      "epoch": 21.00184672206833,
      "grad_norm": 0.7238561511039734,
      "learning_rate": 2.9540000000000002e-05,
      "loss": 0.1525,
      "step": 11383
    },
    {
      "epoch": 21.00369344413666,
      "grad_norm": 0.5545807480812073,
      "learning_rate": 2.9539666666666664e-05,
      "loss": 0.0998,
      "step": 11384
    },
    {
      "epoch": 21.005540166204987,
      "grad_norm": 0.521634578704834,
      "learning_rate": 2.9539333333333334e-05,
      "loss": 0.0941,
      "step": 11385
    },
    {
      "epoch": 21.007386888273317,
      "grad_norm": 0.44180136919021606,
      "learning_rate": 2.9539000000000003e-05,
      "loss": 0.0553,
      "step": 11386
    },
    {
      "epoch": 21.009233610341642,
      "grad_norm": 0.4523581862449646,
      "learning_rate": 2.9538666666666666e-05,
      "loss": 0.0617,
      "step": 11387
    },
    {
      "epoch": 21.01108033240997,
      "grad_norm": 0.6714923977851868,
      "learning_rate": 2.9538333333333335e-05,
      "loss": 0.0931,
      "step": 11388
    },
    {
      "epoch": 21.0129270544783,
      "grad_norm": 0.6149570345878601,
      "learning_rate": 2.9538e-05,
      "loss": 0.054,
      "step": 11389
    },
    {
      "epoch": 21.01477377654663,
      "grad_norm": 0.6411060690879822,
      "learning_rate": 2.9537666666666667e-05,
      "loss": 0.063,
      "step": 11390
    },
    {
      "epoch": 21.01662049861496,
      "grad_norm": 0.49264752864837646,
      "learning_rate": 2.9537333333333333e-05,
      "loss": 0.0485,
      "step": 11391
    },
    {
      "epoch": 21.018467220683288,
      "grad_norm": 0.6387502551078796,
      "learning_rate": 2.9537000000000002e-05,
      "loss": 0.0811,
      "step": 11392
    },
    {
      "epoch": 21.020313942751617,
      "grad_norm": 0.6217536926269531,
      "learning_rate": 2.9536666666666668e-05,
      "loss": 0.0336,
      "step": 11393
    },
    {
      "epoch": 21.022160664819946,
      "grad_norm": 0.41189178824424744,
      "learning_rate": 2.9536333333333334e-05,
      "loss": 0.0546,
      "step": 11394
    },
    {
      "epoch": 21.02400738688827,
      "grad_norm": 0.29428115487098694,
      "learning_rate": 2.9536000000000003e-05,
      "loss": 0.0751,
      "step": 11395
    },
    {
      "epoch": 21.0258541089566,
      "grad_norm": 0.4770805537700653,
      "learning_rate": 2.9535666666666666e-05,
      "loss": 0.0325,
      "step": 11396
    },
    {
      "epoch": 21.02770083102493,
      "grad_norm": 0.5486339330673218,
      "learning_rate": 2.9535333333333335e-05,
      "loss": 0.0211,
      "step": 11397
    },
    {
      "epoch": 21.02954755309326,
      "grad_norm": 0.18714967370033264,
      "learning_rate": 2.9535e-05,
      "loss": 0.0144,
      "step": 11398
    },
    {
      "epoch": 21.03139427516159,
      "grad_norm": 0.22030745446681976,
      "learning_rate": 2.9534666666666667e-05,
      "loss": 0.0101,
      "step": 11399
    },
    {
      "epoch": 21.033240997229917,
      "grad_norm": 0.22807270288467407,
      "learning_rate": 2.9534333333333333e-05,
      "loss": 0.0138,
      "step": 11400
    },
    {
      "epoch": 21.035087719298247,
      "grad_norm": 0.3241746723651886,
      "learning_rate": 2.9534000000000002e-05,
      "loss": 0.0083,
      "step": 11401
    },
    {
      "epoch": 21.036934441366576,
      "grad_norm": 0.21546830236911774,
      "learning_rate": 2.9533666666666668e-05,
      "loss": 0.0074,
      "step": 11402
    },
    {
      "epoch": 21.0387811634349,
      "grad_norm": 0.19861456751823425,
      "learning_rate": 2.9533333333333334e-05,
      "loss": 0.0111,
      "step": 11403
    },
    {
      "epoch": 21.04062788550323,
      "grad_norm": 0.39724117517471313,
      "learning_rate": 2.9533000000000003e-05,
      "loss": 0.0189,
      "step": 11404
    },
    {
      "epoch": 21.04247460757156,
      "grad_norm": 0.21916146576404572,
      "learning_rate": 2.9532666666666666e-05,
      "loss": 0.0055,
      "step": 11405
    },
    {
      "epoch": 21.04432132963989,
      "grad_norm": 0.24507205188274384,
      "learning_rate": 2.9532333333333335e-05,
      "loss": 0.0112,
      "step": 11406
    },
    {
      "epoch": 21.046168051708218,
      "grad_norm": 0.35915157198905945,
      "learning_rate": 2.9532e-05,
      "loss": 0.0137,
      "step": 11407
    },
    {
      "epoch": 21.048014773776547,
      "grad_norm": 0.3480655550956726,
      "learning_rate": 2.9531666666666667e-05,
      "loss": 0.0326,
      "step": 11408
    },
    {
      "epoch": 21.049861495844876,
      "grad_norm": 0.17546714842319489,
      "learning_rate": 2.9531333333333333e-05,
      "loss": 0.0078,
      "step": 11409
    },
    {
      "epoch": 21.051708217913205,
      "grad_norm": 0.22640059888362885,
      "learning_rate": 2.9531e-05,
      "loss": 0.0067,
      "step": 11410
    },
    {
      "epoch": 21.053554939981534,
      "grad_norm": 0.1389237493276596,
      "learning_rate": 2.953066666666667e-05,
      "loss": 0.004,
      "step": 11411
    },
    {
      "epoch": 21.05540166204986,
      "grad_norm": 0.15366314351558685,
      "learning_rate": 2.9530333333333334e-05,
      "loss": 0.0042,
      "step": 11412
    },
    {
      "epoch": 21.05724838411819,
      "grad_norm": 0.21289034187793732,
      "learning_rate": 2.953e-05,
      "loss": 0.0067,
      "step": 11413
    },
    {
      "epoch": 21.05909510618652,
      "grad_norm": 0.1926611363887787,
      "learning_rate": 2.9529666666666666e-05,
      "loss": 0.0062,
      "step": 11414
    },
    {
      "epoch": 21.060941828254848,
      "grad_norm": 0.2734687626361847,
      "learning_rate": 2.9529333333333335e-05,
      "loss": 0.0103,
      "step": 11415
    },
    {
      "epoch": 21.062788550323177,
      "grad_norm": 0.1464875340461731,
      "learning_rate": 2.9528999999999998e-05,
      "loss": 0.0036,
      "step": 11416
    },
    {
      "epoch": 21.064635272391506,
      "grad_norm": 0.2204812616109848,
      "learning_rate": 2.9528666666666667e-05,
      "loss": 0.0053,
      "step": 11417
    },
    {
      "epoch": 21.066481994459835,
      "grad_norm": 0.2089698612689972,
      "learning_rate": 2.9528333333333337e-05,
      "loss": 0.0042,
      "step": 11418
    },
    {
      "epoch": 21.068328716528164,
      "grad_norm": 0.26201918721199036,
      "learning_rate": 2.9528e-05,
      "loss": 0.0074,
      "step": 11419
    },
    {
      "epoch": 21.07017543859649,
      "grad_norm": 0.41449442505836487,
      "learning_rate": 2.952766666666667e-05,
      "loss": 0.0113,
      "step": 11420
    },
    {
      "epoch": 21.07202216066482,
      "grad_norm": 0.3523827791213989,
      "learning_rate": 2.9527333333333334e-05,
      "loss": 0.0106,
      "step": 11421
    },
    {
      "epoch": 21.073868882733148,
      "grad_norm": 0.11771949380636215,
      "learning_rate": 2.9527e-05,
      "loss": 0.003,
      "step": 11422
    },
    {
      "epoch": 21.075715604801477,
      "grad_norm": 0.15286806225776672,
      "learning_rate": 2.9526666666666666e-05,
      "loss": 0.0043,
      "step": 11423
    },
    {
      "epoch": 21.077562326869806,
      "grad_norm": 0.28174591064453125,
      "learning_rate": 2.9526333333333336e-05,
      "loss": 0.006,
      "step": 11424
    },
    {
      "epoch": 21.079409048938135,
      "grad_norm": 0.2134329378604889,
      "learning_rate": 2.9525999999999998e-05,
      "loss": 0.0085,
      "step": 11425
    },
    {
      "epoch": 21.081255771006465,
      "grad_norm": 0.22803491353988647,
      "learning_rate": 2.9525666666666668e-05,
      "loss": 0.0091,
      "step": 11426
    },
    {
      "epoch": 21.083102493074794,
      "grad_norm": 0.3291338384151459,
      "learning_rate": 2.9525333333333337e-05,
      "loss": 0.0092,
      "step": 11427
    },
    {
      "epoch": 21.08494921514312,
      "grad_norm": 0.31323766708374023,
      "learning_rate": 2.9525e-05,
      "loss": 0.015,
      "step": 11428
    },
    {
      "epoch": 21.08679593721145,
      "grad_norm": 0.2662685811519623,
      "learning_rate": 2.952466666666667e-05,
      "loss": 0.0099,
      "step": 11429
    },
    {
      "epoch": 21.088642659279778,
      "grad_norm": 0.4421330392360687,
      "learning_rate": 2.9524333333333335e-05,
      "loss": 0.0069,
      "step": 11430
    },
    {
      "epoch": 21.090489381348107,
      "grad_norm": 0.2042483538389206,
      "learning_rate": 2.9524e-05,
      "loss": 0.0039,
      "step": 11431
    },
    {
      "epoch": 21.092336103416436,
      "grad_norm": 0.29195424914360046,
      "learning_rate": 2.9523666666666667e-05,
      "loss": 0.021,
      "step": 11432
    },
    {
      "epoch": 21.094182825484765,
      "grad_norm": 0.5284271240234375,
      "learning_rate": 2.9523333333333336e-05,
      "loss": 0.1348,
      "step": 11433
    },
    {
      "epoch": 21.096029547553094,
      "grad_norm": 0.6061310172080994,
      "learning_rate": 2.9523e-05,
      "loss": 0.1549,
      "step": 11434
    },
    {
      "epoch": 21.097876269621423,
      "grad_norm": 0.7104319334030151,
      "learning_rate": 2.9522666666666668e-05,
      "loss": 0.1185,
      "step": 11435
    },
    {
      "epoch": 21.099722991689752,
      "grad_norm": 0.40811780095100403,
      "learning_rate": 2.9522333333333337e-05,
      "loss": 0.0631,
      "step": 11436
    },
    {
      "epoch": 21.101569713758078,
      "grad_norm": 0.48598840832710266,
      "learning_rate": 2.9522e-05,
      "loss": 0.1222,
      "step": 11437
    },
    {
      "epoch": 21.103416435826407,
      "grad_norm": 0.5141649842262268,
      "learning_rate": 2.952166666666667e-05,
      "loss": 0.0749,
      "step": 11438
    },
    {
      "epoch": 21.105263157894736,
      "grad_norm": 0.3259578049182892,
      "learning_rate": 2.9521333333333335e-05,
      "loss": 0.0342,
      "step": 11439
    },
    {
      "epoch": 21.107109879963065,
      "grad_norm": 0.4215473532676697,
      "learning_rate": 2.9521e-05,
      "loss": 0.042,
      "step": 11440
    },
    {
      "epoch": 21.108956602031395,
      "grad_norm": 1.668999433517456,
      "learning_rate": 2.9520666666666667e-05,
      "loss": 0.0775,
      "step": 11441
    },
    {
      "epoch": 21.110803324099724,
      "grad_norm": 0.3182021975517273,
      "learning_rate": 2.9520333333333333e-05,
      "loss": 0.0447,
      "step": 11442
    },
    {
      "epoch": 21.112650046168053,
      "grad_norm": 0.44492602348327637,
      "learning_rate": 2.9520000000000002e-05,
      "loss": 0.0343,
      "step": 11443
    },
    {
      "epoch": 21.114496768236382,
      "grad_norm": 0.32682621479034424,
      "learning_rate": 2.9519666666666668e-05,
      "loss": 0.0295,
      "step": 11444
    },
    {
      "epoch": 21.116343490304708,
      "grad_norm": 0.4603782892227173,
      "learning_rate": 2.9519333333333334e-05,
      "loss": 0.0817,
      "step": 11445
    },
    {
      "epoch": 21.118190212373037,
      "grad_norm": 0.24359837174415588,
      "learning_rate": 2.9519e-05,
      "loss": 0.0164,
      "step": 11446
    },
    {
      "epoch": 21.120036934441366,
      "grad_norm": 0.4243770241737366,
      "learning_rate": 2.951866666666667e-05,
      "loss": 0.0193,
      "step": 11447
    },
    {
      "epoch": 21.121883656509695,
      "grad_norm": 0.2748834788799286,
      "learning_rate": 2.951833333333333e-05,
      "loss": 0.011,
      "step": 11448
    },
    {
      "epoch": 21.123730378578024,
      "grad_norm": 0.45617279410362244,
      "learning_rate": 2.9518e-05,
      "loss": 0.0105,
      "step": 11449
    },
    {
      "epoch": 21.125577100646353,
      "grad_norm": 0.32819056510925293,
      "learning_rate": 2.9517666666666667e-05,
      "loss": 0.0462,
      "step": 11450
    },
    {
      "epoch": 21.127423822714682,
      "grad_norm": 0.1864812821149826,
      "learning_rate": 2.9517333333333333e-05,
      "loss": 0.0091,
      "step": 11451
    },
    {
      "epoch": 21.12927054478301,
      "grad_norm": 0.14667195081710815,
      "learning_rate": 2.9517000000000002e-05,
      "loss": 0.0064,
      "step": 11452
    },
    {
      "epoch": 21.131117266851337,
      "grad_norm": 0.3102090358734131,
      "learning_rate": 2.9516666666666668e-05,
      "loss": 0.0107,
      "step": 11453
    },
    {
      "epoch": 21.132963988919666,
      "grad_norm": 0.44857123494148254,
      "learning_rate": 2.9516333333333334e-05,
      "loss": 0.0055,
      "step": 11454
    },
    {
      "epoch": 21.134810710987995,
      "grad_norm": 0.34422293305397034,
      "learning_rate": 2.9516e-05,
      "loss": 0.01,
      "step": 11455
    },
    {
      "epoch": 21.136657433056325,
      "grad_norm": 0.14523011445999146,
      "learning_rate": 2.951566666666667e-05,
      "loss": 0.0063,
      "step": 11456
    },
    {
      "epoch": 21.138504155124654,
      "grad_norm": 0.26662322878837585,
      "learning_rate": 2.9515333333333332e-05,
      "loss": 0.0095,
      "step": 11457
    },
    {
      "epoch": 21.140350877192983,
      "grad_norm": 0.22227898240089417,
      "learning_rate": 2.9515e-05,
      "loss": 0.0104,
      "step": 11458
    },
    {
      "epoch": 21.142197599261312,
      "grad_norm": 0.1880018413066864,
      "learning_rate": 2.9514666666666667e-05,
      "loss": 0.0073,
      "step": 11459
    },
    {
      "epoch": 21.14404432132964,
      "grad_norm": 0.19501319527626038,
      "learning_rate": 2.9514333333333333e-05,
      "loss": 0.0049,
      "step": 11460
    },
    {
      "epoch": 21.14589104339797,
      "grad_norm": 0.13313603401184082,
      "learning_rate": 2.9514000000000002e-05,
      "loss": 0.0045,
      "step": 11461
    },
    {
      "epoch": 21.147737765466296,
      "grad_norm": 0.36689451336860657,
      "learning_rate": 2.9513666666666668e-05,
      "loss": 0.0086,
      "step": 11462
    },
    {
      "epoch": 21.149584487534625,
      "grad_norm": 0.3137074112892151,
      "learning_rate": 2.9513333333333334e-05,
      "loss": 0.0089,
      "step": 11463
    },
    {
      "epoch": 21.151431209602954,
      "grad_norm": 0.16122853755950928,
      "learning_rate": 2.9513e-05,
      "loss": 0.0048,
      "step": 11464
    },
    {
      "epoch": 21.153277931671283,
      "grad_norm": 0.1358654499053955,
      "learning_rate": 2.951266666666667e-05,
      "loss": 0.0025,
      "step": 11465
    },
    {
      "epoch": 21.155124653739612,
      "grad_norm": 0.33823901414871216,
      "learning_rate": 2.9512333333333332e-05,
      "loss": 0.0124,
      "step": 11466
    },
    {
      "epoch": 21.15697137580794,
      "grad_norm": 0.2347034364938736,
      "learning_rate": 2.9512e-05,
      "loss": 0.0055,
      "step": 11467
    },
    {
      "epoch": 21.15881809787627,
      "grad_norm": 0.3039151132106781,
      "learning_rate": 2.951166666666667e-05,
      "loss": 0.0078,
      "step": 11468
    },
    {
      "epoch": 21.1606648199446,
      "grad_norm": 0.12040606886148453,
      "learning_rate": 2.9511333333333333e-05,
      "loss": 0.0043,
      "step": 11469
    },
    {
      "epoch": 21.162511542012926,
      "grad_norm": 0.47133681178092957,
      "learning_rate": 2.9511000000000003e-05,
      "loss": 0.0048,
      "step": 11470
    },
    {
      "epoch": 21.164358264081255,
      "grad_norm": 0.39657193422317505,
      "learning_rate": 2.9510666666666665e-05,
      "loss": 0.0042,
      "step": 11471
    },
    {
      "epoch": 21.166204986149584,
      "grad_norm": 0.23207911849021912,
      "learning_rate": 2.9510333333333334e-05,
      "loss": 0.0082,
      "step": 11472
    },
    {
      "epoch": 21.168051708217913,
      "grad_norm": 0.7355784177780151,
      "learning_rate": 2.951e-05,
      "loss": 0.0089,
      "step": 11473
    },
    {
      "epoch": 21.169898430286242,
      "grad_norm": 0.1707821637392044,
      "learning_rate": 2.9509666666666666e-05,
      "loss": 0.005,
      "step": 11474
    },
    {
      "epoch": 21.17174515235457,
      "grad_norm": 0.21882984042167664,
      "learning_rate": 2.9509333333333332e-05,
      "loss": 0.008,
      "step": 11475
    },
    {
      "epoch": 21.1735918744229,
      "grad_norm": 0.4786779284477234,
      "learning_rate": 2.9509e-05,
      "loss": 0.0116,
      "step": 11476
    },
    {
      "epoch": 21.17543859649123,
      "grad_norm": 0.9925299882888794,
      "learning_rate": 2.9508666666666668e-05,
      "loss": 0.0112,
      "step": 11477
    },
    {
      "epoch": 21.177285318559555,
      "grad_norm": 0.28563570976257324,
      "learning_rate": 2.9508333333333333e-05,
      "loss": 0.0059,
      "step": 11478
    },
    {
      "epoch": 21.179132040627884,
      "grad_norm": 0.19293378293514252,
      "learning_rate": 2.9508000000000003e-05,
      "loss": 0.0057,
      "step": 11479
    },
    {
      "epoch": 21.180978762696213,
      "grad_norm": 0.2053467184305191,
      "learning_rate": 2.9507666666666665e-05,
      "loss": 0.0053,
      "step": 11480
    },
    {
      "epoch": 21.182825484764543,
      "grad_norm": 0.44861987233161926,
      "learning_rate": 2.9507333333333335e-05,
      "loss": 0.0086,
      "step": 11481
    },
    {
      "epoch": 21.18467220683287,
      "grad_norm": 0.42980897426605225,
      "learning_rate": 2.9507e-05,
      "loss": 0.0099,
      "step": 11482
    },
    {
      "epoch": 21.1865189289012,
      "grad_norm": 0.6594889163970947,
      "learning_rate": 2.9506666666666667e-05,
      "loss": 0.1313,
      "step": 11483
    },
    {
      "epoch": 21.18836565096953,
      "grad_norm": 0.7568273544311523,
      "learning_rate": 2.9506333333333332e-05,
      "loss": 0.1434,
      "step": 11484
    },
    {
      "epoch": 21.19021237303786,
      "grad_norm": 0.681679368019104,
      "learning_rate": 2.9506000000000002e-05,
      "loss": 0.1567,
      "step": 11485
    },
    {
      "epoch": 21.19205909510619,
      "grad_norm": 0.45565226674079895,
      "learning_rate": 2.9505666666666668e-05,
      "loss": 0.0674,
      "step": 11486
    },
    {
      "epoch": 21.193905817174514,
      "grad_norm": 0.5891532897949219,
      "learning_rate": 2.9505333333333334e-05,
      "loss": 0.1301,
      "step": 11487
    },
    {
      "epoch": 21.195752539242843,
      "grad_norm": 0.5234256982803345,
      "learning_rate": 2.9505000000000003e-05,
      "loss": 0.0551,
      "step": 11488
    },
    {
      "epoch": 21.197599261311172,
      "grad_norm": 0.557822048664093,
      "learning_rate": 2.9504666666666666e-05,
      "loss": 0.089,
      "step": 11489
    },
    {
      "epoch": 21.1994459833795,
      "grad_norm": 0.3232654929161072,
      "learning_rate": 2.9504333333333335e-05,
      "loss": 0.0384,
      "step": 11490
    },
    {
      "epoch": 21.20129270544783,
      "grad_norm": 0.5436959266662598,
      "learning_rate": 2.9504e-05,
      "loss": 0.0568,
      "step": 11491
    },
    {
      "epoch": 21.20313942751616,
      "grad_norm": 0.3208896517753601,
      "learning_rate": 2.9503666666666667e-05,
      "loss": 0.0287,
      "step": 11492
    },
    {
      "epoch": 21.20498614958449,
      "grad_norm": 0.331940233707428,
      "learning_rate": 2.9503333333333336e-05,
      "loss": 0.0384,
      "step": 11493
    },
    {
      "epoch": 21.206832871652818,
      "grad_norm": 0.3176758587360382,
      "learning_rate": 2.9503000000000002e-05,
      "loss": 0.043,
      "step": 11494
    },
    {
      "epoch": 21.208679593721143,
      "grad_norm": 0.6546593308448792,
      "learning_rate": 2.9502666666666668e-05,
      "loss": 0.0594,
      "step": 11495
    },
    {
      "epoch": 21.210526315789473,
      "grad_norm": 0.2821478247642517,
      "learning_rate": 2.9502333333333334e-05,
      "loss": 0.0339,
      "step": 11496
    },
    {
      "epoch": 21.2123730378578,
      "grad_norm": 0.3861061632633209,
      "learning_rate": 2.9502000000000003e-05,
      "loss": 0.0249,
      "step": 11497
    },
    {
      "epoch": 21.21421975992613,
      "grad_norm": 0.27039122581481934,
      "learning_rate": 2.9501666666666666e-05,
      "loss": 0.0144,
      "step": 11498
    },
    {
      "epoch": 21.21606648199446,
      "grad_norm": 0.18123827874660492,
      "learning_rate": 2.9501333333333335e-05,
      "loss": 0.008,
      "step": 11499
    },
    {
      "epoch": 21.21791320406279,
      "grad_norm": 0.2717362642288208,
      "learning_rate": 2.9500999999999998e-05,
      "loss": 0.014,
      "step": 11500
    },
    {
      "epoch": 21.21975992613112,
      "grad_norm": 0.2388276308774948,
      "learning_rate": 2.9500666666666667e-05,
      "loss": 0.0132,
      "step": 11501
    },
    {
      "epoch": 21.221606648199447,
      "grad_norm": 0.22589102387428284,
      "learning_rate": 2.9500333333333336e-05,
      "loss": 0.0091,
      "step": 11502
    },
    {
      "epoch": 21.223453370267773,
      "grad_norm": 0.21260640025138855,
      "learning_rate": 2.95e-05,
      "loss": 0.0344,
      "step": 11503
    },
    {
      "epoch": 21.225300092336102,
      "grad_norm": 0.6328578591346741,
      "learning_rate": 2.9499666666666668e-05,
      "loss": 0.0065,
      "step": 11504
    },
    {
      "epoch": 21.22714681440443,
      "grad_norm": 0.2761143147945404,
      "learning_rate": 2.9499333333333334e-05,
      "loss": 0.0135,
      "step": 11505
    },
    {
      "epoch": 21.22899353647276,
      "grad_norm": 0.46611031889915466,
      "learning_rate": 2.9499e-05,
      "loss": 0.0186,
      "step": 11506
    },
    {
      "epoch": 21.23084025854109,
      "grad_norm": 0.8633270263671875,
      "learning_rate": 2.9498666666666666e-05,
      "loss": 0.0133,
      "step": 11507
    },
    {
      "epoch": 21.23268698060942,
      "grad_norm": 0.2508547604084015,
      "learning_rate": 2.9498333333333335e-05,
      "loss": 0.0045,
      "step": 11508
    },
    {
      "epoch": 21.234533702677748,
      "grad_norm": 0.18956667184829712,
      "learning_rate": 2.9497999999999998e-05,
      "loss": 0.0054,
      "step": 11509
    },
    {
      "epoch": 21.236380424746077,
      "grad_norm": 0.2119181603193283,
      "learning_rate": 2.9497666666666667e-05,
      "loss": 0.0062,
      "step": 11510
    },
    {
      "epoch": 21.238227146814406,
      "grad_norm": 0.1728789210319519,
      "learning_rate": 2.9497333333333336e-05,
      "loss": 0.0068,
      "step": 11511
    },
    {
      "epoch": 21.24007386888273,
      "grad_norm": 0.5325166583061218,
      "learning_rate": 2.9497e-05,
      "loss": 0.0138,
      "step": 11512
    },
    {
      "epoch": 21.24192059095106,
      "grad_norm": 0.6290391087532043,
      "learning_rate": 2.9496666666666668e-05,
      "loss": 0.0138,
      "step": 11513
    },
    {
      "epoch": 21.24376731301939,
      "grad_norm": 0.16870111227035522,
      "learning_rate": 2.9496333333333334e-05,
      "loss": 0.008,
      "step": 11514
    },
    {
      "epoch": 21.24561403508772,
      "grad_norm": 0.2603950500488281,
      "learning_rate": 2.9496e-05,
      "loss": 0.0101,
      "step": 11515
    },
    {
      "epoch": 21.24746075715605,
      "grad_norm": 0.4347416162490845,
      "learning_rate": 2.9495666666666666e-05,
      "loss": 0.0159,
      "step": 11516
    },
    {
      "epoch": 21.249307479224377,
      "grad_norm": 0.21180957555770874,
      "learning_rate": 2.9495333333333335e-05,
      "loss": 0.0123,
      "step": 11517
    },
    {
      "epoch": 21.251154201292707,
      "grad_norm": 0.20680710673332214,
      "learning_rate": 2.9495e-05,
      "loss": 0.0074,
      "step": 11518
    },
    {
      "epoch": 21.253000923361036,
      "grad_norm": 0.23356780409812927,
      "learning_rate": 2.9494666666666667e-05,
      "loss": 0.0055,
      "step": 11519
    },
    {
      "epoch": 21.25484764542936,
      "grad_norm": 0.5864790678024292,
      "learning_rate": 2.9494333333333337e-05,
      "loss": 0.0047,
      "step": 11520
    },
    {
      "epoch": 21.25669436749769,
      "grad_norm": 0.38270673155784607,
      "learning_rate": 2.9494e-05,
      "loss": 0.0081,
      "step": 11521
    },
    {
      "epoch": 21.25854108956602,
      "grad_norm": 0.34858450293540955,
      "learning_rate": 2.949366666666667e-05,
      "loss": 0.0122,
      "step": 11522
    },
    {
      "epoch": 21.26038781163435,
      "grad_norm": 0.1718364804983139,
      "learning_rate": 2.9493333333333334e-05,
      "loss": 0.0043,
      "step": 11523
    },
    {
      "epoch": 21.262234533702678,
      "grad_norm": 0.4161956310272217,
      "learning_rate": 2.9493e-05,
      "loss": 0.0064,
      "step": 11524
    },
    {
      "epoch": 21.264081255771007,
      "grad_norm": 0.1917291134595871,
      "learning_rate": 2.9492666666666666e-05,
      "loss": 0.0049,
      "step": 11525
    },
    {
      "epoch": 21.265927977839336,
      "grad_norm": 0.20704177021980286,
      "learning_rate": 2.9492333333333336e-05,
      "loss": 0.0056,
      "step": 11526
    },
    {
      "epoch": 21.267774699907665,
      "grad_norm": 0.14914242923259735,
      "learning_rate": 2.9492e-05,
      "loss": 0.0041,
      "step": 11527
    },
    {
      "epoch": 21.26962142197599,
      "grad_norm": 0.2752462327480316,
      "learning_rate": 2.9491666666666667e-05,
      "loss": 0.0056,
      "step": 11528
    },
    {
      "epoch": 21.27146814404432,
      "grad_norm": 0.16511335968971252,
      "learning_rate": 2.9491333333333337e-05,
      "loss": 0.0024,
      "step": 11529
    },
    {
      "epoch": 21.27331486611265,
      "grad_norm": 0.31737568974494934,
      "learning_rate": 2.9491e-05,
      "loss": 0.0085,
      "step": 11530
    },
    {
      "epoch": 21.27516158818098,
      "grad_norm": 0.2316114753484726,
      "learning_rate": 2.949066666666667e-05,
      "loss": 0.0044,
      "step": 11531
    },
    {
      "epoch": 21.277008310249307,
      "grad_norm": 0.3384943902492523,
      "learning_rate": 2.949033333333333e-05,
      "loss": 0.0185,
      "step": 11532
    },
    {
      "epoch": 21.278855032317637,
      "grad_norm": 0.6905454993247986,
      "learning_rate": 2.949e-05,
      "loss": 0.1296,
      "step": 11533
    },
    {
      "epoch": 21.280701754385966,
      "grad_norm": 0.6204212307929993,
      "learning_rate": 2.9489666666666666e-05,
      "loss": 0.1165,
      "step": 11534
    },
    {
      "epoch": 21.282548476454295,
      "grad_norm": 0.45431268215179443,
      "learning_rate": 2.9489333333333332e-05,
      "loss": 0.0895,
      "step": 11535
    },
    {
      "epoch": 21.284395198522624,
      "grad_norm": 0.5600978136062622,
      "learning_rate": 2.9489000000000002e-05,
      "loss": 0.106,
      "step": 11536
    },
    {
      "epoch": 21.28624192059095,
      "grad_norm": 0.5091732144355774,
      "learning_rate": 2.9488666666666668e-05,
      "loss": 0.0879,
      "step": 11537
    },
    {
      "epoch": 21.28808864265928,
      "grad_norm": 0.5164522528648376,
      "learning_rate": 2.9488333333333334e-05,
      "loss": 0.0856,
      "step": 11538
    },
    {
      "epoch": 21.289935364727608,
      "grad_norm": 0.3764141798019409,
      "learning_rate": 2.9488e-05,
      "loss": 0.0866,
      "step": 11539
    },
    {
      "epoch": 21.291782086795937,
      "grad_norm": 0.3163754343986511,
      "learning_rate": 2.948766666666667e-05,
      "loss": 0.049,
      "step": 11540
    },
    {
      "epoch": 21.293628808864266,
      "grad_norm": 0.48345544934272766,
      "learning_rate": 2.948733333333333e-05,
      "loss": 0.0335,
      "step": 11541
    },
    {
      "epoch": 21.295475530932595,
      "grad_norm": 0.3480796217918396,
      "learning_rate": 2.9487e-05,
      "loss": 0.047,
      "step": 11542
    },
    {
      "epoch": 21.297322253000925,
      "grad_norm": 0.7356621623039246,
      "learning_rate": 2.948666666666667e-05,
      "loss": 0.0324,
      "step": 11543
    },
    {
      "epoch": 21.299168975069254,
      "grad_norm": 0.24600377678871155,
      "learning_rate": 2.9486333333333333e-05,
      "loss": 0.0218,
      "step": 11544
    },
    {
      "epoch": 21.30101569713758,
      "grad_norm": 0.242000550031662,
      "learning_rate": 2.9486000000000002e-05,
      "loss": 0.0216,
      "step": 11545
    },
    {
      "epoch": 21.30286241920591,
      "grad_norm": 0.31422117352485657,
      "learning_rate": 2.9485666666666668e-05,
      "loss": 0.0299,
      "step": 11546
    },
    {
      "epoch": 21.304709141274238,
      "grad_norm": 0.2681136727333069,
      "learning_rate": 2.9485333333333334e-05,
      "loss": 0.0118,
      "step": 11547
    },
    {
      "epoch": 21.306555863342567,
      "grad_norm": 0.46410465240478516,
      "learning_rate": 2.9485e-05,
      "loss": 0.0594,
      "step": 11548
    },
    {
      "epoch": 21.308402585410896,
      "grad_norm": 1.1594105958938599,
      "learning_rate": 2.948466666666667e-05,
      "loss": 0.0341,
      "step": 11549
    },
    {
      "epoch": 21.310249307479225,
      "grad_norm": 0.18428504467010498,
      "learning_rate": 2.948433333333333e-05,
      "loss": 0.0106,
      "step": 11550
    },
    {
      "epoch": 21.312096029547554,
      "grad_norm": 0.2524779140949249,
      "learning_rate": 2.9484e-05,
      "loss": 0.0124,
      "step": 11551
    },
    {
      "epoch": 21.313942751615883,
      "grad_norm": 0.4212198555469513,
      "learning_rate": 2.948366666666667e-05,
      "loss": 0.0152,
      "step": 11552
    },
    {
      "epoch": 21.31578947368421,
      "grad_norm": 0.2710402011871338,
      "learning_rate": 2.9483333333333333e-05,
      "loss": 0.016,
      "step": 11553
    },
    {
      "epoch": 21.317636195752538,
      "grad_norm": 0.1798483282327652,
      "learning_rate": 2.9483000000000002e-05,
      "loss": 0.0068,
      "step": 11554
    },
    {
      "epoch": 21.319482917820867,
      "grad_norm": 0.3181666135787964,
      "learning_rate": 2.9482666666666668e-05,
      "loss": 0.0061,
      "step": 11555
    },
    {
      "epoch": 21.321329639889196,
      "grad_norm": 0.1473083347082138,
      "learning_rate": 2.9482333333333334e-05,
      "loss": 0.0082,
      "step": 11556
    },
    {
      "epoch": 21.323176361957525,
      "grad_norm": 0.302139550447464,
      "learning_rate": 2.9482e-05,
      "loss": 0.0076,
      "step": 11557
    },
    {
      "epoch": 21.325023084025855,
      "grad_norm": 0.21949274837970734,
      "learning_rate": 2.948166666666667e-05,
      "loss": 0.0084,
      "step": 11558
    },
    {
      "epoch": 21.326869806094184,
      "grad_norm": 0.1488589644432068,
      "learning_rate": 2.9481333333333332e-05,
      "loss": 0.0055,
      "step": 11559
    },
    {
      "epoch": 21.328716528162513,
      "grad_norm": 0.4395733177661896,
      "learning_rate": 2.9481e-05,
      "loss": 0.0079,
      "step": 11560
    },
    {
      "epoch": 21.330563250230842,
      "grad_norm": 0.38077491521835327,
      "learning_rate": 2.9480666666666667e-05,
      "loss": 0.009,
      "step": 11561
    },
    {
      "epoch": 21.332409972299168,
      "grad_norm": 0.17946629226207733,
      "learning_rate": 2.9480333333333333e-05,
      "loss": 0.0046,
      "step": 11562
    },
    {
      "epoch": 21.334256694367497,
      "grad_norm": 0.25635477900505066,
      "learning_rate": 2.9480000000000002e-05,
      "loss": 0.0087,
      "step": 11563
    },
    {
      "epoch": 21.336103416435826,
      "grad_norm": 0.18848296999931335,
      "learning_rate": 2.9479666666666665e-05,
      "loss": 0.0073,
      "step": 11564
    },
    {
      "epoch": 21.337950138504155,
      "grad_norm": 0.3364658057689667,
      "learning_rate": 2.9479333333333334e-05,
      "loss": 0.0089,
      "step": 11565
    },
    {
      "epoch": 21.339796860572484,
      "grad_norm": 0.16113604605197906,
      "learning_rate": 2.9479e-05,
      "loss": 0.0046,
      "step": 11566
    },
    {
      "epoch": 21.341643582640813,
      "grad_norm": 0.42057058215141296,
      "learning_rate": 2.9478666666666666e-05,
      "loss": 0.012,
      "step": 11567
    },
    {
      "epoch": 21.343490304709142,
      "grad_norm": 0.27273300290107727,
      "learning_rate": 2.9478333333333335e-05,
      "loss": 0.0057,
      "step": 11568
    },
    {
      "epoch": 21.34533702677747,
      "grad_norm": 0.21287663280963898,
      "learning_rate": 2.9478e-05,
      "loss": 0.0045,
      "step": 11569
    },
    {
      "epoch": 21.347183748845797,
      "grad_norm": 0.3194565176963806,
      "learning_rate": 2.9477666666666667e-05,
      "loss": 0.0107,
      "step": 11570
    },
    {
      "epoch": 21.349030470914126,
      "grad_norm": 0.3554210662841797,
      "learning_rate": 2.9477333333333333e-05,
      "loss": 0.0066,
      "step": 11571
    },
    {
      "epoch": 21.350877192982455,
      "grad_norm": 0.3108913004398346,
      "learning_rate": 2.9477000000000003e-05,
      "loss": 0.008,
      "step": 11572
    },
    {
      "epoch": 21.352723915050785,
      "grad_norm": 0.1508089303970337,
      "learning_rate": 2.9476666666666665e-05,
      "loss": 0.0059,
      "step": 11573
    },
    {
      "epoch": 21.354570637119114,
      "grad_norm": 0.41390153765678406,
      "learning_rate": 2.9476333333333334e-05,
      "loss": 0.0156,
      "step": 11574
    },
    {
      "epoch": 21.356417359187443,
      "grad_norm": 0.20888638496398926,
      "learning_rate": 2.9476e-05,
      "loss": 0.0032,
      "step": 11575
    },
    {
      "epoch": 21.358264081255772,
      "grad_norm": 0.29900670051574707,
      "learning_rate": 2.9475666666666666e-05,
      "loss": 0.0112,
      "step": 11576
    },
    {
      "epoch": 21.3601108033241,
      "grad_norm": 0.18397219479084015,
      "learning_rate": 2.9475333333333336e-05,
      "loss": 0.007,
      "step": 11577
    },
    {
      "epoch": 21.361957525392427,
      "grad_norm": 0.2705393135547638,
      "learning_rate": 2.9475e-05,
      "loss": 0.0085,
      "step": 11578
    },
    {
      "epoch": 21.363804247460756,
      "grad_norm": 0.6648091673851013,
      "learning_rate": 2.9474666666666667e-05,
      "loss": 0.0213,
      "step": 11579
    },
    {
      "epoch": 21.365650969529085,
      "grad_norm": 0.5839139819145203,
      "learning_rate": 2.9474333333333333e-05,
      "loss": 0.0153,
      "step": 11580
    },
    {
      "epoch": 21.367497691597414,
      "grad_norm": 0.2984607219696045,
      "learning_rate": 2.9474000000000003e-05,
      "loss": 0.0066,
      "step": 11581
    },
    {
      "epoch": 21.369344413665743,
      "grad_norm": 0.2725304365158081,
      "learning_rate": 2.9473666666666665e-05,
      "loss": 0.005,
      "step": 11582
    },
    {
      "epoch": 21.371191135734072,
      "grad_norm": 0.4688223600387573,
      "learning_rate": 2.9473333333333335e-05,
      "loss": 0.1274,
      "step": 11583
    },
    {
      "epoch": 21.3730378578024,
      "grad_norm": 0.5752362608909607,
      "learning_rate": 2.9473e-05,
      "loss": 0.1327,
      "step": 11584
    },
    {
      "epoch": 21.37488457987073,
      "grad_norm": 0.45737379789352417,
      "learning_rate": 2.9472666666666666e-05,
      "loss": 0.1189,
      "step": 11585
    },
    {
      "epoch": 21.37673130193906,
      "grad_norm": 0.3827018141746521,
      "learning_rate": 2.9472333333333336e-05,
      "loss": 0.0666,
      "step": 11586
    },
    {
      "epoch": 21.378578024007385,
      "grad_norm": 0.5348204970359802,
      "learning_rate": 2.9472000000000002e-05,
      "loss": 0.102,
      "step": 11587
    },
    {
      "epoch": 21.380424746075715,
      "grad_norm": 0.5674592852592468,
      "learning_rate": 2.9471666666666668e-05,
      "loss": 0.0716,
      "step": 11588
    },
    {
      "epoch": 21.382271468144044,
      "grad_norm": 0.3822150230407715,
      "learning_rate": 2.9471333333333334e-05,
      "loss": 0.0597,
      "step": 11589
    },
    {
      "epoch": 21.384118190212373,
      "grad_norm": 0.37607574462890625,
      "learning_rate": 2.9471000000000003e-05,
      "loss": 0.0414,
      "step": 11590
    },
    {
      "epoch": 21.385964912280702,
      "grad_norm": 0.36966872215270996,
      "learning_rate": 2.9470666666666665e-05,
      "loss": 0.0591,
      "step": 11591
    },
    {
      "epoch": 21.38781163434903,
      "grad_norm": 0.6840993762016296,
      "learning_rate": 2.9470333333333335e-05,
      "loss": 0.0338,
      "step": 11592
    },
    {
      "epoch": 21.38965835641736,
      "grad_norm": 0.3434557020664215,
      "learning_rate": 2.947e-05,
      "loss": 0.0374,
      "step": 11593
    },
    {
      "epoch": 21.39150507848569,
      "grad_norm": 0.3484346568584442,
      "learning_rate": 2.9469666666666667e-05,
      "loss": 0.0627,
      "step": 11594
    },
    {
      "epoch": 21.393351800554015,
      "grad_norm": 0.24304430186748505,
      "learning_rate": 2.9469333333333336e-05,
      "loss": 0.0188,
      "step": 11595
    },
    {
      "epoch": 21.395198522622344,
      "grad_norm": 0.2664783000946045,
      "learning_rate": 2.9469e-05,
      "loss": 0.0175,
      "step": 11596
    },
    {
      "epoch": 21.397045244690673,
      "grad_norm": 0.23857802152633667,
      "learning_rate": 2.9468666666666668e-05,
      "loss": 0.0339,
      "step": 11597
    },
    {
      "epoch": 21.398891966759003,
      "grad_norm": 0.2604442238807678,
      "learning_rate": 2.9468333333333334e-05,
      "loss": 0.0162,
      "step": 11598
    },
    {
      "epoch": 21.40073868882733,
      "grad_norm": 0.28916192054748535,
      "learning_rate": 2.9468e-05,
      "loss": 0.0106,
      "step": 11599
    },
    {
      "epoch": 21.40258541089566,
      "grad_norm": 0.1989450752735138,
      "learning_rate": 2.9467666666666666e-05,
      "loss": 0.0076,
      "step": 11600
    },
    {
      "epoch": 21.40443213296399,
      "grad_norm": 0.1816091239452362,
      "learning_rate": 2.9467333333333335e-05,
      "loss": 0.0064,
      "step": 11601
    },
    {
      "epoch": 21.40627885503232,
      "grad_norm": 0.3587622046470642,
      "learning_rate": 2.9467e-05,
      "loss": 0.0159,
      "step": 11602
    },
    {
      "epoch": 21.408125577100645,
      "grad_norm": 0.3580322265625,
      "learning_rate": 2.9466666666666667e-05,
      "loss": 0.009,
      "step": 11603
    },
    {
      "epoch": 21.409972299168974,
      "grad_norm": 0.2690403163433075,
      "learning_rate": 2.9466333333333336e-05,
      "loss": 0.0068,
      "step": 11604
    },
    {
      "epoch": 21.411819021237303,
      "grad_norm": 0.20682458579540253,
      "learning_rate": 2.9466e-05,
      "loss": 0.01,
      "step": 11605
    },
    {
      "epoch": 21.413665743305632,
      "grad_norm": 0.32515549659729004,
      "learning_rate": 2.9465666666666668e-05,
      "loss": 0.0132,
      "step": 11606
    },
    {
      "epoch": 21.41551246537396,
      "grad_norm": 0.2933991253376007,
      "learning_rate": 2.9465333333333334e-05,
      "loss": 0.01,
      "step": 11607
    },
    {
      "epoch": 21.41735918744229,
      "grad_norm": 0.3860311508178711,
      "learning_rate": 2.9465e-05,
      "loss": 0.006,
      "step": 11608
    },
    {
      "epoch": 21.41920590951062,
      "grad_norm": 0.40069878101348877,
      "learning_rate": 2.9464666666666666e-05,
      "loss": 0.0135,
      "step": 11609
    },
    {
      "epoch": 21.42105263157895,
      "grad_norm": 0.23569025099277496,
      "learning_rate": 2.9464333333333335e-05,
      "loss": 0.0064,
      "step": 11610
    },
    {
      "epoch": 21.422899353647278,
      "grad_norm": 0.33610761165618896,
      "learning_rate": 2.9464e-05,
      "loss": 0.0145,
      "step": 11611
    },
    {
      "epoch": 21.424746075715603,
      "grad_norm": 0.645300030708313,
      "learning_rate": 2.9463666666666667e-05,
      "loss": 0.018,
      "step": 11612
    },
    {
      "epoch": 21.426592797783933,
      "grad_norm": 0.5435053706169128,
      "learning_rate": 2.9463333333333336e-05,
      "loss": 0.0312,
      "step": 11613
    },
    {
      "epoch": 21.42843951985226,
      "grad_norm": 0.24929502606391907,
      "learning_rate": 2.9463e-05,
      "loss": 0.0065,
      "step": 11614
    },
    {
      "epoch": 21.43028624192059,
      "grad_norm": 0.3996683657169342,
      "learning_rate": 2.9462666666666668e-05,
      "loss": 0.0093,
      "step": 11615
    },
    {
      "epoch": 21.43213296398892,
      "grad_norm": 0.21856895089149475,
      "learning_rate": 2.9462333333333334e-05,
      "loss": 0.0103,
      "step": 11616
    },
    {
      "epoch": 21.43397968605725,
      "grad_norm": 0.386295348405838,
      "learning_rate": 2.9462e-05,
      "loss": 0.0129,
      "step": 11617
    },
    {
      "epoch": 21.43582640812558,
      "grad_norm": 0.14483925700187683,
      "learning_rate": 2.946166666666667e-05,
      "loss": 0.0049,
      "step": 11618
    },
    {
      "epoch": 21.437673130193907,
      "grad_norm": 0.14828045666217804,
      "learning_rate": 2.9461333333333335e-05,
      "loss": 0.0048,
      "step": 11619
    },
    {
      "epoch": 21.439519852262233,
      "grad_norm": 0.4934016764163971,
      "learning_rate": 2.9461e-05,
      "loss": 0.0209,
      "step": 11620
    },
    {
      "epoch": 21.441366574330562,
      "grad_norm": 0.35568687319755554,
      "learning_rate": 2.9460666666666667e-05,
      "loss": 0.0139,
      "step": 11621
    },
    {
      "epoch": 21.44321329639889,
      "grad_norm": 0.29019680619239807,
      "learning_rate": 2.9460333333333333e-05,
      "loss": 0.0047,
      "step": 11622
    },
    {
      "epoch": 21.44506001846722,
      "grad_norm": 0.3380444347858429,
      "learning_rate": 2.946e-05,
      "loss": 0.014,
      "step": 11623
    },
    {
      "epoch": 21.44690674053555,
      "grad_norm": 0.17599837481975555,
      "learning_rate": 2.945966666666667e-05,
      "loss": 0.0031,
      "step": 11624
    },
    {
      "epoch": 21.44875346260388,
      "grad_norm": 0.46131882071495056,
      "learning_rate": 2.945933333333333e-05,
      "loss": 0.0119,
      "step": 11625
    },
    {
      "epoch": 21.450600184672208,
      "grad_norm": 0.33762016892433167,
      "learning_rate": 2.9459e-05,
      "loss": 0.0108,
      "step": 11626
    },
    {
      "epoch": 21.452446906740537,
      "grad_norm": 0.2296239137649536,
      "learning_rate": 2.945866666666667e-05,
      "loss": 0.0047,
      "step": 11627
    },
    {
      "epoch": 21.454293628808863,
      "grad_norm": 0.09554245322942734,
      "learning_rate": 2.9458333333333332e-05,
      "loss": 0.0034,
      "step": 11628
    },
    {
      "epoch": 21.45614035087719,
      "grad_norm": 0.30348119139671326,
      "learning_rate": 2.9458e-05,
      "loss": 0.0156,
      "step": 11629
    },
    {
      "epoch": 21.45798707294552,
      "grad_norm": 0.5296110510826111,
      "learning_rate": 2.9457666666666667e-05,
      "loss": 0.011,
      "step": 11630
    },
    {
      "epoch": 21.45983379501385,
      "grad_norm": 0.33812788128852844,
      "learning_rate": 2.9457333333333333e-05,
      "loss": 0.0112,
      "step": 11631
    },
    {
      "epoch": 21.46168051708218,
      "grad_norm": 0.17619480192661285,
      "learning_rate": 2.9457e-05,
      "loss": 0.0038,
      "step": 11632
    },
    {
      "epoch": 21.46352723915051,
      "grad_norm": 0.5979994535446167,
      "learning_rate": 2.945666666666667e-05,
      "loss": 0.1385,
      "step": 11633
    },
    {
      "epoch": 21.465373961218837,
      "grad_norm": 0.5296694040298462,
      "learning_rate": 2.945633333333333e-05,
      "loss": 0.1222,
      "step": 11634
    },
    {
      "epoch": 21.467220683287167,
      "grad_norm": 0.8114538192749023,
      "learning_rate": 2.9456e-05,
      "loss": 0.0708,
      "step": 11635
    },
    {
      "epoch": 21.469067405355496,
      "grad_norm": 0.46399155259132385,
      "learning_rate": 2.945566666666667e-05,
      "loss": 0.069,
      "step": 11636
    },
    {
      "epoch": 21.47091412742382,
      "grad_norm": 0.6047453284263611,
      "learning_rate": 2.9455333333333332e-05,
      "loss": 0.075,
      "step": 11637
    },
    {
      "epoch": 21.47276084949215,
      "grad_norm": 0.47024619579315186,
      "learning_rate": 2.9455000000000002e-05,
      "loss": 0.0392,
      "step": 11638
    },
    {
      "epoch": 21.47460757156048,
      "grad_norm": 0.36386826634407043,
      "learning_rate": 2.9454666666666668e-05,
      "loss": 0.0485,
      "step": 11639
    },
    {
      "epoch": 21.47645429362881,
      "grad_norm": 0.4516194462776184,
      "learning_rate": 2.9454333333333334e-05,
      "loss": 0.0745,
      "step": 11640
    },
    {
      "epoch": 21.478301015697138,
      "grad_norm": 0.6910102367401123,
      "learning_rate": 2.9454e-05,
      "loss": 0.0718,
      "step": 11641
    },
    {
      "epoch": 21.480147737765467,
      "grad_norm": 0.2954343259334564,
      "learning_rate": 2.945366666666667e-05,
      "loss": 0.0422,
      "step": 11642
    },
    {
      "epoch": 21.481994459833796,
      "grad_norm": 0.44146180152893066,
      "learning_rate": 2.9453333333333335e-05,
      "loss": 0.0371,
      "step": 11643
    },
    {
      "epoch": 21.483841181902125,
      "grad_norm": 0.46557775139808655,
      "learning_rate": 2.9453e-05,
      "loss": 0.0684,
      "step": 11644
    },
    {
      "epoch": 21.48568790397045,
      "grad_norm": 0.3252936005592346,
      "learning_rate": 2.945266666666667e-05,
      "loss": 0.0207,
      "step": 11645
    },
    {
      "epoch": 21.48753462603878,
      "grad_norm": 0.47686776518821716,
      "learning_rate": 2.9452333333333333e-05,
      "loss": 0.0228,
      "step": 11646
    },
    {
      "epoch": 21.48938134810711,
      "grad_norm": 0.26284104585647583,
      "learning_rate": 2.9452000000000002e-05,
      "loss": 0.0237,
      "step": 11647
    },
    {
      "epoch": 21.49122807017544,
      "grad_norm": 0.20137442648410797,
      "learning_rate": 2.9451666666666668e-05,
      "loss": 0.0117,
      "step": 11648
    },
    {
      "epoch": 21.493074792243767,
      "grad_norm": 0.23116926848888397,
      "learning_rate": 2.9451333333333334e-05,
      "loss": 0.0108,
      "step": 11649
    },
    {
      "epoch": 21.494921514312097,
      "grad_norm": 0.1661728322505951,
      "learning_rate": 2.9451e-05,
      "loss": 0.0075,
      "step": 11650
    },
    {
      "epoch": 21.496768236380426,
      "grad_norm": 0.25384411215782166,
      "learning_rate": 2.945066666666667e-05,
      "loss": 0.0082,
      "step": 11651
    },
    {
      "epoch": 21.498614958448755,
      "grad_norm": 0.16077283024787903,
      "learning_rate": 2.9450333333333335e-05,
      "loss": 0.0109,
      "step": 11652
    },
    {
      "epoch": 21.50046168051708,
      "grad_norm": 0.31073248386383057,
      "learning_rate": 2.945e-05,
      "loss": 0.0081,
      "step": 11653
    },
    {
      "epoch": 21.50230840258541,
      "grad_norm": 0.32054486870765686,
      "learning_rate": 2.9449666666666667e-05,
      "loss": 0.007,
      "step": 11654
    },
    {
      "epoch": 21.50415512465374,
      "grad_norm": 0.23431295156478882,
      "learning_rate": 2.9449333333333333e-05,
      "loss": 0.0133,
      "step": 11655
    },
    {
      "epoch": 21.506001846722068,
      "grad_norm": 0.26101458072662354,
      "learning_rate": 2.9449000000000002e-05,
      "loss": 0.0107,
      "step": 11656
    },
    {
      "epoch": 21.507848568790397,
      "grad_norm": 0.22934867441654205,
      "learning_rate": 2.9448666666666665e-05,
      "loss": 0.0088,
      "step": 11657
    },
    {
      "epoch": 21.509695290858726,
      "grad_norm": 0.31780898571014404,
      "learning_rate": 2.9448333333333334e-05,
      "loss": 0.0093,
      "step": 11658
    },
    {
      "epoch": 21.511542012927055,
      "grad_norm": 0.3254026174545288,
      "learning_rate": 2.9448e-05,
      "loss": 0.0087,
      "step": 11659
    },
    {
      "epoch": 21.513388734995385,
      "grad_norm": 0.11738292127847672,
      "learning_rate": 2.9447666666666666e-05,
      "loss": 0.0049,
      "step": 11660
    },
    {
      "epoch": 21.51523545706371,
      "grad_norm": 0.3099237382411957,
      "learning_rate": 2.9447333333333335e-05,
      "loss": 0.0099,
      "step": 11661
    },
    {
      "epoch": 21.51708217913204,
      "grad_norm": 0.09953440725803375,
      "learning_rate": 2.9447e-05,
      "loss": 0.0033,
      "step": 11662
    },
    {
      "epoch": 21.51892890120037,
      "grad_norm": 0.6011154651641846,
      "learning_rate": 2.9446666666666667e-05,
      "loss": 0.0159,
      "step": 11663
    },
    {
      "epoch": 21.520775623268698,
      "grad_norm": 0.298566609621048,
      "learning_rate": 2.9446333333333333e-05,
      "loss": 0.0161,
      "step": 11664
    },
    {
      "epoch": 21.522622345337027,
      "grad_norm": 0.36138394474983215,
      "learning_rate": 2.9446000000000002e-05,
      "loss": 0.0054,
      "step": 11665
    },
    {
      "epoch": 21.524469067405356,
      "grad_norm": 0.26504313945770264,
      "learning_rate": 2.9445666666666665e-05,
      "loss": 0.0073,
      "step": 11666
    },
    {
      "epoch": 21.526315789473685,
      "grad_norm": 0.15179431438446045,
      "learning_rate": 2.9445333333333334e-05,
      "loss": 0.0036,
      "step": 11667
    },
    {
      "epoch": 21.528162511542014,
      "grad_norm": 0.34152984619140625,
      "learning_rate": 2.9445000000000004e-05,
      "loss": 0.006,
      "step": 11668
    },
    {
      "epoch": 21.530009233610343,
      "grad_norm": 0.1816384345293045,
      "learning_rate": 2.9444666666666666e-05,
      "loss": 0.0035,
      "step": 11669
    },
    {
      "epoch": 21.53185595567867,
      "grad_norm": 0.4318518340587616,
      "learning_rate": 2.9444333333333335e-05,
      "loss": 0.0088,
      "step": 11670
    },
    {
      "epoch": 21.533702677746998,
      "grad_norm": 0.2734701931476593,
      "learning_rate": 2.9444e-05,
      "loss": 0.0063,
      "step": 11671
    },
    {
      "epoch": 21.535549399815327,
      "grad_norm": 0.1769401878118515,
      "learning_rate": 2.9443666666666667e-05,
      "loss": 0.0046,
      "step": 11672
    },
    {
      "epoch": 21.537396121883656,
      "grad_norm": 0.3240947127342224,
      "learning_rate": 2.9443333333333333e-05,
      "loss": 0.0066,
      "step": 11673
    },
    {
      "epoch": 21.539242843951985,
      "grad_norm": 0.40373358130455017,
      "learning_rate": 2.9443000000000003e-05,
      "loss": 0.013,
      "step": 11674
    },
    {
      "epoch": 21.541089566020315,
      "grad_norm": 0.41496601700782776,
      "learning_rate": 2.9442666666666665e-05,
      "loss": 0.0065,
      "step": 11675
    },
    {
      "epoch": 21.542936288088644,
      "grad_norm": 0.32355672121047974,
      "learning_rate": 2.9442333333333334e-05,
      "loss": 0.0101,
      "step": 11676
    },
    {
      "epoch": 21.544783010156973,
      "grad_norm": 0.26241594552993774,
      "learning_rate": 2.9442000000000004e-05,
      "loss": 0.0078,
      "step": 11677
    },
    {
      "epoch": 21.5466297322253,
      "grad_norm": 0.10817354172468185,
      "learning_rate": 2.9441666666666666e-05,
      "loss": 0.0037,
      "step": 11678
    },
    {
      "epoch": 21.548476454293628,
      "grad_norm": 0.05409083515405655,
      "learning_rate": 2.9441333333333336e-05,
      "loss": 0.0009,
      "step": 11679
    },
    {
      "epoch": 21.550323176361957,
      "grad_norm": 0.25125715136528015,
      "learning_rate": 2.9441e-05,
      "loss": 0.003,
      "step": 11680
    },
    {
      "epoch": 21.552169898430286,
      "grad_norm": 0.7134154438972473,
      "learning_rate": 2.9440666666666667e-05,
      "loss": 0.0158,
      "step": 11681
    },
    {
      "epoch": 21.554016620498615,
      "grad_norm": 0.40545278787612915,
      "learning_rate": 2.9440333333333333e-05,
      "loss": 0.0083,
      "step": 11682
    },
    {
      "epoch": 21.555863342566944,
      "grad_norm": 0.6952413320541382,
      "learning_rate": 2.944e-05,
      "loss": 0.2054,
      "step": 11683
    },
    {
      "epoch": 21.557710064635273,
      "grad_norm": 0.5398013591766357,
      "learning_rate": 2.9439666666666665e-05,
      "loss": 0.1455,
      "step": 11684
    },
    {
      "epoch": 21.559556786703602,
      "grad_norm": 0.4659029543399811,
      "learning_rate": 2.9439333333333335e-05,
      "loss": 0.0803,
      "step": 11685
    },
    {
      "epoch": 21.56140350877193,
      "grad_norm": 0.48628848791122437,
      "learning_rate": 2.9439e-05,
      "loss": 0.1113,
      "step": 11686
    },
    {
      "epoch": 21.563250230840257,
      "grad_norm": 0.6104269623756409,
      "learning_rate": 2.9438666666666666e-05,
      "loss": 0.1125,
      "step": 11687
    },
    {
      "epoch": 21.565096952908586,
      "grad_norm": 0.5309699177742004,
      "learning_rate": 2.9438333333333336e-05,
      "loss": 0.0874,
      "step": 11688
    },
    {
      "epoch": 21.566943674976915,
      "grad_norm": 0.4424571394920349,
      "learning_rate": 2.9438e-05,
      "loss": 0.0724,
      "step": 11689
    },
    {
      "epoch": 21.568790397045245,
      "grad_norm": 0.4594380259513855,
      "learning_rate": 2.9437666666666668e-05,
      "loss": 0.0628,
      "step": 11690
    },
    {
      "epoch": 21.570637119113574,
      "grad_norm": 0.7726163864135742,
      "learning_rate": 2.9437333333333334e-05,
      "loss": 0.0555,
      "step": 11691
    },
    {
      "epoch": 21.572483841181903,
      "grad_norm": 0.3133237361907959,
      "learning_rate": 2.9437e-05,
      "loss": 0.0476,
      "step": 11692
    },
    {
      "epoch": 21.574330563250232,
      "grad_norm": 0.39666256308555603,
      "learning_rate": 2.943666666666667e-05,
      "loss": 0.0348,
      "step": 11693
    },
    {
      "epoch": 21.57617728531856,
      "grad_norm": 0.3721177577972412,
      "learning_rate": 2.9436333333333335e-05,
      "loss": 0.0317,
      "step": 11694
    },
    {
      "epoch": 21.578024007386887,
      "grad_norm": 0.23958668112754822,
      "learning_rate": 2.9436e-05,
      "loss": 0.0363,
      "step": 11695
    },
    {
      "epoch": 21.579870729455216,
      "grad_norm": 0.4093874990940094,
      "learning_rate": 2.9435666666666667e-05,
      "loss": 0.0421,
      "step": 11696
    },
    {
      "epoch": 21.581717451523545,
      "grad_norm": 0.24053242802619934,
      "learning_rate": 2.9435333333333336e-05,
      "loss": 0.0342,
      "step": 11697
    },
    {
      "epoch": 21.583564173591874,
      "grad_norm": 0.3315214216709137,
      "learning_rate": 2.9435e-05,
      "loss": 0.0284,
      "step": 11698
    },
    {
      "epoch": 21.585410895660203,
      "grad_norm": 0.18025685846805573,
      "learning_rate": 2.9434666666666668e-05,
      "loss": 0.0115,
      "step": 11699
    },
    {
      "epoch": 21.587257617728532,
      "grad_norm": 0.37826457619667053,
      "learning_rate": 2.9434333333333334e-05,
      "loss": 0.0154,
      "step": 11700
    },
    {
      "epoch": 21.58910433979686,
      "grad_norm": 0.5563769340515137,
      "learning_rate": 2.9434e-05,
      "loss": 0.0287,
      "step": 11701
    },
    {
      "epoch": 21.59095106186519,
      "grad_norm": 0.3572843670845032,
      "learning_rate": 2.943366666666667e-05,
      "loss": 0.0363,
      "step": 11702
    },
    {
      "epoch": 21.592797783933516,
      "grad_norm": 0.2566000521183014,
      "learning_rate": 2.9433333333333335e-05,
      "loss": 0.0102,
      "step": 11703
    },
    {
      "epoch": 21.594644506001845,
      "grad_norm": 0.16738413274288177,
      "learning_rate": 2.9433e-05,
      "loss": 0.0049,
      "step": 11704
    },
    {
      "epoch": 21.596491228070175,
      "grad_norm": 0.31159695982933044,
      "learning_rate": 2.9432666666666667e-05,
      "loss": 0.0089,
      "step": 11705
    },
    {
      "epoch": 21.598337950138504,
      "grad_norm": 0.17356759309768677,
      "learning_rate": 2.9432333333333336e-05,
      "loss": 0.0072,
      "step": 11706
    },
    {
      "epoch": 21.600184672206833,
      "grad_norm": 0.2614094614982605,
      "learning_rate": 2.9432e-05,
      "loss": 0.0084,
      "step": 11707
    },
    {
      "epoch": 21.602031394275162,
      "grad_norm": 0.6964510083198547,
      "learning_rate": 2.9431666666666668e-05,
      "loss": 0.0116,
      "step": 11708
    },
    {
      "epoch": 21.60387811634349,
      "grad_norm": 0.30797988176345825,
      "learning_rate": 2.9431333333333334e-05,
      "loss": 0.0083,
      "step": 11709
    },
    {
      "epoch": 21.60572483841182,
      "grad_norm": 0.23163697123527527,
      "learning_rate": 2.9431e-05,
      "loss": 0.009,
      "step": 11710
    },
    {
      "epoch": 21.607571560480146,
      "grad_norm": 0.4534711241722107,
      "learning_rate": 2.943066666666667e-05,
      "loss": 0.0165,
      "step": 11711
    },
    {
      "epoch": 21.609418282548475,
      "grad_norm": 0.29898273944854736,
      "learning_rate": 2.9430333333333332e-05,
      "loss": 0.0096,
      "step": 11712
    },
    {
      "epoch": 21.611265004616804,
      "grad_norm": 0.3022543787956238,
      "learning_rate": 2.943e-05,
      "loss": 0.0371,
      "step": 11713
    },
    {
      "epoch": 21.613111726685133,
      "grad_norm": 0.241398423910141,
      "learning_rate": 2.9429666666666667e-05,
      "loss": 0.0057,
      "step": 11714
    },
    {
      "epoch": 21.614958448753463,
      "grad_norm": 0.15573015809059143,
      "learning_rate": 2.9429333333333333e-05,
      "loss": 0.0061,
      "step": 11715
    },
    {
      "epoch": 21.61680517082179,
      "grad_norm": 0.15038542449474335,
      "learning_rate": 2.9429e-05,
      "loss": 0.0046,
      "step": 11716
    },
    {
      "epoch": 21.61865189289012,
      "grad_norm": 0.3136591911315918,
      "learning_rate": 2.9428666666666668e-05,
      "loss": 0.0084,
      "step": 11717
    },
    {
      "epoch": 21.62049861495845,
      "grad_norm": 0.14817532896995544,
      "learning_rate": 2.9428333333333334e-05,
      "loss": 0.0038,
      "step": 11718
    },
    {
      "epoch": 21.62234533702678,
      "grad_norm": 0.1666664481163025,
      "learning_rate": 2.9428e-05,
      "loss": 0.0081,
      "step": 11719
    },
    {
      "epoch": 21.624192059095105,
      "grad_norm": 0.10841329395771027,
      "learning_rate": 2.942766666666667e-05,
      "loss": 0.0065,
      "step": 11720
    },
    {
      "epoch": 21.626038781163434,
      "grad_norm": 0.28235870599746704,
      "learning_rate": 2.9427333333333332e-05,
      "loss": 0.0103,
      "step": 11721
    },
    {
      "epoch": 21.627885503231763,
      "grad_norm": 0.12024784833192825,
      "learning_rate": 2.9427e-05,
      "loss": 0.0035,
      "step": 11722
    },
    {
      "epoch": 21.629732225300092,
      "grad_norm": 0.4908735752105713,
      "learning_rate": 2.9426666666666667e-05,
      "loss": 0.0067,
      "step": 11723
    },
    {
      "epoch": 21.63157894736842,
      "grad_norm": 0.21267184615135193,
      "learning_rate": 2.9426333333333333e-05,
      "loss": 0.0056,
      "step": 11724
    },
    {
      "epoch": 21.63342566943675,
      "grad_norm": 0.42651426792144775,
      "learning_rate": 2.9426e-05,
      "loss": 0.0078,
      "step": 11725
    },
    {
      "epoch": 21.63527239150508,
      "grad_norm": 0.16636477410793304,
      "learning_rate": 2.942566666666667e-05,
      "loss": 0.008,
      "step": 11726
    },
    {
      "epoch": 21.63711911357341,
      "grad_norm": 0.26599639654159546,
      "learning_rate": 2.9425333333333334e-05,
      "loss": 0.0101,
      "step": 11727
    },
    {
      "epoch": 21.638965835641734,
      "grad_norm": 0.2585376799106598,
      "learning_rate": 2.9425e-05,
      "loss": 0.0079,
      "step": 11728
    },
    {
      "epoch": 21.640812557710063,
      "grad_norm": 0.5444880723953247,
      "learning_rate": 2.942466666666667e-05,
      "loss": 0.0138,
      "step": 11729
    },
    {
      "epoch": 21.642659279778393,
      "grad_norm": 0.9459370970726013,
      "learning_rate": 2.9424333333333332e-05,
      "loss": 0.0199,
      "step": 11730
    },
    {
      "epoch": 21.64450600184672,
      "grad_norm": 0.5683426260948181,
      "learning_rate": 2.9424e-05,
      "loss": 0.0075,
      "step": 11731
    },
    {
      "epoch": 21.64635272391505,
      "grad_norm": 0.42870140075683594,
      "learning_rate": 2.9423666666666667e-05,
      "loss": 0.0194,
      "step": 11732
    },
    {
      "epoch": 21.64819944598338,
      "grad_norm": 0.48324406147003174,
      "learning_rate": 2.9423333333333333e-05,
      "loss": 0.1129,
      "step": 11733
    },
    {
      "epoch": 21.65004616805171,
      "grad_norm": 0.5933670997619629,
      "learning_rate": 2.9423e-05,
      "loss": 0.0805,
      "step": 11734
    },
    {
      "epoch": 21.65189289012004,
      "grad_norm": 0.9176157116889954,
      "learning_rate": 2.942266666666667e-05,
      "loss": 0.0999,
      "step": 11735
    },
    {
      "epoch": 21.653739612188367,
      "grad_norm": 0.5372908115386963,
      "learning_rate": 2.9422333333333335e-05,
      "loss": 0.1085,
      "step": 11736
    },
    {
      "epoch": 21.655586334256693,
      "grad_norm": 0.9991260766983032,
      "learning_rate": 2.9422e-05,
      "loss": 0.1169,
      "step": 11737
    },
    {
      "epoch": 21.657433056325022,
      "grad_norm": 0.4064606726169586,
      "learning_rate": 2.942166666666667e-05,
      "loss": 0.0742,
      "step": 11738
    },
    {
      "epoch": 21.65927977839335,
      "grad_norm": 0.406076580286026,
      "learning_rate": 2.9421333333333332e-05,
      "loss": 0.0398,
      "step": 11739
    },
    {
      "epoch": 21.66112650046168,
      "grad_norm": 0.976673424243927,
      "learning_rate": 2.9421000000000002e-05,
      "loss": 0.0725,
      "step": 11740
    },
    {
      "epoch": 21.66297322253001,
      "grad_norm": 0.4736194908618927,
      "learning_rate": 2.9420666666666668e-05,
      "loss": 0.0688,
      "step": 11741
    },
    {
      "epoch": 21.66481994459834,
      "grad_norm": 0.374492347240448,
      "learning_rate": 2.9420333333333334e-05,
      "loss": 0.0372,
      "step": 11742
    },
    {
      "epoch": 21.666666666666668,
      "grad_norm": 0.31480005383491516,
      "learning_rate": 2.9420000000000003e-05,
      "loss": 0.0285,
      "step": 11743
    },
    {
      "epoch": 21.668513388734997,
      "grad_norm": 0.40599334239959717,
      "learning_rate": 2.9419666666666665e-05,
      "loss": 0.0283,
      "step": 11744
    },
    {
      "epoch": 21.670360110803323,
      "grad_norm": 0.594463050365448,
      "learning_rate": 2.9419333333333335e-05,
      "loss": 0.0769,
      "step": 11745
    },
    {
      "epoch": 21.67220683287165,
      "grad_norm": 0.35580018162727356,
      "learning_rate": 2.9419e-05,
      "loss": 0.0331,
      "step": 11746
    },
    {
      "epoch": 21.67405355493998,
      "grad_norm": 0.43213754892349243,
      "learning_rate": 2.9418666666666667e-05,
      "loss": 0.013,
      "step": 11747
    },
    {
      "epoch": 21.67590027700831,
      "grad_norm": 0.5789579749107361,
      "learning_rate": 2.9418333333333333e-05,
      "loss": 0.0095,
      "step": 11748
    },
    {
      "epoch": 21.67774699907664,
      "grad_norm": 0.26206323504447937,
      "learning_rate": 2.9418000000000002e-05,
      "loss": 0.0112,
      "step": 11749
    },
    {
      "epoch": 21.67959372114497,
      "grad_norm": 0.30900248885154724,
      "learning_rate": 2.9417666666666664e-05,
      "loss": 0.0095,
      "step": 11750
    },
    {
      "epoch": 21.681440443213297,
      "grad_norm": 0.1787906438112259,
      "learning_rate": 2.9417333333333334e-05,
      "loss": 0.0087,
      "step": 11751
    },
    {
      "epoch": 21.683287165281627,
      "grad_norm": 0.15706923604011536,
      "learning_rate": 2.9417000000000003e-05,
      "loss": 0.0085,
      "step": 11752
    },
    {
      "epoch": 21.685133887349952,
      "grad_norm": 0.160160630941391,
      "learning_rate": 2.9416666666666666e-05,
      "loss": 0.0179,
      "step": 11753
    },
    {
      "epoch": 21.68698060941828,
      "grad_norm": 0.3976205587387085,
      "learning_rate": 2.9416333333333335e-05,
      "loss": 0.0166,
      "step": 11754
    },
    {
      "epoch": 21.68882733148661,
      "grad_norm": 0.45126911997795105,
      "learning_rate": 2.9416e-05,
      "loss": 0.0144,
      "step": 11755
    },
    {
      "epoch": 21.69067405355494,
      "grad_norm": 0.24339599907398224,
      "learning_rate": 2.9415666666666667e-05,
      "loss": 0.0042,
      "step": 11756
    },
    {
      "epoch": 21.69252077562327,
      "grad_norm": 0.32987111806869507,
      "learning_rate": 2.9415333333333333e-05,
      "loss": 0.0105,
      "step": 11757
    },
    {
      "epoch": 21.694367497691598,
      "grad_norm": 0.39428460597991943,
      "learning_rate": 2.9415000000000002e-05,
      "loss": 0.0236,
      "step": 11758
    },
    {
      "epoch": 21.696214219759927,
      "grad_norm": 0.3154822289943695,
      "learning_rate": 2.9414666666666668e-05,
      "loss": 0.0122,
      "step": 11759
    },
    {
      "epoch": 21.698060941828256,
      "grad_norm": 0.3474328815937042,
      "learning_rate": 2.9414333333333334e-05,
      "loss": 0.0129,
      "step": 11760
    },
    {
      "epoch": 21.69990766389658,
      "grad_norm": 0.09453830868005753,
      "learning_rate": 2.9414000000000003e-05,
      "loss": 0.0029,
      "step": 11761
    },
    {
      "epoch": 21.70175438596491,
      "grad_norm": 0.32519862055778503,
      "learning_rate": 2.9413666666666666e-05,
      "loss": 0.0547,
      "step": 11762
    },
    {
      "epoch": 21.70360110803324,
      "grad_norm": 0.2847834825515747,
      "learning_rate": 2.9413333333333335e-05,
      "loss": 0.0051,
      "step": 11763
    },
    {
      "epoch": 21.70544783010157,
      "grad_norm": 0.4314049184322357,
      "learning_rate": 2.9413e-05,
      "loss": 0.0189,
      "step": 11764
    },
    {
      "epoch": 21.7072945521699,
      "grad_norm": 0.1875690221786499,
      "learning_rate": 2.9412666666666667e-05,
      "loss": 0.004,
      "step": 11765
    },
    {
      "epoch": 21.709141274238227,
      "grad_norm": 0.118055060505867,
      "learning_rate": 2.9412333333333333e-05,
      "loss": 0.0031,
      "step": 11766
    },
    {
      "epoch": 21.710987996306557,
      "grad_norm": 0.11226514726877213,
      "learning_rate": 2.9412000000000002e-05,
      "loss": 0.0041,
      "step": 11767
    },
    {
      "epoch": 21.712834718374886,
      "grad_norm": 0.3839562237262726,
      "learning_rate": 2.9411666666666668e-05,
      "loss": 0.0067,
      "step": 11768
    },
    {
      "epoch": 21.714681440443215,
      "grad_norm": 0.06365108489990234,
      "learning_rate": 2.9411333333333334e-05,
      "loss": 0.002,
      "step": 11769
    },
    {
      "epoch": 21.71652816251154,
      "grad_norm": 0.3811405301094055,
      "learning_rate": 2.9411000000000004e-05,
      "loss": 0.0096,
      "step": 11770
    },
    {
      "epoch": 21.71837488457987,
      "grad_norm": 0.4699449837207794,
      "learning_rate": 2.9410666666666666e-05,
      "loss": 0.0139,
      "step": 11771
    },
    {
      "epoch": 21.7202216066482,
      "grad_norm": 0.26368317008018494,
      "learning_rate": 2.9410333333333335e-05,
      "loss": 0.0096,
      "step": 11772
    },
    {
      "epoch": 21.722068328716528,
      "grad_norm": 0.14234060049057007,
      "learning_rate": 2.9409999999999998e-05,
      "loss": 0.003,
      "step": 11773
    },
    {
      "epoch": 21.723915050784857,
      "grad_norm": 0.6422532200813293,
      "learning_rate": 2.9409666666666667e-05,
      "loss": 0.0093,
      "step": 11774
    },
    {
      "epoch": 21.725761772853186,
      "grad_norm": 0.22176691889762878,
      "learning_rate": 2.9409333333333333e-05,
      "loss": 0.0046,
      "step": 11775
    },
    {
      "epoch": 21.727608494921515,
      "grad_norm": 0.30540087819099426,
      "learning_rate": 2.9409e-05,
      "loss": 0.0061,
      "step": 11776
    },
    {
      "epoch": 21.729455216989845,
      "grad_norm": 0.5490514039993286,
      "learning_rate": 2.940866666666667e-05,
      "loss": 0.0117,
      "step": 11777
    },
    {
      "epoch": 21.73130193905817,
      "grad_norm": 0.3425533175468445,
      "learning_rate": 2.9408333333333334e-05,
      "loss": 0.0123,
      "step": 11778
    },
    {
      "epoch": 21.7331486611265,
      "grad_norm": 0.2820558249950409,
      "learning_rate": 2.9408e-05,
      "loss": 0.0091,
      "step": 11779
    },
    {
      "epoch": 21.73499538319483,
      "grad_norm": 0.20925366878509521,
      "learning_rate": 2.9407666666666666e-05,
      "loss": 0.0083,
      "step": 11780
    },
    {
      "epoch": 21.736842105263158,
      "grad_norm": 0.22122719883918762,
      "learning_rate": 2.9407333333333336e-05,
      "loss": 0.0082,
      "step": 11781
    },
    {
      "epoch": 21.738688827331487,
      "grad_norm": 0.36879807710647583,
      "learning_rate": 2.9406999999999998e-05,
      "loss": 0.0145,
      "step": 11782
    },
    {
      "epoch": 21.740535549399816,
      "grad_norm": 0.7262082695960999,
      "learning_rate": 2.9406666666666667e-05,
      "loss": 0.1864,
      "step": 11783
    },
    {
      "epoch": 21.742382271468145,
      "grad_norm": 0.5911471247673035,
      "learning_rate": 2.9406333333333333e-05,
      "loss": 0.1308,
      "step": 11784
    },
    {
      "epoch": 21.744228993536474,
      "grad_norm": 0.5021706819534302,
      "learning_rate": 2.9406e-05,
      "loss": 0.1253,
      "step": 11785
    },
    {
      "epoch": 21.746075715604803,
      "grad_norm": 0.42258352041244507,
      "learning_rate": 2.940566666666667e-05,
      "loss": 0.1105,
      "step": 11786
    },
    {
      "epoch": 21.74792243767313,
      "grad_norm": 0.4874754250049591,
      "learning_rate": 2.9405333333333335e-05,
      "loss": 0.0881,
      "step": 11787
    },
    {
      "epoch": 21.749769159741458,
      "grad_norm": 0.41108840703964233,
      "learning_rate": 2.9405e-05,
      "loss": 0.0996,
      "step": 11788
    },
    {
      "epoch": 21.751615881809787,
      "grad_norm": 0.4531548321247101,
      "learning_rate": 2.9404666666666666e-05,
      "loss": 0.0555,
      "step": 11789
    },
    {
      "epoch": 21.753462603878116,
      "grad_norm": 0.6766632795333862,
      "learning_rate": 2.9404333333333336e-05,
      "loss": 0.055,
      "step": 11790
    },
    {
      "epoch": 21.755309325946445,
      "grad_norm": 0.36510834097862244,
      "learning_rate": 2.9404e-05,
      "loss": 0.0607,
      "step": 11791
    },
    {
      "epoch": 21.757156048014775,
      "grad_norm": 0.38905462622642517,
      "learning_rate": 2.9403666666666668e-05,
      "loss": 0.0764,
      "step": 11792
    },
    {
      "epoch": 21.759002770083104,
      "grad_norm": 0.33395567536354065,
      "learning_rate": 2.9403333333333337e-05,
      "loss": 0.0285,
      "step": 11793
    },
    {
      "epoch": 21.760849492151433,
      "grad_norm": 0.5528409481048584,
      "learning_rate": 2.9403e-05,
      "loss": 0.0547,
      "step": 11794
    },
    {
      "epoch": 21.76269621421976,
      "grad_norm": 0.3216876685619354,
      "learning_rate": 2.940266666666667e-05,
      "loss": 0.017,
      "step": 11795
    },
    {
      "epoch": 21.764542936288088,
      "grad_norm": 0.33285337686538696,
      "learning_rate": 2.9402333333333335e-05,
      "loss": 0.0276,
      "step": 11796
    },
    {
      "epoch": 21.766389658356417,
      "grad_norm": 0.33139219880104065,
      "learning_rate": 2.9402e-05,
      "loss": 0.0302,
      "step": 11797
    },
    {
      "epoch": 21.768236380424746,
      "grad_norm": 0.19674275815486908,
      "learning_rate": 2.9401666666666667e-05,
      "loss": 0.0115,
      "step": 11798
    },
    {
      "epoch": 21.770083102493075,
      "grad_norm": 0.3641238510608673,
      "learning_rate": 2.9401333333333336e-05,
      "loss": 0.0119,
      "step": 11799
    },
    {
      "epoch": 21.771929824561404,
      "grad_norm": 0.24481379985809326,
      "learning_rate": 2.9401e-05,
      "loss": 0.0083,
      "step": 11800
    },
    {
      "epoch": 21.773776546629733,
      "grad_norm": 0.3730131685733795,
      "learning_rate": 2.9400666666666668e-05,
      "loss": 0.041,
      "step": 11801
    },
    {
      "epoch": 21.775623268698062,
      "grad_norm": 0.24020326137542725,
      "learning_rate": 2.9400333333333337e-05,
      "loss": 0.0117,
      "step": 11802
    },
    {
      "epoch": 21.777469990766388,
      "grad_norm": 0.16403193771839142,
      "learning_rate": 2.94e-05,
      "loss": 0.0082,
      "step": 11803
    },
    {
      "epoch": 21.779316712834717,
      "grad_norm": 0.19418635964393616,
      "learning_rate": 2.939966666666667e-05,
      "loss": 0.0092,
      "step": 11804
    },
    {
      "epoch": 21.781163434903046,
      "grad_norm": 0.2227737009525299,
      "learning_rate": 2.939933333333333e-05,
      "loss": 0.0083,
      "step": 11805
    },
    {
      "epoch": 21.783010156971375,
      "grad_norm": 0.2907804250717163,
      "learning_rate": 2.9399e-05,
      "loss": 0.0072,
      "step": 11806
    },
    {
      "epoch": 21.784856879039705,
      "grad_norm": 0.24682292342185974,
      "learning_rate": 2.9398666666666667e-05,
      "loss": 0.0126,
      "step": 11807
    },
    {
      "epoch": 21.786703601108034,
      "grad_norm": 0.31712350249290466,
      "learning_rate": 2.9398333333333333e-05,
      "loss": 0.0087,
      "step": 11808
    },
    {
      "epoch": 21.788550323176363,
      "grad_norm": 0.7004305720329285,
      "learning_rate": 2.9398000000000002e-05,
      "loss": 0.0182,
      "step": 11809
    },
    {
      "epoch": 21.790397045244692,
      "grad_norm": 0.204494446516037,
      "learning_rate": 2.9397666666666668e-05,
      "loss": 0.0048,
      "step": 11810
    },
    {
      "epoch": 21.792243767313018,
      "grad_norm": 0.3658459782600403,
      "learning_rate": 2.9397333333333334e-05,
      "loss": 0.0273,
      "step": 11811
    },
    {
      "epoch": 21.794090489381347,
      "grad_norm": 0.20716539025306702,
      "learning_rate": 2.9397e-05,
      "loss": 0.0075,
      "step": 11812
    },
    {
      "epoch": 21.795937211449676,
      "grad_norm": 0.1901579350233078,
      "learning_rate": 2.939666666666667e-05,
      "loss": 0.0103,
      "step": 11813
    },
    {
      "epoch": 21.797783933518005,
      "grad_norm": 0.07081008702516556,
      "learning_rate": 2.9396333333333332e-05,
      "loss": 0.0021,
      "step": 11814
    },
    {
      "epoch": 21.799630655586334,
      "grad_norm": 0.7699052095413208,
      "learning_rate": 2.9396e-05,
      "loss": 0.0122,
      "step": 11815
    },
    {
      "epoch": 21.801477377654663,
      "grad_norm": 0.12213122844696045,
      "learning_rate": 2.9395666666666667e-05,
      "loss": 0.0033,
      "step": 11816
    },
    {
      "epoch": 21.803324099722992,
      "grad_norm": 0.3606746792793274,
      "learning_rate": 2.9395333333333333e-05,
      "loss": 0.0054,
      "step": 11817
    },
    {
      "epoch": 21.80517082179132,
      "grad_norm": 0.24104273319244385,
      "learning_rate": 2.9395000000000002e-05,
      "loss": 0.0064,
      "step": 11818
    },
    {
      "epoch": 21.80701754385965,
      "grad_norm": 0.2115449160337448,
      "learning_rate": 2.9394666666666668e-05,
      "loss": 0.029,
      "step": 11819
    },
    {
      "epoch": 21.808864265927976,
      "grad_norm": 0.8519856929779053,
      "learning_rate": 2.9394333333333334e-05,
      "loss": 0.0106,
      "step": 11820
    },
    {
      "epoch": 21.810710987996305,
      "grad_norm": 0.44638770818710327,
      "learning_rate": 2.9394e-05,
      "loss": 0.0109,
      "step": 11821
    },
    {
      "epoch": 21.812557710064635,
      "grad_norm": 0.33324283361434937,
      "learning_rate": 2.939366666666667e-05,
      "loss": 0.0129,
      "step": 11822
    },
    {
      "epoch": 21.814404432132964,
      "grad_norm": 0.10796014219522476,
      "learning_rate": 2.9393333333333332e-05,
      "loss": 0.0025,
      "step": 11823
    },
    {
      "epoch": 21.816251154201293,
      "grad_norm": 0.3488299250602722,
      "learning_rate": 2.9393e-05,
      "loss": 0.0109,
      "step": 11824
    },
    {
      "epoch": 21.818097876269622,
      "grad_norm": 0.17116694152355194,
      "learning_rate": 2.9392666666666667e-05,
      "loss": 0.0043,
      "step": 11825
    },
    {
      "epoch": 21.81994459833795,
      "grad_norm": 0.44072869420051575,
      "learning_rate": 2.9392333333333333e-05,
      "loss": 0.0129,
      "step": 11826
    },
    {
      "epoch": 21.82179132040628,
      "grad_norm": 0.2730376422405243,
      "learning_rate": 2.9392000000000003e-05,
      "loss": 0.0067,
      "step": 11827
    },
    {
      "epoch": 21.823638042474606,
      "grad_norm": 0.5067234635353088,
      "learning_rate": 2.939166666666667e-05,
      "loss": 0.0134,
      "step": 11828
    },
    {
      "epoch": 21.825484764542935,
      "grad_norm": 0.21032480895519257,
      "learning_rate": 2.9391333333333334e-05,
      "loss": 0.007,
      "step": 11829
    },
    {
      "epoch": 21.827331486611264,
      "grad_norm": 0.5482940077781677,
      "learning_rate": 2.9391e-05,
      "loss": 0.0253,
      "step": 11830
    },
    {
      "epoch": 21.829178208679593,
      "grad_norm": 1.3305041790008545,
      "learning_rate": 2.939066666666667e-05,
      "loss": 0.0256,
      "step": 11831
    },
    {
      "epoch": 21.831024930747922,
      "grad_norm": 0.6527359485626221,
      "learning_rate": 2.9390333333333332e-05,
      "loss": 0.0083,
      "step": 11832
    },
    {
      "epoch": 21.83287165281625,
      "grad_norm": 0.6475838422775269,
      "learning_rate": 2.939e-05,
      "loss": 0.202,
      "step": 11833
    },
    {
      "epoch": 21.83471837488458,
      "grad_norm": 0.5826760530471802,
      "learning_rate": 2.9389666666666667e-05,
      "loss": 0.1203,
      "step": 11834
    },
    {
      "epoch": 21.83656509695291,
      "grad_norm": 0.5271478295326233,
      "learning_rate": 2.9389333333333333e-05,
      "loss": 0.0983,
      "step": 11835
    },
    {
      "epoch": 21.83841181902124,
      "grad_norm": 0.4728860557079315,
      "learning_rate": 2.9389000000000003e-05,
      "loss": 0.0835,
      "step": 11836
    },
    {
      "epoch": 21.840258541089565,
      "grad_norm": 0.4163111746311188,
      "learning_rate": 2.9388666666666665e-05,
      "loss": 0.0544,
      "step": 11837
    },
    {
      "epoch": 21.842105263157894,
      "grad_norm": 0.8589546084403992,
      "learning_rate": 2.9388333333333335e-05,
      "loss": 0.056,
      "step": 11838
    },
    {
      "epoch": 21.843951985226223,
      "grad_norm": 0.3603813350200653,
      "learning_rate": 2.9388e-05,
      "loss": 0.0418,
      "step": 11839
    },
    {
      "epoch": 21.845798707294552,
      "grad_norm": 0.4066253900527954,
      "learning_rate": 2.9387666666666666e-05,
      "loss": 0.0759,
      "step": 11840
    },
    {
      "epoch": 21.84764542936288,
      "grad_norm": 0.35721030831336975,
      "learning_rate": 2.9387333333333332e-05,
      "loss": 0.0442,
      "step": 11841
    },
    {
      "epoch": 21.84949215143121,
      "grad_norm": 0.36084261536598206,
      "learning_rate": 2.9387000000000002e-05,
      "loss": 0.0407,
      "step": 11842
    },
    {
      "epoch": 21.85133887349954,
      "grad_norm": 0.7102795839309692,
      "learning_rate": 2.9386666666666668e-05,
      "loss": 0.0429,
      "step": 11843
    },
    {
      "epoch": 21.85318559556787,
      "grad_norm": 0.29509997367858887,
      "learning_rate": 2.9386333333333334e-05,
      "loss": 0.0271,
      "step": 11844
    },
    {
      "epoch": 21.855032317636194,
      "grad_norm": 0.4171082079410553,
      "learning_rate": 2.9386000000000003e-05,
      "loss": 0.0446,
      "step": 11845
    },
    {
      "epoch": 21.856879039704523,
      "grad_norm": 0.30098018050193787,
      "learning_rate": 2.9385666666666665e-05,
      "loss": 0.0382,
      "step": 11846
    },
    {
      "epoch": 21.858725761772853,
      "grad_norm": 0.3387888967990875,
      "learning_rate": 2.9385333333333335e-05,
      "loss": 0.0571,
      "step": 11847
    },
    {
      "epoch": 21.86057248384118,
      "grad_norm": 0.25202545523643494,
      "learning_rate": 2.9385e-05,
      "loss": 0.0177,
      "step": 11848
    },
    {
      "epoch": 21.86241920590951,
      "grad_norm": 0.3939182758331299,
      "learning_rate": 2.9384666666666667e-05,
      "loss": 0.0096,
      "step": 11849
    },
    {
      "epoch": 21.86426592797784,
      "grad_norm": 0.16526532173156738,
      "learning_rate": 2.9384333333333333e-05,
      "loss": 0.0074,
      "step": 11850
    },
    {
      "epoch": 21.86611265004617,
      "grad_norm": 0.3572503626346588,
      "learning_rate": 2.9384000000000002e-05,
      "loss": 0.0159,
      "step": 11851
    },
    {
      "epoch": 21.8679593721145,
      "grad_norm": 0.4337249994277954,
      "learning_rate": 2.9383666666666668e-05,
      "loss": 0.041,
      "step": 11852
    },
    {
      "epoch": 21.869806094182824,
      "grad_norm": 0.26018592715263367,
      "learning_rate": 2.9383333333333334e-05,
      "loss": 0.0259,
      "step": 11853
    },
    {
      "epoch": 21.871652816251153,
      "grad_norm": 0.2274974286556244,
      "learning_rate": 2.9383000000000003e-05,
      "loss": 0.012,
      "step": 11854
    },
    {
      "epoch": 21.873499538319482,
      "grad_norm": 0.4401700496673584,
      "learning_rate": 2.9382666666666666e-05,
      "loss": 0.0077,
      "step": 11855
    },
    {
      "epoch": 21.87534626038781,
      "grad_norm": 0.4011286795139313,
      "learning_rate": 2.9382333333333335e-05,
      "loss": 0.0135,
      "step": 11856
    },
    {
      "epoch": 21.87719298245614,
      "grad_norm": 0.1697901487350464,
      "learning_rate": 2.9382e-05,
      "loss": 0.007,
      "step": 11857
    },
    {
      "epoch": 21.87903970452447,
      "grad_norm": 0.5393462181091309,
      "learning_rate": 2.9381666666666667e-05,
      "loss": 0.0145,
      "step": 11858
    },
    {
      "epoch": 21.8808864265928,
      "grad_norm": 0.2966288626194,
      "learning_rate": 2.9381333333333336e-05,
      "loss": 0.008,
      "step": 11859
    },
    {
      "epoch": 21.882733148661128,
      "grad_norm": 0.35856500267982483,
      "learning_rate": 2.9381000000000002e-05,
      "loss": 0.0425,
      "step": 11860
    },
    {
      "epoch": 21.884579870729453,
      "grad_norm": 0.23906317353248596,
      "learning_rate": 2.9380666666666668e-05,
      "loss": 0.0113,
      "step": 11861
    },
    {
      "epoch": 21.886426592797783,
      "grad_norm": 0.2855496108531952,
      "learning_rate": 2.9380333333333334e-05,
      "loss": 0.0099,
      "step": 11862
    },
    {
      "epoch": 21.88827331486611,
      "grad_norm": 0.5273758769035339,
      "learning_rate": 2.938e-05,
      "loss": 0.0099,
      "step": 11863
    },
    {
      "epoch": 21.89012003693444,
      "grad_norm": 0.2559138536453247,
      "learning_rate": 2.9379666666666666e-05,
      "loss": 0.0084,
      "step": 11864
    },
    {
      "epoch": 21.89196675900277,
      "grad_norm": 0.31710177659988403,
      "learning_rate": 2.9379333333333335e-05,
      "loss": 0.0111,
      "step": 11865
    },
    {
      "epoch": 21.8938134810711,
      "grad_norm": 0.32185274362564087,
      "learning_rate": 2.9378999999999998e-05,
      "loss": 0.0124,
      "step": 11866
    },
    {
      "epoch": 21.89566020313943,
      "grad_norm": 0.21019087731838226,
      "learning_rate": 2.9378666666666667e-05,
      "loss": 0.0072,
      "step": 11867
    },
    {
      "epoch": 21.897506925207757,
      "grad_norm": 0.25909486413002014,
      "learning_rate": 2.9378333333333336e-05,
      "loss": 0.0059,
      "step": 11868
    },
    {
      "epoch": 21.899353647276087,
      "grad_norm": 0.3391138017177582,
      "learning_rate": 2.9378e-05,
      "loss": 0.0408,
      "step": 11869
    },
    {
      "epoch": 21.901200369344412,
      "grad_norm": 0.4207199215888977,
      "learning_rate": 2.9377666666666668e-05,
      "loss": 0.0093,
      "step": 11870
    },
    {
      "epoch": 21.90304709141274,
      "grad_norm": 0.4389612078666687,
      "learning_rate": 2.9377333333333334e-05,
      "loss": 0.0103,
      "step": 11871
    },
    {
      "epoch": 21.90489381348107,
      "grad_norm": 0.4361323416233063,
      "learning_rate": 2.9377e-05,
      "loss": 0.0186,
      "step": 11872
    },
    {
      "epoch": 21.9067405355494,
      "grad_norm": 0.2915911078453064,
      "learning_rate": 2.9376666666666666e-05,
      "loss": 0.008,
      "step": 11873
    },
    {
      "epoch": 21.90858725761773,
      "grad_norm": 0.43618321418762207,
      "learning_rate": 2.9376333333333335e-05,
      "loss": 0.0159,
      "step": 11874
    },
    {
      "epoch": 21.910433979686058,
      "grad_norm": 0.2591547966003418,
      "learning_rate": 2.9375999999999998e-05,
      "loss": 0.0112,
      "step": 11875
    },
    {
      "epoch": 21.912280701754387,
      "grad_norm": 0.1294482797384262,
      "learning_rate": 2.9375666666666667e-05,
      "loss": 0.0045,
      "step": 11876
    },
    {
      "epoch": 21.914127423822716,
      "grad_norm": 0.23369941115379333,
      "learning_rate": 2.9375333333333337e-05,
      "loss": 0.0107,
      "step": 11877
    },
    {
      "epoch": 21.91597414589104,
      "grad_norm": 0.2956857979297638,
      "learning_rate": 2.9375e-05,
      "loss": 0.0092,
      "step": 11878
    },
    {
      "epoch": 21.91782086795937,
      "grad_norm": 0.21449799835681915,
      "learning_rate": 2.937466666666667e-05,
      "loss": 0.0053,
      "step": 11879
    },
    {
      "epoch": 21.9196675900277,
      "grad_norm": 0.2647559344768524,
      "learning_rate": 2.9374333333333334e-05,
      "loss": 0.0087,
      "step": 11880
    },
    {
      "epoch": 21.92151431209603,
      "grad_norm": 0.16165274381637573,
      "learning_rate": 2.9374e-05,
      "loss": 0.0032,
      "step": 11881
    },
    {
      "epoch": 21.92336103416436,
      "grad_norm": 0.5226666331291199,
      "learning_rate": 2.9373666666666666e-05,
      "loss": 0.0183,
      "step": 11882
    },
    {
      "epoch": 21.925207756232687,
      "grad_norm": 0.6835513114929199,
      "learning_rate": 2.9373333333333336e-05,
      "loss": 0.1625,
      "step": 11883
    },
    {
      "epoch": 21.927054478301017,
      "grad_norm": 0.7763351202011108,
      "learning_rate": 2.9373e-05,
      "loss": 0.1565,
      "step": 11884
    },
    {
      "epoch": 21.928901200369346,
      "grad_norm": 0.43676137924194336,
      "learning_rate": 2.9372666666666667e-05,
      "loss": 0.0685,
      "step": 11885
    },
    {
      "epoch": 21.930747922437675,
      "grad_norm": 0.47941815853118896,
      "learning_rate": 2.9372333333333337e-05,
      "loss": 0.0891,
      "step": 11886
    },
    {
      "epoch": 21.932594644506,
      "grad_norm": 0.4046361744403839,
      "learning_rate": 2.9372e-05,
      "loss": 0.0809,
      "step": 11887
    },
    {
      "epoch": 21.93444136657433,
      "grad_norm": 0.39610689878463745,
      "learning_rate": 2.937166666666667e-05,
      "loss": 0.0526,
      "step": 11888
    },
    {
      "epoch": 21.93628808864266,
      "grad_norm": 0.6374268531799316,
      "learning_rate": 2.9371333333333335e-05,
      "loss": 0.0754,
      "step": 11889
    },
    {
      "epoch": 21.938134810710988,
      "grad_norm": 0.3975745439529419,
      "learning_rate": 2.9371e-05,
      "loss": 0.0746,
      "step": 11890
    },
    {
      "epoch": 21.939981532779317,
      "grad_norm": 0.5438203811645508,
      "learning_rate": 2.9370666666666666e-05,
      "loss": 0.0442,
      "step": 11891
    },
    {
      "epoch": 21.941828254847646,
      "grad_norm": 0.883966326713562,
      "learning_rate": 2.9370333333333336e-05,
      "loss": 0.0496,
      "step": 11892
    },
    {
      "epoch": 21.943674976915975,
      "grad_norm": 0.3531246781349182,
      "learning_rate": 2.9370000000000002e-05,
      "loss": 0.0282,
      "step": 11893
    },
    {
      "epoch": 21.945521698984304,
      "grad_norm": 0.2855888307094574,
      "learning_rate": 2.9369666666666668e-05,
      "loss": 0.0433,
      "step": 11894
    },
    {
      "epoch": 21.94736842105263,
      "grad_norm": 0.3364258110523224,
      "learning_rate": 2.9369333333333334e-05,
      "loss": 0.0136,
      "step": 11895
    },
    {
      "epoch": 21.94921514312096,
      "grad_norm": 0.1887637972831726,
      "learning_rate": 2.9369e-05,
      "loss": 0.0134,
      "step": 11896
    },
    {
      "epoch": 21.95106186518929,
      "grad_norm": 0.4127003252506256,
      "learning_rate": 2.936866666666667e-05,
      "loss": 0.0344,
      "step": 11897
    },
    {
      "epoch": 21.952908587257618,
      "grad_norm": 0.22459253668785095,
      "learning_rate": 2.936833333333333e-05,
      "loss": 0.0133,
      "step": 11898
    },
    {
      "epoch": 21.954755309325947,
      "grad_norm": 0.33327922224998474,
      "learning_rate": 2.9368e-05,
      "loss": 0.0099,
      "step": 11899
    },
    {
      "epoch": 21.956602031394276,
      "grad_norm": 0.26645731925964355,
      "learning_rate": 2.9367666666666667e-05,
      "loss": 0.0152,
      "step": 11900
    },
    {
      "epoch": 21.958448753462605,
      "grad_norm": 0.34843745827674866,
      "learning_rate": 2.9367333333333333e-05,
      "loss": 0.0189,
      "step": 11901
    },
    {
      "epoch": 21.960295475530934,
      "grad_norm": 0.26357918977737427,
      "learning_rate": 2.9367000000000002e-05,
      "loss": 0.0073,
      "step": 11902
    },
    {
      "epoch": 21.96214219759926,
      "grad_norm": 0.31214573979377747,
      "learning_rate": 2.9366666666666668e-05,
      "loss": 0.0131,
      "step": 11903
    },
    {
      "epoch": 21.96398891966759,
      "grad_norm": 0.2717858850955963,
      "learning_rate": 2.9366333333333334e-05,
      "loss": 0.0073,
      "step": 11904
    },
    {
      "epoch": 21.965835641735918,
      "grad_norm": 0.16268011927604675,
      "learning_rate": 2.9366e-05,
      "loss": 0.0041,
      "step": 11905
    },
    {
      "epoch": 21.967682363804247,
      "grad_norm": 0.23915761709213257,
      "learning_rate": 2.936566666666667e-05,
      "loss": 0.0089,
      "step": 11906
    },
    {
      "epoch": 21.969529085872576,
      "grad_norm": 0.1611839383840561,
      "learning_rate": 2.936533333333333e-05,
      "loss": 0.0058,
      "step": 11907
    },
    {
      "epoch": 21.971375807940905,
      "grad_norm": 0.6211763024330139,
      "learning_rate": 2.9365e-05,
      "loss": 0.0098,
      "step": 11908
    },
    {
      "epoch": 21.973222530009235,
      "grad_norm": 0.8325933814048767,
      "learning_rate": 2.936466666666667e-05,
      "loss": 0.0118,
      "step": 11909
    },
    {
      "epoch": 21.975069252077564,
      "grad_norm": 0.17792974412441254,
      "learning_rate": 2.9364333333333333e-05,
      "loss": 0.0048,
      "step": 11910
    },
    {
      "epoch": 21.97691597414589,
      "grad_norm": 0.17959211766719818,
      "learning_rate": 2.9364000000000002e-05,
      "loss": 0.0082,
      "step": 11911
    },
    {
      "epoch": 21.97876269621422,
      "grad_norm": 0.23362372815608978,
      "learning_rate": 2.9363666666666668e-05,
      "loss": 0.0053,
      "step": 11912
    },
    {
      "epoch": 21.980609418282548,
      "grad_norm": 0.39807364344596863,
      "learning_rate": 2.9363333333333334e-05,
      "loss": 0.0072,
      "step": 11913
    },
    {
      "epoch": 21.982456140350877,
      "grad_norm": 0.411865770816803,
      "learning_rate": 2.9363e-05,
      "loss": 0.0095,
      "step": 11914
    },
    {
      "epoch": 21.984302862419206,
      "grad_norm": 0.41350361704826355,
      "learning_rate": 2.936266666666667e-05,
      "loss": 0.0128,
      "step": 11915
    },
    {
      "epoch": 21.986149584487535,
      "grad_norm": 0.24316683411598206,
      "learning_rate": 2.9362333333333332e-05,
      "loss": 0.0054,
      "step": 11916
    },
    {
      "epoch": 21.987996306555864,
      "grad_norm": 0.21313484013080597,
      "learning_rate": 2.9362e-05,
      "loss": 0.0042,
      "step": 11917
    },
    {
      "epoch": 21.989843028624193,
      "grad_norm": 0.11858999729156494,
      "learning_rate": 2.936166666666667e-05,
      "loss": 0.0046,
      "step": 11918
    },
    {
      "epoch": 21.991689750692522,
      "grad_norm": 0.13938215374946594,
      "learning_rate": 2.9361333333333333e-05,
      "loss": 0.004,
      "step": 11919
    },
    {
      "epoch": 21.993536472760848,
      "grad_norm": 0.23660410940647125,
      "learning_rate": 2.9361000000000002e-05,
      "loss": 0.0073,
      "step": 11920
    },
    {
      "epoch": 21.995383194829177,
      "grad_norm": 0.29818493127822876,
      "learning_rate": 2.9360666666666668e-05,
      "loss": 0.0094,
      "step": 11921
    },
    {
      "epoch": 21.997229916897506,
      "grad_norm": 0.2225433886051178,
      "learning_rate": 2.9360333333333334e-05,
      "loss": 0.006,
      "step": 11922
    },
    {
      "epoch": 21.999076638965835,
      "grad_norm": 0.5248429775238037,
      "learning_rate": 2.936e-05,
      "loss": 0.0093,
      "step": 11923
    },
    {
      "epoch": 22.0,
      "grad_norm": 0.4427490532398224,
      "learning_rate": 2.9359666666666666e-05,
      "loss": 0.0058,
      "step": 11924
    },
    {
      "epoch": 22.00184672206833,
      "grad_norm": 0.8038821220397949,
      "learning_rate": 2.9359333333333332e-05,
      "loss": 0.2321,
      "step": 11925
    },
    {
      "epoch": 22.00369344413666,
      "grad_norm": 0.7266762256622314,
      "learning_rate": 2.9359e-05,
      "loss": 0.1044,
      "step": 11926
    },
    {
      "epoch": 22.005540166204987,
      "grad_norm": 0.5080312490463257,
      "learning_rate": 2.9358666666666667e-05,
      "loss": 0.1741,
      "step": 11927
    },
    {
      "epoch": 22.007386888273317,
      "grad_norm": 0.4087110459804535,
      "learning_rate": 2.9358333333333333e-05,
      "loss": 0.091,
      "step": 11928
    },
    {
      "epoch": 22.009233610341642,
      "grad_norm": 0.4159901440143585,
      "learning_rate": 2.9358000000000003e-05,
      "loss": 0.0538,
      "step": 11929
    },
    {
      "epoch": 22.01108033240997,
      "grad_norm": 0.429823637008667,
      "learning_rate": 2.9357666666666665e-05,
      "loss": 0.0617,
      "step": 11930
    },
    {
      "epoch": 22.0129270544783,
      "grad_norm": 0.3844093084335327,
      "learning_rate": 2.9357333333333334e-05,
      "loss": 0.0551,
      "step": 11931
    },
    {
      "epoch": 22.01477377654663,
      "grad_norm": 0.34892767667770386,
      "learning_rate": 2.9357e-05,
      "loss": 0.0541,
      "step": 11932
    },
    {
      "epoch": 22.01662049861496,
      "grad_norm": 0.383372962474823,
      "learning_rate": 2.9356666666666666e-05,
      "loss": 0.048,
      "step": 11933
    },
    {
      "epoch": 22.018467220683288,
      "grad_norm": 0.45354166626930237,
      "learning_rate": 2.9356333333333336e-05,
      "loss": 0.0946,
      "step": 11934
    },
    {
      "epoch": 22.020313942751617,
      "grad_norm": 0.3834838569164276,
      "learning_rate": 2.9356e-05,
      "loss": 0.0244,
      "step": 11935
    },
    {
      "epoch": 22.022160664819946,
      "grad_norm": 0.2780103385448456,
      "learning_rate": 2.9355666666666667e-05,
      "loss": 0.0156,
      "step": 11936
    },
    {
      "epoch": 22.02400738688827,
      "grad_norm": 0.29520273208618164,
      "learning_rate": 2.9355333333333333e-05,
      "loss": 0.0245,
      "step": 11937
    },
    {
      "epoch": 22.0258541089566,
      "grad_norm": 0.4779583215713501,
      "learning_rate": 2.9355000000000003e-05,
      "loss": 0.0202,
      "step": 11938
    },
    {
      "epoch": 22.02770083102493,
      "grad_norm": 0.22411291301250458,
      "learning_rate": 2.9354666666666665e-05,
      "loss": 0.01,
      "step": 11939
    },
    {
      "epoch": 22.02954755309326,
      "grad_norm": 0.5006150007247925,
      "learning_rate": 2.9354333333333335e-05,
      "loss": 0.0118,
      "step": 11940
    },
    {
      "epoch": 22.03139427516159,
      "grad_norm": 0.6376142501831055,
      "learning_rate": 2.9354e-05,
      "loss": 0.0134,
      "step": 11941
    },
    {
      "epoch": 22.033240997229917,
      "grad_norm": 0.2224431335926056,
      "learning_rate": 2.9353666666666666e-05,
      "loss": 0.0117,
      "step": 11942
    },
    {
      "epoch": 22.035087719298247,
      "grad_norm": 0.24049821496009827,
      "learning_rate": 2.9353333333333336e-05,
      "loss": 0.0058,
      "step": 11943
    },
    {
      "epoch": 22.036934441366576,
      "grad_norm": 0.6087546348571777,
      "learning_rate": 2.9353000000000002e-05,
      "loss": 0.0123,
      "step": 11944
    },
    {
      "epoch": 22.0387811634349,
      "grad_norm": 0.3148014545440674,
      "learning_rate": 2.9352666666666668e-05,
      "loss": 0.0105,
      "step": 11945
    },
    {
      "epoch": 22.04062788550323,
      "grad_norm": 0.3352877199649811,
      "learning_rate": 2.9352333333333334e-05,
      "loss": 0.017,
      "step": 11946
    },
    {
      "epoch": 22.04247460757156,
      "grad_norm": 0.1324327439069748,
      "learning_rate": 2.9352000000000003e-05,
      "loss": 0.0046,
      "step": 11947
    },
    {
      "epoch": 22.04432132963989,
      "grad_norm": 0.40338218212127686,
      "learning_rate": 2.9351666666666665e-05,
      "loss": 0.0093,
      "step": 11948
    },
    {
      "epoch": 22.046168051708218,
      "grad_norm": 0.32593637704849243,
      "learning_rate": 2.9351333333333335e-05,
      "loss": 0.0119,
      "step": 11949
    },
    {
      "epoch": 22.048014773776547,
      "grad_norm": 0.0924089178442955,
      "learning_rate": 2.9351e-05,
      "loss": 0.0031,
      "step": 11950
    },
    {
      "epoch": 22.049861495844876,
      "grad_norm": 0.3839172124862671,
      "learning_rate": 2.9350666666666667e-05,
      "loss": 0.0126,
      "step": 11951
    },
    {
      "epoch": 22.051708217913205,
      "grad_norm": 0.1773180216550827,
      "learning_rate": 2.9350333333333336e-05,
      "loss": 0.004,
      "step": 11952
    },
    {
      "epoch": 22.053554939981534,
      "grad_norm": 0.23860660195350647,
      "learning_rate": 2.9350000000000002e-05,
      "loss": 0.0068,
      "step": 11953
    },
    {
      "epoch": 22.05540166204986,
      "grad_norm": 0.3832785487174988,
      "learning_rate": 2.9349666666666668e-05,
      "loss": 0.0615,
      "step": 11954
    },
    {
      "epoch": 22.05724838411819,
      "grad_norm": 1.8486205339431763,
      "learning_rate": 2.9349333333333334e-05,
      "loss": 0.0121,
      "step": 11955
    },
    {
      "epoch": 22.05909510618652,
      "grad_norm": 0.26004332304000854,
      "learning_rate": 2.9349e-05,
      "loss": 0.0065,
      "step": 11956
    },
    {
      "epoch": 22.060941828254848,
      "grad_norm": 0.35053035616874695,
      "learning_rate": 2.9348666666666666e-05,
      "loss": 0.0068,
      "step": 11957
    },
    {
      "epoch": 22.062788550323177,
      "grad_norm": 0.14764510095119476,
      "learning_rate": 2.9348333333333335e-05,
      "loss": 0.006,
      "step": 11958
    },
    {
      "epoch": 22.064635272391506,
      "grad_norm": 0.3242139220237732,
      "learning_rate": 2.9348e-05,
      "loss": 0.0076,
      "step": 11959
    },
    {
      "epoch": 22.066481994459835,
      "grad_norm": 0.2089972198009491,
      "learning_rate": 2.9347666666666667e-05,
      "loss": 0.0058,
      "step": 11960
    },
    {
      "epoch": 22.068328716528164,
      "grad_norm": 0.7421181797981262,
      "learning_rate": 2.9347333333333336e-05,
      "loss": 0.009,
      "step": 11961
    },
    {
      "epoch": 22.07017543859649,
      "grad_norm": 0.11591432243585587,
      "learning_rate": 2.9347e-05,
      "loss": 0.0026,
      "step": 11962
    },
    {
      "epoch": 22.07202216066482,
      "grad_norm": 0.4323197901248932,
      "learning_rate": 2.9346666666666668e-05,
      "loss": 0.0148,
      "step": 11963
    },
    {
      "epoch": 22.073868882733148,
      "grad_norm": 0.3520541489124298,
      "learning_rate": 2.9346333333333334e-05,
      "loss": 0.0121,
      "step": 11964
    },
    {
      "epoch": 22.075715604801477,
      "grad_norm": 0.3193742334842682,
      "learning_rate": 2.9346e-05,
      "loss": 0.0059,
      "step": 11965
    },
    {
      "epoch": 22.077562326869806,
      "grad_norm": 0.21015574038028717,
      "learning_rate": 2.9345666666666666e-05,
      "loss": 0.0057,
      "step": 11966
    },
    {
      "epoch": 22.079409048938135,
      "grad_norm": 1.1927063465118408,
      "learning_rate": 2.9345333333333335e-05,
      "loss": 0.0139,
      "step": 11967
    },
    {
      "epoch": 22.081255771006465,
      "grad_norm": 0.2794034481048584,
      "learning_rate": 2.9345e-05,
      "loss": 0.0066,
      "step": 11968
    },
    {
      "epoch": 22.083102493074794,
      "grad_norm": 0.16369950771331787,
      "learning_rate": 2.9344666666666667e-05,
      "loss": 0.0024,
      "step": 11969
    },
    {
      "epoch": 22.08494921514312,
      "grad_norm": 0.6939327716827393,
      "learning_rate": 2.9344333333333336e-05,
      "loss": 0.0132,
      "step": 11970
    },
    {
      "epoch": 22.08679593721145,
      "grad_norm": 0.5204373002052307,
      "learning_rate": 2.9344e-05,
      "loss": 0.0091,
      "step": 11971
    },
    {
      "epoch": 22.088642659279778,
      "grad_norm": 0.6986927390098572,
      "learning_rate": 2.9343666666666668e-05,
      "loss": 0.0167,
      "step": 11972
    },
    {
      "epoch": 22.090489381348107,
      "grad_norm": 0.2969571053981781,
      "learning_rate": 2.9343333333333334e-05,
      "loss": 0.0136,
      "step": 11973
    },
    {
      "epoch": 22.092336103416436,
      "grad_norm": 0.49085596203804016,
      "learning_rate": 2.9343e-05,
      "loss": 0.0067,
      "step": 11974
    },
    {
      "epoch": 22.094182825484765,
      "grad_norm": 0.610810399055481,
      "learning_rate": 2.9342666666666666e-05,
      "loss": 0.1319,
      "step": 11975
    },
    {
      "epoch": 22.096029547553094,
      "grad_norm": 0.5385555028915405,
      "learning_rate": 2.9342333333333335e-05,
      "loss": 0.1277,
      "step": 11976
    },
    {
      "epoch": 22.097876269621423,
      "grad_norm": 0.4245043098926544,
      "learning_rate": 2.9342e-05,
      "loss": 0.0807,
      "step": 11977
    },
    {
      "epoch": 22.099722991689752,
      "grad_norm": 0.5050533413887024,
      "learning_rate": 2.9341666666666667e-05,
      "loss": 0.0656,
      "step": 11978
    },
    {
      "epoch": 22.101569713758078,
      "grad_norm": 0.5482106804847717,
      "learning_rate": 2.9341333333333337e-05,
      "loss": 0.0574,
      "step": 11979
    },
    {
      "epoch": 22.103416435826407,
      "grad_norm": 0.4365978538990021,
      "learning_rate": 2.9341e-05,
      "loss": 0.0619,
      "step": 11980
    },
    {
      "epoch": 22.105263157894736,
      "grad_norm": 0.4204999506473541,
      "learning_rate": 2.934066666666667e-05,
      "loss": 0.0534,
      "step": 11981
    },
    {
      "epoch": 22.107109879963065,
      "grad_norm": 0.44639596343040466,
      "learning_rate": 2.9340333333333334e-05,
      "loss": 0.0629,
      "step": 11982
    },
    {
      "epoch": 22.108956602031395,
      "grad_norm": 0.37912020087242126,
      "learning_rate": 2.934e-05,
      "loss": 0.0423,
      "step": 11983
    },
    {
      "epoch": 22.110803324099724,
      "grad_norm": 0.4707135260105133,
      "learning_rate": 2.933966666666667e-05,
      "loss": 0.0384,
      "step": 11984
    },
    {
      "epoch": 22.112650046168053,
      "grad_norm": 0.3049370348453522,
      "learning_rate": 2.9339333333333332e-05,
      "loss": 0.0268,
      "step": 11985
    },
    {
      "epoch": 22.114496768236382,
      "grad_norm": 0.5002748370170593,
      "learning_rate": 2.9339e-05,
      "loss": 0.0265,
      "step": 11986
    },
    {
      "epoch": 22.116343490304708,
      "grad_norm": 0.2719067633152008,
      "learning_rate": 2.9338666666666667e-05,
      "loss": 0.0373,
      "step": 11987
    },
    {
      "epoch": 22.118190212373037,
      "grad_norm": 0.4024271070957184,
      "learning_rate": 2.9338333333333333e-05,
      "loss": 0.034,
      "step": 11988
    },
    {
      "epoch": 22.120036934441366,
      "grad_norm": 0.7035671472549438,
      "learning_rate": 2.9338e-05,
      "loss": 0.0112,
      "step": 11989
    },
    {
      "epoch": 22.121883656509695,
      "grad_norm": 0.9298180937767029,
      "learning_rate": 2.933766666666667e-05,
      "loss": 0.0209,
      "step": 11990
    },
    {
      "epoch": 22.123730378578024,
      "grad_norm": 0.27738744020462036,
      "learning_rate": 2.933733333333333e-05,
      "loss": 0.0384,
      "step": 11991
    },
    {
      "epoch": 22.125577100646353,
      "grad_norm": 0.554492712020874,
      "learning_rate": 2.9337e-05,
      "loss": 0.016,
      "step": 11992
    },
    {
      "epoch": 22.127423822714682,
      "grad_norm": 0.19704753160476685,
      "learning_rate": 2.933666666666667e-05,
      "loss": 0.0064,
      "step": 11993
    },
    {
      "epoch": 22.12927054478301,
      "grad_norm": 0.24221715331077576,
      "learning_rate": 2.9336333333333332e-05,
      "loss": 0.0455,
      "step": 11994
    },
    {
      "epoch": 22.131117266851337,
      "grad_norm": 0.25139182806015015,
      "learning_rate": 2.9336000000000002e-05,
      "loss": 0.0084,
      "step": 11995
    },
    {
      "epoch": 22.132963988919666,
      "grad_norm": 0.18211950361728668,
      "learning_rate": 2.9335666666666668e-05,
      "loss": 0.0073,
      "step": 11996
    },
    {
      "epoch": 22.134810710987995,
      "grad_norm": 0.2991064786911011,
      "learning_rate": 2.9335333333333334e-05,
      "loss": 0.0063,
      "step": 11997
    },
    {
      "epoch": 22.136657433056325,
      "grad_norm": 0.1840602308511734,
      "learning_rate": 2.9335e-05,
      "loss": 0.0088,
      "step": 11998
    },
    {
      "epoch": 22.138504155124654,
      "grad_norm": 0.25436100363731384,
      "learning_rate": 2.933466666666667e-05,
      "loss": 0.005,
      "step": 11999
    },
    {
      "epoch": 22.140350877192983,
      "grad_norm": 0.16578309237957,
      "learning_rate": 2.933433333333333e-05,
      "loss": 0.0057,
      "step": 12000
    },
    {
      "epoch": 22.140350877192983,
      "eval_cer": 0.11143945874484797,
      "eval_loss": 0.3457863926887512,
      "eval_runtime": 16.4823,
      "eval_samples_per_second": 18.444,
      "eval_steps_per_second": 0.607,
      "eval_wer": 0.38468917881811204,
      "step": 12000
    },
    {
      "epoch": 22.142197599261312,
      "grad_norm": 0.1465679258108139,
      "learning_rate": 2.9334e-05,
      "loss": 0.0074,
      "step": 12001
    },
    {
      "epoch": 22.14404432132964,
      "grad_norm": 0.35484927892684937,
      "learning_rate": 2.933366666666667e-05,
      "loss": 0.0123,
      "step": 12002
    },
    {
      "epoch": 22.14589104339797,
      "grad_norm": 0.22260144352912903,
      "learning_rate": 2.9333333333333333e-05,
      "loss": 0.0095,
      "step": 12003
    },
    {
      "epoch": 22.147737765466296,
      "grad_norm": 0.13370046019554138,
      "learning_rate": 2.9333000000000002e-05,
      "loss": 0.0053,
      "step": 12004
    },
    {
      "epoch": 22.149584487534625,
      "grad_norm": 0.585996150970459,
      "learning_rate": 2.9332666666666668e-05,
      "loss": 0.007,
      "step": 12005
    },
    {
      "epoch": 22.151431209602954,
      "grad_norm": 0.16085104644298553,
      "learning_rate": 2.9332333333333334e-05,
      "loss": 0.0037,
      "step": 12006
    },
    {
      "epoch": 22.153277931671283,
      "grad_norm": 0.2741457223892212,
      "learning_rate": 2.9332e-05,
      "loss": 0.0094,
      "step": 12007
    },
    {
      "epoch": 22.155124653739612,
      "grad_norm": 0.35082828998565674,
      "learning_rate": 2.933166666666667e-05,
      "loss": 0.0087,
      "step": 12008
    },
    {
      "epoch": 22.15697137580794,
      "grad_norm": 0.7393125295639038,
      "learning_rate": 2.9331333333333335e-05,
      "loss": 0.0208,
      "step": 12009
    },
    {
      "epoch": 22.15881809787627,
      "grad_norm": 0.37203606963157654,
      "learning_rate": 2.9331e-05,
      "loss": 0.0075,
      "step": 12010
    },
    {
      "epoch": 22.1606648199446,
      "grad_norm": 1.1032886505126953,
      "learning_rate": 2.933066666666667e-05,
      "loss": 0.0181,
      "step": 12011
    },
    {
      "epoch": 22.162511542012926,
      "grad_norm": 0.24382264912128448,
      "learning_rate": 2.9330333333333333e-05,
      "loss": 0.0086,
      "step": 12012
    },
    {
      "epoch": 22.164358264081255,
      "grad_norm": 0.20161493122577667,
      "learning_rate": 2.9330000000000002e-05,
      "loss": 0.0069,
      "step": 12013
    },
    {
      "epoch": 22.166204986149584,
      "grad_norm": 0.6993123292922974,
      "learning_rate": 2.9329666666666668e-05,
      "loss": 0.0114,
      "step": 12014
    },
    {
      "epoch": 22.168051708217913,
      "grad_norm": 0.6505698561668396,
      "learning_rate": 2.9329333333333334e-05,
      "loss": 0.0132,
      "step": 12015
    },
    {
      "epoch": 22.169898430286242,
      "grad_norm": 0.21036770939826965,
      "learning_rate": 2.9329e-05,
      "loss": 0.0067,
      "step": 12016
    },
    {
      "epoch": 22.17174515235457,
      "grad_norm": 0.3262076675891876,
      "learning_rate": 2.9328666666666666e-05,
      "loss": 0.0095,
      "step": 12017
    },
    {
      "epoch": 22.1735918744229,
      "grad_norm": 0.42080506682395935,
      "learning_rate": 2.9328333333333335e-05,
      "loss": 0.0061,
      "step": 12018
    },
    {
      "epoch": 22.17543859649123,
      "grad_norm": 0.38495534658432007,
      "learning_rate": 2.9328e-05,
      "loss": 0.0161,
      "step": 12019
    },
    {
      "epoch": 22.177285318559555,
      "grad_norm": 0.31718382239341736,
      "learning_rate": 2.9327666666666667e-05,
      "loss": 0.0072,
      "step": 12020
    },
    {
      "epoch": 22.179132040627884,
      "grad_norm": 0.19104288518428802,
      "learning_rate": 2.9327333333333333e-05,
      "loss": 0.0053,
      "step": 12021
    },
    {
      "epoch": 22.180978762696213,
      "grad_norm": 0.38349059224128723,
      "learning_rate": 2.9327000000000002e-05,
      "loss": 0.0101,
      "step": 12022
    },
    {
      "epoch": 22.182825484764543,
      "grad_norm": 0.9522183537483215,
      "learning_rate": 2.9326666666666665e-05,
      "loss": 0.0163,
      "step": 12023
    },
    {
      "epoch": 22.18467220683287,
      "grad_norm": 0.3559163510799408,
      "learning_rate": 2.9326333333333334e-05,
      "loss": 0.0066,
      "step": 12024
    },
    {
      "epoch": 22.1865189289012,
      "grad_norm": 0.6288875341415405,
      "learning_rate": 2.9326e-05,
      "loss": 0.1693,
      "step": 12025
    },
    {
      "epoch": 22.18836565096953,
      "grad_norm": 0.6415272951126099,
      "learning_rate": 2.9325666666666666e-05,
      "loss": 0.1282,
      "step": 12026
    },
    {
      "epoch": 22.19021237303786,
      "grad_norm": 0.4664081931114197,
      "learning_rate": 2.9325333333333335e-05,
      "loss": 0.092,
      "step": 12027
    },
    {
      "epoch": 22.19205909510619,
      "grad_norm": 0.8116239309310913,
      "learning_rate": 2.9325e-05,
      "loss": 0.1315,
      "step": 12028
    },
    {
      "epoch": 22.193905817174514,
      "grad_norm": 0.4563292860984802,
      "learning_rate": 2.9324666666666667e-05,
      "loss": 0.0755,
      "step": 12029
    },
    {
      "epoch": 22.195752539242843,
      "grad_norm": 0.5227322578430176,
      "learning_rate": 2.9324333333333333e-05,
      "loss": 0.0758,
      "step": 12030
    },
    {
      "epoch": 22.197599261311172,
      "grad_norm": 0.438588410615921,
      "learning_rate": 2.9324000000000002e-05,
      "loss": 0.0496,
      "step": 12031
    },
    {
      "epoch": 22.1994459833795,
      "grad_norm": 0.43619734048843384,
      "learning_rate": 2.9323666666666665e-05,
      "loss": 0.0529,
      "step": 12032
    },
    {
      "epoch": 22.20129270544783,
      "grad_norm": 0.2944128215312958,
      "learning_rate": 2.9323333333333334e-05,
      "loss": 0.0584,
      "step": 12033
    },
    {
      "epoch": 22.20313942751616,
      "grad_norm": 0.4487317204475403,
      "learning_rate": 2.9323000000000004e-05,
      "loss": 0.0547,
      "step": 12034
    },
    {
      "epoch": 22.20498614958449,
      "grad_norm": 0.5154827833175659,
      "learning_rate": 2.9322666666666666e-05,
      "loss": 0.0434,
      "step": 12035
    },
    {
      "epoch": 22.206832871652818,
      "grad_norm": 1.1755114793777466,
      "learning_rate": 2.9322333333333336e-05,
      "loss": 0.0644,
      "step": 12036
    },
    {
      "epoch": 22.208679593721143,
      "grad_norm": 0.29100021719932556,
      "learning_rate": 2.9322e-05,
      "loss": 0.0423,
      "step": 12037
    },
    {
      "epoch": 22.210526315789473,
      "grad_norm": 0.4555946886539459,
      "learning_rate": 2.9321666666666667e-05,
      "loss": 0.0637,
      "step": 12038
    },
    {
      "epoch": 22.2123730378578,
      "grad_norm": 0.49455103278160095,
      "learning_rate": 2.9321333333333333e-05,
      "loss": 0.025,
      "step": 12039
    },
    {
      "epoch": 22.21421975992613,
      "grad_norm": 0.28613951802253723,
      "learning_rate": 2.9321000000000003e-05,
      "loss": 0.0136,
      "step": 12040
    },
    {
      "epoch": 22.21606648199446,
      "grad_norm": 0.2727801501750946,
      "learning_rate": 2.9320666666666665e-05,
      "loss": 0.0154,
      "step": 12041
    },
    {
      "epoch": 22.21791320406279,
      "grad_norm": 0.3077845871448517,
      "learning_rate": 2.9320333333333335e-05,
      "loss": 0.0201,
      "step": 12042
    },
    {
      "epoch": 22.21975992613112,
      "grad_norm": 0.19647696614265442,
      "learning_rate": 2.9320000000000004e-05,
      "loss": 0.008,
      "step": 12043
    },
    {
      "epoch": 22.221606648199447,
      "grad_norm": 0.297075092792511,
      "learning_rate": 2.9319666666666666e-05,
      "loss": 0.0082,
      "step": 12044
    },
    {
      "epoch": 22.223453370267773,
      "grad_norm": 1.275119662284851,
      "learning_rate": 2.9319333333333336e-05,
      "loss": 0.0216,
      "step": 12045
    },
    {
      "epoch": 22.225300092336102,
      "grad_norm": 0.4337138235569,
      "learning_rate": 2.9318999999999998e-05,
      "loss": 0.0153,
      "step": 12046
    },
    {
      "epoch": 22.22714681440443,
      "grad_norm": 0.2728216350078583,
      "learning_rate": 2.9318666666666668e-05,
      "loss": 0.0084,
      "step": 12047
    },
    {
      "epoch": 22.22899353647276,
      "grad_norm": 0.33013466000556946,
      "learning_rate": 2.9318333333333334e-05,
      "loss": 0.0157,
      "step": 12048
    },
    {
      "epoch": 22.23084025854109,
      "grad_norm": 0.19151604175567627,
      "learning_rate": 2.9318e-05,
      "loss": 0.0073,
      "step": 12049
    },
    {
      "epoch": 22.23268698060942,
      "grad_norm": 0.2966844141483307,
      "learning_rate": 2.9317666666666665e-05,
      "loss": 0.0093,
      "step": 12050
    },
    {
      "epoch": 22.234533702677748,
      "grad_norm": 0.33475998044013977,
      "learning_rate": 2.9317333333333335e-05,
      "loss": 0.009,
      "step": 12051
    },
    {
      "epoch": 22.236380424746077,
      "grad_norm": 0.16185058653354645,
      "learning_rate": 2.9317e-05,
      "loss": 0.0049,
      "step": 12052
    },
    {
      "epoch": 22.238227146814406,
      "grad_norm": 0.25134894251823425,
      "learning_rate": 2.9316666666666667e-05,
      "loss": 0.0105,
      "step": 12053
    },
    {
      "epoch": 22.24007386888273,
      "grad_norm": 0.21591123938560486,
      "learning_rate": 2.9316333333333336e-05,
      "loss": 0.0085,
      "step": 12054
    },
    {
      "epoch": 22.24192059095106,
      "grad_norm": 0.36200016736984253,
      "learning_rate": 2.9316e-05,
      "loss": 0.007,
      "step": 12055
    },
    {
      "epoch": 22.24376731301939,
      "grad_norm": 0.20390160381793976,
      "learning_rate": 2.9315666666666668e-05,
      "loss": 0.0078,
      "step": 12056
    },
    {
      "epoch": 22.24561403508772,
      "grad_norm": 0.5636849999427795,
      "learning_rate": 2.9315333333333334e-05,
      "loss": 0.0074,
      "step": 12057
    },
    {
      "epoch": 22.24746075715605,
      "grad_norm": 0.3945927619934082,
      "learning_rate": 2.9315e-05,
      "loss": 0.0136,
      "step": 12058
    },
    {
      "epoch": 22.249307479224377,
      "grad_norm": 0.20777307450771332,
      "learning_rate": 2.931466666666667e-05,
      "loss": 0.0047,
      "step": 12059
    },
    {
      "epoch": 22.251154201292707,
      "grad_norm": 0.7784906029701233,
      "learning_rate": 2.9314333333333335e-05,
      "loss": 0.0208,
      "step": 12060
    },
    {
      "epoch": 22.253000923361036,
      "grad_norm": 0.16684590280056,
      "learning_rate": 2.9314e-05,
      "loss": 0.0034,
      "step": 12061
    },
    {
      "epoch": 22.25484764542936,
      "grad_norm": 0.2453305870294571,
      "learning_rate": 2.9313666666666667e-05,
      "loss": 0.0059,
      "step": 12062
    },
    {
      "epoch": 22.25669436749769,
      "grad_norm": 0.4265834093093872,
      "learning_rate": 2.9313333333333336e-05,
      "loss": 0.0111,
      "step": 12063
    },
    {
      "epoch": 22.25854108956602,
      "grad_norm": 0.25547710061073303,
      "learning_rate": 2.9313e-05,
      "loss": 0.0041,
      "step": 12064
    },
    {
      "epoch": 22.26038781163435,
      "grad_norm": 0.6966615319252014,
      "learning_rate": 2.9312666666666668e-05,
      "loss": 0.0074,
      "step": 12065
    },
    {
      "epoch": 22.262234533702678,
      "grad_norm": 0.23735538125038147,
      "learning_rate": 2.9312333333333334e-05,
      "loss": 0.0038,
      "step": 12066
    },
    {
      "epoch": 22.264081255771007,
      "grad_norm": 0.24145880341529846,
      "learning_rate": 2.9312e-05,
      "loss": 0.0062,
      "step": 12067
    },
    {
      "epoch": 22.265927977839336,
      "grad_norm": 0.23477429151535034,
      "learning_rate": 2.931166666666667e-05,
      "loss": 0.0054,
      "step": 12068
    },
    {
      "epoch": 22.267774699907665,
      "grad_norm": 0.22543811798095703,
      "learning_rate": 2.9311333333333335e-05,
      "loss": 0.0115,
      "step": 12069
    },
    {
      "epoch": 22.26962142197599,
      "grad_norm": 0.21335336565971375,
      "learning_rate": 2.9311e-05,
      "loss": 0.0071,
      "step": 12070
    },
    {
      "epoch": 22.27146814404432,
      "grad_norm": 0.4495827853679657,
      "learning_rate": 2.9310666666666667e-05,
      "loss": 0.0107,
      "step": 12071
    },
    {
      "epoch": 22.27331486611265,
      "grad_norm": 0.5423740744590759,
      "learning_rate": 2.9310333333333336e-05,
      "loss": 0.0125,
      "step": 12072
    },
    {
      "epoch": 22.27516158818098,
      "grad_norm": 0.2808704376220703,
      "learning_rate": 2.931e-05,
      "loss": 0.0047,
      "step": 12073
    },
    {
      "epoch": 22.277008310249307,
      "grad_norm": 0.40028220415115356,
      "learning_rate": 2.9309666666666668e-05,
      "loss": 0.0144,
      "step": 12074
    },
    {
      "epoch": 22.278855032317637,
      "grad_norm": 0.6234591603279114,
      "learning_rate": 2.930933333333333e-05,
      "loss": 0.1581,
      "step": 12075
    },
    {
      "epoch": 22.280701754385966,
      "grad_norm": 0.5651040077209473,
      "learning_rate": 2.9309e-05,
      "loss": 0.1046,
      "step": 12076
    },
    {
      "epoch": 22.282548476454295,
      "grad_norm": 1.1081950664520264,
      "learning_rate": 2.930866666666667e-05,
      "loss": 0.1222,
      "step": 12077
    },
    {
      "epoch": 22.284395198522624,
      "grad_norm": 0.39549520611763,
      "learning_rate": 2.9308333333333332e-05,
      "loss": 0.0584,
      "step": 12078
    },
    {
      "epoch": 22.28624192059095,
      "grad_norm": 0.49096477031707764,
      "learning_rate": 2.9308e-05,
      "loss": 0.1001,
      "step": 12079
    },
    {
      "epoch": 22.28808864265928,
      "grad_norm": 0.3321293592453003,
      "learning_rate": 2.9307666666666667e-05,
      "loss": 0.0656,
      "step": 12080
    },
    {
      "epoch": 22.289935364727608,
      "grad_norm": 0.39119619131088257,
      "learning_rate": 2.9307333333333333e-05,
      "loss": 0.0507,
      "step": 12081
    },
    {
      "epoch": 22.291782086795937,
      "grad_norm": 0.6798700094223022,
      "learning_rate": 2.9307e-05,
      "loss": 0.052,
      "step": 12082
    },
    {
      "epoch": 22.293628808864266,
      "grad_norm": 0.4221579432487488,
      "learning_rate": 2.930666666666667e-05,
      "loss": 0.0662,
      "step": 12083
    },
    {
      "epoch": 22.295475530932595,
      "grad_norm": 0.33786314725875854,
      "learning_rate": 2.9306333333333334e-05,
      "loss": 0.0566,
      "step": 12084
    },
    {
      "epoch": 22.297322253000925,
      "grad_norm": 0.2803206741809845,
      "learning_rate": 2.9306e-05,
      "loss": 0.0293,
      "step": 12085
    },
    {
      "epoch": 22.299168975069254,
      "grad_norm": 0.5887284874916077,
      "learning_rate": 2.930566666666667e-05,
      "loss": 0.0489,
      "step": 12086
    },
    {
      "epoch": 22.30101569713758,
      "grad_norm": 0.4942704439163208,
      "learning_rate": 2.9305333333333332e-05,
      "loss": 0.0543,
      "step": 12087
    },
    {
      "epoch": 22.30286241920591,
      "grad_norm": 0.34390565752983093,
      "learning_rate": 2.9305e-05,
      "loss": 0.0362,
      "step": 12088
    },
    {
      "epoch": 22.304709141274238,
      "grad_norm": 0.3027403950691223,
      "learning_rate": 2.9304666666666667e-05,
      "loss": 0.0288,
      "step": 12089
    },
    {
      "epoch": 22.306555863342567,
      "grad_norm": 0.334240198135376,
      "learning_rate": 2.9304333333333333e-05,
      "loss": 0.013,
      "step": 12090
    },
    {
      "epoch": 22.308402585410896,
      "grad_norm": 0.2725203335285187,
      "learning_rate": 2.9304e-05,
      "loss": 0.0095,
      "step": 12091
    },
    {
      "epoch": 22.310249307479225,
      "grad_norm": 0.3006215989589691,
      "learning_rate": 2.930366666666667e-05,
      "loss": 0.0138,
      "step": 12092
    },
    {
      "epoch": 22.312096029547554,
      "grad_norm": 0.22081726789474487,
      "learning_rate": 2.9303333333333335e-05,
      "loss": 0.0087,
      "step": 12093
    },
    {
      "epoch": 22.313942751615883,
      "grad_norm": 0.2759893238544464,
      "learning_rate": 2.9303e-05,
      "loss": 0.0117,
      "step": 12094
    },
    {
      "epoch": 22.31578947368421,
      "grad_norm": 0.2866109311580658,
      "learning_rate": 2.930266666666667e-05,
      "loss": 0.0095,
      "step": 12095
    },
    {
      "epoch": 22.317636195752538,
      "grad_norm": 0.14103403687477112,
      "learning_rate": 2.9302333333333332e-05,
      "loss": 0.0044,
      "step": 12096
    },
    {
      "epoch": 22.319482917820867,
      "grad_norm": 0.2929133474826813,
      "learning_rate": 2.9302e-05,
      "loss": 0.0256,
      "step": 12097
    },
    {
      "epoch": 22.321329639889196,
      "grad_norm": 0.1882898211479187,
      "learning_rate": 2.9301666666666668e-05,
      "loss": 0.0093,
      "step": 12098
    },
    {
      "epoch": 22.323176361957525,
      "grad_norm": 0.16777978837490082,
      "learning_rate": 2.9301333333333334e-05,
      "loss": 0.0093,
      "step": 12099
    },
    {
      "epoch": 22.325023084025855,
      "grad_norm": 0.1872076392173767,
      "learning_rate": 2.9301e-05,
      "loss": 0.0081,
      "step": 12100
    },
    {
      "epoch": 22.326869806094184,
      "grad_norm": 0.49100226163864136,
      "learning_rate": 2.930066666666667e-05,
      "loss": 0.0141,
      "step": 12101
    },
    {
      "epoch": 22.328716528162513,
      "grad_norm": 0.21988950669765472,
      "learning_rate": 2.9300333333333335e-05,
      "loss": 0.035,
      "step": 12102
    },
    {
      "epoch": 22.330563250230842,
      "grad_norm": 0.21322642266750336,
      "learning_rate": 2.93e-05,
      "loss": 0.0083,
      "step": 12103
    },
    {
      "epoch": 22.332409972299168,
      "grad_norm": 0.28099849820137024,
      "learning_rate": 2.929966666666667e-05,
      "loss": 0.0067,
      "step": 12104
    },
    {
      "epoch": 22.334256694367497,
      "grad_norm": 0.13967318832874298,
      "learning_rate": 2.9299333333333333e-05,
      "loss": 0.004,
      "step": 12105
    },
    {
      "epoch": 22.336103416435826,
      "grad_norm": 0.38747766613960266,
      "learning_rate": 2.9299000000000002e-05,
      "loss": 0.0175,
      "step": 12106
    },
    {
      "epoch": 22.337950138504155,
      "grad_norm": 0.1795417219400406,
      "learning_rate": 2.9298666666666664e-05,
      "loss": 0.0046,
      "step": 12107
    },
    {
      "epoch": 22.339796860572484,
      "grad_norm": 0.21303912997245789,
      "learning_rate": 2.9298333333333334e-05,
      "loss": 0.0058,
      "step": 12108
    },
    {
      "epoch": 22.341643582640813,
      "grad_norm": 0.1777993142604828,
      "learning_rate": 2.9298000000000003e-05,
      "loss": 0.0056,
      "step": 12109
    },
    {
      "epoch": 22.343490304709142,
      "grad_norm": 0.09992282837629318,
      "learning_rate": 2.9297666666666666e-05,
      "loss": 0.0026,
      "step": 12110
    },
    {
      "epoch": 22.34533702677747,
      "grad_norm": 0.34188637137413025,
      "learning_rate": 2.9297333333333335e-05,
      "loss": 0.0091,
      "step": 12111
    },
    {
      "epoch": 22.347183748845797,
      "grad_norm": 0.2958120107650757,
      "learning_rate": 2.9297e-05,
      "loss": 0.0076,
      "step": 12112
    },
    {
      "epoch": 22.349030470914126,
      "grad_norm": 0.27804356813430786,
      "learning_rate": 2.9296666666666667e-05,
      "loss": 0.0065,
      "step": 12113
    },
    {
      "epoch": 22.350877192982455,
      "grad_norm": 0.23710592091083527,
      "learning_rate": 2.9296333333333333e-05,
      "loss": 0.0056,
      "step": 12114
    },
    {
      "epoch": 22.352723915050785,
      "grad_norm": 0.17549476027488708,
      "learning_rate": 2.9296000000000002e-05,
      "loss": 0.0044,
      "step": 12115
    },
    {
      "epoch": 22.354570637119114,
      "grad_norm": 0.15604440867900848,
      "learning_rate": 2.9295666666666665e-05,
      "loss": 0.0046,
      "step": 12116
    },
    {
      "epoch": 22.356417359187443,
      "grad_norm": 0.3873245120048523,
      "learning_rate": 2.9295333333333334e-05,
      "loss": 0.009,
      "step": 12117
    },
    {
      "epoch": 22.358264081255772,
      "grad_norm": 0.22579556703567505,
      "learning_rate": 2.9295000000000003e-05,
      "loss": 0.0049,
      "step": 12118
    },
    {
      "epoch": 22.3601108033241,
      "grad_norm": 0.5002829432487488,
      "learning_rate": 2.9294666666666666e-05,
      "loss": 0.0226,
      "step": 12119
    },
    {
      "epoch": 22.361957525392427,
      "grad_norm": 0.5181460380554199,
      "learning_rate": 2.9294333333333335e-05,
      "loss": 0.0115,
      "step": 12120
    },
    {
      "epoch": 22.363804247460756,
      "grad_norm": 0.27780306339263916,
      "learning_rate": 2.9294e-05,
      "loss": 0.0215,
      "step": 12121
    },
    {
      "epoch": 22.365650969529085,
      "grad_norm": 0.4311021566390991,
      "learning_rate": 2.9293666666666667e-05,
      "loss": 0.006,
      "step": 12122
    },
    {
      "epoch": 22.367497691597414,
      "grad_norm": 0.30893686413764954,
      "learning_rate": 2.9293333333333333e-05,
      "loss": 0.0045,
      "step": 12123
    },
    {
      "epoch": 22.369344413665743,
      "grad_norm": 0.26298242807388306,
      "learning_rate": 2.9293000000000002e-05,
      "loss": 0.0069,
      "step": 12124
    },
    {
      "epoch": 22.371191135734072,
      "grad_norm": 0.6145325899124146,
      "learning_rate": 2.9292666666666665e-05,
      "loss": 0.146,
      "step": 12125
    },
    {
      "epoch": 22.3730378578024,
      "grad_norm": 0.551490068435669,
      "learning_rate": 2.9292333333333334e-05,
      "loss": 0.1247,
      "step": 12126
    },
    {
      "epoch": 22.37488457987073,
      "grad_norm": 0.4913572072982788,
      "learning_rate": 2.9292000000000003e-05,
      "loss": 0.0839,
      "step": 12127
    },
    {
      "epoch": 22.37673130193906,
      "grad_norm": 0.4714609682559967,
      "learning_rate": 2.9291666666666666e-05,
      "loss": 0.0949,
      "step": 12128
    },
    {
      "epoch": 22.378578024007385,
      "grad_norm": 0.5117810964584351,
      "learning_rate": 2.9291333333333335e-05,
      "loss": 0.0748,
      "step": 12129
    },
    {
      "epoch": 22.380424746075715,
      "grad_norm": 0.5069825053215027,
      "learning_rate": 2.9291e-05,
      "loss": 0.0757,
      "step": 12130
    },
    {
      "epoch": 22.382271468144044,
      "grad_norm": 0.39257165789604187,
      "learning_rate": 2.9290666666666667e-05,
      "loss": 0.0428,
      "step": 12131
    },
    {
      "epoch": 22.384118190212373,
      "grad_norm": 0.3679068386554718,
      "learning_rate": 2.9290333333333333e-05,
      "loss": 0.0406,
      "step": 12132
    },
    {
      "epoch": 22.385964912280702,
      "grad_norm": 0.5544233322143555,
      "learning_rate": 2.9290000000000002e-05,
      "loss": 0.0572,
      "step": 12133
    },
    {
      "epoch": 22.38781163434903,
      "grad_norm": 0.37580254673957825,
      "learning_rate": 2.928966666666667e-05,
      "loss": 0.0317,
      "step": 12134
    },
    {
      "epoch": 22.38965835641736,
      "grad_norm": 0.3190949261188507,
      "learning_rate": 2.9289333333333334e-05,
      "loss": 0.0368,
      "step": 12135
    },
    {
      "epoch": 22.39150507848569,
      "grad_norm": 0.8851911425590515,
      "learning_rate": 2.9289e-05,
      "loss": 0.0398,
      "step": 12136
    },
    {
      "epoch": 22.393351800554015,
      "grad_norm": 0.27544206380844116,
      "learning_rate": 2.9288666666666666e-05,
      "loss": 0.0435,
      "step": 12137
    },
    {
      "epoch": 22.395198522622344,
      "grad_norm": 0.23330990970134735,
      "learning_rate": 2.9288333333333336e-05,
      "loss": 0.0127,
      "step": 12138
    },
    {
      "epoch": 22.397045244690673,
      "grad_norm": 0.3330238461494446,
      "learning_rate": 2.9287999999999998e-05,
      "loss": 0.0211,
      "step": 12139
    },
    {
      "epoch": 22.398891966759003,
      "grad_norm": 0.3399524390697479,
      "learning_rate": 2.9287666666666667e-05,
      "loss": 0.0272,
      "step": 12140
    },
    {
      "epoch": 22.40073868882733,
      "grad_norm": 0.23550830781459808,
      "learning_rate": 2.9287333333333333e-05,
      "loss": 0.009,
      "step": 12141
    },
    {
      "epoch": 22.40258541089566,
      "grad_norm": 0.2656426429748535,
      "learning_rate": 2.9287e-05,
      "loss": 0.0082,
      "step": 12142
    },
    {
      "epoch": 22.40443213296399,
      "grad_norm": 0.23883987963199615,
      "learning_rate": 2.928666666666667e-05,
      "loss": 0.0256,
      "step": 12143
    },
    {
      "epoch": 22.40627885503232,
      "grad_norm": 0.29757219552993774,
      "learning_rate": 2.9286333333333335e-05,
      "loss": 0.0359,
      "step": 12144
    },
    {
      "epoch": 22.408125577100645,
      "grad_norm": 0.4060526192188263,
      "learning_rate": 2.9286e-05,
      "loss": 0.0257,
      "step": 12145
    },
    {
      "epoch": 22.409972299168974,
      "grad_norm": 0.7830715775489807,
      "learning_rate": 2.9285666666666666e-05,
      "loss": 0.0142,
      "step": 12146
    },
    {
      "epoch": 22.411819021237303,
      "grad_norm": 0.22146490216255188,
      "learning_rate": 2.9285333333333336e-05,
      "loss": 0.011,
      "step": 12147
    },
    {
      "epoch": 22.413665743305632,
      "grad_norm": 0.31865814328193665,
      "learning_rate": 2.9284999999999998e-05,
      "loss": 0.013,
      "step": 12148
    },
    {
      "epoch": 22.41551246537396,
      "grad_norm": 0.15901601314544678,
      "learning_rate": 2.9284666666666668e-05,
      "loss": 0.005,
      "step": 12149
    },
    {
      "epoch": 22.41735918744229,
      "grad_norm": 0.3411862850189209,
      "learning_rate": 2.9284333333333334e-05,
      "loss": 0.0087,
      "step": 12150
    },
    {
      "epoch": 22.41920590951062,
      "grad_norm": 0.40633055567741394,
      "learning_rate": 2.9284e-05,
      "loss": 0.0125,
      "step": 12151
    },
    {
      "epoch": 22.42105263157895,
      "grad_norm": 0.6341411471366882,
      "learning_rate": 2.928366666666667e-05,
      "loss": 0.0102,
      "step": 12152
    },
    {
      "epoch": 22.422899353647278,
      "grad_norm": 0.18682046234607697,
      "learning_rate": 2.9283333333333335e-05,
      "loss": 0.007,
      "step": 12153
    },
    {
      "epoch": 22.424746075715603,
      "grad_norm": 0.17826856672763824,
      "learning_rate": 2.9283e-05,
      "loss": 0.0052,
      "step": 12154
    },
    {
      "epoch": 22.426592797783933,
      "grad_norm": 0.22934889793395996,
      "learning_rate": 2.9282666666666667e-05,
      "loss": 0.0055,
      "step": 12155
    },
    {
      "epoch": 22.42843951985226,
      "grad_norm": 0.20013564825057983,
      "learning_rate": 2.9282333333333336e-05,
      "loss": 0.0066,
      "step": 12156
    },
    {
      "epoch": 22.43028624192059,
      "grad_norm": 0.31481319665908813,
      "learning_rate": 2.9282e-05,
      "loss": 0.0065,
      "step": 12157
    },
    {
      "epoch": 22.43213296398892,
      "grad_norm": 0.24155759811401367,
      "learning_rate": 2.9281666666666668e-05,
      "loss": 0.0077,
      "step": 12158
    },
    {
      "epoch": 22.43397968605725,
      "grad_norm": 0.3853037655353546,
      "learning_rate": 2.9281333333333337e-05,
      "loss": 0.0069,
      "step": 12159
    },
    {
      "epoch": 22.43582640812558,
      "grad_norm": 0.42370277643203735,
      "learning_rate": 2.9281e-05,
      "loss": 0.009,
      "step": 12160
    },
    {
      "epoch": 22.437673130193907,
      "grad_norm": 0.27092936635017395,
      "learning_rate": 2.928066666666667e-05,
      "loss": 0.0079,
      "step": 12161
    },
    {
      "epoch": 22.439519852262233,
      "grad_norm": 0.27856534719467163,
      "learning_rate": 2.9280333333333335e-05,
      "loss": 0.0111,
      "step": 12162
    },
    {
      "epoch": 22.441366574330562,
      "grad_norm": 1.4294368028640747,
      "learning_rate": 2.928e-05,
      "loss": 0.0193,
      "step": 12163
    },
    {
      "epoch": 22.44321329639889,
      "grad_norm": 0.217570498585701,
      "learning_rate": 2.9279666666666667e-05,
      "loss": 0.0051,
      "step": 12164
    },
    {
      "epoch": 22.44506001846722,
      "grad_norm": 0.3201679587364197,
      "learning_rate": 2.9279333333333336e-05,
      "loss": 0.0093,
      "step": 12165
    },
    {
      "epoch": 22.44690674053555,
      "grad_norm": 0.0861976221203804,
      "learning_rate": 2.9279e-05,
      "loss": 0.0029,
      "step": 12166
    },
    {
      "epoch": 22.44875346260388,
      "grad_norm": 0.27611875534057617,
      "learning_rate": 2.9278666666666668e-05,
      "loss": 0.0094,
      "step": 12167
    },
    {
      "epoch": 22.450600184672208,
      "grad_norm": 0.5227705240249634,
      "learning_rate": 2.9278333333333334e-05,
      "loss": 0.0118,
      "step": 12168
    },
    {
      "epoch": 22.452446906740537,
      "grad_norm": 0.28735342621803284,
      "learning_rate": 2.9278e-05,
      "loss": 0.0103,
      "step": 12169
    },
    {
      "epoch": 22.454293628808863,
      "grad_norm": 0.2518790066242218,
      "learning_rate": 2.927766666666667e-05,
      "loss": 0.006,
      "step": 12170
    },
    {
      "epoch": 22.45614035087719,
      "grad_norm": 0.21540474891662598,
      "learning_rate": 2.9277333333333332e-05,
      "loss": 0.0067,
      "step": 12171
    },
    {
      "epoch": 22.45798707294552,
      "grad_norm": 0.17142251133918762,
      "learning_rate": 2.9277e-05,
      "loss": 0.005,
      "step": 12172
    },
    {
      "epoch": 22.45983379501385,
      "grad_norm": 0.544914960861206,
      "learning_rate": 2.9276666666666667e-05,
      "loss": 0.007,
      "step": 12173
    },
    {
      "epoch": 22.46168051708218,
      "grad_norm": 0.5627887845039368,
      "learning_rate": 2.9276333333333333e-05,
      "loss": 0.0282,
      "step": 12174
    },
    {
      "epoch": 22.46352723915051,
      "grad_norm": 0.5315102338790894,
      "learning_rate": 2.9276e-05,
      "loss": 0.1304,
      "step": 12175
    },
    {
      "epoch": 22.465373961218837,
      "grad_norm": 0.6638376712799072,
      "learning_rate": 2.9275666666666668e-05,
      "loss": 0.1575,
      "step": 12176
    },
    {
      "epoch": 22.467220683287167,
      "grad_norm": 0.543993353843689,
      "learning_rate": 2.9275333333333334e-05,
      "loss": 0.0801,
      "step": 12177
    },
    {
      "epoch": 22.469067405355496,
      "grad_norm": 0.48404455184936523,
      "learning_rate": 2.9275e-05,
      "loss": 0.1037,
      "step": 12178
    },
    {
      "epoch": 22.47091412742382,
      "grad_norm": 0.5442525744438171,
      "learning_rate": 2.927466666666667e-05,
      "loss": 0.1334,
      "step": 12179
    },
    {
      "epoch": 22.47276084949215,
      "grad_norm": 0.3651285767555237,
      "learning_rate": 2.9274333333333332e-05,
      "loss": 0.0792,
      "step": 12180
    },
    {
      "epoch": 22.47460757156048,
      "grad_norm": 0.6731127500534058,
      "learning_rate": 2.9274e-05,
      "loss": 0.0938,
      "step": 12181
    },
    {
      "epoch": 22.47645429362881,
      "grad_norm": 0.34531083703041077,
      "learning_rate": 2.9273666666666667e-05,
      "loss": 0.0419,
      "step": 12182
    },
    {
      "epoch": 22.478301015697138,
      "grad_norm": 0.4028952717781067,
      "learning_rate": 2.9273333333333333e-05,
      "loss": 0.0674,
      "step": 12183
    },
    {
      "epoch": 22.480147737765467,
      "grad_norm": 0.358442485332489,
      "learning_rate": 2.9273000000000002e-05,
      "loss": 0.0481,
      "step": 12184
    },
    {
      "epoch": 22.481994459833796,
      "grad_norm": 0.44311586022377014,
      "learning_rate": 2.927266666666667e-05,
      "loss": 0.0316,
      "step": 12185
    },
    {
      "epoch": 22.483841181902125,
      "grad_norm": 0.32679107785224915,
      "learning_rate": 2.9272333333333334e-05,
      "loss": 0.0895,
      "step": 12186
    },
    {
      "epoch": 22.48568790397045,
      "grad_norm": 0.44924479722976685,
      "learning_rate": 2.9272e-05,
      "loss": 0.0197,
      "step": 12187
    },
    {
      "epoch": 22.48753462603878,
      "grad_norm": 0.22197957336902618,
      "learning_rate": 2.927166666666667e-05,
      "loss": 0.0312,
      "step": 12188
    },
    {
      "epoch": 22.48938134810711,
      "grad_norm": 0.22441096603870392,
      "learning_rate": 2.9271333333333332e-05,
      "loss": 0.0378,
      "step": 12189
    },
    {
      "epoch": 22.49122807017544,
      "grad_norm": 0.24108120799064636,
      "learning_rate": 2.9271e-05,
      "loss": 0.0108,
      "step": 12190
    },
    {
      "epoch": 22.493074792243767,
      "grad_norm": 0.22874818742275238,
      "learning_rate": 2.9270666666666667e-05,
      "loss": 0.017,
      "step": 12191
    },
    {
      "epoch": 22.494921514312097,
      "grad_norm": 0.3547055423259735,
      "learning_rate": 2.9270333333333333e-05,
      "loss": 0.0107,
      "step": 12192
    },
    {
      "epoch": 22.496768236380426,
      "grad_norm": 0.1598084270954132,
      "learning_rate": 2.9270000000000003e-05,
      "loss": 0.0065,
      "step": 12193
    },
    {
      "epoch": 22.498614958448755,
      "grad_norm": 0.1760522723197937,
      "learning_rate": 2.926966666666667e-05,
      "loss": 0.0057,
      "step": 12194
    },
    {
      "epoch": 22.50046168051708,
      "grad_norm": 0.1980983018875122,
      "learning_rate": 2.9269333333333335e-05,
      "loss": 0.0042,
      "step": 12195
    },
    {
      "epoch": 22.50230840258541,
      "grad_norm": 0.24898752570152283,
      "learning_rate": 2.9269e-05,
      "loss": 0.0054,
      "step": 12196
    },
    {
      "epoch": 22.50415512465374,
      "grad_norm": 0.42561447620391846,
      "learning_rate": 2.9268666666666666e-05,
      "loss": 0.0061,
      "step": 12197
    },
    {
      "epoch": 22.506001846722068,
      "grad_norm": 0.3103271424770355,
      "learning_rate": 2.9268333333333332e-05,
      "loss": 0.0088,
      "step": 12198
    },
    {
      "epoch": 22.507848568790397,
      "grad_norm": 0.2575102746486664,
      "learning_rate": 2.9268e-05,
      "loss": 0.008,
      "step": 12199
    },
    {
      "epoch": 22.509695290858726,
      "grad_norm": 0.14075084030628204,
      "learning_rate": 2.9267666666666664e-05,
      "loss": 0.0051,
      "step": 12200
    },
    {
      "epoch": 22.511542012927055,
      "grad_norm": 0.38890862464904785,
      "learning_rate": 2.9267333333333334e-05,
      "loss": 0.0121,
      "step": 12201
    },
    {
      "epoch": 22.513388734995385,
      "grad_norm": 0.15616172552108765,
      "learning_rate": 2.9267000000000003e-05,
      "loss": 0.0037,
      "step": 12202
    },
    {
      "epoch": 22.51523545706371,
      "grad_norm": 0.5177579522132874,
      "learning_rate": 2.9266666666666665e-05,
      "loss": 0.0101,
      "step": 12203
    },
    {
      "epoch": 22.51708217913204,
      "grad_norm": 0.20640073716640472,
      "learning_rate": 2.9266333333333335e-05,
      "loss": 0.0072,
      "step": 12204
    },
    {
      "epoch": 22.51892890120037,
      "grad_norm": 0.2602652907371521,
      "learning_rate": 2.9266e-05,
      "loss": 0.01,
      "step": 12205
    },
    {
      "epoch": 22.520775623268698,
      "grad_norm": 0.29666656255722046,
      "learning_rate": 2.9265666666666667e-05,
      "loss": 0.0082,
      "step": 12206
    },
    {
      "epoch": 22.522622345337027,
      "grad_norm": 0.33792397379875183,
      "learning_rate": 2.9265333333333333e-05,
      "loss": 0.0111,
      "step": 12207
    },
    {
      "epoch": 22.524469067405356,
      "grad_norm": 0.22936464846134186,
      "learning_rate": 2.9265000000000002e-05,
      "loss": 0.0093,
      "step": 12208
    },
    {
      "epoch": 22.526315789473685,
      "grad_norm": 0.270159512758255,
      "learning_rate": 2.9264666666666668e-05,
      "loss": 0.0086,
      "step": 12209
    },
    {
      "epoch": 22.528162511542014,
      "grad_norm": 0.14591240882873535,
      "learning_rate": 2.9264333333333334e-05,
      "loss": 0.0035,
      "step": 12210
    },
    {
      "epoch": 22.530009233610343,
      "grad_norm": 0.15915511548519135,
      "learning_rate": 2.9264000000000003e-05,
      "loss": 0.0059,
      "step": 12211
    },
    {
      "epoch": 22.53185595567867,
      "grad_norm": 0.20213396847248077,
      "learning_rate": 2.9263666666666666e-05,
      "loss": 0.0053,
      "step": 12212
    },
    {
      "epoch": 22.533702677746998,
      "grad_norm": 0.26984938979148865,
      "learning_rate": 2.9263333333333335e-05,
      "loss": 0.008,
      "step": 12213
    },
    {
      "epoch": 22.535549399815327,
      "grad_norm": 0.3322078287601471,
      "learning_rate": 2.9263e-05,
      "loss": 0.009,
      "step": 12214
    },
    {
      "epoch": 22.537396121883656,
      "grad_norm": 0.7859511375427246,
      "learning_rate": 2.9262666666666667e-05,
      "loss": 0.0101,
      "step": 12215
    },
    {
      "epoch": 22.539242843951985,
      "grad_norm": 0.18283531069755554,
      "learning_rate": 2.9262333333333333e-05,
      "loss": 0.0044,
      "step": 12216
    },
    {
      "epoch": 22.541089566020315,
      "grad_norm": 0.3152421712875366,
      "learning_rate": 2.9262000000000002e-05,
      "loss": 0.0115,
      "step": 12217
    },
    {
      "epoch": 22.542936288088644,
      "grad_norm": 0.24157525599002838,
      "learning_rate": 2.9261666666666668e-05,
      "loss": 0.0057,
      "step": 12218
    },
    {
      "epoch": 22.544783010156973,
      "grad_norm": 0.26296094059944153,
      "learning_rate": 2.9261333333333334e-05,
      "loss": 0.0093,
      "step": 12219
    },
    {
      "epoch": 22.5466297322253,
      "grad_norm": 0.4341523051261902,
      "learning_rate": 2.9261000000000003e-05,
      "loss": 0.0098,
      "step": 12220
    },
    {
      "epoch": 22.548476454293628,
      "grad_norm": 0.2000827044248581,
      "learning_rate": 2.9260666666666666e-05,
      "loss": 0.0048,
      "step": 12221
    },
    {
      "epoch": 22.550323176361957,
      "grad_norm": 0.4425687789916992,
      "learning_rate": 2.9260333333333335e-05,
      "loss": 0.0083,
      "step": 12222
    },
    {
      "epoch": 22.552169898430286,
      "grad_norm": 0.3309001922607422,
      "learning_rate": 2.926e-05,
      "loss": 0.0098,
      "step": 12223
    },
    {
      "epoch": 22.554016620498615,
      "grad_norm": 0.18561774492263794,
      "learning_rate": 2.9259666666666667e-05,
      "loss": 0.0151,
      "step": 12224
    },
    {
      "epoch": 22.555863342566944,
      "grad_norm": 0.5745847225189209,
      "learning_rate": 2.9259333333333333e-05,
      "loss": 0.1422,
      "step": 12225
    },
    {
      "epoch": 22.557710064635273,
      "grad_norm": 0.4258785545825958,
      "learning_rate": 2.9259e-05,
      "loss": 0.1035,
      "step": 12226
    },
    {
      "epoch": 22.559556786703602,
      "grad_norm": 0.554884672164917,
      "learning_rate": 2.9258666666666668e-05,
      "loss": 0.093,
      "step": 12227
    },
    {
      "epoch": 22.56140350877193,
      "grad_norm": 0.35626792907714844,
      "learning_rate": 2.9258333333333334e-05,
      "loss": 0.0755,
      "step": 12228
    },
    {
      "epoch": 22.563250230840257,
      "grad_norm": 0.5393857955932617,
      "learning_rate": 2.9258e-05,
      "loss": 0.0659,
      "step": 12229
    },
    {
      "epoch": 22.565096952908586,
      "grad_norm": 0.43814486265182495,
      "learning_rate": 2.9257666666666666e-05,
      "loss": 0.0474,
      "step": 12230
    },
    {
      "epoch": 22.566943674976915,
      "grad_norm": 0.6040318012237549,
      "learning_rate": 2.9257333333333335e-05,
      "loss": 0.0505,
      "step": 12231
    },
    {
      "epoch": 22.568790397045245,
      "grad_norm": 0.4554961919784546,
      "learning_rate": 2.9256999999999998e-05,
      "loss": 0.064,
      "step": 12232
    },
    {
      "epoch": 22.570637119113574,
      "grad_norm": 0.36496084928512573,
      "learning_rate": 2.9256666666666667e-05,
      "loss": 0.0463,
      "step": 12233
    },
    {
      "epoch": 22.572483841181903,
      "grad_norm": 0.8404495716094971,
      "learning_rate": 2.9256333333333337e-05,
      "loss": 0.038,
      "step": 12234
    },
    {
      "epoch": 22.574330563250232,
      "grad_norm": 0.37269464135169983,
      "learning_rate": 2.9256e-05,
      "loss": 0.0317,
      "step": 12235
    },
    {
      "epoch": 22.57617728531856,
      "grad_norm": 0.659870445728302,
      "learning_rate": 2.925566666666667e-05,
      "loss": 0.0335,
      "step": 12236
    },
    {
      "epoch": 22.578024007386887,
      "grad_norm": 0.22030824422836304,
      "learning_rate": 2.9255333333333334e-05,
      "loss": 0.0308,
      "step": 12237
    },
    {
      "epoch": 22.579870729455216,
      "grad_norm": 0.3626607358455658,
      "learning_rate": 2.9255e-05,
      "loss": 0.0477,
      "step": 12238
    },
    {
      "epoch": 22.581717451523545,
      "grad_norm": 0.2966305911540985,
      "learning_rate": 2.9254666666666666e-05,
      "loss": 0.0113,
      "step": 12239
    },
    {
      "epoch": 22.583564173591874,
      "grad_norm": 0.3302536904811859,
      "learning_rate": 2.9254333333333336e-05,
      "loss": 0.0311,
      "step": 12240
    },
    {
      "epoch": 22.585410895660203,
      "grad_norm": 0.1487927883863449,
      "learning_rate": 2.9253999999999998e-05,
      "loss": 0.0076,
      "step": 12241
    },
    {
      "epoch": 22.587257617728532,
      "grad_norm": 0.22927799820899963,
      "learning_rate": 2.9253666666666667e-05,
      "loss": 0.0106,
      "step": 12242
    },
    {
      "epoch": 22.58910433979686,
      "grad_norm": 0.1772453337907791,
      "learning_rate": 2.9253333333333337e-05,
      "loss": 0.0095,
      "step": 12243
    },
    {
      "epoch": 22.59095106186519,
      "grad_norm": 0.6161147356033325,
      "learning_rate": 2.9253e-05,
      "loss": 0.0099,
      "step": 12244
    },
    {
      "epoch": 22.592797783933516,
      "grad_norm": 0.16344311833381653,
      "learning_rate": 2.925266666666667e-05,
      "loss": 0.0298,
      "step": 12245
    },
    {
      "epoch": 22.594644506001845,
      "grad_norm": 0.299240380525589,
      "learning_rate": 2.9252333333333335e-05,
      "loss": 0.0086,
      "step": 12246
    },
    {
      "epoch": 22.596491228070175,
      "grad_norm": 0.23759780824184418,
      "learning_rate": 2.9252e-05,
      "loss": 0.0075,
      "step": 12247
    },
    {
      "epoch": 22.598337950138504,
      "grad_norm": 0.14949437975883484,
      "learning_rate": 2.9251666666666666e-05,
      "loss": 0.0048,
      "step": 12248
    },
    {
      "epoch": 22.600184672206833,
      "grad_norm": 0.2368757277727127,
      "learning_rate": 2.9251333333333336e-05,
      "loss": 0.0061,
      "step": 12249
    },
    {
      "epoch": 22.602031394275162,
      "grad_norm": 0.2707505226135254,
      "learning_rate": 2.9250999999999998e-05,
      "loss": 0.0055,
      "step": 12250
    },
    {
      "epoch": 22.60387811634349,
      "grad_norm": 0.34671154618263245,
      "learning_rate": 2.9250666666666668e-05,
      "loss": 0.023,
      "step": 12251
    },
    {
      "epoch": 22.60572483841182,
      "grad_norm": 0.14770708978176117,
      "learning_rate": 2.9250333333333337e-05,
      "loss": 0.0025,
      "step": 12252
    },
    {
      "epoch": 22.607571560480146,
      "grad_norm": 0.30914875864982605,
      "learning_rate": 2.925e-05,
      "loss": 0.0089,
      "step": 12253
    },
    {
      "epoch": 22.609418282548475,
      "grad_norm": 0.3850860893726349,
      "learning_rate": 2.924966666666667e-05,
      "loss": 0.0096,
      "step": 12254
    },
    {
      "epoch": 22.611265004616804,
      "grad_norm": 0.7597085237503052,
      "learning_rate": 2.9249333333333335e-05,
      "loss": 0.0407,
      "step": 12255
    },
    {
      "epoch": 22.613111726685133,
      "grad_norm": 0.13821229338645935,
      "learning_rate": 2.9249e-05,
      "loss": 0.0022,
      "step": 12256
    },
    {
      "epoch": 22.614958448753463,
      "grad_norm": 0.2871163487434387,
      "learning_rate": 2.9248666666666667e-05,
      "loss": 0.0092,
      "step": 12257
    },
    {
      "epoch": 22.61680517082179,
      "grad_norm": 0.17912018299102783,
      "learning_rate": 2.9248333333333333e-05,
      "loss": 0.0068,
      "step": 12258
    },
    {
      "epoch": 22.61865189289012,
      "grad_norm": 0.21821361780166626,
      "learning_rate": 2.9248000000000002e-05,
      "loss": 0.0055,
      "step": 12259
    },
    {
      "epoch": 22.62049861495845,
      "grad_norm": 0.09547823667526245,
      "learning_rate": 2.9247666666666668e-05,
      "loss": 0.0034,
      "step": 12260
    },
    {
      "epoch": 22.62234533702678,
      "grad_norm": 0.2564859092235565,
      "learning_rate": 2.9247333333333334e-05,
      "loss": 0.0101,
      "step": 12261
    },
    {
      "epoch": 22.624192059095105,
      "grad_norm": 0.37441202998161316,
      "learning_rate": 2.9247e-05,
      "loss": 0.0173,
      "step": 12262
    },
    {
      "epoch": 22.626038781163434,
      "grad_norm": 0.3169558644294739,
      "learning_rate": 2.924666666666667e-05,
      "loss": 0.0136,
      "step": 12263
    },
    {
      "epoch": 22.627885503231763,
      "grad_norm": 0.4918834865093231,
      "learning_rate": 2.924633333333333e-05,
      "loss": 0.0113,
      "step": 12264
    },
    {
      "epoch": 22.629732225300092,
      "grad_norm": 0.4788316786289215,
      "learning_rate": 2.9246e-05,
      "loss": 0.0171,
      "step": 12265
    },
    {
      "epoch": 22.63157894736842,
      "grad_norm": 0.18662941455841064,
      "learning_rate": 2.9245666666666667e-05,
      "loss": 0.0084,
      "step": 12266
    },
    {
      "epoch": 22.63342566943675,
      "grad_norm": 0.2884886860847473,
      "learning_rate": 2.9245333333333333e-05,
      "loss": 0.0075,
      "step": 12267
    },
    {
      "epoch": 22.63527239150508,
      "grad_norm": 0.32492247223854065,
      "learning_rate": 2.9245000000000002e-05,
      "loss": 0.0111,
      "step": 12268
    },
    {
      "epoch": 22.63711911357341,
      "grad_norm": 0.16241343319416046,
      "learning_rate": 2.9244666666666668e-05,
      "loss": 0.0052,
      "step": 12269
    },
    {
      "epoch": 22.638965835641734,
      "grad_norm": 0.8470730781555176,
      "learning_rate": 2.9244333333333334e-05,
      "loss": 0.0159,
      "step": 12270
    },
    {
      "epoch": 22.640812557710063,
      "grad_norm": 0.6832287311553955,
      "learning_rate": 2.9244e-05,
      "loss": 0.0084,
      "step": 12271
    },
    {
      "epoch": 22.642659279778393,
      "grad_norm": 0.7469265460968018,
      "learning_rate": 2.924366666666667e-05,
      "loss": 0.0117,
      "step": 12272
    },
    {
      "epoch": 22.64450600184672,
      "grad_norm": 0.5027961134910583,
      "learning_rate": 2.9243333333333332e-05,
      "loss": 0.013,
      "step": 12273
    },
    {
      "epoch": 22.64635272391505,
      "grad_norm": 0.536918580532074,
      "learning_rate": 2.9243e-05,
      "loss": 0.0101,
      "step": 12274
    },
    {
      "epoch": 22.64819944598338,
      "grad_norm": 0.6451272368431091,
      "learning_rate": 2.9242666666666667e-05,
      "loss": 0.111,
      "step": 12275
    },
    {
      "epoch": 22.65004616805171,
      "grad_norm": 0.576657235622406,
      "learning_rate": 2.9242333333333333e-05,
      "loss": 0.1548,
      "step": 12276
    },
    {
      "epoch": 22.65189289012004,
      "grad_norm": 0.4048967957496643,
      "learning_rate": 2.9242000000000002e-05,
      "loss": 0.0777,
      "step": 12277
    },
    {
      "epoch": 22.653739612188367,
      "grad_norm": 0.6146196722984314,
      "learning_rate": 2.9241666666666668e-05,
      "loss": 0.0728,
      "step": 12278
    },
    {
      "epoch": 22.655586334256693,
      "grad_norm": 0.541233479976654,
      "learning_rate": 2.9241333333333334e-05,
      "loss": 0.0761,
      "step": 12279
    },
    {
      "epoch": 22.657433056325022,
      "grad_norm": 0.49906066060066223,
      "learning_rate": 2.9241e-05,
      "loss": 0.0583,
      "step": 12280
    },
    {
      "epoch": 22.65927977839335,
      "grad_norm": 0.3858031928539276,
      "learning_rate": 2.924066666666667e-05,
      "loss": 0.0799,
      "step": 12281
    },
    {
      "epoch": 22.66112650046168,
      "grad_norm": 0.34499648213386536,
      "learning_rate": 2.9240333333333332e-05,
      "loss": 0.0415,
      "step": 12282
    },
    {
      "epoch": 22.66297322253001,
      "grad_norm": 0.5997113585472107,
      "learning_rate": 2.924e-05,
      "loss": 0.0449,
      "step": 12283
    },
    {
      "epoch": 22.66481994459834,
      "grad_norm": 0.35634171962738037,
      "learning_rate": 2.923966666666667e-05,
      "loss": 0.0362,
      "step": 12284
    },
    {
      "epoch": 22.666666666666668,
      "grad_norm": 0.37014394998550415,
      "learning_rate": 2.9239333333333333e-05,
      "loss": 0.0327,
      "step": 12285
    },
    {
      "epoch": 22.668513388734997,
      "grad_norm": 0.49967631697654724,
      "learning_rate": 2.9239000000000002e-05,
      "loss": 0.0628,
      "step": 12286
    },
    {
      "epoch": 22.670360110803323,
      "grad_norm": 0.2549503743648529,
      "learning_rate": 2.9238666666666665e-05,
      "loss": 0.0292,
      "step": 12287
    },
    {
      "epoch": 22.67220683287165,
      "grad_norm": 0.5967966318130493,
      "learning_rate": 2.9238333333333334e-05,
      "loss": 0.015,
      "step": 12288
    },
    {
      "epoch": 22.67405355493998,
      "grad_norm": 0.36082586646080017,
      "learning_rate": 2.9238e-05,
      "loss": 0.0207,
      "step": 12289
    },
    {
      "epoch": 22.67590027700831,
      "grad_norm": 0.5506548285484314,
      "learning_rate": 2.9237666666666666e-05,
      "loss": 0.0234,
      "step": 12290
    },
    {
      "epoch": 22.67774699907664,
      "grad_norm": 0.3026949465274811,
      "learning_rate": 2.9237333333333332e-05,
      "loss": 0.0488,
      "step": 12291
    },
    {
      "epoch": 22.67959372114497,
      "grad_norm": 0.1955622285604477,
      "learning_rate": 2.9237e-05,
      "loss": 0.0094,
      "step": 12292
    },
    {
      "epoch": 22.681440443213297,
      "grad_norm": 0.1570124328136444,
      "learning_rate": 2.9236666666666667e-05,
      "loss": 0.0082,
      "step": 12293
    },
    {
      "epoch": 22.683287165281627,
      "grad_norm": 0.2552327811717987,
      "learning_rate": 2.9236333333333333e-05,
      "loss": 0.0144,
      "step": 12294
    },
    {
      "epoch": 22.685133887349952,
      "grad_norm": 0.16709206998348236,
      "learning_rate": 2.9236000000000003e-05,
      "loss": 0.0078,
      "step": 12295
    },
    {
      "epoch": 22.68698060941828,
      "grad_norm": 0.3998396694660187,
      "learning_rate": 2.9235666666666665e-05,
      "loss": 0.0126,
      "step": 12296
    },
    {
      "epoch": 22.68882733148661,
      "grad_norm": 0.30569109320640564,
      "learning_rate": 2.9235333333333335e-05,
      "loss": 0.0095,
      "step": 12297
    },
    {
      "epoch": 22.69067405355494,
      "grad_norm": 0.14271549880504608,
      "learning_rate": 2.9235e-05,
      "loss": 0.0078,
      "step": 12298
    },
    {
      "epoch": 22.69252077562327,
      "grad_norm": 0.2638216018676758,
      "learning_rate": 2.9234666666666666e-05,
      "loss": 0.0067,
      "step": 12299
    },
    {
      "epoch": 22.694367497691598,
      "grad_norm": 0.2199135720729828,
      "learning_rate": 2.9234333333333332e-05,
      "loss": 0.0054,
      "step": 12300
    },
    {
      "epoch": 22.696214219759927,
      "grad_norm": 0.22184860706329346,
      "learning_rate": 2.9234e-05,
      "loss": 0.0059,
      "step": 12301
    },
    {
      "epoch": 22.698060941828256,
      "grad_norm": 0.2710758447647095,
      "learning_rate": 2.9233666666666668e-05,
      "loss": 0.0062,
      "step": 12302
    },
    {
      "epoch": 22.69990766389658,
      "grad_norm": 0.38386720418930054,
      "learning_rate": 2.9233333333333334e-05,
      "loss": 0.0252,
      "step": 12303
    },
    {
      "epoch": 22.70175438596491,
      "grad_norm": 0.23074857890605927,
      "learning_rate": 2.9233000000000003e-05,
      "loss": 0.0105,
      "step": 12304
    },
    {
      "epoch": 22.70360110803324,
      "grad_norm": 0.12127603590488434,
      "learning_rate": 2.9232666666666665e-05,
      "loss": 0.0064,
      "step": 12305
    },
    {
      "epoch": 22.70544783010157,
      "grad_norm": 0.29814377427101135,
      "learning_rate": 2.9232333333333335e-05,
      "loss": 0.0078,
      "step": 12306
    },
    {
      "epoch": 22.7072945521699,
      "grad_norm": 0.18820050358772278,
      "learning_rate": 2.9232e-05,
      "loss": 0.0058,
      "step": 12307
    },
    {
      "epoch": 22.709141274238227,
      "grad_norm": 0.3828445374965668,
      "learning_rate": 2.9231666666666667e-05,
      "loss": 0.0119,
      "step": 12308
    },
    {
      "epoch": 22.710987996306557,
      "grad_norm": 0.3817533850669861,
      "learning_rate": 2.9231333333333336e-05,
      "loss": 0.0095,
      "step": 12309
    },
    {
      "epoch": 22.712834718374886,
      "grad_norm": 0.270891398191452,
      "learning_rate": 2.9231000000000002e-05,
      "loss": 0.0106,
      "step": 12310
    },
    {
      "epoch": 22.714681440443215,
      "grad_norm": 0.09616384655237198,
      "learning_rate": 2.9230666666666668e-05,
      "loss": 0.0034,
      "step": 12311
    },
    {
      "epoch": 22.71652816251154,
      "grad_norm": 0.16380581259727478,
      "learning_rate": 2.9230333333333334e-05,
      "loss": 0.0027,
      "step": 12312
    },
    {
      "epoch": 22.71837488457987,
      "grad_norm": 0.35364049673080444,
      "learning_rate": 2.9230000000000003e-05,
      "loss": 0.0121,
      "step": 12313
    },
    {
      "epoch": 22.7202216066482,
      "grad_norm": 0.4222791790962219,
      "learning_rate": 2.9229666666666666e-05,
      "loss": 0.0134,
      "step": 12314
    },
    {
      "epoch": 22.722068328716528,
      "grad_norm": 0.3655267655849457,
      "learning_rate": 2.9229333333333335e-05,
      "loss": 0.0047,
      "step": 12315
    },
    {
      "epoch": 22.723915050784857,
      "grad_norm": 0.1754109412431717,
      "learning_rate": 2.9229e-05,
      "loss": 0.0056,
      "step": 12316
    },
    {
      "epoch": 22.725761772853186,
      "grad_norm": 0.21043196320533752,
      "learning_rate": 2.9228666666666667e-05,
      "loss": 0.0084,
      "step": 12317
    },
    {
      "epoch": 22.727608494921515,
      "grad_norm": 0.26307398080825806,
      "learning_rate": 2.9228333333333336e-05,
      "loss": 0.008,
      "step": 12318
    },
    {
      "epoch": 22.729455216989845,
      "grad_norm": 0.43681734800338745,
      "learning_rate": 2.9228e-05,
      "loss": 0.0064,
      "step": 12319
    },
    {
      "epoch": 22.73130193905817,
      "grad_norm": 0.267784059047699,
      "learning_rate": 2.9227666666666668e-05,
      "loss": 0.0069,
      "step": 12320
    },
    {
      "epoch": 22.7331486611265,
      "grad_norm": 0.22806259989738464,
      "learning_rate": 2.9227333333333334e-05,
      "loss": 0.0081,
      "step": 12321
    },
    {
      "epoch": 22.73499538319483,
      "grad_norm": 0.16762927174568176,
      "learning_rate": 2.9227e-05,
      "loss": 0.0072,
      "step": 12322
    },
    {
      "epoch": 22.736842105263158,
      "grad_norm": 0.41946953535079956,
      "learning_rate": 2.9226666666666666e-05,
      "loss": 0.0075,
      "step": 12323
    },
    {
      "epoch": 22.738688827331487,
      "grad_norm": 0.3906378448009491,
      "learning_rate": 2.9226333333333335e-05,
      "loss": 0.0139,
      "step": 12324
    },
    {
      "epoch": 22.740535549399816,
      "grad_norm": 0.5999815464019775,
      "learning_rate": 2.9226e-05,
      "loss": 0.1396,
      "step": 12325
    },
    {
      "epoch": 22.742382271468145,
      "grad_norm": 0.3809105157852173,
      "learning_rate": 2.9225666666666667e-05,
      "loss": 0.0924,
      "step": 12326
    },
    {
      "epoch": 22.744228993536474,
      "grad_norm": 0.3599500060081482,
      "learning_rate": 2.9225333333333336e-05,
      "loss": 0.053,
      "step": 12327
    },
    {
      "epoch": 22.746075715604803,
      "grad_norm": 0.529497504234314,
      "learning_rate": 2.9225e-05,
      "loss": 0.0831,
      "step": 12328
    },
    {
      "epoch": 22.74792243767313,
      "grad_norm": 0.5233191847801208,
      "learning_rate": 2.9224666666666668e-05,
      "loss": 0.0952,
      "step": 12329
    },
    {
      "epoch": 22.749769159741458,
      "grad_norm": 0.5627424716949463,
      "learning_rate": 2.9224333333333334e-05,
      "loss": 0.1162,
      "step": 12330
    },
    {
      "epoch": 22.751615881809787,
      "grad_norm": 0.5519795417785645,
      "learning_rate": 2.9224e-05,
      "loss": 0.0477,
      "step": 12331
    },
    {
      "epoch": 22.753462603878116,
      "grad_norm": 0.6382800936698914,
      "learning_rate": 2.9223666666666666e-05,
      "loss": 0.0557,
      "step": 12332
    },
    {
      "epoch": 22.755309325946445,
      "grad_norm": 0.3135432004928589,
      "learning_rate": 2.9223333333333335e-05,
      "loss": 0.037,
      "step": 12333
    },
    {
      "epoch": 22.757156048014775,
      "grad_norm": 0.26903706789016724,
      "learning_rate": 2.9223e-05,
      "loss": 0.0289,
      "step": 12334
    },
    {
      "epoch": 22.759002770083104,
      "grad_norm": 0.3219813108444214,
      "learning_rate": 2.9222666666666667e-05,
      "loss": 0.0197,
      "step": 12335
    },
    {
      "epoch": 22.760849492151433,
      "grad_norm": 0.611541211605072,
      "learning_rate": 2.9222333333333337e-05,
      "loss": 0.0809,
      "step": 12336
    },
    {
      "epoch": 22.76269621421976,
      "grad_norm": 0.5747827291488647,
      "learning_rate": 2.9222e-05,
      "loss": 0.0344,
      "step": 12337
    },
    {
      "epoch": 22.764542936288088,
      "grad_norm": 0.2298043668270111,
      "learning_rate": 2.922166666666667e-05,
      "loss": 0.022,
      "step": 12338
    },
    {
      "epoch": 22.766389658356417,
      "grad_norm": 0.3402498960494995,
      "learning_rate": 2.9221333333333334e-05,
      "loss": 0.0213,
      "step": 12339
    },
    {
      "epoch": 22.768236380424746,
      "grad_norm": 0.2586126923561096,
      "learning_rate": 2.9221e-05,
      "loss": 0.0232,
      "step": 12340
    },
    {
      "epoch": 22.770083102493075,
      "grad_norm": 0.15024542808532715,
      "learning_rate": 2.9220666666666666e-05,
      "loss": 0.0079,
      "step": 12341
    },
    {
      "epoch": 22.771929824561404,
      "grad_norm": 0.3457852900028229,
      "learning_rate": 2.9220333333333336e-05,
      "loss": 0.0245,
      "step": 12342
    },
    {
      "epoch": 22.773776546629733,
      "grad_norm": 0.1955130249261856,
      "learning_rate": 2.922e-05,
      "loss": 0.0227,
      "step": 12343
    },
    {
      "epoch": 22.775623268698062,
      "grad_norm": 0.16288869082927704,
      "learning_rate": 2.9219666666666667e-05,
      "loss": 0.0053,
      "step": 12344
    },
    {
      "epoch": 22.777469990766388,
      "grad_norm": 0.17608100175857544,
      "learning_rate": 2.9219333333333337e-05,
      "loss": 0.0078,
      "step": 12345
    },
    {
      "epoch": 22.779316712834717,
      "grad_norm": 0.3658122420310974,
      "learning_rate": 2.9219e-05,
      "loss": 0.013,
      "step": 12346
    },
    {
      "epoch": 22.781163434903046,
      "grad_norm": 0.130655437707901,
      "learning_rate": 2.921866666666667e-05,
      "loss": 0.004,
      "step": 12347
    },
    {
      "epoch": 22.783010156971375,
      "grad_norm": 0.4952363967895508,
      "learning_rate": 2.921833333333333e-05,
      "loss": 0.0066,
      "step": 12348
    },
    {
      "epoch": 22.784856879039705,
      "grad_norm": 0.278836727142334,
      "learning_rate": 2.9218e-05,
      "loss": 0.009,
      "step": 12349
    },
    {
      "epoch": 22.786703601108034,
      "grad_norm": 0.13057789206504822,
      "learning_rate": 2.9217666666666666e-05,
      "loss": 0.0044,
      "step": 12350
    },
    {
      "epoch": 22.788550323176363,
      "grad_norm": 0.15483197569847107,
      "learning_rate": 2.9217333333333332e-05,
      "loss": 0.006,
      "step": 12351
    },
    {
      "epoch": 22.790397045244692,
      "grad_norm": 0.6134659051895142,
      "learning_rate": 2.9217e-05,
      "loss": 0.0126,
      "step": 12352
    },
    {
      "epoch": 22.792243767313018,
      "grad_norm": 0.2991887927055359,
      "learning_rate": 2.9216666666666668e-05,
      "loss": 0.0077,
      "step": 12353
    },
    {
      "epoch": 22.794090489381347,
      "grad_norm": 0.1424655318260193,
      "learning_rate": 2.9216333333333334e-05,
      "loss": 0.003,
      "step": 12354
    },
    {
      "epoch": 22.795937211449676,
      "grad_norm": 1.4153794050216675,
      "learning_rate": 2.9216e-05,
      "loss": 0.0358,
      "step": 12355
    },
    {
      "epoch": 22.797783933518005,
      "grad_norm": 0.13688984513282776,
      "learning_rate": 2.921566666666667e-05,
      "loss": 0.004,
      "step": 12356
    },
    {
      "epoch": 22.799630655586334,
      "grad_norm": 0.38424474000930786,
      "learning_rate": 2.921533333333333e-05,
      "loss": 0.0066,
      "step": 12357
    },
    {
      "epoch": 22.801477377654663,
      "grad_norm": 0.19820640981197357,
      "learning_rate": 2.9215e-05,
      "loss": 0.0051,
      "step": 12358
    },
    {
      "epoch": 22.803324099722992,
      "grad_norm": 0.20740287005901337,
      "learning_rate": 2.921466666666667e-05,
      "loss": 0.007,
      "step": 12359
    },
    {
      "epoch": 22.80517082179132,
      "grad_norm": 0.14528685808181763,
      "learning_rate": 2.9214333333333333e-05,
      "loss": 0.0036,
      "step": 12360
    },
    {
      "epoch": 22.80701754385965,
      "grad_norm": 0.1387391835451126,
      "learning_rate": 2.9214000000000002e-05,
      "loss": 0.0283,
      "step": 12361
    },
    {
      "epoch": 22.808864265927976,
      "grad_norm": 0.3198137581348419,
      "learning_rate": 2.9213666666666668e-05,
      "loss": 0.0107,
      "step": 12362
    },
    {
      "epoch": 22.810710987996305,
      "grad_norm": 0.6842020750045776,
      "learning_rate": 2.9213333333333334e-05,
      "loss": 0.0076,
      "step": 12363
    },
    {
      "epoch": 22.812557710064635,
      "grad_norm": 0.3192748725414276,
      "learning_rate": 2.9213e-05,
      "loss": 0.0049,
      "step": 12364
    },
    {
      "epoch": 22.814404432132964,
      "grad_norm": 0.15512192249298096,
      "learning_rate": 2.921266666666667e-05,
      "loss": 0.0048,
      "step": 12365
    },
    {
      "epoch": 22.816251154201293,
      "grad_norm": 0.15249937772750854,
      "learning_rate": 2.921233333333333e-05,
      "loss": 0.0041,
      "step": 12366
    },
    {
      "epoch": 22.818097876269622,
      "grad_norm": 0.44840556383132935,
      "learning_rate": 2.9212e-05,
      "loss": 0.0116,
      "step": 12367
    },
    {
      "epoch": 22.81994459833795,
      "grad_norm": 0.5910830497741699,
      "learning_rate": 2.921166666666667e-05,
      "loss": 0.0089,
      "step": 12368
    },
    {
      "epoch": 22.82179132040628,
      "grad_norm": 0.25855201482772827,
      "learning_rate": 2.9211333333333333e-05,
      "loss": 0.0104,
      "step": 12369
    },
    {
      "epoch": 22.823638042474606,
      "grad_norm": 0.25850245356559753,
      "learning_rate": 2.9211000000000002e-05,
      "loss": 0.0112,
      "step": 12370
    },
    {
      "epoch": 22.825484764542935,
      "grad_norm": 0.40904897451400757,
      "learning_rate": 2.9210666666666668e-05,
      "loss": 0.0103,
      "step": 12371
    },
    {
      "epoch": 22.827331486611264,
      "grad_norm": 0.4205116927623749,
      "learning_rate": 2.9210333333333334e-05,
      "loss": 0.0101,
      "step": 12372
    },
    {
      "epoch": 22.829178208679593,
      "grad_norm": 0.2385093718767166,
      "learning_rate": 2.921e-05,
      "loss": 0.0056,
      "step": 12373
    },
    {
      "epoch": 22.831024930747922,
      "grad_norm": 0.3131072223186493,
      "learning_rate": 2.920966666666667e-05,
      "loss": 0.0096,
      "step": 12374
    },
    {
      "epoch": 22.83287165281625,
      "grad_norm": 0.4860314130783081,
      "learning_rate": 2.9209333333333335e-05,
      "loss": 0.0993,
      "step": 12375
    },
    {
      "epoch": 22.83471837488458,
      "grad_norm": 0.5568736791610718,
      "learning_rate": 2.9209e-05,
      "loss": 0.1497,
      "step": 12376
    },
    {
      "epoch": 22.83656509695291,
      "grad_norm": 0.649315595626831,
      "learning_rate": 2.9208666666666667e-05,
      "loss": 0.1676,
      "step": 12377
    },
    {
      "epoch": 22.83841181902124,
      "grad_norm": 0.46940651535987854,
      "learning_rate": 2.9208333333333333e-05,
      "loss": 0.0722,
      "step": 12378
    },
    {
      "epoch": 22.840258541089565,
      "grad_norm": 0.43727776408195496,
      "learning_rate": 2.9208000000000002e-05,
      "loss": 0.0828,
      "step": 12379
    },
    {
      "epoch": 22.842105263157894,
      "grad_norm": 0.6778110265731812,
      "learning_rate": 2.9207666666666665e-05,
      "loss": 0.05,
      "step": 12380
    },
    {
      "epoch": 22.843951985226223,
      "grad_norm": 0.3043400049209595,
      "learning_rate": 2.9207333333333334e-05,
      "loss": 0.0321,
      "step": 12381
    },
    {
      "epoch": 22.845798707294552,
      "grad_norm": 0.590112566947937,
      "learning_rate": 2.9207e-05,
      "loss": 0.0585,
      "step": 12382
    },
    {
      "epoch": 22.84764542936288,
      "grad_norm": 0.3127261996269226,
      "learning_rate": 2.9206666666666666e-05,
      "loss": 0.0292,
      "step": 12383
    },
    {
      "epoch": 22.84949215143121,
      "grad_norm": 0.3524312973022461,
      "learning_rate": 2.9206333333333335e-05,
      "loss": 0.0343,
      "step": 12384
    },
    {
      "epoch": 22.85133887349954,
      "grad_norm": 0.3275330662727356,
      "learning_rate": 2.9206e-05,
      "loss": 0.021,
      "step": 12385
    },
    {
      "epoch": 22.85318559556787,
      "grad_norm": 0.2704507112503052,
      "learning_rate": 2.9205666666666667e-05,
      "loss": 0.0518,
      "step": 12386
    },
    {
      "epoch": 22.855032317636194,
      "grad_norm": 0.3582041263580322,
      "learning_rate": 2.9205333333333333e-05,
      "loss": 0.0167,
      "step": 12387
    },
    {
      "epoch": 22.856879039704523,
      "grad_norm": 0.334196537733078,
      "learning_rate": 2.9205000000000002e-05,
      "loss": 0.0154,
      "step": 12388
    },
    {
      "epoch": 22.858725761772853,
      "grad_norm": 0.23437385261058807,
      "learning_rate": 2.9204666666666665e-05,
      "loss": 0.0097,
      "step": 12389
    },
    {
      "epoch": 22.86057248384118,
      "grad_norm": 0.5138218402862549,
      "learning_rate": 2.9204333333333334e-05,
      "loss": 0.0144,
      "step": 12390
    },
    {
      "epoch": 22.86241920590951,
      "grad_norm": 0.422588586807251,
      "learning_rate": 2.9204e-05,
      "loss": 0.0121,
      "step": 12391
    },
    {
      "epoch": 22.86426592797784,
      "grad_norm": 0.2288866490125656,
      "learning_rate": 2.9203666666666666e-05,
      "loss": 0.0129,
      "step": 12392
    },
    {
      "epoch": 22.86611265004617,
      "grad_norm": 0.1804754137992859,
      "learning_rate": 2.9203333333333336e-05,
      "loss": 0.0082,
      "step": 12393
    },
    {
      "epoch": 22.8679593721145,
      "grad_norm": 0.2894505262374878,
      "learning_rate": 2.9203e-05,
      "loss": 0.01,
      "step": 12394
    },
    {
      "epoch": 22.869806094182824,
      "grad_norm": 0.2467958778142929,
      "learning_rate": 2.9202666666666667e-05,
      "loss": 0.0109,
      "step": 12395
    },
    {
      "epoch": 22.871652816251153,
      "grad_norm": 0.2496487945318222,
      "learning_rate": 2.9202333333333333e-05,
      "loss": 0.0096,
      "step": 12396
    },
    {
      "epoch": 22.873499538319482,
      "grad_norm": 0.22739478945732117,
      "learning_rate": 2.9202000000000003e-05,
      "loss": 0.0033,
      "step": 12397
    },
    {
      "epoch": 22.87534626038781,
      "grad_norm": 0.1910630315542221,
      "learning_rate": 2.9201666666666665e-05,
      "loss": 0.0071,
      "step": 12398
    },
    {
      "epoch": 22.87719298245614,
      "grad_norm": 0.1774922013282776,
      "learning_rate": 2.9201333333333335e-05,
      "loss": 0.0067,
      "step": 12399
    },
    {
      "epoch": 22.87903970452447,
      "grad_norm": 0.20055580139160156,
      "learning_rate": 2.9201e-05,
      "loss": 0.0056,
      "step": 12400
    },
    {
      "epoch": 22.8808864265928,
      "grad_norm": 0.311154842376709,
      "learning_rate": 2.9200666666666666e-05,
      "loss": 0.0077,
      "step": 12401
    },
    {
      "epoch": 22.882733148661128,
      "grad_norm": 0.23825916647911072,
      "learning_rate": 2.9200333333333336e-05,
      "loss": 0.0113,
      "step": 12402
    },
    {
      "epoch": 22.884579870729453,
      "grad_norm": 0.1427665799856186,
      "learning_rate": 2.92e-05,
      "loss": 0.0048,
      "step": 12403
    },
    {
      "epoch": 22.886426592797783,
      "grad_norm": 0.27265864610671997,
      "learning_rate": 2.9199666666666668e-05,
      "loss": 0.0062,
      "step": 12404
    },
    {
      "epoch": 22.88827331486611,
      "grad_norm": 0.2476198524236679,
      "learning_rate": 2.9199333333333334e-05,
      "loss": 0.0093,
      "step": 12405
    },
    {
      "epoch": 22.89012003693444,
      "grad_norm": 0.23540052771568298,
      "learning_rate": 2.9199000000000003e-05,
      "loss": 0.0033,
      "step": 12406
    },
    {
      "epoch": 22.89196675900277,
      "grad_norm": 0.14567874372005463,
      "learning_rate": 2.9198666666666665e-05,
      "loss": 0.0049,
      "step": 12407
    },
    {
      "epoch": 22.8938134810711,
      "grad_norm": 0.20529676973819733,
      "learning_rate": 2.9198333333333335e-05,
      "loss": 0.0038,
      "step": 12408
    },
    {
      "epoch": 22.89566020313943,
      "grad_norm": 0.1766471415758133,
      "learning_rate": 2.9198e-05,
      "loss": 0.0055,
      "step": 12409
    },
    {
      "epoch": 22.897506925207757,
      "grad_norm": 0.3108919858932495,
      "learning_rate": 2.9197666666666667e-05,
      "loss": 0.0057,
      "step": 12410
    },
    {
      "epoch": 22.899353647276087,
      "grad_norm": 0.44312068819999695,
      "learning_rate": 2.9197333333333336e-05,
      "loss": 0.0169,
      "step": 12411
    },
    {
      "epoch": 22.901200369344412,
      "grad_norm": 0.9064782857894897,
      "learning_rate": 2.9197e-05,
      "loss": 0.0138,
      "step": 12412
    },
    {
      "epoch": 22.90304709141274,
      "grad_norm": 0.39348894357681274,
      "learning_rate": 2.9196666666666668e-05,
      "loss": 0.0108,
      "step": 12413
    },
    {
      "epoch": 22.90489381348107,
      "grad_norm": 0.11488889157772064,
      "learning_rate": 2.9196333333333334e-05,
      "loss": 0.004,
      "step": 12414
    },
    {
      "epoch": 22.9067405355494,
      "grad_norm": 0.2089713215827942,
      "learning_rate": 2.9196e-05,
      "loss": 0.0065,
      "step": 12415
    },
    {
      "epoch": 22.90858725761773,
      "grad_norm": 0.20302043855190277,
      "learning_rate": 2.9195666666666666e-05,
      "loss": 0.0036,
      "step": 12416
    },
    {
      "epoch": 22.910433979686058,
      "grad_norm": 0.2970274090766907,
      "learning_rate": 2.9195333333333335e-05,
      "loss": 0.0065,
      "step": 12417
    },
    {
      "epoch": 22.912280701754387,
      "grad_norm": 1.0791375637054443,
      "learning_rate": 2.9195e-05,
      "loss": 0.0047,
      "step": 12418
    },
    {
      "epoch": 22.914127423822716,
      "grad_norm": 0.742299497127533,
      "learning_rate": 2.9194666666666667e-05,
      "loss": 0.0157,
      "step": 12419
    },
    {
      "epoch": 22.91597414589104,
      "grad_norm": 0.1627863496541977,
      "learning_rate": 2.9194333333333336e-05,
      "loss": 0.0041,
      "step": 12420
    },
    {
      "epoch": 22.91782086795937,
      "grad_norm": 0.10822322219610214,
      "learning_rate": 2.9194e-05,
      "loss": 0.0026,
      "step": 12421
    },
    {
      "epoch": 22.9196675900277,
      "grad_norm": 0.6408373713493347,
      "learning_rate": 2.9193666666666668e-05,
      "loss": 0.0173,
      "step": 12422
    },
    {
      "epoch": 22.92151431209603,
      "grad_norm": 0.3392656445503235,
      "learning_rate": 2.9193333333333334e-05,
      "loss": 0.0089,
      "step": 12423
    },
    {
      "epoch": 22.92336103416436,
      "grad_norm": 0.6655851006507874,
      "learning_rate": 2.9193e-05,
      "loss": 0.0062,
      "step": 12424
    },
    {
      "epoch": 22.925207756232687,
      "grad_norm": 0.5653405785560608,
      "learning_rate": 2.919266666666667e-05,
      "loss": 0.1226,
      "step": 12425
    },
    {
      "epoch": 22.927054478301017,
      "grad_norm": 0.8246968388557434,
      "learning_rate": 2.9192333333333335e-05,
      "loss": 0.1196,
      "step": 12426
    },
    {
      "epoch": 22.928901200369346,
      "grad_norm": 0.5462274551391602,
      "learning_rate": 2.9192e-05,
      "loss": 0.0869,
      "step": 12427
    },
    {
      "epoch": 22.930747922437675,
      "grad_norm": 0.31987103819847107,
      "learning_rate": 2.9191666666666667e-05,
      "loss": 0.0669,
      "step": 12428
    },
    {
      "epoch": 22.932594644506,
      "grad_norm": 0.4289247989654541,
      "learning_rate": 2.9191333333333336e-05,
      "loss": 0.0557,
      "step": 12429
    },
    {
      "epoch": 22.93444136657433,
      "grad_norm": 0.31459957361221313,
      "learning_rate": 2.9191e-05,
      "loss": 0.0436,
      "step": 12430
    },
    {
      "epoch": 22.93628808864266,
      "grad_norm": 0.7228941917419434,
      "learning_rate": 2.9190666666666668e-05,
      "loss": 0.0688,
      "step": 12431
    },
    {
      "epoch": 22.938134810710988,
      "grad_norm": 0.3625168800354004,
      "learning_rate": 2.9190333333333334e-05,
      "loss": 0.0388,
      "step": 12432
    },
    {
      "epoch": 22.939981532779317,
      "grad_norm": 0.3639623820781708,
      "learning_rate": 2.919e-05,
      "loss": 0.0273,
      "step": 12433
    },
    {
      "epoch": 22.941828254847646,
      "grad_norm": 0.391364723443985,
      "learning_rate": 2.918966666666667e-05,
      "loss": 0.0201,
      "step": 12434
    },
    {
      "epoch": 22.943674976915975,
      "grad_norm": 0.25264233350753784,
      "learning_rate": 2.9189333333333335e-05,
      "loss": 0.0196,
      "step": 12435
    },
    {
      "epoch": 22.945521698984304,
      "grad_norm": 0.24215036630630493,
      "learning_rate": 2.9189e-05,
      "loss": 0.0312,
      "step": 12436
    },
    {
      "epoch": 22.94736842105263,
      "grad_norm": 0.3020641803741455,
      "learning_rate": 2.9188666666666667e-05,
      "loss": 0.034,
      "step": 12437
    },
    {
      "epoch": 22.94921514312096,
      "grad_norm": 0.19730018079280853,
      "learning_rate": 2.9188333333333333e-05,
      "loss": 0.0124,
      "step": 12438
    },
    {
      "epoch": 22.95106186518929,
      "grad_norm": 0.23389595746994019,
      "learning_rate": 2.9188e-05,
      "loss": 0.0103,
      "step": 12439
    },
    {
      "epoch": 22.952908587257618,
      "grad_norm": 0.18454474210739136,
      "learning_rate": 2.918766666666667e-05,
      "loss": 0.0073,
      "step": 12440
    },
    {
      "epoch": 22.954755309325947,
      "grad_norm": 0.4759189486503601,
      "learning_rate": 2.918733333333333e-05,
      "loss": 0.0075,
      "step": 12441
    },
    {
      "epoch": 22.956602031394276,
      "grad_norm": 0.37919747829437256,
      "learning_rate": 2.9187e-05,
      "loss": 0.0115,
      "step": 12442
    },
    {
      "epoch": 22.958448753462605,
      "grad_norm": 0.2569182813167572,
      "learning_rate": 2.918666666666667e-05,
      "loss": 0.0079,
      "step": 12443
    },
    {
      "epoch": 22.960295475530934,
      "grad_norm": 0.3278430104255676,
      "learning_rate": 2.9186333333333332e-05,
      "loss": 0.0053,
      "step": 12444
    },
    {
      "epoch": 22.96214219759926,
      "grad_norm": 0.20840658247470856,
      "learning_rate": 2.9186e-05,
      "loss": 0.0179,
      "step": 12445
    },
    {
      "epoch": 22.96398891966759,
      "grad_norm": 0.15534381568431854,
      "learning_rate": 2.9185666666666667e-05,
      "loss": 0.0057,
      "step": 12446
    },
    {
      "epoch": 22.965835641735918,
      "grad_norm": 0.1486392617225647,
      "learning_rate": 2.9185333333333333e-05,
      "loss": 0.007,
      "step": 12447
    },
    {
      "epoch": 22.967682363804247,
      "grad_norm": 0.07954803854227066,
      "learning_rate": 2.9185e-05,
      "loss": 0.0029,
      "step": 12448
    },
    {
      "epoch": 22.969529085872576,
      "grad_norm": 0.13395774364471436,
      "learning_rate": 2.918466666666667e-05,
      "loss": 0.0049,
      "step": 12449
    },
    {
      "epoch": 22.971375807940905,
      "grad_norm": 0.17729364335536957,
      "learning_rate": 2.9184333333333335e-05,
      "loss": 0.0053,
      "step": 12450
    },
    {
      "epoch": 22.973222530009235,
      "grad_norm": 0.16160070896148682,
      "learning_rate": 2.9184e-05,
      "loss": 0.0042,
      "step": 12451
    },
    {
      "epoch": 22.975069252077564,
      "grad_norm": 0.28334203362464905,
      "learning_rate": 2.918366666666667e-05,
      "loss": 0.0061,
      "step": 12452
    },
    {
      "epoch": 22.97691597414589,
      "grad_norm": 0.3061280846595764,
      "learning_rate": 2.9183333333333332e-05,
      "loss": 0.0086,
      "step": 12453
    },
    {
      "epoch": 22.97876269621422,
      "grad_norm": 0.09037530422210693,
      "learning_rate": 2.9183e-05,
      "loss": 0.0019,
      "step": 12454
    },
    {
      "epoch": 22.980609418282548,
      "grad_norm": 0.1277635246515274,
      "learning_rate": 2.9182666666666668e-05,
      "loss": 0.004,
      "step": 12455
    },
    {
      "epoch": 22.982456140350877,
      "grad_norm": 0.20603525638580322,
      "learning_rate": 2.9182333333333334e-05,
      "loss": 0.009,
      "step": 12456
    },
    {
      "epoch": 22.984302862419206,
      "grad_norm": 0.23812738060951233,
      "learning_rate": 2.9182e-05,
      "loss": 0.0052,
      "step": 12457
    },
    {
      "epoch": 22.986149584487535,
      "grad_norm": 0.2609906494617462,
      "learning_rate": 2.918166666666667e-05,
      "loss": 0.004,
      "step": 12458
    },
    {
      "epoch": 22.987996306555864,
      "grad_norm": 0.3261003792285919,
      "learning_rate": 2.9181333333333335e-05,
      "loss": 0.008,
      "step": 12459
    },
    {
      "epoch": 22.989843028624193,
      "grad_norm": 0.08258796483278275,
      "learning_rate": 2.9181e-05,
      "loss": 0.0019,
      "step": 12460
    },
    {
      "epoch": 22.991689750692522,
      "grad_norm": 0.28730282187461853,
      "learning_rate": 2.918066666666667e-05,
      "loss": 0.0064,
      "step": 12461
    },
    {
      "epoch": 22.993536472760848,
      "grad_norm": 0.1670047640800476,
      "learning_rate": 2.9180333333333333e-05,
      "loss": 0.0064,
      "step": 12462
    },
    {
      "epoch": 22.995383194829177,
      "grad_norm": 0.23746854066848755,
      "learning_rate": 2.9180000000000002e-05,
      "loss": 0.0067,
      "step": 12463
    },
    {
      "epoch": 22.997229916897506,
      "grad_norm": 0.2047201544046402,
      "learning_rate": 2.9179666666666668e-05,
      "loss": 0.0051,
      "step": 12464
    },
    {
      "epoch": 22.999076638965835,
      "grad_norm": 0.2234080284833908,
      "learning_rate": 2.9179333333333334e-05,
      "loss": 0.0064,
      "step": 12465
    },
    {
      "epoch": 23.0,
      "grad_norm": 0.013559604994952679,
      "learning_rate": 2.9179e-05,
      "loss": 0.0002,
      "step": 12466
    },
    {
      "epoch": 23.00184672206833,
      "grad_norm": 1.3079710006713867,
      "learning_rate": 2.917866666666667e-05,
      "loss": 0.1453,
      "step": 12467
    },
    {
      "epoch": 23.00369344413666,
      "grad_norm": 0.37919649481773376,
      "learning_rate": 2.9178333333333335e-05,
      "loss": 0.0945,
      "step": 12468
    },
    {
      "epoch": 23.005540166204987,
      "grad_norm": 0.5010470151901245,
      "learning_rate": 2.9178e-05,
      "loss": 0.0713,
      "step": 12469
    },
    {
      "epoch": 23.007386888273317,
      "grad_norm": 0.5508612990379333,
      "learning_rate": 2.9177666666666667e-05,
      "loss": 0.0923,
      "step": 12470
    },
    {
      "epoch": 23.009233610341642,
      "grad_norm": 0.37328410148620605,
      "learning_rate": 2.9177333333333333e-05,
      "loss": 0.0611,
      "step": 12471
    },
    {
      "epoch": 23.01108033240997,
      "grad_norm": 0.4111389219760895,
      "learning_rate": 2.9177000000000002e-05,
      "loss": 0.081,
      "step": 12472
    },
    {
      "epoch": 23.0129270544783,
      "grad_norm": 0.37702229619026184,
      "learning_rate": 2.9176666666666665e-05,
      "loss": 0.0854,
      "step": 12473
    },
    {
      "epoch": 23.01477377654663,
      "grad_norm": 0.7613016963005066,
      "learning_rate": 2.9176333333333334e-05,
      "loss": 0.0359,
      "step": 12474
    },
    {
      "epoch": 23.01662049861496,
      "grad_norm": 0.5037769675254822,
      "learning_rate": 2.9176000000000003e-05,
      "loss": 0.0771,
      "step": 12475
    },
    {
      "epoch": 23.018467220683288,
      "grad_norm": 1.4689644575119019,
      "learning_rate": 2.9175666666666666e-05,
      "loss": 0.0912,
      "step": 12476
    },
    {
      "epoch": 23.020313942751617,
      "grad_norm": 0.408502995967865,
      "learning_rate": 2.9175333333333335e-05,
      "loss": 0.0248,
      "step": 12477
    },
    {
      "epoch": 23.022160664819946,
      "grad_norm": 0.375881165266037,
      "learning_rate": 2.9175e-05,
      "loss": 0.0332,
      "step": 12478
    },
    {
      "epoch": 23.02400738688827,
      "grad_norm": 0.19678357243537903,
      "learning_rate": 2.9174666666666667e-05,
      "loss": 0.0225,
      "step": 12479
    },
    {
      "epoch": 23.0258541089566,
      "grad_norm": 0.2557147741317749,
      "learning_rate": 2.9174333333333333e-05,
      "loss": 0.0119,
      "step": 12480
    },
    {
      "epoch": 23.02770083102493,
      "grad_norm": 1.2409764528274536,
      "learning_rate": 2.9174000000000002e-05,
      "loss": 0.058,
      "step": 12481
    },
    {
      "epoch": 23.02954755309326,
      "grad_norm": 0.23163054883480072,
      "learning_rate": 2.9173666666666665e-05,
      "loss": 0.0096,
      "step": 12482
    },
    {
      "epoch": 23.03139427516159,
      "grad_norm": 0.4650649130344391,
      "learning_rate": 2.9173333333333334e-05,
      "loss": 0.0358,
      "step": 12483
    },
    {
      "epoch": 23.033240997229917,
      "grad_norm": 0.16126181185245514,
      "learning_rate": 2.9173000000000003e-05,
      "loss": 0.0067,
      "step": 12484
    },
    {
      "epoch": 23.035087719298247,
      "grad_norm": 0.20258918404579163,
      "learning_rate": 2.9172666666666666e-05,
      "loss": 0.0099,
      "step": 12485
    },
    {
      "epoch": 23.036934441366576,
      "grad_norm": 0.23453587293624878,
      "learning_rate": 2.9172333333333335e-05,
      "loss": 0.0086,
      "step": 12486
    },
    {
      "epoch": 23.0387811634349,
      "grad_norm": 0.33570969104766846,
      "learning_rate": 2.9172e-05,
      "loss": 0.0076,
      "step": 12487
    },
    {
      "epoch": 23.04062788550323,
      "grad_norm": 0.20703759789466858,
      "learning_rate": 2.9171666666666667e-05,
      "loss": 0.005,
      "step": 12488
    },
    {
      "epoch": 23.04247460757156,
      "grad_norm": 0.29281994700431824,
      "learning_rate": 2.9171333333333333e-05,
      "loss": 0.0093,
      "step": 12489
    },
    {
      "epoch": 23.04432132963989,
      "grad_norm": 0.18278279900550842,
      "learning_rate": 2.9171000000000002e-05,
      "loss": 0.0055,
      "step": 12490
    },
    {
      "epoch": 23.046168051708218,
      "grad_norm": 0.15609000623226166,
      "learning_rate": 2.9170666666666665e-05,
      "loss": 0.0033,
      "step": 12491
    },
    {
      "epoch": 23.048014773776547,
      "grad_norm": 0.27524250745773315,
      "learning_rate": 2.9170333333333334e-05,
      "loss": 0.0065,
      "step": 12492
    },
    {
      "epoch": 23.049861495844876,
      "grad_norm": 0.11792323738336563,
      "learning_rate": 2.9170000000000004e-05,
      "loss": 0.0031,
      "step": 12493
    },
    {
      "epoch": 23.051708217913205,
      "grad_norm": 0.6204906702041626,
      "learning_rate": 2.9169666666666666e-05,
      "loss": 0.0385,
      "step": 12494
    },
    {
      "epoch": 23.053554939981534,
      "grad_norm": 0.13774995505809784,
      "learning_rate": 2.9169333333333335e-05,
      "loss": 0.0047,
      "step": 12495
    },
    {
      "epoch": 23.05540166204986,
      "grad_norm": 0.0980542004108429,
      "learning_rate": 2.9169e-05,
      "loss": 0.0032,
      "step": 12496
    },
    {
      "epoch": 23.05724838411819,
      "grad_norm": 0.5431872010231018,
      "learning_rate": 2.9168666666666667e-05,
      "loss": 0.0131,
      "step": 12497
    },
    {
      "epoch": 23.05909510618652,
      "grad_norm": 0.19743503630161285,
      "learning_rate": 2.9168333333333333e-05,
      "loss": 0.0043,
      "step": 12498
    },
    {
      "epoch": 23.060941828254848,
      "grad_norm": 0.2094327211380005,
      "learning_rate": 2.9168e-05,
      "loss": 0.0029,
      "step": 12499
    },
    {
      "epoch": 23.062788550323177,
      "grad_norm": 0.2092137336730957,
      "learning_rate": 2.916766666666667e-05,
      "loss": 0.0048,
      "step": 12500
    },
    {
      "epoch": 23.064635272391506,
      "grad_norm": 0.1424982249736786,
      "learning_rate": 2.9167333333333334e-05,
      "loss": 0.0054,
      "step": 12501
    },
    {
      "epoch": 23.066481994459835,
      "grad_norm": 0.20085524022579193,
      "learning_rate": 2.9167e-05,
      "loss": 0.0033,
      "step": 12502
    },
    {
      "epoch": 23.068328716528164,
      "grad_norm": 0.24300634860992432,
      "learning_rate": 2.9166666666666666e-05,
      "loss": 0.0124,
      "step": 12503
    },
    {
      "epoch": 23.07017543859649,
      "grad_norm": 0.24897274374961853,
      "learning_rate": 2.9166333333333336e-05,
      "loss": 0.0106,
      "step": 12504
    },
    {
      "epoch": 23.07202216066482,
      "grad_norm": 0.7429565787315369,
      "learning_rate": 2.9165999999999998e-05,
      "loss": 0.0112,
      "step": 12505
    },
    {
      "epoch": 23.073868882733148,
      "grad_norm": 0.12227975577116013,
      "learning_rate": 2.9165666666666668e-05,
      "loss": 0.0058,
      "step": 12506
    },
    {
      "epoch": 23.075715604801477,
      "grad_norm": 0.21596293151378632,
      "learning_rate": 2.9165333333333334e-05,
      "loss": 0.0071,
      "step": 12507
    },
    {
      "epoch": 23.077562326869806,
      "grad_norm": 0.14186695218086243,
      "learning_rate": 2.9165e-05,
      "loss": 0.0026,
      "step": 12508
    },
    {
      "epoch": 23.079409048938135,
      "grad_norm": 0.307436466217041,
      "learning_rate": 2.916466666666667e-05,
      "loss": 0.0049,
      "step": 12509
    },
    {
      "epoch": 23.081255771006465,
      "grad_norm": 0.42267316579818726,
      "learning_rate": 2.9164333333333335e-05,
      "loss": 0.0093,
      "step": 12510
    },
    {
      "epoch": 23.083102493074794,
      "grad_norm": 0.1583794504404068,
      "learning_rate": 2.9164e-05,
      "loss": 0.0048,
      "step": 12511
    },
    {
      "epoch": 23.08494921514312,
      "grad_norm": 0.30443325638771057,
      "learning_rate": 2.9163666666666667e-05,
      "loss": 0.0095,
      "step": 12512
    },
    {
      "epoch": 23.08679593721145,
      "grad_norm": 0.276859849691391,
      "learning_rate": 2.9163333333333336e-05,
      "loss": 0.0065,
      "step": 12513
    },
    {
      "epoch": 23.088642659279778,
      "grad_norm": 0.49172136187553406,
      "learning_rate": 2.9163e-05,
      "loss": 0.0111,
      "step": 12514
    },
    {
      "epoch": 23.090489381348107,
      "grad_norm": 0.47428908944129944,
      "learning_rate": 2.9162666666666668e-05,
      "loss": 0.0068,
      "step": 12515
    },
    {
      "epoch": 23.092336103416436,
      "grad_norm": 0.16608187556266785,
      "learning_rate": 2.9162333333333334e-05,
      "loss": 0.0046,
      "step": 12516
    },
    {
      "epoch": 23.094182825484765,
      "grad_norm": 0.6573468446731567,
      "learning_rate": 2.9162e-05,
      "loss": 0.1058,
      "step": 12517
    },
    {
      "epoch": 23.096029547553094,
      "grad_norm": 0.545746386051178,
      "learning_rate": 2.916166666666667e-05,
      "loss": 0.0954,
      "step": 12518
    },
    {
      "epoch": 23.097876269621423,
      "grad_norm": 0.5941588878631592,
      "learning_rate": 2.9161333333333335e-05,
      "loss": 0.1491,
      "step": 12519
    },
    {
      "epoch": 23.099722991689752,
      "grad_norm": 0.4453190863132477,
      "learning_rate": 2.9161e-05,
      "loss": 0.0645,
      "step": 12520
    },
    {
      "epoch": 23.101569713758078,
      "grad_norm": 0.3122742176055908,
      "learning_rate": 2.9160666666666667e-05,
      "loss": 0.0887,
      "step": 12521
    },
    {
      "epoch": 23.103416435826407,
      "grad_norm": 0.9442075490951538,
      "learning_rate": 2.9160333333333336e-05,
      "loss": 0.0764,
      "step": 12522
    },
    {
      "epoch": 23.105263157894736,
      "grad_norm": 0.2535693049430847,
      "learning_rate": 2.916e-05,
      "loss": 0.0327,
      "step": 12523
    },
    {
      "epoch": 23.107109879963065,
      "grad_norm": 0.45661574602127075,
      "learning_rate": 2.9159666666666668e-05,
      "loss": 0.0532,
      "step": 12524
    },
    {
      "epoch": 23.108956602031395,
      "grad_norm": 0.6820226311683655,
      "learning_rate": 2.9159333333333337e-05,
      "loss": 0.0371,
      "step": 12525
    },
    {
      "epoch": 23.110803324099724,
      "grad_norm": 0.4742131233215332,
      "learning_rate": 2.9159e-05,
      "loss": 0.0313,
      "step": 12526
    },
    {
      "epoch": 23.112650046168053,
      "grad_norm": 0.8153624534606934,
      "learning_rate": 2.915866666666667e-05,
      "loss": 0.0468,
      "step": 12527
    },
    {
      "epoch": 23.114496768236382,
      "grad_norm": 0.3805418908596039,
      "learning_rate": 2.9158333333333335e-05,
      "loss": 0.0377,
      "step": 12528
    },
    {
      "epoch": 23.116343490304708,
      "grad_norm": 0.30506157875061035,
      "learning_rate": 2.9158e-05,
      "loss": 0.0674,
      "step": 12529
    },
    {
      "epoch": 23.118190212373037,
      "grad_norm": 0.7386313676834106,
      "learning_rate": 2.9157666666666667e-05,
      "loss": 0.0401,
      "step": 12530
    },
    {
      "epoch": 23.120036934441366,
      "grad_norm": 0.22930021584033966,
      "learning_rate": 2.9157333333333333e-05,
      "loss": 0.0078,
      "step": 12531
    },
    {
      "epoch": 23.121883656509695,
      "grad_norm": 0.22462855279445648,
      "learning_rate": 2.9157e-05,
      "loss": 0.0218,
      "step": 12532
    },
    {
      "epoch": 23.123730378578024,
      "grad_norm": 0.11458956450223923,
      "learning_rate": 2.9156666666666668e-05,
      "loss": 0.0035,
      "step": 12533
    },
    {
      "epoch": 23.125577100646353,
      "grad_norm": 0.18596674501895905,
      "learning_rate": 2.9156333333333334e-05,
      "loss": 0.0059,
      "step": 12534
    },
    {
      "epoch": 23.127423822714682,
      "grad_norm": 0.2123582661151886,
      "learning_rate": 2.9156e-05,
      "loss": 0.0067,
      "step": 12535
    },
    {
      "epoch": 23.12927054478301,
      "grad_norm": 0.2569544315338135,
      "learning_rate": 2.915566666666667e-05,
      "loss": 0.0405,
      "step": 12536
    },
    {
      "epoch": 23.131117266851337,
      "grad_norm": 0.22042743861675262,
      "learning_rate": 2.9155333333333332e-05,
      "loss": 0.0086,
      "step": 12537
    },
    {
      "epoch": 23.132963988919666,
      "grad_norm": 0.2655360698699951,
      "learning_rate": 2.9155e-05,
      "loss": 0.006,
      "step": 12538
    },
    {
      "epoch": 23.134810710987995,
      "grad_norm": 0.24769195914268494,
      "learning_rate": 2.9154666666666667e-05,
      "loss": 0.008,
      "step": 12539
    },
    {
      "epoch": 23.136657433056325,
      "grad_norm": 0.25198256969451904,
      "learning_rate": 2.9154333333333333e-05,
      "loss": 0.0057,
      "step": 12540
    },
    {
      "epoch": 23.138504155124654,
      "grad_norm": 0.19294026494026184,
      "learning_rate": 2.9154e-05,
      "loss": 0.0054,
      "step": 12541
    },
    {
      "epoch": 23.140350877192983,
      "grad_norm": 0.2432713508605957,
      "learning_rate": 2.915366666666667e-05,
      "loss": 0.008,
      "step": 12542
    },
    {
      "epoch": 23.142197599261312,
      "grad_norm": 0.351470410823822,
      "learning_rate": 2.9153333333333334e-05,
      "loss": 0.0145,
      "step": 12543
    },
    {
      "epoch": 23.14404432132964,
      "grad_norm": 0.3115994930267334,
      "learning_rate": 2.9153e-05,
      "loss": 0.0134,
      "step": 12544
    },
    {
      "epoch": 23.14589104339797,
      "grad_norm": 0.39242538809776306,
      "learning_rate": 2.915266666666667e-05,
      "loss": 0.0063,
      "step": 12545
    },
    {
      "epoch": 23.147737765466296,
      "grad_norm": 0.11153402179479599,
      "learning_rate": 2.9152333333333332e-05,
      "loss": 0.0024,
      "step": 12546
    },
    {
      "epoch": 23.149584487534625,
      "grad_norm": 0.30442219972610474,
      "learning_rate": 2.9152e-05,
      "loss": 0.0049,
      "step": 12547
    },
    {
      "epoch": 23.151431209602954,
      "grad_norm": 0.2781001031398773,
      "learning_rate": 2.9151666666666667e-05,
      "loss": 0.0356,
      "step": 12548
    },
    {
      "epoch": 23.153277931671283,
      "grad_norm": 0.20190612971782684,
      "learning_rate": 2.9151333333333333e-05,
      "loss": 0.0047,
      "step": 12549
    },
    {
      "epoch": 23.155124653739612,
      "grad_norm": 0.3549395799636841,
      "learning_rate": 2.9151000000000003e-05,
      "loss": 0.0073,
      "step": 12550
    },
    {
      "epoch": 23.15697137580794,
      "grad_norm": 0.39812251925468445,
      "learning_rate": 2.915066666666667e-05,
      "loss": 0.0173,
      "step": 12551
    },
    {
      "epoch": 23.15881809787627,
      "grad_norm": 0.22466783225536346,
      "learning_rate": 2.9150333333333334e-05,
      "loss": 0.0056,
      "step": 12552
    },
    {
      "epoch": 23.1606648199446,
      "grad_norm": 0.06262242048978806,
      "learning_rate": 2.915e-05,
      "loss": 0.0017,
      "step": 12553
    },
    {
      "epoch": 23.162511542012926,
      "grad_norm": 0.5167449116706848,
      "learning_rate": 2.914966666666667e-05,
      "loss": 0.0075,
      "step": 12554
    },
    {
      "epoch": 23.164358264081255,
      "grad_norm": 0.1647091954946518,
      "learning_rate": 2.9149333333333332e-05,
      "loss": 0.0044,
      "step": 12555
    },
    {
      "epoch": 23.166204986149584,
      "grad_norm": 0.13090741634368896,
      "learning_rate": 2.9149e-05,
      "loss": 0.0031,
      "step": 12556
    },
    {
      "epoch": 23.168051708217913,
      "grad_norm": 0.3570130169391632,
      "learning_rate": 2.9148666666666668e-05,
      "loss": 0.0042,
      "step": 12557
    },
    {
      "epoch": 23.169898430286242,
      "grad_norm": 0.3519172966480255,
      "learning_rate": 2.9148333333333333e-05,
      "loss": 0.0079,
      "step": 12558
    },
    {
      "epoch": 23.17174515235457,
      "grad_norm": 0.22465786337852478,
      "learning_rate": 2.9148000000000003e-05,
      "loss": 0.0029,
      "step": 12559
    },
    {
      "epoch": 23.1735918744229,
      "grad_norm": 0.10870680212974548,
      "learning_rate": 2.9147666666666665e-05,
      "loss": 0.0026,
      "step": 12560
    },
    {
      "epoch": 23.17543859649123,
      "grad_norm": 0.19479309022426605,
      "learning_rate": 2.9147333333333335e-05,
      "loss": 0.006,
      "step": 12561
    },
    {
      "epoch": 23.177285318559555,
      "grad_norm": 0.26224997639656067,
      "learning_rate": 2.9147e-05,
      "loss": 0.0031,
      "step": 12562
    },
    {
      "epoch": 23.179132040627884,
      "grad_norm": 0.22105391323566437,
      "learning_rate": 2.9146666666666667e-05,
      "loss": 0.0039,
      "step": 12563
    },
    {
      "epoch": 23.180978762696213,
      "grad_norm": 0.3528079390525818,
      "learning_rate": 2.9146333333333332e-05,
      "loss": 0.006,
      "step": 12564
    },
    {
      "epoch": 23.182825484764543,
      "grad_norm": 0.19657863676548004,
      "learning_rate": 2.9146000000000002e-05,
      "loss": 0.0039,
      "step": 12565
    },
    {
      "epoch": 23.18467220683287,
      "grad_norm": 0.14275223016738892,
      "learning_rate": 2.9145666666666664e-05,
      "loss": 0.0017,
      "step": 12566
    },
    {
      "epoch": 23.1865189289012,
      "grad_norm": 0.6491601467132568,
      "learning_rate": 2.9145333333333334e-05,
      "loss": 0.1199,
      "step": 12567
    },
    {
      "epoch": 23.18836565096953,
      "grad_norm": 0.7567947506904602,
      "learning_rate": 2.9145000000000003e-05,
      "loss": 0.1151,
      "step": 12568
    },
    {
      "epoch": 23.19021237303786,
      "grad_norm": 0.3431910574436188,
      "learning_rate": 2.9144666666666666e-05,
      "loss": 0.068,
      "step": 12569
    },
    {
      "epoch": 23.19205909510619,
      "grad_norm": 0.40085679292678833,
      "learning_rate": 2.9144333333333335e-05,
      "loss": 0.0507,
      "step": 12570
    },
    {
      "epoch": 23.193905817174514,
      "grad_norm": 0.5359059572219849,
      "learning_rate": 2.9144e-05,
      "loss": 0.0545,
      "step": 12571
    },
    {
      "epoch": 23.195752539242843,
      "grad_norm": 0.7867258787155151,
      "learning_rate": 2.9143666666666667e-05,
      "loss": 0.1062,
      "step": 12572
    },
    {
      "epoch": 23.197599261311172,
      "grad_norm": 0.3034769892692566,
      "learning_rate": 2.9143333333333333e-05,
      "loss": 0.0363,
      "step": 12573
    },
    {
      "epoch": 23.1994459833795,
      "grad_norm": 1.3643563985824585,
      "learning_rate": 2.9143000000000002e-05,
      "loss": 0.0534,
      "step": 12574
    },
    {
      "epoch": 23.20129270544783,
      "grad_norm": 0.30873343348503113,
      "learning_rate": 2.9142666666666668e-05,
      "loss": 0.0444,
      "step": 12575
    },
    {
      "epoch": 23.20313942751616,
      "grad_norm": 0.3701000213623047,
      "learning_rate": 2.9142333333333334e-05,
      "loss": 0.0196,
      "step": 12576
    },
    {
      "epoch": 23.20498614958449,
      "grad_norm": 0.29086777567863464,
      "learning_rate": 2.9142000000000003e-05,
      "loss": 0.0192,
      "step": 12577
    },
    {
      "epoch": 23.206832871652818,
      "grad_norm": 0.3202081024646759,
      "learning_rate": 2.9141666666666666e-05,
      "loss": 0.0642,
      "step": 12578
    },
    {
      "epoch": 23.208679593721143,
      "grad_norm": 0.37413352727890015,
      "learning_rate": 2.9141333333333335e-05,
      "loss": 0.0289,
      "step": 12579
    },
    {
      "epoch": 23.210526315789473,
      "grad_norm": 0.23711729049682617,
      "learning_rate": 2.9141e-05,
      "loss": 0.0334,
      "step": 12580
    },
    {
      "epoch": 23.2123730378578,
      "grad_norm": 0.2957731783390045,
      "learning_rate": 2.9140666666666667e-05,
      "loss": 0.0108,
      "step": 12581
    },
    {
      "epoch": 23.21421975992613,
      "grad_norm": 1.019913911819458,
      "learning_rate": 2.9140333333333333e-05,
      "loss": 0.027,
      "step": 12582
    },
    {
      "epoch": 23.21606648199446,
      "grad_norm": 0.29194220900535583,
      "learning_rate": 2.9140000000000002e-05,
      "loss": 0.0112,
      "step": 12583
    },
    {
      "epoch": 23.21791320406279,
      "grad_norm": 0.33904239535331726,
      "learning_rate": 2.9139666666666668e-05,
      "loss": 0.0122,
      "step": 12584
    },
    {
      "epoch": 23.21975992613112,
      "grad_norm": 0.23502930998802185,
      "learning_rate": 2.9139333333333334e-05,
      "loss": 0.0105,
      "step": 12585
    },
    {
      "epoch": 23.221606648199447,
      "grad_norm": 0.08390850573778152,
      "learning_rate": 2.9139000000000003e-05,
      "loss": 0.0032,
      "step": 12586
    },
    {
      "epoch": 23.223453370267773,
      "grad_norm": 0.3864550292491913,
      "learning_rate": 2.9138666666666666e-05,
      "loss": 0.0066,
      "step": 12587
    },
    {
      "epoch": 23.225300092336102,
      "grad_norm": 0.37667298316955566,
      "learning_rate": 2.9138333333333335e-05,
      "loss": 0.0087,
      "step": 12588
    },
    {
      "epoch": 23.22714681440443,
      "grad_norm": 0.3234068751335144,
      "learning_rate": 2.9137999999999998e-05,
      "loss": 0.0078,
      "step": 12589
    },
    {
      "epoch": 23.22899353647276,
      "grad_norm": 0.145706906914711,
      "learning_rate": 2.9137666666666667e-05,
      "loss": 0.0045,
      "step": 12590
    },
    {
      "epoch": 23.23084025854109,
      "grad_norm": 0.1408831626176834,
      "learning_rate": 2.9137333333333333e-05,
      "loss": 0.0049,
      "step": 12591
    },
    {
      "epoch": 23.23268698060942,
      "grad_norm": 0.36585304141044617,
      "learning_rate": 2.9137e-05,
      "loss": 0.0084,
      "step": 12592
    },
    {
      "epoch": 23.234533702677748,
      "grad_norm": 0.1734265834093094,
      "learning_rate": 2.913666666666667e-05,
      "loss": 0.0043,
      "step": 12593
    },
    {
      "epoch": 23.236380424746077,
      "grad_norm": 0.24555036425590515,
      "learning_rate": 2.9136333333333334e-05,
      "loss": 0.0069,
      "step": 12594
    },
    {
      "epoch": 23.238227146814406,
      "grad_norm": 0.1656181663274765,
      "learning_rate": 2.9136e-05,
      "loss": 0.0042,
      "step": 12595
    },
    {
      "epoch": 23.24007386888273,
      "grad_norm": 0.27070939540863037,
      "learning_rate": 2.9135666666666666e-05,
      "loss": 0.0057,
      "step": 12596
    },
    {
      "epoch": 23.24192059095106,
      "grad_norm": 0.27400118112564087,
      "learning_rate": 2.9135333333333335e-05,
      "loss": 0.008,
      "step": 12597
    },
    {
      "epoch": 23.24376731301939,
      "grad_norm": 0.30686214566230774,
      "learning_rate": 2.9134999999999998e-05,
      "loss": 0.0042,
      "step": 12598
    },
    {
      "epoch": 23.24561403508772,
      "grad_norm": 0.46688786149024963,
      "learning_rate": 2.9134666666666667e-05,
      "loss": 0.0075,
      "step": 12599
    },
    {
      "epoch": 23.24746075715605,
      "grad_norm": 0.30184173583984375,
      "learning_rate": 2.9134333333333337e-05,
      "loss": 0.0058,
      "step": 12600
    },
    {
      "epoch": 23.249307479224377,
      "grad_norm": 0.15009987354278564,
      "learning_rate": 2.9134e-05,
      "loss": 0.0033,
      "step": 12601
    },
    {
      "epoch": 23.251154201292707,
      "grad_norm": 0.48980292677879333,
      "learning_rate": 2.913366666666667e-05,
      "loss": 0.0133,
      "step": 12602
    },
    {
      "epoch": 23.253000923361036,
      "grad_norm": 0.1655283272266388,
      "learning_rate": 2.9133333333333334e-05,
      "loss": 0.0046,
      "step": 12603
    },
    {
      "epoch": 23.25484764542936,
      "grad_norm": 0.665810763835907,
      "learning_rate": 2.9133e-05,
      "loss": 0.0266,
      "step": 12604
    },
    {
      "epoch": 23.25669436749769,
      "grad_norm": 1.3942642211914062,
      "learning_rate": 2.9132666666666666e-05,
      "loss": 0.0121,
      "step": 12605
    },
    {
      "epoch": 23.25854108956602,
      "grad_norm": 0.11559968441724777,
      "learning_rate": 2.9132333333333336e-05,
      "loss": 0.0024,
      "step": 12606
    },
    {
      "epoch": 23.26038781163435,
      "grad_norm": 0.14520831406116486,
      "learning_rate": 2.9131999999999998e-05,
      "loss": 0.0042,
      "step": 12607
    },
    {
      "epoch": 23.262234533702678,
      "grad_norm": 0.3412821590900421,
      "learning_rate": 2.9131666666666668e-05,
      "loss": 0.0064,
      "step": 12608
    },
    {
      "epoch": 23.264081255771007,
      "grad_norm": 0.28790393471717834,
      "learning_rate": 2.9131333333333337e-05,
      "loss": 0.0051,
      "step": 12609
    },
    {
      "epoch": 23.265927977839336,
      "grad_norm": 0.231611967086792,
      "learning_rate": 2.9131e-05,
      "loss": 0.0083,
      "step": 12610
    },
    {
      "epoch": 23.267774699907665,
      "grad_norm": 0.28216591477394104,
      "learning_rate": 2.913066666666667e-05,
      "loss": 0.0062,
      "step": 12611
    },
    {
      "epoch": 23.26962142197599,
      "grad_norm": 0.24869054555892944,
      "learning_rate": 2.9130333333333335e-05,
      "loss": 0.0048,
      "step": 12612
    },
    {
      "epoch": 23.27146814404432,
      "grad_norm": 0.27544718980789185,
      "learning_rate": 2.913e-05,
      "loss": 0.0252,
      "step": 12613
    },
    {
      "epoch": 23.27331486611265,
      "grad_norm": 0.6152492165565491,
      "learning_rate": 2.9129666666666667e-05,
      "loss": 0.0046,
      "step": 12614
    },
    {
      "epoch": 23.27516158818098,
      "grad_norm": 0.26752281188964844,
      "learning_rate": 2.9129333333333336e-05,
      "loss": 0.0058,
      "step": 12615
    },
    {
      "epoch": 23.277008310249307,
      "grad_norm": 0.31082385778427124,
      "learning_rate": 2.9129e-05,
      "loss": 0.0076,
      "step": 12616
    },
    {
      "epoch": 23.278855032317637,
      "grad_norm": 0.5576523542404175,
      "learning_rate": 2.9128666666666668e-05,
      "loss": 0.1092,
      "step": 12617
    },
    {
      "epoch": 23.280701754385966,
      "grad_norm": 0.5595287084579468,
      "learning_rate": 2.9128333333333337e-05,
      "loss": 0.1438,
      "step": 12618
    },
    {
      "epoch": 23.282548476454295,
      "grad_norm": 0.4700721800327301,
      "learning_rate": 2.9128e-05,
      "loss": 0.0899,
      "step": 12619
    },
    {
      "epoch": 23.284395198522624,
      "grad_norm": 0.5074760317802429,
      "learning_rate": 2.912766666666667e-05,
      "loss": 0.1053,
      "step": 12620
    },
    {
      "epoch": 23.28624192059095,
      "grad_norm": 0.3127371668815613,
      "learning_rate": 2.912733333333333e-05,
      "loss": 0.0583,
      "step": 12621
    },
    {
      "epoch": 23.28808864265928,
      "grad_norm": 0.510924756526947,
      "learning_rate": 2.9127e-05,
      "loss": 0.101,
      "step": 12622
    },
    {
      "epoch": 23.289935364727608,
      "grad_norm": 0.46187469363212585,
      "learning_rate": 2.9126666666666667e-05,
      "loss": 0.0423,
      "step": 12623
    },
    {
      "epoch": 23.291782086795937,
      "grad_norm": 0.37520408630371094,
      "learning_rate": 2.9126333333333333e-05,
      "loss": 0.073,
      "step": 12624
    },
    {
      "epoch": 23.293628808864266,
      "grad_norm": 0.3077435791492462,
      "learning_rate": 2.9126000000000002e-05,
      "loss": 0.0263,
      "step": 12625
    },
    {
      "epoch": 23.295475530932595,
      "grad_norm": 0.32046741247177124,
      "learning_rate": 2.9125666666666668e-05,
      "loss": 0.058,
      "step": 12626
    },
    {
      "epoch": 23.297322253000925,
      "grad_norm": 0.5632917284965515,
      "learning_rate": 2.9125333333333334e-05,
      "loss": 0.0277,
      "step": 12627
    },
    {
      "epoch": 23.299168975069254,
      "grad_norm": 0.7035937905311584,
      "learning_rate": 2.9125e-05,
      "loss": 0.0307,
      "step": 12628
    },
    {
      "epoch": 23.30101569713758,
      "grad_norm": 0.794179379940033,
      "learning_rate": 2.912466666666667e-05,
      "loss": 0.022,
      "step": 12629
    },
    {
      "epoch": 23.30286241920591,
      "grad_norm": 0.3423013389110565,
      "learning_rate": 2.912433333333333e-05,
      "loss": 0.0282,
      "step": 12630
    },
    {
      "epoch": 23.304709141274238,
      "grad_norm": 0.2058839648962021,
      "learning_rate": 2.9124e-05,
      "loss": 0.0126,
      "step": 12631
    },
    {
      "epoch": 23.306555863342567,
      "grad_norm": 0.270408570766449,
      "learning_rate": 2.9123666666666667e-05,
      "loss": 0.0207,
      "step": 12632
    },
    {
      "epoch": 23.308402585410896,
      "grad_norm": 0.3184267580509186,
      "learning_rate": 2.9123333333333333e-05,
      "loss": 0.0294,
      "step": 12633
    },
    {
      "epoch": 23.310249307479225,
      "grad_norm": 0.41973748803138733,
      "learning_rate": 2.9123000000000002e-05,
      "loss": 0.0124,
      "step": 12634
    },
    {
      "epoch": 23.312096029547554,
      "grad_norm": 0.1604977697134018,
      "learning_rate": 2.9122666666666668e-05,
      "loss": 0.0039,
      "step": 12635
    },
    {
      "epoch": 23.313942751615883,
      "grad_norm": 0.5372335314750671,
      "learning_rate": 2.9122333333333334e-05,
      "loss": 0.0087,
      "step": 12636
    },
    {
      "epoch": 23.31578947368421,
      "grad_norm": 0.17940805852413177,
      "learning_rate": 2.9122e-05,
      "loss": 0.0081,
      "step": 12637
    },
    {
      "epoch": 23.317636195752538,
      "grad_norm": 0.187593013048172,
      "learning_rate": 2.912166666666667e-05,
      "loss": 0.0061,
      "step": 12638
    },
    {
      "epoch": 23.319482917820867,
      "grad_norm": 0.27255645394325256,
      "learning_rate": 2.9121333333333332e-05,
      "loss": 0.012,
      "step": 12639
    },
    {
      "epoch": 23.321329639889196,
      "grad_norm": 0.5173263549804688,
      "learning_rate": 2.9121e-05,
      "loss": 0.0131,
      "step": 12640
    },
    {
      "epoch": 23.323176361957525,
      "grad_norm": 0.3303021788597107,
      "learning_rate": 2.9120666666666667e-05,
      "loss": 0.013,
      "step": 12641
    },
    {
      "epoch": 23.325023084025855,
      "grad_norm": 0.14484189450740814,
      "learning_rate": 2.9120333333333333e-05,
      "loss": 0.0036,
      "step": 12642
    },
    {
      "epoch": 23.326869806094184,
      "grad_norm": 0.19277650117874146,
      "learning_rate": 2.9120000000000002e-05,
      "loss": 0.0054,
      "step": 12643
    },
    {
      "epoch": 23.328716528162513,
      "grad_norm": 0.39073553681373596,
      "learning_rate": 2.911966666666667e-05,
      "loss": 0.0081,
      "step": 12644
    },
    {
      "epoch": 23.330563250230842,
      "grad_norm": 0.23671211302280426,
      "learning_rate": 2.9119333333333334e-05,
      "loss": 0.0036,
      "step": 12645
    },
    {
      "epoch": 23.332409972299168,
      "grad_norm": 0.13604755699634552,
      "learning_rate": 2.9119e-05,
      "loss": 0.004,
      "step": 12646
    },
    {
      "epoch": 23.334256694367497,
      "grad_norm": 0.12806524336338043,
      "learning_rate": 2.911866666666667e-05,
      "loss": 0.0036,
      "step": 12647
    },
    {
      "epoch": 23.336103416435826,
      "grad_norm": 0.26302891969680786,
      "learning_rate": 2.9118333333333332e-05,
      "loss": 0.0084,
      "step": 12648
    },
    {
      "epoch": 23.337950138504155,
      "grad_norm": 0.32138726115226746,
      "learning_rate": 2.9118e-05,
      "loss": 0.0082,
      "step": 12649
    },
    {
      "epoch": 23.339796860572484,
      "grad_norm": 0.2769556939601898,
      "learning_rate": 2.9117666666666667e-05,
      "loss": 0.006,
      "step": 12650
    },
    {
      "epoch": 23.341643582640813,
      "grad_norm": 0.22334329783916473,
      "learning_rate": 2.9117333333333333e-05,
      "loss": 0.007,
      "step": 12651
    },
    {
      "epoch": 23.343490304709142,
      "grad_norm": 0.5758256316184998,
      "learning_rate": 2.9117000000000003e-05,
      "loss": 0.0087,
      "step": 12652
    },
    {
      "epoch": 23.34533702677747,
      "grad_norm": 1.134762167930603,
      "learning_rate": 2.9116666666666665e-05,
      "loss": 0.0071,
      "step": 12653
    },
    {
      "epoch": 23.347183748845797,
      "grad_norm": 0.18429134786128998,
      "learning_rate": 2.9116333333333334e-05,
      "loss": 0.0043,
      "step": 12654
    },
    {
      "epoch": 23.349030470914126,
      "grad_norm": 0.13381122052669525,
      "learning_rate": 2.9116e-05,
      "loss": 0.0041,
      "step": 12655
    },
    {
      "epoch": 23.350877192982455,
      "grad_norm": 0.37146177887916565,
      "learning_rate": 2.9115666666666666e-05,
      "loss": 0.0133,
      "step": 12656
    },
    {
      "epoch": 23.352723915050785,
      "grad_norm": 0.1729261875152588,
      "learning_rate": 2.9115333333333332e-05,
      "loss": 0.0048,
      "step": 12657
    },
    {
      "epoch": 23.354570637119114,
      "grad_norm": 0.21925175189971924,
      "learning_rate": 2.9115e-05,
      "loss": 0.0078,
      "step": 12658
    },
    {
      "epoch": 23.356417359187443,
      "grad_norm": 0.3351660370826721,
      "learning_rate": 2.9114666666666668e-05,
      "loss": 0.0081,
      "step": 12659
    },
    {
      "epoch": 23.358264081255772,
      "grad_norm": 0.150557741522789,
      "learning_rate": 2.9114333333333333e-05,
      "loss": 0.005,
      "step": 12660
    },
    {
      "epoch": 23.3601108033241,
      "grad_norm": 0.2575150430202484,
      "learning_rate": 2.9114000000000003e-05,
      "loss": 0.009,
      "step": 12661
    },
    {
      "epoch": 23.361957525392427,
      "grad_norm": 0.24913664162158966,
      "learning_rate": 2.9113666666666665e-05,
      "loss": 0.0057,
      "step": 12662
    },
    {
      "epoch": 23.363804247460756,
      "grad_norm": 0.13307268917560577,
      "learning_rate": 2.9113333333333335e-05,
      "loss": 0.0021,
      "step": 12663
    },
    {
      "epoch": 23.365650969529085,
      "grad_norm": 0.47522854804992676,
      "learning_rate": 2.9113e-05,
      "loss": 0.0133,
      "step": 12664
    },
    {
      "epoch": 23.367497691597414,
      "grad_norm": 0.2812003791332245,
      "learning_rate": 2.9112666666666667e-05,
      "loss": 0.0066,
      "step": 12665
    },
    {
      "epoch": 23.369344413665743,
      "grad_norm": 0.33105483651161194,
      "learning_rate": 2.9112333333333332e-05,
      "loss": 0.0053,
      "step": 12666
    },
    {
      "epoch": 23.371191135734072,
      "grad_norm": 0.5941159725189209,
      "learning_rate": 2.9112000000000002e-05,
      "loss": 0.1752,
      "step": 12667
    },
    {
      "epoch": 23.3730378578024,
      "grad_norm": 0.5663772821426392,
      "learning_rate": 2.9111666666666668e-05,
      "loss": 0.1289,
      "step": 12668
    },
    {
      "epoch": 23.37488457987073,
      "grad_norm": 0.4650385081768036,
      "learning_rate": 2.9111333333333334e-05,
      "loss": 0.0951,
      "step": 12669
    },
    {
      "epoch": 23.37673130193906,
      "grad_norm": 0.659968376159668,
      "learning_rate": 2.9111000000000003e-05,
      "loss": 0.0744,
      "step": 12670
    },
    {
      "epoch": 23.378578024007385,
      "grad_norm": 0.44175758957862854,
      "learning_rate": 2.9110666666666666e-05,
      "loss": 0.0556,
      "step": 12671
    },
    {
      "epoch": 23.380424746075715,
      "grad_norm": 0.5107433795928955,
      "learning_rate": 2.9110333333333335e-05,
      "loss": 0.0617,
      "step": 12672
    },
    {
      "epoch": 23.382271468144044,
      "grad_norm": 0.3211396634578705,
      "learning_rate": 2.911e-05,
      "loss": 0.0402,
      "step": 12673
    },
    {
      "epoch": 23.384118190212373,
      "grad_norm": 0.37351423501968384,
      "learning_rate": 2.9109666666666667e-05,
      "loss": 0.0478,
      "step": 12674
    },
    {
      "epoch": 23.385964912280702,
      "grad_norm": 0.2717868685722351,
      "learning_rate": 2.9109333333333336e-05,
      "loss": 0.024,
      "step": 12675
    },
    {
      "epoch": 23.38781163434903,
      "grad_norm": 0.40558385848999023,
      "learning_rate": 2.9109000000000002e-05,
      "loss": 0.0448,
      "step": 12676
    },
    {
      "epoch": 23.38965835641736,
      "grad_norm": 0.4204607605934143,
      "learning_rate": 2.9108666666666668e-05,
      "loss": 0.0393,
      "step": 12677
    },
    {
      "epoch": 23.39150507848569,
      "grad_norm": 0.3483840823173523,
      "learning_rate": 2.9108333333333334e-05,
      "loss": 0.0241,
      "step": 12678
    },
    {
      "epoch": 23.393351800554015,
      "grad_norm": 0.4850672483444214,
      "learning_rate": 2.9108000000000003e-05,
      "loss": 0.0657,
      "step": 12679
    },
    {
      "epoch": 23.395198522622344,
      "grad_norm": 0.31261759996414185,
      "learning_rate": 2.9107666666666666e-05,
      "loss": 0.0128,
      "step": 12680
    },
    {
      "epoch": 23.397045244690673,
      "grad_norm": 0.4061281085014343,
      "learning_rate": 2.9107333333333335e-05,
      "loss": 0.0239,
      "step": 12681
    },
    {
      "epoch": 23.398891966759003,
      "grad_norm": 0.22250206768512726,
      "learning_rate": 2.9106999999999998e-05,
      "loss": 0.0144,
      "step": 12682
    },
    {
      "epoch": 23.40073868882733,
      "grad_norm": 0.2188418060541153,
      "learning_rate": 2.9106666666666667e-05,
      "loss": 0.0074,
      "step": 12683
    },
    {
      "epoch": 23.40258541089566,
      "grad_norm": 0.18531258404254913,
      "learning_rate": 2.9106333333333336e-05,
      "loss": 0.0123,
      "step": 12684
    },
    {
      "epoch": 23.40443213296399,
      "grad_norm": 0.16225002706050873,
      "learning_rate": 2.9106e-05,
      "loss": 0.0046,
      "step": 12685
    },
    {
      "epoch": 23.40627885503232,
      "grad_norm": 0.37271827459335327,
      "learning_rate": 2.9105666666666668e-05,
      "loss": 0.0058,
      "step": 12686
    },
    {
      "epoch": 23.408125577100645,
      "grad_norm": 0.3324749767780304,
      "learning_rate": 2.9105333333333334e-05,
      "loss": 0.0076,
      "step": 12687
    },
    {
      "epoch": 23.409972299168974,
      "grad_norm": 0.07640788704156876,
      "learning_rate": 2.9105e-05,
      "loss": 0.0031,
      "step": 12688
    },
    {
      "epoch": 23.411819021237303,
      "grad_norm": 0.36134073138237,
      "learning_rate": 2.9104666666666666e-05,
      "loss": 0.0119,
      "step": 12689
    },
    {
      "epoch": 23.413665743305632,
      "grad_norm": 0.12606969475746155,
      "learning_rate": 2.9104333333333335e-05,
      "loss": 0.0025,
      "step": 12690
    },
    {
      "epoch": 23.41551246537396,
      "grad_norm": 0.38095346093177795,
      "learning_rate": 2.9103999999999998e-05,
      "loss": 0.0088,
      "step": 12691
    },
    {
      "epoch": 23.41735918744229,
      "grad_norm": 0.18201282620429993,
      "learning_rate": 2.9103666666666667e-05,
      "loss": 0.0078,
      "step": 12692
    },
    {
      "epoch": 23.41920590951062,
      "grad_norm": 0.24027188122272491,
      "learning_rate": 2.9103333333333336e-05,
      "loss": 0.0129,
      "step": 12693
    },
    {
      "epoch": 23.42105263157895,
      "grad_norm": 0.12497299909591675,
      "learning_rate": 2.9103e-05,
      "loss": 0.0044,
      "step": 12694
    },
    {
      "epoch": 23.422899353647278,
      "grad_norm": 0.3064292371273041,
      "learning_rate": 2.910266666666667e-05,
      "loss": 0.003,
      "step": 12695
    },
    {
      "epoch": 23.424746075715603,
      "grad_norm": 0.2261001467704773,
      "learning_rate": 2.9102333333333334e-05,
      "loss": 0.0072,
      "step": 12696
    },
    {
      "epoch": 23.426592797783933,
      "grad_norm": 0.2631829082965851,
      "learning_rate": 2.9102e-05,
      "loss": 0.0062,
      "step": 12697
    },
    {
      "epoch": 23.42843951985226,
      "grad_norm": 0.23572014272212982,
      "learning_rate": 2.9101666666666666e-05,
      "loss": 0.0045,
      "step": 12698
    },
    {
      "epoch": 23.43028624192059,
      "grad_norm": 0.21679142117500305,
      "learning_rate": 2.9101333333333335e-05,
      "loss": 0.0063,
      "step": 12699
    },
    {
      "epoch": 23.43213296398892,
      "grad_norm": 0.16654612123966217,
      "learning_rate": 2.9101e-05,
      "loss": 0.0059,
      "step": 12700
    },
    {
      "epoch": 23.43397968605725,
      "grad_norm": 0.39357325434684753,
      "learning_rate": 2.9100666666666667e-05,
      "loss": 0.0092,
      "step": 12701
    },
    {
      "epoch": 23.43582640812558,
      "grad_norm": 0.2027686983346939,
      "learning_rate": 2.9100333333333337e-05,
      "loss": 0.0027,
      "step": 12702
    },
    {
      "epoch": 23.437673130193907,
      "grad_norm": 0.18773828446865082,
      "learning_rate": 2.91e-05,
      "loss": 0.0043,
      "step": 12703
    },
    {
      "epoch": 23.439519852262233,
      "grad_norm": 0.1085064485669136,
      "learning_rate": 2.909966666666667e-05,
      "loss": 0.0034,
      "step": 12704
    },
    {
      "epoch": 23.441366574330562,
      "grad_norm": 0.24215655028820038,
      "learning_rate": 2.9099333333333334e-05,
      "loss": 0.0073,
      "step": 12705
    },
    {
      "epoch": 23.44321329639889,
      "grad_norm": 0.3163291811943054,
      "learning_rate": 2.9099e-05,
      "loss": 0.0096,
      "step": 12706
    },
    {
      "epoch": 23.44506001846722,
      "grad_norm": 0.1995258331298828,
      "learning_rate": 2.9098666666666666e-05,
      "loss": 0.0059,
      "step": 12707
    },
    {
      "epoch": 23.44690674053555,
      "grad_norm": 0.09423418343067169,
      "learning_rate": 2.9098333333333336e-05,
      "loss": 0.0024,
      "step": 12708
    },
    {
      "epoch": 23.44875346260388,
      "grad_norm": 0.1298067569732666,
      "learning_rate": 2.9098e-05,
      "loss": 0.0034,
      "step": 12709
    },
    {
      "epoch": 23.450600184672208,
      "grad_norm": 0.16497434675693512,
      "learning_rate": 2.9097666666666668e-05,
      "loss": 0.0068,
      "step": 12710
    },
    {
      "epoch": 23.452446906740537,
      "grad_norm": 0.16174855828285217,
      "learning_rate": 2.9097333333333333e-05,
      "loss": 0.0036,
      "step": 12711
    },
    {
      "epoch": 23.454293628808863,
      "grad_norm": 0.08075518906116486,
      "learning_rate": 2.9097e-05,
      "loss": 0.0018,
      "step": 12712
    },
    {
      "epoch": 23.45614035087719,
      "grad_norm": 0.2373785823583603,
      "learning_rate": 2.909666666666667e-05,
      "loss": 0.0108,
      "step": 12713
    },
    {
      "epoch": 23.45798707294552,
      "grad_norm": 0.23240111768245697,
      "learning_rate": 2.909633333333333e-05,
      "loss": 0.0214,
      "step": 12714
    },
    {
      "epoch": 23.45983379501385,
      "grad_norm": 0.19462589919567108,
      "learning_rate": 2.9096e-05,
      "loss": 0.006,
      "step": 12715
    },
    {
      "epoch": 23.46168051708218,
      "grad_norm": 0.2087223082780838,
      "learning_rate": 2.9095666666666667e-05,
      "loss": 0.0174,
      "step": 12716
    },
    {
      "epoch": 23.46352723915051,
      "grad_norm": 0.4997592866420746,
      "learning_rate": 2.9095333333333332e-05,
      "loss": 0.1128,
      "step": 12717
    },
    {
      "epoch": 23.465373961218837,
      "grad_norm": 0.5647820830345154,
      "learning_rate": 2.9095000000000002e-05,
      "loss": 0.1108,
      "step": 12718
    },
    {
      "epoch": 23.467220683287167,
      "grad_norm": 0.5153325796127319,
      "learning_rate": 2.9094666666666668e-05,
      "loss": 0.0906,
      "step": 12719
    },
    {
      "epoch": 23.469067405355496,
      "grad_norm": 0.560015082359314,
      "learning_rate": 2.9094333333333334e-05,
      "loss": 0.1163,
      "step": 12720
    },
    {
      "epoch": 23.47091412742382,
      "grad_norm": 0.5469334721565247,
      "learning_rate": 2.9094e-05,
      "loss": 0.0669,
      "step": 12721
    },
    {
      "epoch": 23.47276084949215,
      "grad_norm": 0.5907200574874878,
      "learning_rate": 2.909366666666667e-05,
      "loss": 0.0516,
      "step": 12722
    },
    {
      "epoch": 23.47460757156048,
      "grad_norm": 0.40143343806266785,
      "learning_rate": 2.909333333333333e-05,
      "loss": 0.0358,
      "step": 12723
    },
    {
      "epoch": 23.47645429362881,
      "grad_norm": 0.37248697876930237,
      "learning_rate": 2.9093e-05,
      "loss": 0.031,
      "step": 12724
    },
    {
      "epoch": 23.478301015697138,
      "grad_norm": 0.6797778010368347,
      "learning_rate": 2.909266666666667e-05,
      "loss": 0.0435,
      "step": 12725
    },
    {
      "epoch": 23.480147737765467,
      "grad_norm": 1.5801701545715332,
      "learning_rate": 2.9092333333333333e-05,
      "loss": 0.0214,
      "step": 12726
    },
    {
      "epoch": 23.481994459833796,
      "grad_norm": 0.26523342728614807,
      "learning_rate": 2.9092000000000002e-05,
      "loss": 0.029,
      "step": 12727
    },
    {
      "epoch": 23.483841181902125,
      "grad_norm": 0.2564677298069,
      "learning_rate": 2.9091666666666668e-05,
      "loss": 0.0188,
      "step": 12728
    },
    {
      "epoch": 23.48568790397045,
      "grad_norm": 0.31995120644569397,
      "learning_rate": 2.9091333333333334e-05,
      "loss": 0.0285,
      "step": 12729
    },
    {
      "epoch": 23.48753462603878,
      "grad_norm": 0.2464805245399475,
      "learning_rate": 2.9091e-05,
      "loss": 0.0092,
      "step": 12730
    },
    {
      "epoch": 23.48938134810711,
      "grad_norm": 0.30991116166114807,
      "learning_rate": 2.909066666666667e-05,
      "loss": 0.0168,
      "step": 12731
    },
    {
      "epoch": 23.49122807017544,
      "grad_norm": 0.2520284056663513,
      "learning_rate": 2.909033333333333e-05,
      "loss": 0.0107,
      "step": 12732
    },
    {
      "epoch": 23.493074792243767,
      "grad_norm": 0.2349097579717636,
      "learning_rate": 2.909e-05,
      "loss": 0.0176,
      "step": 12733
    },
    {
      "epoch": 23.494921514312097,
      "grad_norm": 0.4836427867412567,
      "learning_rate": 2.908966666666667e-05,
      "loss": 0.0101,
      "step": 12734
    },
    {
      "epoch": 23.496768236380426,
      "grad_norm": 1.5186824798583984,
      "learning_rate": 2.9089333333333333e-05,
      "loss": 0.0418,
      "step": 12735
    },
    {
      "epoch": 23.498614958448755,
      "grad_norm": 0.25052887201309204,
      "learning_rate": 2.9089000000000002e-05,
      "loss": 0.0119,
      "step": 12736
    },
    {
      "epoch": 23.50046168051708,
      "grad_norm": 0.2631968855857849,
      "learning_rate": 2.9088666666666668e-05,
      "loss": 0.0337,
      "step": 12737
    },
    {
      "epoch": 23.50230840258541,
      "grad_norm": 0.18188248574733734,
      "learning_rate": 2.9088333333333334e-05,
      "loss": 0.0042,
      "step": 12738
    },
    {
      "epoch": 23.50415512465374,
      "grad_norm": 0.5120291709899902,
      "learning_rate": 2.9088e-05,
      "loss": 0.0093,
      "step": 12739
    },
    {
      "epoch": 23.506001846722068,
      "grad_norm": 0.12005823105573654,
      "learning_rate": 2.9087666666666666e-05,
      "loss": 0.0039,
      "step": 12740
    },
    {
      "epoch": 23.507848568790397,
      "grad_norm": 0.28853529691696167,
      "learning_rate": 2.9087333333333332e-05,
      "loss": 0.0082,
      "step": 12741
    },
    {
      "epoch": 23.509695290858726,
      "grad_norm": 0.1488683521747589,
      "learning_rate": 2.9087e-05,
      "loss": 0.0051,
      "step": 12742
    },
    {
      "epoch": 23.511542012927055,
      "grad_norm": 0.6102681159973145,
      "learning_rate": 2.9086666666666667e-05,
      "loss": 0.0086,
      "step": 12743
    },
    {
      "epoch": 23.513388734995385,
      "grad_norm": 0.30232924222946167,
      "learning_rate": 2.9086333333333333e-05,
      "loss": 0.0079,
      "step": 12744
    },
    {
      "epoch": 23.51523545706371,
      "grad_norm": 0.3770429790019989,
      "learning_rate": 2.9086000000000002e-05,
      "loss": 0.0087,
      "step": 12745
    },
    {
      "epoch": 23.51708217913204,
      "grad_norm": 0.27687808871269226,
      "learning_rate": 2.9085666666666665e-05,
      "loss": 0.0072,
      "step": 12746
    },
    {
      "epoch": 23.51892890120037,
      "grad_norm": 0.4040433168411255,
      "learning_rate": 2.9085333333333334e-05,
      "loss": 0.0291,
      "step": 12747
    },
    {
      "epoch": 23.520775623268698,
      "grad_norm": 0.4633181095123291,
      "learning_rate": 2.9085e-05,
      "loss": 0.01,
      "step": 12748
    },
    {
      "epoch": 23.522622345337027,
      "grad_norm": 0.21011197566986084,
      "learning_rate": 2.9084666666666666e-05,
      "loss": 0.0058,
      "step": 12749
    },
    {
      "epoch": 23.524469067405356,
      "grad_norm": 0.09449068456888199,
      "learning_rate": 2.9084333333333335e-05,
      "loss": 0.0026,
      "step": 12750
    },
    {
      "epoch": 23.526315789473685,
      "grad_norm": 0.22294382750988007,
      "learning_rate": 2.9084e-05,
      "loss": 0.0052,
      "step": 12751
    },
    {
      "epoch": 23.528162511542014,
      "grad_norm": 0.6660040616989136,
      "learning_rate": 2.9083666666666667e-05,
      "loss": 0.0324,
      "step": 12752
    },
    {
      "epoch": 23.530009233610343,
      "grad_norm": 0.20625843107700348,
      "learning_rate": 2.9083333333333333e-05,
      "loss": 0.0064,
      "step": 12753
    },
    {
      "epoch": 23.53185595567867,
      "grad_norm": 0.21546894311904907,
      "learning_rate": 2.9083000000000003e-05,
      "loss": 0.0051,
      "step": 12754
    },
    {
      "epoch": 23.533702677746998,
      "grad_norm": 0.21883495151996613,
      "learning_rate": 2.9082666666666665e-05,
      "loss": 0.0059,
      "step": 12755
    },
    {
      "epoch": 23.535549399815327,
      "grad_norm": 1.180564522743225,
      "learning_rate": 2.9082333333333334e-05,
      "loss": 0.0141,
      "step": 12756
    },
    {
      "epoch": 23.537396121883656,
      "grad_norm": 0.13864454627037048,
      "learning_rate": 2.9082e-05,
      "loss": 0.0026,
      "step": 12757
    },
    {
      "epoch": 23.539242843951985,
      "grad_norm": 0.2007858157157898,
      "learning_rate": 2.9081666666666666e-05,
      "loss": 0.0038,
      "step": 12758
    },
    {
      "epoch": 23.541089566020315,
      "grad_norm": 0.27783122658729553,
      "learning_rate": 2.9081333333333336e-05,
      "loss": 0.007,
      "step": 12759
    },
    {
      "epoch": 23.542936288088644,
      "grad_norm": 0.19886739552021027,
      "learning_rate": 2.9081e-05,
      "loss": 0.0057,
      "step": 12760
    },
    {
      "epoch": 23.544783010156973,
      "grad_norm": 0.23898394405841827,
      "learning_rate": 2.9080666666666668e-05,
      "loss": 0.0054,
      "step": 12761
    },
    {
      "epoch": 23.5466297322253,
      "grad_norm": 0.22545062005519867,
      "learning_rate": 2.9080333333333333e-05,
      "loss": 0.0046,
      "step": 12762
    },
    {
      "epoch": 23.548476454293628,
      "grad_norm": 0.16324621438980103,
      "learning_rate": 2.9080000000000003e-05,
      "loss": 0.002,
      "step": 12763
    },
    {
      "epoch": 23.550323176361957,
      "grad_norm": 0.19871051609516144,
      "learning_rate": 2.9079666666666665e-05,
      "loss": 0.0041,
      "step": 12764
    },
    {
      "epoch": 23.552169898430286,
      "grad_norm": 0.31745415925979614,
      "learning_rate": 2.9079333333333335e-05,
      "loss": 0.0033,
      "step": 12765
    },
    {
      "epoch": 23.554016620498615,
      "grad_norm": 0.6388939619064331,
      "learning_rate": 2.9079e-05,
      "loss": 0.0151,
      "step": 12766
    },
    {
      "epoch": 23.555863342566944,
      "grad_norm": 0.46394413709640503,
      "learning_rate": 2.9078666666666667e-05,
      "loss": 0.1668,
      "step": 12767
    },
    {
      "epoch": 23.557710064635273,
      "grad_norm": 0.5517489910125732,
      "learning_rate": 2.9078333333333336e-05,
      "loss": 0.0812,
      "step": 12768
    },
    {
      "epoch": 23.559556786703602,
      "grad_norm": 0.46741852164268494,
      "learning_rate": 2.9078000000000002e-05,
      "loss": 0.0696,
      "step": 12769
    },
    {
      "epoch": 23.56140350877193,
      "grad_norm": 0.44784027338027954,
      "learning_rate": 2.9077666666666668e-05,
      "loss": 0.0636,
      "step": 12770
    },
    {
      "epoch": 23.563250230840257,
      "grad_norm": 0.4539169669151306,
      "learning_rate": 2.9077333333333334e-05,
      "loss": 0.0528,
      "step": 12771
    },
    {
      "epoch": 23.565096952908586,
      "grad_norm": 0.3409072458744049,
      "learning_rate": 2.9077e-05,
      "loss": 0.0333,
      "step": 12772
    },
    {
      "epoch": 23.566943674976915,
      "grad_norm": 0.4125320017337799,
      "learning_rate": 2.9076666666666666e-05,
      "loss": 0.0608,
      "step": 12773
    },
    {
      "epoch": 23.568790397045245,
      "grad_norm": 0.4366224408149719,
      "learning_rate": 2.9076333333333335e-05,
      "loss": 0.0594,
      "step": 12774
    },
    {
      "epoch": 23.570637119113574,
      "grad_norm": 0.32634279131889343,
      "learning_rate": 2.9076e-05,
      "loss": 0.0346,
      "step": 12775
    },
    {
      "epoch": 23.572483841181903,
      "grad_norm": 0.4740084409713745,
      "learning_rate": 2.9075666666666667e-05,
      "loss": 0.0271,
      "step": 12776
    },
    {
      "epoch": 23.574330563250232,
      "grad_norm": 0.45157182216644287,
      "learning_rate": 2.9075333333333336e-05,
      "loss": 0.0275,
      "step": 12777
    },
    {
      "epoch": 23.57617728531856,
      "grad_norm": 0.4033220410346985,
      "learning_rate": 2.9075e-05,
      "loss": 0.0467,
      "step": 12778
    },
    {
      "epoch": 23.578024007386887,
      "grad_norm": 0.5562022924423218,
      "learning_rate": 2.9074666666666668e-05,
      "loss": 0.0194,
      "step": 12779
    },
    {
      "epoch": 23.579870729455216,
      "grad_norm": 0.30805903673171997,
      "learning_rate": 2.9074333333333334e-05,
      "loss": 0.0141,
      "step": 12780
    },
    {
      "epoch": 23.581717451523545,
      "grad_norm": 0.45251429080963135,
      "learning_rate": 2.9074e-05,
      "loss": 0.024,
      "step": 12781
    },
    {
      "epoch": 23.583564173591874,
      "grad_norm": 0.3250104784965515,
      "learning_rate": 2.9073666666666666e-05,
      "loss": 0.0216,
      "step": 12782
    },
    {
      "epoch": 23.585410895660203,
      "grad_norm": 0.11664088815450668,
      "learning_rate": 2.9073333333333335e-05,
      "loss": 0.0041,
      "step": 12783
    },
    {
      "epoch": 23.587257617728532,
      "grad_norm": 0.5777392387390137,
      "learning_rate": 2.9073e-05,
      "loss": 0.0263,
      "step": 12784
    },
    {
      "epoch": 23.58910433979686,
      "grad_norm": 0.32111817598342896,
      "learning_rate": 2.9072666666666667e-05,
      "loss": 0.0114,
      "step": 12785
    },
    {
      "epoch": 23.59095106186519,
      "grad_norm": 0.34782543778419495,
      "learning_rate": 2.9072333333333336e-05,
      "loss": 0.0136,
      "step": 12786
    },
    {
      "epoch": 23.592797783933516,
      "grad_norm": 0.539177656173706,
      "learning_rate": 2.9072e-05,
      "loss": 0.0094,
      "step": 12787
    },
    {
      "epoch": 23.594644506001845,
      "grad_norm": 0.1888371706008911,
      "learning_rate": 2.9071666666666668e-05,
      "loss": 0.0069,
      "step": 12788
    },
    {
      "epoch": 23.596491228070175,
      "grad_norm": 0.3934464156627655,
      "learning_rate": 2.9071333333333334e-05,
      "loss": 0.0128,
      "step": 12789
    },
    {
      "epoch": 23.598337950138504,
      "grad_norm": 0.21380853652954102,
      "learning_rate": 2.9071e-05,
      "loss": 0.0054,
      "step": 12790
    },
    {
      "epoch": 23.600184672206833,
      "grad_norm": 0.18468987941741943,
      "learning_rate": 2.9070666666666666e-05,
      "loss": 0.0048,
      "step": 12791
    },
    {
      "epoch": 23.602031394275162,
      "grad_norm": 0.10190550237894058,
      "learning_rate": 2.9070333333333335e-05,
      "loss": 0.0044,
      "step": 12792
    },
    {
      "epoch": 23.60387811634349,
      "grad_norm": 0.31460437178611755,
      "learning_rate": 2.907e-05,
      "loss": 0.0057,
      "step": 12793
    },
    {
      "epoch": 23.60572483841182,
      "grad_norm": 0.20910628139972687,
      "learning_rate": 2.9069666666666667e-05,
      "loss": 0.0061,
      "step": 12794
    },
    {
      "epoch": 23.607571560480146,
      "grad_norm": 0.2906193435192108,
      "learning_rate": 2.9069333333333336e-05,
      "loss": 0.0111,
      "step": 12795
    },
    {
      "epoch": 23.609418282548475,
      "grad_norm": 0.3938979506492615,
      "learning_rate": 2.9069e-05,
      "loss": 0.055,
      "step": 12796
    },
    {
      "epoch": 23.611265004616804,
      "grad_norm": 0.27934545278549194,
      "learning_rate": 2.906866666666667e-05,
      "loss": 0.0339,
      "step": 12797
    },
    {
      "epoch": 23.613111726685133,
      "grad_norm": 0.4780665338039398,
      "learning_rate": 2.9068333333333334e-05,
      "loss": 0.007,
      "step": 12798
    },
    {
      "epoch": 23.614958448753463,
      "grad_norm": 0.14447623491287231,
      "learning_rate": 2.9068e-05,
      "loss": 0.0028,
      "step": 12799
    },
    {
      "epoch": 23.61680517082179,
      "grad_norm": 0.4102858901023865,
      "learning_rate": 2.906766666666667e-05,
      "loss": 0.0126,
      "step": 12800
    },
    {
      "epoch": 23.61865189289012,
      "grad_norm": 0.41097262501716614,
      "learning_rate": 2.9067333333333332e-05,
      "loss": 0.0072,
      "step": 12801
    },
    {
      "epoch": 23.62049861495845,
      "grad_norm": 0.11427780240774155,
      "learning_rate": 2.9067e-05,
      "loss": 0.0031,
      "step": 12802
    },
    {
      "epoch": 23.62234533702678,
      "grad_norm": 0.3202766180038452,
      "learning_rate": 2.9066666666666667e-05,
      "loss": 0.0076,
      "step": 12803
    },
    {
      "epoch": 23.624192059095105,
      "grad_norm": 0.21148407459259033,
      "learning_rate": 2.9066333333333333e-05,
      "loss": 0.0045,
      "step": 12804
    },
    {
      "epoch": 23.626038781163434,
      "grad_norm": 0.4769347608089447,
      "learning_rate": 2.9066e-05,
      "loss": 0.0062,
      "step": 12805
    },
    {
      "epoch": 23.627885503231763,
      "grad_norm": 0.19845016300678253,
      "learning_rate": 2.906566666666667e-05,
      "loss": 0.0055,
      "step": 12806
    },
    {
      "epoch": 23.629732225300092,
      "grad_norm": 0.2139330506324768,
      "learning_rate": 2.906533333333333e-05,
      "loss": 0.0058,
      "step": 12807
    },
    {
      "epoch": 23.63157894736842,
      "grad_norm": 0.27285897731781006,
      "learning_rate": 2.9065e-05,
      "loss": 0.0104,
      "step": 12808
    },
    {
      "epoch": 23.63342566943675,
      "grad_norm": 0.4506590962409973,
      "learning_rate": 2.906466666666667e-05,
      "loss": 0.009,
      "step": 12809
    },
    {
      "epoch": 23.63527239150508,
      "grad_norm": 0.3636559247970581,
      "learning_rate": 2.9064333333333332e-05,
      "loss": 0.0044,
      "step": 12810
    },
    {
      "epoch": 23.63711911357341,
      "grad_norm": 0.10463052988052368,
      "learning_rate": 2.9064e-05,
      "loss": 0.0025,
      "step": 12811
    },
    {
      "epoch": 23.638965835641734,
      "grad_norm": 0.47499239444732666,
      "learning_rate": 2.9063666666666668e-05,
      "loss": 0.0088,
      "step": 12812
    },
    {
      "epoch": 23.640812557710063,
      "grad_norm": 0.23979461193084717,
      "learning_rate": 2.9063333333333333e-05,
      "loss": 0.0046,
      "step": 12813
    },
    {
      "epoch": 23.642659279778393,
      "grad_norm": 0.3647124171257019,
      "learning_rate": 2.9063e-05,
      "loss": 0.0087,
      "step": 12814
    },
    {
      "epoch": 23.64450600184672,
      "grad_norm": 0.25506895780563354,
      "learning_rate": 2.906266666666667e-05,
      "loss": 0.0048,
      "step": 12815
    },
    {
      "epoch": 23.64635272391505,
      "grad_norm": 0.3625432252883911,
      "learning_rate": 2.9062333333333335e-05,
      "loss": 0.0099,
      "step": 12816
    },
    {
      "epoch": 23.64819944598338,
      "grad_norm": 0.5960791110992432,
      "learning_rate": 2.9062e-05,
      "loss": 0.1273,
      "step": 12817
    },
    {
      "epoch": 23.65004616805171,
      "grad_norm": 0.46018001437187195,
      "learning_rate": 2.906166666666667e-05,
      "loss": 0.1252,
      "step": 12818
    },
    {
      "epoch": 23.65189289012004,
      "grad_norm": 0.44288110733032227,
      "learning_rate": 2.9061333333333332e-05,
      "loss": 0.1154,
      "step": 12819
    },
    {
      "epoch": 23.653739612188367,
      "grad_norm": 0.4618895649909973,
      "learning_rate": 2.9061000000000002e-05,
      "loss": 0.0859,
      "step": 12820
    },
    {
      "epoch": 23.655586334256693,
      "grad_norm": 0.4462798237800598,
      "learning_rate": 2.9060666666666668e-05,
      "loss": 0.1258,
      "step": 12821
    },
    {
      "epoch": 23.657433056325022,
      "grad_norm": 0.5325228571891785,
      "learning_rate": 2.9060333333333334e-05,
      "loss": 0.0484,
      "step": 12822
    },
    {
      "epoch": 23.65927977839335,
      "grad_norm": 0.3383803069591522,
      "learning_rate": 2.906e-05,
      "loss": 0.041,
      "step": 12823
    },
    {
      "epoch": 23.66112650046168,
      "grad_norm": 0.3687587082386017,
      "learning_rate": 2.905966666666667e-05,
      "loss": 0.0365,
      "step": 12824
    },
    {
      "epoch": 23.66297322253001,
      "grad_norm": 0.4730265438556671,
      "learning_rate": 2.9059333333333335e-05,
      "loss": 0.077,
      "step": 12825
    },
    {
      "epoch": 23.66481994459834,
      "grad_norm": 0.41435232758522034,
      "learning_rate": 2.9059e-05,
      "loss": 0.0335,
      "step": 12826
    },
    {
      "epoch": 23.666666666666668,
      "grad_norm": 0.3603576123714447,
      "learning_rate": 2.905866666666667e-05,
      "loss": 0.0235,
      "step": 12827
    },
    {
      "epoch": 23.668513388734997,
      "grad_norm": 0.22493116557598114,
      "learning_rate": 2.9058333333333333e-05,
      "loss": 0.0158,
      "step": 12828
    },
    {
      "epoch": 23.670360110803323,
      "grad_norm": 0.5691091418266296,
      "learning_rate": 2.9058000000000002e-05,
      "loss": 0.0814,
      "step": 12829
    },
    {
      "epoch": 23.67220683287165,
      "grad_norm": 0.3875930607318878,
      "learning_rate": 2.9057666666666668e-05,
      "loss": 0.0219,
      "step": 12830
    },
    {
      "epoch": 23.67405355493998,
      "grad_norm": 0.7209383249282837,
      "learning_rate": 2.9057333333333334e-05,
      "loss": 0.0278,
      "step": 12831
    },
    {
      "epoch": 23.67590027700831,
      "grad_norm": 0.2960909605026245,
      "learning_rate": 2.9057e-05,
      "loss": 0.0321,
      "step": 12832
    },
    {
      "epoch": 23.67774699907664,
      "grad_norm": 0.3103616535663605,
      "learning_rate": 2.9056666666666666e-05,
      "loss": 0.0115,
      "step": 12833
    },
    {
      "epoch": 23.67959372114497,
      "grad_norm": 0.20014365017414093,
      "learning_rate": 2.9056333333333335e-05,
      "loss": 0.0082,
      "step": 12834
    },
    {
      "epoch": 23.681440443213297,
      "grad_norm": 0.1251969039440155,
      "learning_rate": 2.9056e-05,
      "loss": 0.0062,
      "step": 12835
    },
    {
      "epoch": 23.683287165281627,
      "grad_norm": 0.26539865136146545,
      "learning_rate": 2.9055666666666667e-05,
      "loss": 0.024,
      "step": 12836
    },
    {
      "epoch": 23.685133887349952,
      "grad_norm": 0.26761507987976074,
      "learning_rate": 2.9055333333333333e-05,
      "loss": 0.0085,
      "step": 12837
    },
    {
      "epoch": 23.68698060941828,
      "grad_norm": 0.077335886657238,
      "learning_rate": 2.9055000000000002e-05,
      "loss": 0.0043,
      "step": 12838
    },
    {
      "epoch": 23.68882733148661,
      "grad_norm": 0.24382653832435608,
      "learning_rate": 2.9054666666666665e-05,
      "loss": 0.007,
      "step": 12839
    },
    {
      "epoch": 23.69067405355494,
      "grad_norm": 0.2220059633255005,
      "learning_rate": 2.9054333333333334e-05,
      "loss": 0.0062,
      "step": 12840
    },
    {
      "epoch": 23.69252077562327,
      "grad_norm": 0.28311780095100403,
      "learning_rate": 2.9054e-05,
      "loss": 0.0221,
      "step": 12841
    },
    {
      "epoch": 23.694367497691598,
      "grad_norm": 0.1493522971868515,
      "learning_rate": 2.9053666666666666e-05,
      "loss": 0.0035,
      "step": 12842
    },
    {
      "epoch": 23.696214219759927,
      "grad_norm": 0.17693676054477692,
      "learning_rate": 2.9053333333333335e-05,
      "loss": 0.0089,
      "step": 12843
    },
    {
      "epoch": 23.698060941828256,
      "grad_norm": 0.21312695741653442,
      "learning_rate": 2.9053e-05,
      "loss": 0.0049,
      "step": 12844
    },
    {
      "epoch": 23.69990766389658,
      "grad_norm": 0.31813400983810425,
      "learning_rate": 2.9052666666666667e-05,
      "loss": 0.0073,
      "step": 12845
    },
    {
      "epoch": 23.70175438596491,
      "grad_norm": 0.1391538828611374,
      "learning_rate": 2.9052333333333333e-05,
      "loss": 0.0045,
      "step": 12846
    },
    {
      "epoch": 23.70360110803324,
      "grad_norm": 0.28718283772468567,
      "learning_rate": 2.9052000000000002e-05,
      "loss": 0.0061,
      "step": 12847
    },
    {
      "epoch": 23.70544783010157,
      "grad_norm": 0.27051207423210144,
      "learning_rate": 2.9051666666666665e-05,
      "loss": 0.0049,
      "step": 12848
    },
    {
      "epoch": 23.7072945521699,
      "grad_norm": 0.15699365735054016,
      "learning_rate": 2.9051333333333334e-05,
      "loss": 0.0048,
      "step": 12849
    },
    {
      "epoch": 23.709141274238227,
      "grad_norm": 0.5311880707740784,
      "learning_rate": 2.9051000000000004e-05,
      "loss": 0.0072,
      "step": 12850
    },
    {
      "epoch": 23.710987996306557,
      "grad_norm": 0.5135352611541748,
      "learning_rate": 2.9050666666666666e-05,
      "loss": 0.0092,
      "step": 12851
    },
    {
      "epoch": 23.712834718374886,
      "grad_norm": 0.30122503638267517,
      "learning_rate": 2.9050333333333335e-05,
      "loss": 0.0049,
      "step": 12852
    },
    {
      "epoch": 23.714681440443215,
      "grad_norm": 0.2334645688533783,
      "learning_rate": 2.905e-05,
      "loss": 0.0349,
      "step": 12853
    },
    {
      "epoch": 23.71652816251154,
      "grad_norm": 0.2619197368621826,
      "learning_rate": 2.9049666666666667e-05,
      "loss": 0.0044,
      "step": 12854
    },
    {
      "epoch": 23.71837488457987,
      "grad_norm": 0.23635706305503845,
      "learning_rate": 2.9049333333333333e-05,
      "loss": 0.0091,
      "step": 12855
    },
    {
      "epoch": 23.7202216066482,
      "grad_norm": 0.24201664328575134,
      "learning_rate": 2.9049000000000003e-05,
      "loss": 0.0053,
      "step": 12856
    },
    {
      "epoch": 23.722068328716528,
      "grad_norm": 0.3879227340221405,
      "learning_rate": 2.9048666666666665e-05,
      "loss": 0.004,
      "step": 12857
    },
    {
      "epoch": 23.723915050784857,
      "grad_norm": 0.2202427089214325,
      "learning_rate": 2.9048333333333334e-05,
      "loss": 0.003,
      "step": 12858
    },
    {
      "epoch": 23.725761772853186,
      "grad_norm": 0.21150343120098114,
      "learning_rate": 2.9048000000000004e-05,
      "loss": 0.007,
      "step": 12859
    },
    {
      "epoch": 23.727608494921515,
      "grad_norm": 0.20980283617973328,
      "learning_rate": 2.9047666666666666e-05,
      "loss": 0.0042,
      "step": 12860
    },
    {
      "epoch": 23.729455216989845,
      "grad_norm": 0.22404924035072327,
      "learning_rate": 2.9047333333333336e-05,
      "loss": 0.0074,
      "step": 12861
    },
    {
      "epoch": 23.73130193905817,
      "grad_norm": 0.26028111577033997,
      "learning_rate": 2.9046999999999998e-05,
      "loss": 0.0068,
      "step": 12862
    },
    {
      "epoch": 23.7331486611265,
      "grad_norm": 0.2740020453929901,
      "learning_rate": 2.9046666666666668e-05,
      "loss": 0.0064,
      "step": 12863
    },
    {
      "epoch": 23.73499538319483,
      "grad_norm": 0.22199766337871552,
      "learning_rate": 2.9046333333333333e-05,
      "loss": 0.0032,
      "step": 12864
    },
    {
      "epoch": 23.736842105263158,
      "grad_norm": 0.3249658942222595,
      "learning_rate": 2.9046e-05,
      "loss": 0.0118,
      "step": 12865
    },
    {
      "epoch": 23.738688827331487,
      "grad_norm": 0.2935706377029419,
      "learning_rate": 2.904566666666667e-05,
      "loss": 0.0129,
      "step": 12866
    },
    {
      "epoch": 23.740535549399816,
      "grad_norm": 0.5398800373077393,
      "learning_rate": 2.9045333333333335e-05,
      "loss": 0.1306,
      "step": 12867
    },
    {
      "epoch": 23.742382271468145,
      "grad_norm": 0.3459707796573639,
      "learning_rate": 2.9045e-05,
      "loss": 0.0956,
      "step": 12868
    },
    {
      "epoch": 23.744228993536474,
      "grad_norm": 0.7129822969436646,
      "learning_rate": 2.9044666666666667e-05,
      "loss": 0.099,
      "step": 12869
    },
    {
      "epoch": 23.746075715604803,
      "grad_norm": 0.5271465182304382,
      "learning_rate": 2.9044333333333336e-05,
      "loss": 0.07,
      "step": 12870
    },
    {
      "epoch": 23.74792243767313,
      "grad_norm": 0.5197558403015137,
      "learning_rate": 2.9044e-05,
      "loss": 0.0734,
      "step": 12871
    },
    {
      "epoch": 23.749769159741458,
      "grad_norm": 0.3547579050064087,
      "learning_rate": 2.9043666666666668e-05,
      "loss": 0.0382,
      "step": 12872
    },
    {
      "epoch": 23.751615881809787,
      "grad_norm": 0.38486313819885254,
      "learning_rate": 2.9043333333333334e-05,
      "loss": 0.0638,
      "step": 12873
    },
    {
      "epoch": 23.753462603878116,
      "grad_norm": 0.2941659390926361,
      "learning_rate": 2.9043e-05,
      "loss": 0.0366,
      "step": 12874
    },
    {
      "epoch": 23.755309325946445,
      "grad_norm": 0.37029245495796204,
      "learning_rate": 2.904266666666667e-05,
      "loss": 0.0396,
      "step": 12875
    },
    {
      "epoch": 23.757156048014775,
      "grad_norm": 0.8271615505218506,
      "learning_rate": 2.9042333333333335e-05,
      "loss": 0.0512,
      "step": 12876
    },
    {
      "epoch": 23.759002770083104,
      "grad_norm": 0.5787763595581055,
      "learning_rate": 2.9042e-05,
      "loss": 0.023,
      "step": 12877
    },
    {
      "epoch": 23.760849492151433,
      "grad_norm": 0.3281150162220001,
      "learning_rate": 2.9041666666666667e-05,
      "loss": 0.0432,
      "step": 12878
    },
    {
      "epoch": 23.76269621421976,
      "grad_norm": 0.5119446516036987,
      "learning_rate": 2.9041333333333336e-05,
      "loss": 0.0633,
      "step": 12879
    },
    {
      "epoch": 23.764542936288088,
      "grad_norm": 0.27672407031059265,
      "learning_rate": 2.9041e-05,
      "loss": 0.036,
      "step": 12880
    },
    {
      "epoch": 23.766389658356417,
      "grad_norm": 0.44429293274879456,
      "learning_rate": 2.9040666666666668e-05,
      "loss": 0.0299,
      "step": 12881
    },
    {
      "epoch": 23.768236380424746,
      "grad_norm": 0.24825359880924225,
      "learning_rate": 2.9040333333333334e-05,
      "loss": 0.0329,
      "step": 12882
    },
    {
      "epoch": 23.770083102493075,
      "grad_norm": 0.45071160793304443,
      "learning_rate": 2.904e-05,
      "loss": 0.0113,
      "step": 12883
    },
    {
      "epoch": 23.771929824561404,
      "grad_norm": 0.26706889271736145,
      "learning_rate": 2.903966666666667e-05,
      "loss": 0.0067,
      "step": 12884
    },
    {
      "epoch": 23.773776546629733,
      "grad_norm": 0.41779765486717224,
      "learning_rate": 2.9039333333333335e-05,
      "loss": 0.027,
      "step": 12885
    },
    {
      "epoch": 23.775623268698062,
      "grad_norm": 0.23480276763439178,
      "learning_rate": 2.9039e-05,
      "loss": 0.0097,
      "step": 12886
    },
    {
      "epoch": 23.777469990766388,
      "grad_norm": 0.2854717969894409,
      "learning_rate": 2.9038666666666667e-05,
      "loss": 0.0125,
      "step": 12887
    },
    {
      "epoch": 23.779316712834717,
      "grad_norm": 0.21366217732429504,
      "learning_rate": 2.9038333333333336e-05,
      "loss": 0.0036,
      "step": 12888
    },
    {
      "epoch": 23.781163434903046,
      "grad_norm": 0.1949768364429474,
      "learning_rate": 2.9038e-05,
      "loss": 0.0098,
      "step": 12889
    },
    {
      "epoch": 23.783010156971375,
      "grad_norm": 0.13498984277248383,
      "learning_rate": 2.9037666666666668e-05,
      "loss": 0.0029,
      "step": 12890
    },
    {
      "epoch": 23.784856879039705,
      "grad_norm": 0.25774431228637695,
      "learning_rate": 2.9037333333333334e-05,
      "loss": 0.0073,
      "step": 12891
    },
    {
      "epoch": 23.786703601108034,
      "grad_norm": 0.3333340287208557,
      "learning_rate": 2.9037e-05,
      "loss": 0.0121,
      "step": 12892
    },
    {
      "epoch": 23.788550323176363,
      "grad_norm": 0.18044157326221466,
      "learning_rate": 2.903666666666667e-05,
      "loss": 0.0163,
      "step": 12893
    },
    {
      "epoch": 23.790397045244692,
      "grad_norm": 0.08477292209863663,
      "learning_rate": 2.9036333333333332e-05,
      "loss": 0.0021,
      "step": 12894
    },
    {
      "epoch": 23.792243767313018,
      "grad_norm": 0.11724770069122314,
      "learning_rate": 2.9036e-05,
      "loss": 0.0036,
      "step": 12895
    },
    {
      "epoch": 23.794090489381347,
      "grad_norm": 0.480721116065979,
      "learning_rate": 2.9035666666666667e-05,
      "loss": 0.0112,
      "step": 12896
    },
    {
      "epoch": 23.795937211449676,
      "grad_norm": 0.2716176509857178,
      "learning_rate": 2.9035333333333333e-05,
      "loss": 0.0064,
      "step": 12897
    },
    {
      "epoch": 23.797783933518005,
      "grad_norm": 0.3495538532733917,
      "learning_rate": 2.9035e-05,
      "loss": 0.006,
      "step": 12898
    },
    {
      "epoch": 23.799630655586334,
      "grad_norm": 0.1573408991098404,
      "learning_rate": 2.903466666666667e-05,
      "loss": 0.0038,
      "step": 12899
    },
    {
      "epoch": 23.801477377654663,
      "grad_norm": 0.2714599072933197,
      "learning_rate": 2.9034333333333334e-05,
      "loss": 0.0043,
      "step": 12900
    },
    {
      "epoch": 23.803324099722992,
      "grad_norm": 0.14191612601280212,
      "learning_rate": 2.9034e-05,
      "loss": 0.0038,
      "step": 12901
    },
    {
      "epoch": 23.80517082179132,
      "grad_norm": 0.5580388307571411,
      "learning_rate": 2.903366666666667e-05,
      "loss": 0.0123,
      "step": 12902
    },
    {
      "epoch": 23.80701754385965,
      "grad_norm": 0.3556264340877533,
      "learning_rate": 2.9033333333333332e-05,
      "loss": 0.0073,
      "step": 12903
    },
    {
      "epoch": 23.808864265927976,
      "grad_norm": 0.24814888834953308,
      "learning_rate": 2.9033e-05,
      "loss": 0.01,
      "step": 12904
    },
    {
      "epoch": 23.810710987996305,
      "grad_norm": 0.35184523463249207,
      "learning_rate": 2.9032666666666667e-05,
      "loss": 0.0043,
      "step": 12905
    },
    {
      "epoch": 23.812557710064635,
      "grad_norm": 0.1633293479681015,
      "learning_rate": 2.9032333333333333e-05,
      "loss": 0.0047,
      "step": 12906
    },
    {
      "epoch": 23.814404432132964,
      "grad_norm": 0.1968468427658081,
      "learning_rate": 2.9032e-05,
      "loss": 0.0063,
      "step": 12907
    },
    {
      "epoch": 23.816251154201293,
      "grad_norm": 0.3138710558414459,
      "learning_rate": 2.903166666666667e-05,
      "loss": 0.0059,
      "step": 12908
    },
    {
      "epoch": 23.818097876269622,
      "grad_norm": 0.19953186810016632,
      "learning_rate": 2.9031333333333334e-05,
      "loss": 0.0058,
      "step": 12909
    },
    {
      "epoch": 23.81994459833795,
      "grad_norm": 0.19568628072738647,
      "learning_rate": 2.9031e-05,
      "loss": 0.0069,
      "step": 12910
    },
    {
      "epoch": 23.82179132040628,
      "grad_norm": 0.3259732127189636,
      "learning_rate": 2.903066666666667e-05,
      "loss": 0.0063,
      "step": 12911
    },
    {
      "epoch": 23.823638042474606,
      "grad_norm": 0.1458972543478012,
      "learning_rate": 2.9030333333333332e-05,
      "loss": 0.0044,
      "step": 12912
    },
    {
      "epoch": 23.825484764542935,
      "grad_norm": 0.16559676826000214,
      "learning_rate": 2.903e-05,
      "loss": 0.0051,
      "step": 12913
    },
    {
      "epoch": 23.827331486611264,
      "grad_norm": 0.3490874171257019,
      "learning_rate": 2.9029666666666668e-05,
      "loss": 0.0059,
      "step": 12914
    },
    {
      "epoch": 23.829178208679593,
      "grad_norm": 0.5214719772338867,
      "learning_rate": 2.9029333333333333e-05,
      "loss": 0.0076,
      "step": 12915
    },
    {
      "epoch": 23.831024930747922,
      "grad_norm": 0.2873786985874176,
      "learning_rate": 2.9029000000000003e-05,
      "loss": 0.0113,
      "step": 12916
    },
    {
      "epoch": 23.83287165281625,
      "grad_norm": 0.7002087235450745,
      "learning_rate": 2.902866666666667e-05,
      "loss": 0.151,
      "step": 12917
    },
    {
      "epoch": 23.83471837488458,
      "grad_norm": 0.48249414563179016,
      "learning_rate": 2.9028333333333335e-05,
      "loss": 0.108,
      "step": 12918
    },
    {
      "epoch": 23.83656509695291,
      "grad_norm": 0.5469855070114136,
      "learning_rate": 2.9028e-05,
      "loss": 0.0875,
      "step": 12919
    },
    {
      "epoch": 23.83841181902124,
      "grad_norm": 0.3179090917110443,
      "learning_rate": 2.902766666666667e-05,
      "loss": 0.0698,
      "step": 12920
    },
    {
      "epoch": 23.840258541089565,
      "grad_norm": 0.30969929695129395,
      "learning_rate": 2.9027333333333332e-05,
      "loss": 0.0512,
      "step": 12921
    },
    {
      "epoch": 23.842105263157894,
      "grad_norm": 1.0350627899169922,
      "learning_rate": 2.9027000000000002e-05,
      "loss": 0.0404,
      "step": 12922
    },
    {
      "epoch": 23.843951985226223,
      "grad_norm": 0.5427265763282776,
      "learning_rate": 2.9026666666666664e-05,
      "loss": 0.0655,
      "step": 12923
    },
    {
      "epoch": 23.845798707294552,
      "grad_norm": 0.48509711027145386,
      "learning_rate": 2.9026333333333334e-05,
      "loss": 0.048,
      "step": 12924
    },
    {
      "epoch": 23.84764542936288,
      "grad_norm": 0.5452003479003906,
      "learning_rate": 2.9026000000000003e-05,
      "loss": 0.0208,
      "step": 12925
    },
    {
      "epoch": 23.84949215143121,
      "grad_norm": 0.361055850982666,
      "learning_rate": 2.9025666666666666e-05,
      "loss": 0.0243,
      "step": 12926
    },
    {
      "epoch": 23.85133887349954,
      "grad_norm": 0.3371514081954956,
      "learning_rate": 2.9025333333333335e-05,
      "loss": 0.0268,
      "step": 12927
    },
    {
      "epoch": 23.85318559556787,
      "grad_norm": 0.23774278163909912,
      "learning_rate": 2.9025e-05,
      "loss": 0.0188,
      "step": 12928
    },
    {
      "epoch": 23.855032317636194,
      "grad_norm": 0.364484041929245,
      "learning_rate": 2.9024666666666667e-05,
      "loss": 0.0227,
      "step": 12929
    },
    {
      "epoch": 23.856879039704523,
      "grad_norm": 0.42037510871887207,
      "learning_rate": 2.9024333333333333e-05,
      "loss": 0.0112,
      "step": 12930
    },
    {
      "epoch": 23.858725761772853,
      "grad_norm": 0.29534637928009033,
      "learning_rate": 2.9024000000000002e-05,
      "loss": 0.0173,
      "step": 12931
    },
    {
      "epoch": 23.86057248384118,
      "grad_norm": 3.09122896194458,
      "learning_rate": 2.9023666666666665e-05,
      "loss": 0.0062,
      "step": 12932
    },
    {
      "epoch": 23.86241920590951,
      "grad_norm": 0.28615516424179077,
      "learning_rate": 2.9023333333333334e-05,
      "loss": 0.0058,
      "step": 12933
    },
    {
      "epoch": 23.86426592797784,
      "grad_norm": 1.4320789575576782,
      "learning_rate": 2.9023000000000003e-05,
      "loss": 0.031,
      "step": 12934
    },
    {
      "epoch": 23.86611265004617,
      "grad_norm": 0.24899813532829285,
      "learning_rate": 2.9022666666666666e-05,
      "loss": 0.0061,
      "step": 12935
    },
    {
      "epoch": 23.8679593721145,
      "grad_norm": 0.1240304559469223,
      "learning_rate": 2.9022333333333335e-05,
      "loss": 0.0036,
      "step": 12936
    },
    {
      "epoch": 23.869806094182824,
      "grad_norm": 0.25697949528694153,
      "learning_rate": 2.9022e-05,
      "loss": 0.0068,
      "step": 12937
    },
    {
      "epoch": 23.871652816251153,
      "grad_norm": 0.23376378417015076,
      "learning_rate": 2.9021666666666667e-05,
      "loss": 0.0072,
      "step": 12938
    },
    {
      "epoch": 23.873499538319482,
      "grad_norm": 0.1371239274740219,
      "learning_rate": 2.9021333333333333e-05,
      "loss": 0.0055,
      "step": 12939
    },
    {
      "epoch": 23.87534626038781,
      "grad_norm": 0.3965795636177063,
      "learning_rate": 2.9021000000000002e-05,
      "loss": 0.0067,
      "step": 12940
    },
    {
      "epoch": 23.87719298245614,
      "grad_norm": 0.2709249258041382,
      "learning_rate": 2.9020666666666668e-05,
      "loss": 0.0106,
      "step": 12941
    },
    {
      "epoch": 23.87903970452447,
      "grad_norm": 0.5115156769752502,
      "learning_rate": 2.9020333333333334e-05,
      "loss": 0.0115,
      "step": 12942
    },
    {
      "epoch": 23.8808864265928,
      "grad_norm": 0.2460068166255951,
      "learning_rate": 2.9020000000000003e-05,
      "loss": 0.0069,
      "step": 12943
    },
    {
      "epoch": 23.882733148661128,
      "grad_norm": 0.14581836760044098,
      "learning_rate": 2.9019666666666666e-05,
      "loss": 0.0189,
      "step": 12944
    },
    {
      "epoch": 23.884579870729453,
      "grad_norm": 0.2299008071422577,
      "learning_rate": 2.9019333333333335e-05,
      "loss": 0.0056,
      "step": 12945
    },
    {
      "epoch": 23.886426592797783,
      "grad_norm": 0.15325182676315308,
      "learning_rate": 2.9019e-05,
      "loss": 0.0044,
      "step": 12946
    },
    {
      "epoch": 23.88827331486611,
      "grad_norm": 0.24629101157188416,
      "learning_rate": 2.9018666666666667e-05,
      "loss": 0.0108,
      "step": 12947
    },
    {
      "epoch": 23.89012003693444,
      "grad_norm": 0.25775256752967834,
      "learning_rate": 2.9018333333333333e-05,
      "loss": 0.0145,
      "step": 12948
    },
    {
      "epoch": 23.89196675900277,
      "grad_norm": 0.20524519681930542,
      "learning_rate": 2.9018000000000002e-05,
      "loss": 0.0046,
      "step": 12949
    },
    {
      "epoch": 23.8938134810711,
      "grad_norm": 0.21955180168151855,
      "learning_rate": 2.9017666666666668e-05,
      "loss": 0.0076,
      "step": 12950
    },
    {
      "epoch": 23.89566020313943,
      "grad_norm": 0.44401729106903076,
      "learning_rate": 2.9017333333333334e-05,
      "loss": 0.0039,
      "step": 12951
    },
    {
      "epoch": 23.897506925207757,
      "grad_norm": 0.29561930894851685,
      "learning_rate": 2.9017e-05,
      "loss": 0.0056,
      "step": 12952
    },
    {
      "epoch": 23.899353647276087,
      "grad_norm": 0.17952415347099304,
      "learning_rate": 2.9016666666666666e-05,
      "loss": 0.0042,
      "step": 12953
    },
    {
      "epoch": 23.901200369344412,
      "grad_norm": 0.41808587312698364,
      "learning_rate": 2.9016333333333335e-05,
      "loss": 0.0036,
      "step": 12954
    },
    {
      "epoch": 23.90304709141274,
      "grad_norm": 0.23725056648254395,
      "learning_rate": 2.9015999999999998e-05,
      "loss": 0.0055,
      "step": 12955
    },
    {
      "epoch": 23.90489381348107,
      "grad_norm": 0.372693806886673,
      "learning_rate": 2.9015666666666667e-05,
      "loss": 0.0084,
      "step": 12956
    },
    {
      "epoch": 23.9067405355494,
      "grad_norm": 0.19695430994033813,
      "learning_rate": 2.9015333333333333e-05,
      "loss": 0.007,
      "step": 12957
    },
    {
      "epoch": 23.90858725761773,
      "grad_norm": 0.13928771018981934,
      "learning_rate": 2.9015e-05,
      "loss": 0.0032,
      "step": 12958
    },
    {
      "epoch": 23.910433979686058,
      "grad_norm": 0.43361949920654297,
      "learning_rate": 2.901466666666667e-05,
      "loss": 0.0104,
      "step": 12959
    },
    {
      "epoch": 23.912280701754387,
      "grad_norm": 0.1476796269416809,
      "learning_rate": 2.9014333333333334e-05,
      "loss": 0.0049,
      "step": 12960
    },
    {
      "epoch": 23.914127423822716,
      "grad_norm": 0.2999615967273712,
      "learning_rate": 2.9014e-05,
      "loss": 0.0067,
      "step": 12961
    },
    {
      "epoch": 23.91597414589104,
      "grad_norm": 0.7123618125915527,
      "learning_rate": 2.9013666666666666e-05,
      "loss": 0.0119,
      "step": 12962
    },
    {
      "epoch": 23.91782086795937,
      "grad_norm": 0.18582561612129211,
      "learning_rate": 2.9013333333333336e-05,
      "loss": 0.0033,
      "step": 12963
    },
    {
      "epoch": 23.9196675900277,
      "grad_norm": 0.5258939862251282,
      "learning_rate": 2.9012999999999998e-05,
      "loss": 0.0071,
      "step": 12964
    },
    {
      "epoch": 23.92151431209603,
      "grad_norm": 0.6415529847145081,
      "learning_rate": 2.9012666666666668e-05,
      "loss": 0.0128,
      "step": 12965
    },
    {
      "epoch": 23.92336103416436,
      "grad_norm": 0.29516181349754333,
      "learning_rate": 2.9012333333333337e-05,
      "loss": 0.0132,
      "step": 12966
    },
    {
      "epoch": 23.925207756232687,
      "grad_norm": 0.7700099349021912,
      "learning_rate": 2.9012e-05,
      "loss": 0.1746,
      "step": 12967
    },
    {
      "epoch": 23.927054478301017,
      "grad_norm": 0.46317270398139954,
      "learning_rate": 2.901166666666667e-05,
      "loss": 0.1412,
      "step": 12968
    },
    {
      "epoch": 23.928901200369346,
      "grad_norm": 0.6192085146903992,
      "learning_rate": 2.9011333333333335e-05,
      "loss": 0.1088,
      "step": 12969
    },
    {
      "epoch": 23.930747922437675,
      "grad_norm": 0.6775713562965393,
      "learning_rate": 2.9011e-05,
      "loss": 0.0943,
      "step": 12970
    },
    {
      "epoch": 23.932594644506,
      "grad_norm": 0.6361328959465027,
      "learning_rate": 2.9010666666666667e-05,
      "loss": 0.0614,
      "step": 12971
    },
    {
      "epoch": 23.93444136657433,
      "grad_norm": 0.6036672592163086,
      "learning_rate": 2.9010333333333336e-05,
      "loss": 0.0678,
      "step": 12972
    },
    {
      "epoch": 23.93628808864266,
      "grad_norm": 0.37628716230392456,
      "learning_rate": 2.901e-05,
      "loss": 0.0702,
      "step": 12973
    },
    {
      "epoch": 23.938134810710988,
      "grad_norm": 0.34638822078704834,
      "learning_rate": 2.9009666666666668e-05,
      "loss": 0.0843,
      "step": 12974
    },
    {
      "epoch": 23.939981532779317,
      "grad_norm": 0.4206011891365051,
      "learning_rate": 2.9009333333333337e-05,
      "loss": 0.034,
      "step": 12975
    },
    {
      "epoch": 23.941828254847646,
      "grad_norm": 0.4541192352771759,
      "learning_rate": 2.9009e-05,
      "loss": 0.0732,
      "step": 12976
    },
    {
      "epoch": 23.943674976915975,
      "grad_norm": 0.3251061737537384,
      "learning_rate": 2.900866666666667e-05,
      "loss": 0.0167,
      "step": 12977
    },
    {
      "epoch": 23.945521698984304,
      "grad_norm": 0.4052632451057434,
      "learning_rate": 2.9008333333333335e-05,
      "loss": 0.0125,
      "step": 12978
    },
    {
      "epoch": 23.94736842105263,
      "grad_norm": 0.20792420208454132,
      "learning_rate": 2.9008e-05,
      "loss": 0.0106,
      "step": 12979
    },
    {
      "epoch": 23.94921514312096,
      "grad_norm": 0.2851109206676483,
      "learning_rate": 2.9007666666666667e-05,
      "loss": 0.0157,
      "step": 12980
    },
    {
      "epoch": 23.95106186518929,
      "grad_norm": 0.4680311381816864,
      "learning_rate": 2.9007333333333336e-05,
      "loss": 0.0161,
      "step": 12981
    },
    {
      "epoch": 23.952908587257618,
      "grad_norm": 0.1415628045797348,
      "learning_rate": 2.9007e-05,
      "loss": 0.0334,
      "step": 12982
    },
    {
      "epoch": 23.954755309325947,
      "grad_norm": 0.20743387937545776,
      "learning_rate": 2.9006666666666668e-05,
      "loss": 0.0088,
      "step": 12983
    },
    {
      "epoch": 23.956602031394276,
      "grad_norm": 0.20807644724845886,
      "learning_rate": 2.9006333333333334e-05,
      "loss": 0.0204,
      "step": 12984
    },
    {
      "epoch": 23.958448753462605,
      "grad_norm": 0.2604505121707916,
      "learning_rate": 2.9006e-05,
      "loss": 0.0062,
      "step": 12985
    },
    {
      "epoch": 23.960295475530934,
      "grad_norm": 0.22881126403808594,
      "learning_rate": 2.900566666666667e-05,
      "loss": 0.0053,
      "step": 12986
    },
    {
      "epoch": 23.96214219759926,
      "grad_norm": 0.11627691984176636,
      "learning_rate": 2.900533333333333e-05,
      "loss": 0.0041,
      "step": 12987
    },
    {
      "epoch": 23.96398891966759,
      "grad_norm": 0.24788479506969452,
      "learning_rate": 2.9005e-05,
      "loss": 0.0072,
      "step": 12988
    },
    {
      "epoch": 23.965835641735918,
      "grad_norm": 0.27122530341148376,
      "learning_rate": 2.9004666666666667e-05,
      "loss": 0.006,
      "step": 12989
    },
    {
      "epoch": 23.967682363804247,
      "grad_norm": 0.2032022923231125,
      "learning_rate": 2.9004333333333333e-05,
      "loss": 0.0043,
      "step": 12990
    },
    {
      "epoch": 23.969529085872576,
      "grad_norm": 0.31291407346725464,
      "learning_rate": 2.9004000000000002e-05,
      "loss": 0.004,
      "step": 12991
    },
    {
      "epoch": 23.971375807940905,
      "grad_norm": 0.38916656374931335,
      "learning_rate": 2.9003666666666668e-05,
      "loss": 0.0113,
      "step": 12992
    },
    {
      "epoch": 23.973222530009235,
      "grad_norm": 0.2948424518108368,
      "learning_rate": 2.9003333333333334e-05,
      "loss": 0.0118,
      "step": 12993
    },
    {
      "epoch": 23.975069252077564,
      "grad_norm": 0.223440483212471,
      "learning_rate": 2.9003e-05,
      "loss": 0.0082,
      "step": 12994
    },
    {
      "epoch": 23.97691597414589,
      "grad_norm": 0.4852774739265442,
      "learning_rate": 2.900266666666667e-05,
      "loss": 0.0103,
      "step": 12995
    },
    {
      "epoch": 23.97876269621422,
      "grad_norm": 0.2066236287355423,
      "learning_rate": 2.9002333333333332e-05,
      "loss": 0.0038,
      "step": 12996
    },
    {
      "epoch": 23.980609418282548,
      "grad_norm": 0.32647451758384705,
      "learning_rate": 2.9002e-05,
      "loss": 0.0078,
      "step": 12997
    },
    {
      "epoch": 23.982456140350877,
      "grad_norm": 0.21102245151996613,
      "learning_rate": 2.9001666666666667e-05,
      "loss": 0.005,
      "step": 12998
    },
    {
      "epoch": 23.984302862419206,
      "grad_norm": 0.4090909957885742,
      "learning_rate": 2.9001333333333333e-05,
      "loss": 0.0124,
      "step": 12999
    },
    {
      "epoch": 23.986149584487535,
      "grad_norm": 0.1885944902896881,
      "learning_rate": 2.9001000000000002e-05,
      "loss": 0.0041,
      "step": 13000
    },
    {
      "epoch": 23.986149584487535,
      "eval_cer": 0.11374653290820956,
      "eval_loss": 0.38288652896881104,
      "eval_runtime": 15.7412,
      "eval_samples_per_second": 19.312,
      "eval_steps_per_second": 0.635,
      "eval_wer": 0.3898695318495779,
      "step": 13000
    },
    {
      "epoch": 23.987996306555864,
      "grad_norm": 0.45336973667144775,
      "learning_rate": 2.9000666666666668e-05,
      "loss": 0.0112,
      "step": 13001
    },
    {
      "epoch": 23.989843028624193,
      "grad_norm": 0.5542886257171631,
      "learning_rate": 2.9000333333333334e-05,
      "loss": 0.0076,
      "step": 13002
    },
    {
      "epoch": 23.991689750692522,
      "grad_norm": 0.18405455350875854,
      "learning_rate": 2.9e-05,
      "loss": 0.0044,
      "step": 13003
    },
    {
      "epoch": 23.993536472760848,
      "grad_norm": 0.2096349596977234,
      "learning_rate": 2.899966666666667e-05,
      "loss": 0.006,
      "step": 13004
    },
    {
      "epoch": 23.995383194829177,
      "grad_norm": 0.2750890851020813,
      "learning_rate": 2.8999333333333332e-05,
      "loss": 0.0053,
      "step": 13005
    },
    {
      "epoch": 23.997229916897506,
      "grad_norm": 0.24388429522514343,
      "learning_rate": 2.8999e-05,
      "loss": 0.0051,
      "step": 13006
    },
    {
      "epoch": 23.999076638965835,
      "grad_norm": 0.6541364789009094,
      "learning_rate": 2.8998666666666667e-05,
      "loss": 0.0086,
      "step": 13007
    },
    {
      "epoch": 24.0,
      "grad_norm": 0.0675673708319664,
      "learning_rate": 2.8998333333333333e-05,
      "loss": 0.0008,
      "step": 13008
    },
    {
      "epoch": 24.00184672206833,
      "grad_norm": 0.7535595893859863,
      "learning_rate": 2.8998000000000003e-05,
      "loss": 0.1244,
      "step": 13009
    },
    {
      "epoch": 24.00369344413666,
      "grad_norm": 0.40268534421920776,
      "learning_rate": 2.899766666666667e-05,
      "loss": 0.1111,
      "step": 13010
    },
    {
      "epoch": 24.005540166204987,
      "grad_norm": 0.49000465869903564,
      "learning_rate": 2.8997333333333334e-05,
      "loss": 0.0655,
      "step": 13011
    },
    {
      "epoch": 24.007386888273317,
      "grad_norm": 0.6041421890258789,
      "learning_rate": 2.8997e-05,
      "loss": 0.052,
      "step": 13012
    },
    {
      "epoch": 24.009233610341642,
      "grad_norm": 0.37183427810668945,
      "learning_rate": 2.8996666666666666e-05,
      "loss": 0.0523,
      "step": 13013
    },
    {
      "epoch": 24.01108033240997,
      "grad_norm": 0.5852181315422058,
      "learning_rate": 2.8996333333333332e-05,
      "loss": 0.0852,
      "step": 13014
    },
    {
      "epoch": 24.0129270544783,
      "grad_norm": 0.32598167657852173,
      "learning_rate": 2.8996e-05,
      "loss": 0.0321,
      "step": 13015
    },
    {
      "epoch": 24.01477377654663,
      "grad_norm": 0.342527836561203,
      "learning_rate": 2.8995666666666667e-05,
      "loss": 0.0431,
      "step": 13016
    },
    {
      "epoch": 24.01662049861496,
      "grad_norm": 0.753602147102356,
      "learning_rate": 2.8995333333333333e-05,
      "loss": 0.0439,
      "step": 13017
    },
    {
      "epoch": 24.018467220683288,
      "grad_norm": 0.46627503633499146,
      "learning_rate": 2.8995000000000003e-05,
      "loss": 0.0432,
      "step": 13018
    },
    {
      "epoch": 24.020313942751617,
      "grad_norm": 0.37367913126945496,
      "learning_rate": 2.8994666666666665e-05,
      "loss": 0.0298,
      "step": 13019
    },
    {
      "epoch": 24.022160664819946,
      "grad_norm": 0.5896576046943665,
      "learning_rate": 2.8994333333333335e-05,
      "loss": 0.0205,
      "step": 13020
    },
    {
      "epoch": 24.02400738688827,
      "grad_norm": 0.46239790320396423,
      "learning_rate": 2.8994e-05,
      "loss": 0.0388,
      "step": 13021
    },
    {
      "epoch": 24.0258541089566,
      "grad_norm": 0.25117456912994385,
      "learning_rate": 2.8993666666666667e-05,
      "loss": 0.0259,
      "step": 13022
    },
    {
      "epoch": 24.02770083102493,
      "grad_norm": 0.20962220430374146,
      "learning_rate": 2.8993333333333332e-05,
      "loss": 0.0174,
      "step": 13023
    },
    {
      "epoch": 24.02954755309326,
      "grad_norm": 0.3992897570133209,
      "learning_rate": 2.8993000000000002e-05,
      "loss": 0.0439,
      "step": 13024
    },
    {
      "epoch": 24.03139427516159,
      "grad_norm": 0.14528046548366547,
      "learning_rate": 2.8992666666666668e-05,
      "loss": 0.0058,
      "step": 13025
    },
    {
      "epoch": 24.033240997229917,
      "grad_norm": 0.2308790236711502,
      "learning_rate": 2.8992333333333334e-05,
      "loss": 0.0047,
      "step": 13026
    },
    {
      "epoch": 24.035087719298247,
      "grad_norm": 0.12781424820423126,
      "learning_rate": 2.8992000000000003e-05,
      "loss": 0.0049,
      "step": 13027
    },
    {
      "epoch": 24.036934441366576,
      "grad_norm": 0.2648352384567261,
      "learning_rate": 2.8991666666666666e-05,
      "loss": 0.0061,
      "step": 13028
    },
    {
      "epoch": 24.0387811634349,
      "grad_norm": 0.16820089519023895,
      "learning_rate": 2.8991333333333335e-05,
      "loss": 0.0049,
      "step": 13029
    },
    {
      "epoch": 24.04062788550323,
      "grad_norm": 0.37286460399627686,
      "learning_rate": 2.8991e-05,
      "loss": 0.0055,
      "step": 13030
    },
    {
      "epoch": 24.04247460757156,
      "grad_norm": 0.3106335699558258,
      "learning_rate": 2.8990666666666667e-05,
      "loss": 0.0088,
      "step": 13031
    },
    {
      "epoch": 24.04432132963989,
      "grad_norm": 0.27128398418426514,
      "learning_rate": 2.8990333333333333e-05,
      "loss": 0.005,
      "step": 13032
    },
    {
      "epoch": 24.046168051708218,
      "grad_norm": 0.29108282923698425,
      "learning_rate": 2.8990000000000002e-05,
      "loss": 0.0094,
      "step": 13033
    },
    {
      "epoch": 24.048014773776547,
      "grad_norm": 0.2468157261610031,
      "learning_rate": 2.8989666666666668e-05,
      "loss": 0.006,
      "step": 13034
    },
    {
      "epoch": 24.049861495844876,
      "grad_norm": 0.28584250807762146,
      "learning_rate": 2.8989333333333334e-05,
      "loss": 0.0074,
      "step": 13035
    },
    {
      "epoch": 24.051708217913205,
      "grad_norm": 0.2079189568758011,
      "learning_rate": 2.8989000000000003e-05,
      "loss": 0.0043,
      "step": 13036
    },
    {
      "epoch": 24.053554939981534,
      "grad_norm": 0.4608403146266937,
      "learning_rate": 2.8988666666666666e-05,
      "loss": 0.0071,
      "step": 13037
    },
    {
      "epoch": 24.05540166204986,
      "grad_norm": 0.09350863099098206,
      "learning_rate": 2.8988333333333335e-05,
      "loss": 0.0022,
      "step": 13038
    },
    {
      "epoch": 24.05724838411819,
      "grad_norm": 0.48382362723350525,
      "learning_rate": 2.8988e-05,
      "loss": 0.0173,
      "step": 13039
    },
    {
      "epoch": 24.05909510618652,
      "grad_norm": 0.44276735186576843,
      "learning_rate": 2.8987666666666667e-05,
      "loss": 0.0077,
      "step": 13040
    },
    {
      "epoch": 24.060941828254848,
      "grad_norm": 0.3569282591342926,
      "learning_rate": 2.8987333333333336e-05,
      "loss": 0.0081,
      "step": 13041
    },
    {
      "epoch": 24.062788550323177,
      "grad_norm": 0.11440275609493256,
      "learning_rate": 2.8987000000000002e-05,
      "loss": 0.0015,
      "step": 13042
    },
    {
      "epoch": 24.064635272391506,
      "grad_norm": 0.11149106919765472,
      "learning_rate": 2.8986666666666668e-05,
      "loss": 0.0027,
      "step": 13043
    },
    {
      "epoch": 24.066481994459835,
      "grad_norm": 0.576856255531311,
      "learning_rate": 2.8986333333333334e-05,
      "loss": 0.0067,
      "step": 13044
    },
    {
      "epoch": 24.068328716528164,
      "grad_norm": 0.07895402610301971,
      "learning_rate": 2.8986e-05,
      "loss": 0.0019,
      "step": 13045
    },
    {
      "epoch": 24.07017543859649,
      "grad_norm": 0.15396630764007568,
      "learning_rate": 2.8985666666666666e-05,
      "loss": 0.0052,
      "step": 13046
    },
    {
      "epoch": 24.07202216066482,
      "grad_norm": 0.1727500557899475,
      "learning_rate": 2.8985333333333335e-05,
      "loss": 0.0033,
      "step": 13047
    },
    {
      "epoch": 24.073868882733148,
      "grad_norm": 1.133514165878296,
      "learning_rate": 2.8984999999999998e-05,
      "loss": 0.0144,
      "step": 13048
    },
    {
      "epoch": 24.075715604801477,
      "grad_norm": 0.7800816297531128,
      "learning_rate": 2.8984666666666667e-05,
      "loss": 0.0094,
      "step": 13049
    },
    {
      "epoch": 24.077562326869806,
      "grad_norm": 0.13474318385124207,
      "learning_rate": 2.8984333333333336e-05,
      "loss": 0.0057,
      "step": 13050
    },
    {
      "epoch": 24.079409048938135,
      "grad_norm": 0.20290903747081757,
      "learning_rate": 2.8984e-05,
      "loss": 0.0098,
      "step": 13051
    },
    {
      "epoch": 24.081255771006465,
      "grad_norm": 0.22160065174102783,
      "learning_rate": 2.8983666666666668e-05,
      "loss": 0.0046,
      "step": 13052
    },
    {
      "epoch": 24.083102493074794,
      "grad_norm": 0.5779085755348206,
      "learning_rate": 2.8983333333333334e-05,
      "loss": 0.0111,
      "step": 13053
    },
    {
      "epoch": 24.08494921514312,
      "grad_norm": 0.39063599705696106,
      "learning_rate": 2.8983e-05,
      "loss": 0.0076,
      "step": 13054
    },
    {
      "epoch": 24.08679593721145,
      "grad_norm": 0.34382039308547974,
      "learning_rate": 2.8982666666666666e-05,
      "loss": 0.008,
      "step": 13055
    },
    {
      "epoch": 24.088642659279778,
      "grad_norm": 0.10230110585689545,
      "learning_rate": 2.8982333333333335e-05,
      "loss": 0.0027,
      "step": 13056
    },
    {
      "epoch": 24.090489381348107,
      "grad_norm": 0.6774642467498779,
      "learning_rate": 2.8981999999999998e-05,
      "loss": 0.0165,
      "step": 13057
    },
    {
      "epoch": 24.092336103416436,
      "grad_norm": 0.2824467122554779,
      "learning_rate": 2.8981666666666667e-05,
      "loss": 0.0098,
      "step": 13058
    },
    {
      "epoch": 24.094182825484765,
      "grad_norm": 0.5672895908355713,
      "learning_rate": 2.8981333333333337e-05,
      "loss": 0.1176,
      "step": 13059
    },
    {
      "epoch": 24.096029547553094,
      "grad_norm": 0.3753058910369873,
      "learning_rate": 2.8981e-05,
      "loss": 0.1258,
      "step": 13060
    },
    {
      "epoch": 24.097876269621423,
      "grad_norm": 0.5236627459526062,
      "learning_rate": 2.898066666666667e-05,
      "loss": 0.0992,
      "step": 13061
    },
    {
      "epoch": 24.099722991689752,
      "grad_norm": 0.46382221579551697,
      "learning_rate": 2.8980333333333334e-05,
      "loss": 0.0433,
      "step": 13062
    },
    {
      "epoch": 24.101569713758078,
      "grad_norm": 0.3146475553512573,
      "learning_rate": 2.898e-05,
      "loss": 0.0483,
      "step": 13063
    },
    {
      "epoch": 24.103416435826407,
      "grad_norm": 0.4047550857067108,
      "learning_rate": 2.8979666666666666e-05,
      "loss": 0.0892,
      "step": 13064
    },
    {
      "epoch": 24.105263157894736,
      "grad_norm": 0.38300085067749023,
      "learning_rate": 2.8979333333333336e-05,
      "loss": 0.0531,
      "step": 13065
    },
    {
      "epoch": 24.107109879963065,
      "grad_norm": 0.4566096067428589,
      "learning_rate": 2.8979e-05,
      "loss": 0.0476,
      "step": 13066
    },
    {
      "epoch": 24.108956602031395,
      "grad_norm": 0.36871248483657837,
      "learning_rate": 2.8978666666666667e-05,
      "loss": 0.0597,
      "step": 13067
    },
    {
      "epoch": 24.110803324099724,
      "grad_norm": 0.42504021525382996,
      "learning_rate": 2.8978333333333337e-05,
      "loss": 0.0416,
      "step": 13068
    },
    {
      "epoch": 24.112650046168053,
      "grad_norm": 0.744490385055542,
      "learning_rate": 2.8978e-05,
      "loss": 0.0171,
      "step": 13069
    },
    {
      "epoch": 24.114496768236382,
      "grad_norm": 0.32120198011398315,
      "learning_rate": 2.897766666666667e-05,
      "loss": 0.0374,
      "step": 13070
    },
    {
      "epoch": 24.116343490304708,
      "grad_norm": 0.40598228573799133,
      "learning_rate": 2.8977333333333335e-05,
      "loss": 0.0406,
      "step": 13071
    },
    {
      "epoch": 24.118190212373037,
      "grad_norm": 0.44137951731681824,
      "learning_rate": 2.8977e-05,
      "loss": 0.0151,
      "step": 13072
    },
    {
      "epoch": 24.120036934441366,
      "grad_norm": 0.1768002212047577,
      "learning_rate": 2.8976666666666666e-05,
      "loss": 0.0071,
      "step": 13073
    },
    {
      "epoch": 24.121883656509695,
      "grad_norm": 0.13510826230049133,
      "learning_rate": 2.8976333333333332e-05,
      "loss": 0.0054,
      "step": 13074
    },
    {
      "epoch": 24.123730378578024,
      "grad_norm": 0.21818716824054718,
      "learning_rate": 2.8976000000000002e-05,
      "loss": 0.0068,
      "step": 13075
    },
    {
      "epoch": 24.125577100646353,
      "grad_norm": 0.16706795990467072,
      "learning_rate": 2.8975666666666668e-05,
      "loss": 0.0062,
      "step": 13076
    },
    {
      "epoch": 24.127423822714682,
      "grad_norm": 0.31920814514160156,
      "learning_rate": 2.8975333333333334e-05,
      "loss": 0.0098,
      "step": 13077
    },
    {
      "epoch": 24.12927054478301,
      "grad_norm": 0.2236560881137848,
      "learning_rate": 2.8975e-05,
      "loss": 0.0141,
      "step": 13078
    },
    {
      "epoch": 24.131117266851337,
      "grad_norm": 0.1601002961397171,
      "learning_rate": 2.897466666666667e-05,
      "loss": 0.005,
      "step": 13079
    },
    {
      "epoch": 24.132963988919666,
      "grad_norm": 0.13643929362297058,
      "learning_rate": 2.897433333333333e-05,
      "loss": 0.0048,
      "step": 13080
    },
    {
      "epoch": 24.134810710987995,
      "grad_norm": 0.24548067152500153,
      "learning_rate": 2.8974e-05,
      "loss": 0.0048,
      "step": 13081
    },
    {
      "epoch": 24.136657433056325,
      "grad_norm": 0.3406643569469452,
      "learning_rate": 2.8973666666666667e-05,
      "loss": 0.0129,
      "step": 13082
    },
    {
      "epoch": 24.138504155124654,
      "grad_norm": 0.362541526556015,
      "learning_rate": 2.8973333333333333e-05,
      "loss": 0.0058,
      "step": 13083
    },
    {
      "epoch": 24.140350877192983,
      "grad_norm": 0.21052448451519012,
      "learning_rate": 2.8973000000000002e-05,
      "loss": 0.0081,
      "step": 13084
    },
    {
      "epoch": 24.142197599261312,
      "grad_norm": 0.3793521523475647,
      "learning_rate": 2.8972666666666668e-05,
      "loss": 0.0037,
      "step": 13085
    },
    {
      "epoch": 24.14404432132964,
      "grad_norm": 0.11015784740447998,
      "learning_rate": 2.8972333333333334e-05,
      "loss": 0.0026,
      "step": 13086
    },
    {
      "epoch": 24.14589104339797,
      "grad_norm": 0.08412471413612366,
      "learning_rate": 2.8972e-05,
      "loss": 0.0023,
      "step": 13087
    },
    {
      "epoch": 24.147737765466296,
      "grad_norm": 0.2851986587047577,
      "learning_rate": 2.897166666666667e-05,
      "loss": 0.0031,
      "step": 13088
    },
    {
      "epoch": 24.149584487534625,
      "grad_norm": 0.1866234540939331,
      "learning_rate": 2.897133333333333e-05,
      "loss": 0.0027,
      "step": 13089
    },
    {
      "epoch": 24.151431209602954,
      "grad_norm": 0.10915181040763855,
      "learning_rate": 2.8971e-05,
      "loss": 0.0021,
      "step": 13090
    },
    {
      "epoch": 24.153277931671283,
      "grad_norm": 0.3201306164264679,
      "learning_rate": 2.897066666666667e-05,
      "loss": 0.0069,
      "step": 13091
    },
    {
      "epoch": 24.155124653739612,
      "grad_norm": 0.538478672504425,
      "learning_rate": 2.8970333333333333e-05,
      "loss": 0.0042,
      "step": 13092
    },
    {
      "epoch": 24.15697137580794,
      "grad_norm": 0.12584218382835388,
      "learning_rate": 2.8970000000000002e-05,
      "loss": 0.0036,
      "step": 13093
    },
    {
      "epoch": 24.15881809787627,
      "grad_norm": 0.2512202262878418,
      "learning_rate": 2.8969666666666668e-05,
      "loss": 0.0068,
      "step": 13094
    },
    {
      "epoch": 24.1606648199446,
      "grad_norm": 0.3698013424873352,
      "learning_rate": 2.8969333333333334e-05,
      "loss": 0.0074,
      "step": 13095
    },
    {
      "epoch": 24.162511542012926,
      "grad_norm": 0.13004982471466064,
      "learning_rate": 2.8969e-05,
      "loss": 0.0025,
      "step": 13096
    },
    {
      "epoch": 24.164358264081255,
      "grad_norm": 0.44549521803855896,
      "learning_rate": 2.896866666666667e-05,
      "loss": 0.0107,
      "step": 13097
    },
    {
      "epoch": 24.166204986149584,
      "grad_norm": 0.33616870641708374,
      "learning_rate": 2.8968333333333332e-05,
      "loss": 0.0087,
      "step": 13098
    },
    {
      "epoch": 24.168051708217913,
      "grad_norm": 0.114255391061306,
      "learning_rate": 2.8968e-05,
      "loss": 0.0017,
      "step": 13099
    },
    {
      "epoch": 24.169898430286242,
      "grad_norm": 0.8930944204330444,
      "learning_rate": 2.896766666666667e-05,
      "loss": 0.0139,
      "step": 13100
    },
    {
      "epoch": 24.17174515235457,
      "grad_norm": 0.6546629071235657,
      "learning_rate": 2.8967333333333333e-05,
      "loss": 0.0122,
      "step": 13101
    },
    {
      "epoch": 24.1735918744229,
      "grad_norm": 0.4253278076648712,
      "learning_rate": 2.8967000000000002e-05,
      "loss": 0.0088,
      "step": 13102
    },
    {
      "epoch": 24.17543859649123,
      "grad_norm": 0.2592533528804779,
      "learning_rate": 2.8966666666666665e-05,
      "loss": 0.0072,
      "step": 13103
    },
    {
      "epoch": 24.177285318559555,
      "grad_norm": 0.24567706882953644,
      "learning_rate": 2.8966333333333334e-05,
      "loss": 0.0038,
      "step": 13104
    },
    {
      "epoch": 24.179132040627884,
      "grad_norm": 0.21996194124221802,
      "learning_rate": 2.8966e-05,
      "loss": 0.0035,
      "step": 13105
    },
    {
      "epoch": 24.180978762696213,
      "grad_norm": 0.5467721223831177,
      "learning_rate": 2.8965666666666666e-05,
      "loss": 0.0069,
      "step": 13106
    },
    {
      "epoch": 24.182825484764543,
      "grad_norm": 0.19381636381149292,
      "learning_rate": 2.8965333333333332e-05,
      "loss": 0.0048,
      "step": 13107
    },
    {
      "epoch": 24.18467220683287,
      "grad_norm": 0.15538102388381958,
      "learning_rate": 2.8965e-05,
      "loss": 0.0124,
      "step": 13108
    },
    {
      "epoch": 24.1865189289012,
      "grad_norm": 0.6547185182571411,
      "learning_rate": 2.8964666666666667e-05,
      "loss": 0.1145,
      "step": 13109
    },
    {
      "epoch": 24.18836565096953,
      "grad_norm": 0.612529456615448,
      "learning_rate": 2.8964333333333333e-05,
      "loss": 0.1432,
      "step": 13110
    },
    {
      "epoch": 24.19021237303786,
      "grad_norm": 0.5609632730484009,
      "learning_rate": 2.8964000000000003e-05,
      "loss": 0.1157,
      "step": 13111
    },
    {
      "epoch": 24.19205909510619,
      "grad_norm": 0.3962590992450714,
      "learning_rate": 2.8963666666666665e-05,
      "loss": 0.1006,
      "step": 13112
    },
    {
      "epoch": 24.193905817174514,
      "grad_norm": 0.3930540382862091,
      "learning_rate": 2.8963333333333334e-05,
      "loss": 0.041,
      "step": 13113
    },
    {
      "epoch": 24.195752539242843,
      "grad_norm": 0.5370532870292664,
      "learning_rate": 2.8963e-05,
      "loss": 0.0613,
      "step": 13114
    },
    {
      "epoch": 24.197599261311172,
      "grad_norm": 0.5153535008430481,
      "learning_rate": 2.8962666666666666e-05,
      "loss": 0.0355,
      "step": 13115
    },
    {
      "epoch": 24.1994459833795,
      "grad_norm": 0.4103752374649048,
      "learning_rate": 2.8962333333333336e-05,
      "loss": 0.049,
      "step": 13116
    },
    {
      "epoch": 24.20129270544783,
      "grad_norm": 0.35563552379608154,
      "learning_rate": 2.8962e-05,
      "loss": 0.0612,
      "step": 13117
    },
    {
      "epoch": 24.20313942751616,
      "grad_norm": 0.3967292606830597,
      "learning_rate": 2.8961666666666667e-05,
      "loss": 0.0478,
      "step": 13118
    },
    {
      "epoch": 24.20498614958449,
      "grad_norm": 0.42570221424102783,
      "learning_rate": 2.8961333333333333e-05,
      "loss": 0.0329,
      "step": 13119
    },
    {
      "epoch": 24.206832871652818,
      "grad_norm": 0.34922346472740173,
      "learning_rate": 2.8961000000000003e-05,
      "loss": 0.0428,
      "step": 13120
    },
    {
      "epoch": 24.208679593721143,
      "grad_norm": 0.46752482652664185,
      "learning_rate": 2.8960666666666665e-05,
      "loss": 0.0234,
      "step": 13121
    },
    {
      "epoch": 24.210526315789473,
      "grad_norm": 0.7021492719650269,
      "learning_rate": 2.8960333333333335e-05,
      "loss": 0.015,
      "step": 13122
    },
    {
      "epoch": 24.2123730378578,
      "grad_norm": 0.28904110193252563,
      "learning_rate": 2.896e-05,
      "loss": 0.0157,
      "step": 13123
    },
    {
      "epoch": 24.21421975992613,
      "grad_norm": 0.36432945728302,
      "learning_rate": 2.8959666666666666e-05,
      "loss": 0.0299,
      "step": 13124
    },
    {
      "epoch": 24.21606648199446,
      "grad_norm": 0.26848480105400085,
      "learning_rate": 2.8959333333333336e-05,
      "loss": 0.009,
      "step": 13125
    },
    {
      "epoch": 24.21791320406279,
      "grad_norm": 0.19157792627811432,
      "learning_rate": 2.8959000000000002e-05,
      "loss": 0.0183,
      "step": 13126
    },
    {
      "epoch": 24.21975992613112,
      "grad_norm": 0.35590115189552307,
      "learning_rate": 2.8958666666666668e-05,
      "loss": 0.009,
      "step": 13127
    },
    {
      "epoch": 24.221606648199447,
      "grad_norm": 0.46946394443511963,
      "learning_rate": 2.8958333333333334e-05,
      "loss": 0.0125,
      "step": 13128
    },
    {
      "epoch": 24.223453370267773,
      "grad_norm": 0.3206366300582886,
      "learning_rate": 2.8958000000000003e-05,
      "loss": 0.0122,
      "step": 13129
    },
    {
      "epoch": 24.225300092336102,
      "grad_norm": 0.19763174653053284,
      "learning_rate": 2.8957666666666665e-05,
      "loss": 0.0051,
      "step": 13130
    },
    {
      "epoch": 24.22714681440443,
      "grad_norm": 0.18738523125648499,
      "learning_rate": 2.8957333333333335e-05,
      "loss": 0.0066,
      "step": 13131
    },
    {
      "epoch": 24.22899353647276,
      "grad_norm": 0.554878830909729,
      "learning_rate": 2.8957e-05,
      "loss": 0.0073,
      "step": 13132
    },
    {
      "epoch": 24.23084025854109,
      "grad_norm": 0.3369830548763275,
      "learning_rate": 2.8956666666666667e-05,
      "loss": 0.0064,
      "step": 13133
    },
    {
      "epoch": 24.23268698060942,
      "grad_norm": 0.3536060154438019,
      "learning_rate": 2.8956333333333336e-05,
      "loss": 0.006,
      "step": 13134
    },
    {
      "epoch": 24.234533702677748,
      "grad_norm": 0.30617913603782654,
      "learning_rate": 2.8956e-05,
      "loss": 0.0182,
      "step": 13135
    },
    {
      "epoch": 24.236380424746077,
      "grad_norm": 0.4099351465702057,
      "learning_rate": 2.8955666666666668e-05,
      "loss": 0.0025,
      "step": 13136
    },
    {
      "epoch": 24.238227146814406,
      "grad_norm": 0.2719319760799408,
      "learning_rate": 2.8955333333333334e-05,
      "loss": 0.0076,
      "step": 13137
    },
    {
      "epoch": 24.24007386888273,
      "grad_norm": 0.29288655519485474,
      "learning_rate": 2.8955e-05,
      "loss": 0.0071,
      "step": 13138
    },
    {
      "epoch": 24.24192059095106,
      "grad_norm": 0.36805543303489685,
      "learning_rate": 2.8954666666666666e-05,
      "loss": 0.0101,
      "step": 13139
    },
    {
      "epoch": 24.24376731301939,
      "grad_norm": 0.1612132489681244,
      "learning_rate": 2.8954333333333335e-05,
      "loss": 0.0045,
      "step": 13140
    },
    {
      "epoch": 24.24561403508772,
      "grad_norm": 0.17175914347171783,
      "learning_rate": 2.8954e-05,
      "loss": 0.0039,
      "step": 13141
    },
    {
      "epoch": 24.24746075715605,
      "grad_norm": 0.15694212913513184,
      "learning_rate": 2.8953666666666667e-05,
      "loss": 0.0041,
      "step": 13142
    },
    {
      "epoch": 24.249307479224377,
      "grad_norm": 0.1983942985534668,
      "learning_rate": 2.8953333333333336e-05,
      "loss": 0.0032,
      "step": 13143
    },
    {
      "epoch": 24.251154201292707,
      "grad_norm": 0.42327597737312317,
      "learning_rate": 2.8953e-05,
      "loss": 0.0112,
      "step": 13144
    },
    {
      "epoch": 24.253000923361036,
      "grad_norm": 0.29380321502685547,
      "learning_rate": 2.8952666666666668e-05,
      "loss": 0.0289,
      "step": 13145
    },
    {
      "epoch": 24.25484764542936,
      "grad_norm": 0.21177636086940765,
      "learning_rate": 2.8952333333333334e-05,
      "loss": 0.005,
      "step": 13146
    },
    {
      "epoch": 24.25669436749769,
      "grad_norm": 1.7719300985336304,
      "learning_rate": 2.8952e-05,
      "loss": 0.0124,
      "step": 13147
    },
    {
      "epoch": 24.25854108956602,
      "grad_norm": 0.41852688789367676,
      "learning_rate": 2.8951666666666666e-05,
      "loss": 0.0088,
      "step": 13148
    },
    {
      "epoch": 24.26038781163435,
      "grad_norm": 0.1518332064151764,
      "learning_rate": 2.8951333333333335e-05,
      "loss": 0.0046,
      "step": 13149
    },
    {
      "epoch": 24.262234533702678,
      "grad_norm": 0.09099515527486801,
      "learning_rate": 2.8951e-05,
      "loss": 0.0017,
      "step": 13150
    },
    {
      "epoch": 24.264081255771007,
      "grad_norm": 0.1368519365787506,
      "learning_rate": 2.8950666666666667e-05,
      "loss": 0.0021,
      "step": 13151
    },
    {
      "epoch": 24.265927977839336,
      "grad_norm": 0.27972957491874695,
      "learning_rate": 2.8950333333333336e-05,
      "loss": 0.0131,
      "step": 13152
    },
    {
      "epoch": 24.267774699907665,
      "grad_norm": 0.39006727933883667,
      "learning_rate": 2.895e-05,
      "loss": 0.0109,
      "step": 13153
    },
    {
      "epoch": 24.26962142197599,
      "grad_norm": 0.23203179240226746,
      "learning_rate": 2.8949666666666668e-05,
      "loss": 0.0073,
      "step": 13154
    },
    {
      "epoch": 24.27146814404432,
      "grad_norm": 0.43176040053367615,
      "learning_rate": 2.8949333333333334e-05,
      "loss": 0.0131,
      "step": 13155
    },
    {
      "epoch": 24.27331486611265,
      "grad_norm": 0.1430387645959854,
      "learning_rate": 2.8949e-05,
      "loss": 0.0032,
      "step": 13156
    },
    {
      "epoch": 24.27516158818098,
      "grad_norm": 0.32787269353866577,
      "learning_rate": 2.8948666666666666e-05,
      "loss": 0.008,
      "step": 13157
    },
    {
      "epoch": 24.277008310249307,
      "grad_norm": 0.3612595498561859,
      "learning_rate": 2.8948333333333335e-05,
      "loss": 0.0124,
      "step": 13158
    },
    {
      "epoch": 24.278855032317637,
      "grad_norm": 0.8463873863220215,
      "learning_rate": 2.8948e-05,
      "loss": 0.1127,
      "step": 13159
    },
    {
      "epoch": 24.280701754385966,
      "grad_norm": 0.453067421913147,
      "learning_rate": 2.8947666666666667e-05,
      "loss": 0.0955,
      "step": 13160
    },
    {
      "epoch": 24.282548476454295,
      "grad_norm": 0.4151947796344757,
      "learning_rate": 2.8947333333333337e-05,
      "loss": 0.068,
      "step": 13161
    },
    {
      "epoch": 24.284395198522624,
      "grad_norm": 0.3739778995513916,
      "learning_rate": 2.8947e-05,
      "loss": 0.0635,
      "step": 13162
    },
    {
      "epoch": 24.28624192059095,
      "grad_norm": 0.486072838306427,
      "learning_rate": 2.894666666666667e-05,
      "loss": 0.0461,
      "step": 13163
    },
    {
      "epoch": 24.28808864265928,
      "grad_norm": 0.33261334896087646,
      "learning_rate": 2.894633333333333e-05,
      "loss": 0.048,
      "step": 13164
    },
    {
      "epoch": 24.289935364727608,
      "grad_norm": 0.3986312448978424,
      "learning_rate": 2.8946e-05,
      "loss": 0.0442,
      "step": 13165
    },
    {
      "epoch": 24.291782086795937,
      "grad_norm": 0.5497700572013855,
      "learning_rate": 2.894566666666667e-05,
      "loss": 0.0656,
      "step": 13166
    },
    {
      "epoch": 24.293628808864266,
      "grad_norm": 0.41510388255119324,
      "learning_rate": 2.8945333333333332e-05,
      "loss": 0.058,
      "step": 13167
    },
    {
      "epoch": 24.295475530932595,
      "grad_norm": 0.28671392798423767,
      "learning_rate": 2.8945e-05,
      "loss": 0.031,
      "step": 13168
    },
    {
      "epoch": 24.297322253000925,
      "grad_norm": 0.8106816411018372,
      "learning_rate": 2.8944666666666667e-05,
      "loss": 0.0557,
      "step": 13169
    },
    {
      "epoch": 24.299168975069254,
      "grad_norm": 0.26283910870552063,
      "learning_rate": 2.8944333333333333e-05,
      "loss": 0.021,
      "step": 13170
    },
    {
      "epoch": 24.30101569713758,
      "grad_norm": 0.3150317370891571,
      "learning_rate": 2.8944e-05,
      "loss": 0.0208,
      "step": 13171
    },
    {
      "epoch": 24.30286241920591,
      "grad_norm": 0.448843389749527,
      "learning_rate": 2.894366666666667e-05,
      "loss": 0.0123,
      "step": 13172
    },
    {
      "epoch": 24.304709141274238,
      "grad_norm": 0.19662699103355408,
      "learning_rate": 2.894333333333333e-05,
      "loss": 0.0088,
      "step": 13173
    },
    {
      "epoch": 24.306555863342567,
      "grad_norm": 0.2782691717147827,
      "learning_rate": 2.8943e-05,
      "loss": 0.0099,
      "step": 13174
    },
    {
      "epoch": 24.308402585410896,
      "grad_norm": 0.12477515637874603,
      "learning_rate": 2.894266666666667e-05,
      "loss": 0.0048,
      "step": 13175
    },
    {
      "epoch": 24.310249307479225,
      "grad_norm": 0.44106438755989075,
      "learning_rate": 2.8942333333333332e-05,
      "loss": 0.009,
      "step": 13176
    },
    {
      "epoch": 24.312096029547554,
      "grad_norm": 0.15457753837108612,
      "learning_rate": 2.8942000000000002e-05,
      "loss": 0.0067,
      "step": 13177
    },
    {
      "epoch": 24.313942751615883,
      "grad_norm": 0.14934605360031128,
      "learning_rate": 2.8941666666666668e-05,
      "loss": 0.0059,
      "step": 13178
    },
    {
      "epoch": 24.31578947368421,
      "grad_norm": 0.15137791633605957,
      "learning_rate": 2.8941333333333334e-05,
      "loss": 0.0039,
      "step": 13179
    },
    {
      "epoch": 24.317636195752538,
      "grad_norm": 0.21593965590000153,
      "learning_rate": 2.8941e-05,
      "loss": 0.0085,
      "step": 13180
    },
    {
      "epoch": 24.319482917820867,
      "grad_norm": 0.2960059344768524,
      "learning_rate": 2.894066666666667e-05,
      "loss": 0.0063,
      "step": 13181
    },
    {
      "epoch": 24.321329639889196,
      "grad_norm": 0.13511642813682556,
      "learning_rate": 2.894033333333333e-05,
      "loss": 0.0022,
      "step": 13182
    },
    {
      "epoch": 24.323176361957525,
      "grad_norm": 0.22784291207790375,
      "learning_rate": 2.894e-05,
      "loss": 0.0063,
      "step": 13183
    },
    {
      "epoch": 24.325023084025855,
      "grad_norm": 0.11898264288902283,
      "learning_rate": 2.893966666666667e-05,
      "loss": 0.004,
      "step": 13184
    },
    {
      "epoch": 24.326869806094184,
      "grad_norm": 0.23218907415866852,
      "learning_rate": 2.8939333333333333e-05,
      "loss": 0.0049,
      "step": 13185
    },
    {
      "epoch": 24.328716528162513,
      "grad_norm": 0.2760145664215088,
      "learning_rate": 2.8939000000000002e-05,
      "loss": 0.0036,
      "step": 13186
    },
    {
      "epoch": 24.330563250230842,
      "grad_norm": 0.22240634262561798,
      "learning_rate": 2.8938666666666668e-05,
      "loss": 0.0041,
      "step": 13187
    },
    {
      "epoch": 24.332409972299168,
      "grad_norm": 0.18136060237884521,
      "learning_rate": 2.8938333333333334e-05,
      "loss": 0.0048,
      "step": 13188
    },
    {
      "epoch": 24.334256694367497,
      "grad_norm": 0.31389379501342773,
      "learning_rate": 2.8938e-05,
      "loss": 0.0049,
      "step": 13189
    },
    {
      "epoch": 24.336103416435826,
      "grad_norm": 0.2006513774394989,
      "learning_rate": 2.893766666666667e-05,
      "loss": 0.0063,
      "step": 13190
    },
    {
      "epoch": 24.337950138504155,
      "grad_norm": 0.3250141143798828,
      "learning_rate": 2.8937333333333335e-05,
      "loss": 0.0057,
      "step": 13191
    },
    {
      "epoch": 24.339796860572484,
      "grad_norm": 0.1399124562740326,
      "learning_rate": 2.8937e-05,
      "loss": 0.0055,
      "step": 13192
    },
    {
      "epoch": 24.341643582640813,
      "grad_norm": 0.36624839901924133,
      "learning_rate": 2.893666666666667e-05,
      "loss": 0.008,
      "step": 13193
    },
    {
      "epoch": 24.343490304709142,
      "grad_norm": 0.2624996304512024,
      "learning_rate": 2.8936333333333333e-05,
      "loss": 0.0142,
      "step": 13194
    },
    {
      "epoch": 24.34533702677747,
      "grad_norm": 0.16521060466766357,
      "learning_rate": 2.8936000000000002e-05,
      "loss": 0.0033,
      "step": 13195
    },
    {
      "epoch": 24.347183748845797,
      "grad_norm": 0.1808953732252121,
      "learning_rate": 2.8935666666666665e-05,
      "loss": 0.0018,
      "step": 13196
    },
    {
      "epoch": 24.349030470914126,
      "grad_norm": 0.2830214500427246,
      "learning_rate": 2.8935333333333334e-05,
      "loss": 0.0082,
      "step": 13197
    },
    {
      "epoch": 24.350877192982455,
      "grad_norm": 0.8021502494812012,
      "learning_rate": 2.8935e-05,
      "loss": 0.0118,
      "step": 13198
    },
    {
      "epoch": 24.352723915050785,
      "grad_norm": 0.28621774911880493,
      "learning_rate": 2.8934666666666666e-05,
      "loss": 0.0078,
      "step": 13199
    },
    {
      "epoch": 24.354570637119114,
      "grad_norm": 0.5483892560005188,
      "learning_rate": 2.8934333333333335e-05,
      "loss": 0.005,
      "step": 13200
    },
    {
      "epoch": 24.356417359187443,
      "grad_norm": 0.11460002511739731,
      "learning_rate": 2.8934e-05,
      "loss": 0.0026,
      "step": 13201
    },
    {
      "epoch": 24.358264081255772,
      "grad_norm": 0.22816605865955353,
      "learning_rate": 2.8933666666666667e-05,
      "loss": 0.0056,
      "step": 13202
    },
    {
      "epoch": 24.3601108033241,
      "grad_norm": 0.06393951922655106,
      "learning_rate": 2.8933333333333333e-05,
      "loss": 0.0011,
      "step": 13203
    },
    {
      "epoch": 24.361957525392427,
      "grad_norm": 0.26738423109054565,
      "learning_rate": 2.8933000000000002e-05,
      "loss": 0.0034,
      "step": 13204
    },
    {
      "epoch": 24.363804247460756,
      "grad_norm": 0.4886252284049988,
      "learning_rate": 2.8932666666666665e-05,
      "loss": 0.0043,
      "step": 13205
    },
    {
      "epoch": 24.365650969529085,
      "grad_norm": 0.3189277648925781,
      "learning_rate": 2.8932333333333334e-05,
      "loss": 0.0057,
      "step": 13206
    },
    {
      "epoch": 24.367497691597414,
      "grad_norm": 0.1422029286623001,
      "learning_rate": 2.8932e-05,
      "loss": 0.004,
      "step": 13207
    },
    {
      "epoch": 24.369344413665743,
      "grad_norm": 0.07584530115127563,
      "learning_rate": 2.8931666666666666e-05,
      "loss": 0.001,
      "step": 13208
    },
    {
      "epoch": 24.371191135734072,
      "grad_norm": 0.9531792998313904,
      "learning_rate": 2.8931333333333335e-05,
      "loss": 0.1437,
      "step": 13209
    },
    {
      "epoch": 24.3730378578024,
      "grad_norm": 0.6607851982116699,
      "learning_rate": 2.8931e-05,
      "loss": 0.124,
      "step": 13210
    },
    {
      "epoch": 24.37488457987073,
      "grad_norm": 0.43645626306533813,
      "learning_rate": 2.8930666666666667e-05,
      "loss": 0.0584,
      "step": 13211
    },
    {
      "epoch": 24.37673130193906,
      "grad_norm": 0.7066429853439331,
      "learning_rate": 2.8930333333333333e-05,
      "loss": 0.1331,
      "step": 13212
    },
    {
      "epoch": 24.378578024007385,
      "grad_norm": 0.6537086963653564,
      "learning_rate": 2.8930000000000003e-05,
      "loss": 0.0721,
      "step": 13213
    },
    {
      "epoch": 24.380424746075715,
      "grad_norm": 0.481343150138855,
      "learning_rate": 2.8929666666666665e-05,
      "loss": 0.0539,
      "step": 13214
    },
    {
      "epoch": 24.382271468144044,
      "grad_norm": 1.752601146697998,
      "learning_rate": 2.8929333333333334e-05,
      "loss": 0.0644,
      "step": 13215
    },
    {
      "epoch": 24.384118190212373,
      "grad_norm": 0.49992284178733826,
      "learning_rate": 2.8929000000000004e-05,
      "loss": 0.0564,
      "step": 13216
    },
    {
      "epoch": 24.385964912280702,
      "grad_norm": 0.4543759822845459,
      "learning_rate": 2.8928666666666666e-05,
      "loss": 0.0531,
      "step": 13217
    },
    {
      "epoch": 24.38781163434903,
      "grad_norm": 0.4484714865684509,
      "learning_rate": 2.8928333333333336e-05,
      "loss": 0.0298,
      "step": 13218
    },
    {
      "epoch": 24.38965835641736,
      "grad_norm": 0.5918431878089905,
      "learning_rate": 2.8928e-05,
      "loss": 0.044,
      "step": 13219
    },
    {
      "epoch": 24.39150507848569,
      "grad_norm": 0.29719340801239014,
      "learning_rate": 2.8927666666666667e-05,
      "loss": 0.0324,
      "step": 13220
    },
    {
      "epoch": 24.393351800554015,
      "grad_norm": 0.20278289914131165,
      "learning_rate": 2.8927333333333333e-05,
      "loss": 0.0549,
      "step": 13221
    },
    {
      "epoch": 24.395198522622344,
      "grad_norm": 0.2530817985534668,
      "learning_rate": 2.8927000000000003e-05,
      "loss": 0.0388,
      "step": 13222
    },
    {
      "epoch": 24.397045244690673,
      "grad_norm": 0.5364762544631958,
      "learning_rate": 2.8926666666666665e-05,
      "loss": 0.0477,
      "step": 13223
    },
    {
      "epoch": 24.398891966759003,
      "grad_norm": 0.7806441187858582,
      "learning_rate": 2.8926333333333335e-05,
      "loss": 0.0251,
      "step": 13224
    },
    {
      "epoch": 24.40073868882733,
      "grad_norm": 0.29668524861335754,
      "learning_rate": 2.8926e-05,
      "loss": 0.0114,
      "step": 13225
    },
    {
      "epoch": 24.40258541089566,
      "grad_norm": 0.282160222530365,
      "learning_rate": 2.8925666666666666e-05,
      "loss": 0.0141,
      "step": 13226
    },
    {
      "epoch": 24.40443213296399,
      "grad_norm": 0.24868442118167877,
      "learning_rate": 2.8925333333333336e-05,
      "loss": 0.0102,
      "step": 13227
    },
    {
      "epoch": 24.40627885503232,
      "grad_norm": 0.30800092220306396,
      "learning_rate": 2.8925e-05,
      "loss": 0.0074,
      "step": 13228
    },
    {
      "epoch": 24.408125577100645,
      "grad_norm": 0.23089873790740967,
      "learning_rate": 2.8924666666666668e-05,
      "loss": 0.021,
      "step": 13229
    },
    {
      "epoch": 24.409972299168974,
      "grad_norm": 0.33230042457580566,
      "learning_rate": 2.8924333333333334e-05,
      "loss": 0.0067,
      "step": 13230
    },
    {
      "epoch": 24.411819021237303,
      "grad_norm": 0.40151605010032654,
      "learning_rate": 2.8924e-05,
      "loss": 0.0063,
      "step": 13231
    },
    {
      "epoch": 24.413665743305632,
      "grad_norm": 0.29878801107406616,
      "learning_rate": 2.8923666666666665e-05,
      "loss": 0.0052,
      "step": 13232
    },
    {
      "epoch": 24.41551246537396,
      "grad_norm": 0.1386539191007614,
      "learning_rate": 2.8923333333333335e-05,
      "loss": 0.0035,
      "step": 13233
    },
    {
      "epoch": 24.41735918744229,
      "grad_norm": 0.20161989331245422,
      "learning_rate": 2.8923e-05,
      "loss": 0.0081,
      "step": 13234
    },
    {
      "epoch": 24.41920590951062,
      "grad_norm": 0.23625294864177704,
      "learning_rate": 2.8922666666666667e-05,
      "loss": 0.0042,
      "step": 13235
    },
    {
      "epoch": 24.42105263157895,
      "grad_norm": 0.45596843957901,
      "learning_rate": 2.8922333333333336e-05,
      "loss": 0.0075,
      "step": 13236
    },
    {
      "epoch": 24.422899353647278,
      "grad_norm": 0.2962850034236908,
      "learning_rate": 2.8922e-05,
      "loss": 0.0054,
      "step": 13237
    },
    {
      "epoch": 24.424746075715603,
      "grad_norm": 0.7074769139289856,
      "learning_rate": 2.8921666666666668e-05,
      "loss": 0.0161,
      "step": 13238
    },
    {
      "epoch": 24.426592797783933,
      "grad_norm": 0.22207657992839813,
      "learning_rate": 2.8921333333333334e-05,
      "loss": 0.0047,
      "step": 13239
    },
    {
      "epoch": 24.42843951985226,
      "grad_norm": 0.26032233238220215,
      "learning_rate": 2.8921e-05,
      "loss": 0.0067,
      "step": 13240
    },
    {
      "epoch": 24.43028624192059,
      "grad_norm": 0.15774919092655182,
      "learning_rate": 2.892066666666667e-05,
      "loss": 0.005,
      "step": 13241
    },
    {
      "epoch": 24.43213296398892,
      "grad_norm": 0.20989170670509338,
      "learning_rate": 2.8920333333333335e-05,
      "loss": 0.0039,
      "step": 13242
    },
    {
      "epoch": 24.43397968605725,
      "grad_norm": 0.25299087166786194,
      "learning_rate": 2.892e-05,
      "loss": 0.0044,
      "step": 13243
    },
    {
      "epoch": 24.43582640812558,
      "grad_norm": 0.31932535767555237,
      "learning_rate": 2.8919666666666667e-05,
      "loss": 0.0128,
      "step": 13244
    },
    {
      "epoch": 24.437673130193907,
      "grad_norm": 0.14626465737819672,
      "learning_rate": 2.8919333333333336e-05,
      "loss": 0.0029,
      "step": 13245
    },
    {
      "epoch": 24.439519852262233,
      "grad_norm": 0.2414906620979309,
      "learning_rate": 2.8919e-05,
      "loss": 0.0026,
      "step": 13246
    },
    {
      "epoch": 24.441366574330562,
      "grad_norm": 0.20157982409000397,
      "learning_rate": 2.8918666666666668e-05,
      "loss": 0.0074,
      "step": 13247
    },
    {
      "epoch": 24.44321329639889,
      "grad_norm": 0.14444953203201294,
      "learning_rate": 2.8918333333333334e-05,
      "loss": 0.0018,
      "step": 13248
    },
    {
      "epoch": 24.44506001846722,
      "grad_norm": 0.26020002365112305,
      "learning_rate": 2.8918e-05,
      "loss": 0.0115,
      "step": 13249
    },
    {
      "epoch": 24.44690674053555,
      "grad_norm": 0.18706028163433075,
      "learning_rate": 2.891766666666667e-05,
      "loss": 0.0071,
      "step": 13250
    },
    {
      "epoch": 24.44875346260388,
      "grad_norm": 0.7018864154815674,
      "learning_rate": 2.8917333333333335e-05,
      "loss": 0.0053,
      "step": 13251
    },
    {
      "epoch": 24.450600184672208,
      "grad_norm": 0.20267437398433685,
      "learning_rate": 2.8917e-05,
      "loss": 0.007,
      "step": 13252
    },
    {
      "epoch": 24.452446906740537,
      "grad_norm": 0.22838689386844635,
      "learning_rate": 2.8916666666666667e-05,
      "loss": 0.0072,
      "step": 13253
    },
    {
      "epoch": 24.454293628808863,
      "grad_norm": 0.30894041061401367,
      "learning_rate": 2.8916333333333336e-05,
      "loss": 0.0039,
      "step": 13254
    },
    {
      "epoch": 24.45614035087719,
      "grad_norm": 0.33067917823791504,
      "learning_rate": 2.8916e-05,
      "loss": 0.0084,
      "step": 13255
    },
    {
      "epoch": 24.45798707294552,
      "grad_norm": 0.10167045146226883,
      "learning_rate": 2.8915666666666668e-05,
      "loss": 0.003,
      "step": 13256
    },
    {
      "epoch": 24.45983379501385,
      "grad_norm": 0.37686604261398315,
      "learning_rate": 2.8915333333333334e-05,
      "loss": 0.014,
      "step": 13257
    },
    {
      "epoch": 24.46168051708218,
      "grad_norm": 0.5401965379714966,
      "learning_rate": 2.8915e-05,
      "loss": 0.0052,
      "step": 13258
    },
    {
      "epoch": 24.46352723915051,
      "grad_norm": 0.7932718992233276,
      "learning_rate": 2.891466666666667e-05,
      "loss": 0.1161,
      "step": 13259
    },
    {
      "epoch": 24.465373961218837,
      "grad_norm": 0.6346531510353088,
      "learning_rate": 2.8914333333333332e-05,
      "loss": 0.1108,
      "step": 13260
    },
    {
      "epoch": 24.467220683287167,
      "grad_norm": 0.6250077486038208,
      "learning_rate": 2.8914e-05,
      "loss": 0.1033,
      "step": 13261
    },
    {
      "epoch": 24.469067405355496,
      "grad_norm": 0.5043835639953613,
      "learning_rate": 2.8913666666666667e-05,
      "loss": 0.0862,
      "step": 13262
    },
    {
      "epoch": 24.47091412742382,
      "grad_norm": 0.3136883080005646,
      "learning_rate": 2.8913333333333333e-05,
      "loss": 0.0542,
      "step": 13263
    },
    {
      "epoch": 24.47276084949215,
      "grad_norm": 0.4678816795349121,
      "learning_rate": 2.8913e-05,
      "loss": 0.0719,
      "step": 13264
    },
    {
      "epoch": 24.47460757156048,
      "grad_norm": 0.30108150839805603,
      "learning_rate": 2.891266666666667e-05,
      "loss": 0.0384,
      "step": 13265
    },
    {
      "epoch": 24.47645429362881,
      "grad_norm": 0.33100152015686035,
      "learning_rate": 2.8912333333333334e-05,
      "loss": 0.042,
      "step": 13266
    },
    {
      "epoch": 24.478301015697138,
      "grad_norm": 0.41817301511764526,
      "learning_rate": 2.8912e-05,
      "loss": 0.0338,
      "step": 13267
    },
    {
      "epoch": 24.480147737765467,
      "grad_norm": 0.5045404434204102,
      "learning_rate": 2.891166666666667e-05,
      "loss": 0.0461,
      "step": 13268
    },
    {
      "epoch": 24.481994459833796,
      "grad_norm": 0.3951895236968994,
      "learning_rate": 2.8911333333333332e-05,
      "loss": 0.0282,
      "step": 13269
    },
    {
      "epoch": 24.483841181902125,
      "grad_norm": 0.5995557904243469,
      "learning_rate": 2.8911e-05,
      "loss": 0.0291,
      "step": 13270
    },
    {
      "epoch": 24.48568790397045,
      "grad_norm": 0.39431875944137573,
      "learning_rate": 2.8910666666666667e-05,
      "loss": 0.0178,
      "step": 13271
    },
    {
      "epoch": 24.48753462603878,
      "grad_norm": 0.19993415474891663,
      "learning_rate": 2.8910333333333333e-05,
      "loss": 0.0146,
      "step": 13272
    },
    {
      "epoch": 24.48938134810711,
      "grad_norm": 0.49636518955230713,
      "learning_rate": 2.891e-05,
      "loss": 0.022,
      "step": 13273
    },
    {
      "epoch": 24.49122807017544,
      "grad_norm": 0.16462446749210358,
      "learning_rate": 2.890966666666667e-05,
      "loss": 0.0063,
      "step": 13274
    },
    {
      "epoch": 24.493074792243767,
      "grad_norm": 0.2293635904788971,
      "learning_rate": 2.8909333333333335e-05,
      "loss": 0.0052,
      "step": 13275
    },
    {
      "epoch": 24.494921514312097,
      "grad_norm": 0.2452090084552765,
      "learning_rate": 2.8909e-05,
      "loss": 0.0078,
      "step": 13276
    },
    {
      "epoch": 24.496768236380426,
      "grad_norm": 0.5290648341178894,
      "learning_rate": 2.890866666666667e-05,
      "loss": 0.0172,
      "step": 13277
    },
    {
      "epoch": 24.498614958448755,
      "grad_norm": 0.28430402278900146,
      "learning_rate": 2.8908333333333332e-05,
      "loss": 0.0126,
      "step": 13278
    },
    {
      "epoch": 24.50046168051708,
      "grad_norm": 0.22045278549194336,
      "learning_rate": 2.8908000000000002e-05,
      "loss": 0.0115,
      "step": 13279
    },
    {
      "epoch": 24.50230840258541,
      "grad_norm": 0.28687742352485657,
      "learning_rate": 2.8907666666666668e-05,
      "loss": 0.006,
      "step": 13280
    },
    {
      "epoch": 24.50415512465374,
      "grad_norm": 0.2711004316806793,
      "learning_rate": 2.8907333333333334e-05,
      "loss": 0.0083,
      "step": 13281
    },
    {
      "epoch": 24.506001846722068,
      "grad_norm": 0.1592472940683365,
      "learning_rate": 2.8907e-05,
      "loss": 0.0054,
      "step": 13282
    },
    {
      "epoch": 24.507848568790397,
      "grad_norm": 0.20724567770957947,
      "learning_rate": 2.890666666666667e-05,
      "loss": 0.0056,
      "step": 13283
    },
    {
      "epoch": 24.509695290858726,
      "grad_norm": 0.19441282749176025,
      "learning_rate": 2.8906333333333335e-05,
      "loss": 0.0047,
      "step": 13284
    },
    {
      "epoch": 24.511542012927055,
      "grad_norm": 0.2539184093475342,
      "learning_rate": 2.8906e-05,
      "loss": 0.008,
      "step": 13285
    },
    {
      "epoch": 24.513388734995385,
      "grad_norm": 0.36838990449905396,
      "learning_rate": 2.8905666666666667e-05,
      "loss": 0.0067,
      "step": 13286
    },
    {
      "epoch": 24.51523545706371,
      "grad_norm": 0.23784856498241425,
      "learning_rate": 2.8905333333333333e-05,
      "loss": 0.0058,
      "step": 13287
    },
    {
      "epoch": 24.51708217913204,
      "grad_norm": 0.1111980453133583,
      "learning_rate": 2.8905000000000002e-05,
      "loss": 0.0024,
      "step": 13288
    },
    {
      "epoch": 24.51892890120037,
      "grad_norm": 0.24265041947364807,
      "learning_rate": 2.8904666666666664e-05,
      "loss": 0.0087,
      "step": 13289
    },
    {
      "epoch": 24.520775623268698,
      "grad_norm": 0.21897314488887787,
      "learning_rate": 2.8904333333333334e-05,
      "loss": 0.0052,
      "step": 13290
    },
    {
      "epoch": 24.522622345337027,
      "grad_norm": 0.29665127396583557,
      "learning_rate": 2.8904000000000003e-05,
      "loss": 0.0144,
      "step": 13291
    },
    {
      "epoch": 24.524469067405356,
      "grad_norm": 0.2656313180923462,
      "learning_rate": 2.8903666666666666e-05,
      "loss": 0.008,
      "step": 13292
    },
    {
      "epoch": 24.526315789473685,
      "grad_norm": 0.1815796196460724,
      "learning_rate": 2.8903333333333335e-05,
      "loss": 0.0042,
      "step": 13293
    },
    {
      "epoch": 24.528162511542014,
      "grad_norm": 0.2628864049911499,
      "learning_rate": 2.8903e-05,
      "loss": 0.0061,
      "step": 13294
    },
    {
      "epoch": 24.530009233610343,
      "grad_norm": 0.09849076718091965,
      "learning_rate": 2.8902666666666667e-05,
      "loss": 0.0026,
      "step": 13295
    },
    {
      "epoch": 24.53185595567867,
      "grad_norm": 0.25689443945884705,
      "learning_rate": 2.8902333333333333e-05,
      "loss": 0.0053,
      "step": 13296
    },
    {
      "epoch": 24.533702677746998,
      "grad_norm": 0.28762394189834595,
      "learning_rate": 2.8902000000000002e-05,
      "loss": 0.0081,
      "step": 13297
    },
    {
      "epoch": 24.535549399815327,
      "grad_norm": 0.14530755579471588,
      "learning_rate": 2.8901666666666665e-05,
      "loss": 0.0043,
      "step": 13298
    },
    {
      "epoch": 24.537396121883656,
      "grad_norm": 0.10289536416530609,
      "learning_rate": 2.8901333333333334e-05,
      "loss": 0.0027,
      "step": 13299
    },
    {
      "epoch": 24.539242843951985,
      "grad_norm": 0.33966517448425293,
      "learning_rate": 2.8901000000000003e-05,
      "loss": 0.007,
      "step": 13300
    },
    {
      "epoch": 24.541089566020315,
      "grad_norm": 0.21148870885372162,
      "learning_rate": 2.8900666666666666e-05,
      "loss": 0.0036,
      "step": 13301
    },
    {
      "epoch": 24.542936288088644,
      "grad_norm": 0.24400590360164642,
      "learning_rate": 2.8900333333333335e-05,
      "loss": 0.0048,
      "step": 13302
    },
    {
      "epoch": 24.544783010156973,
      "grad_norm": 0.2181837111711502,
      "learning_rate": 2.89e-05,
      "loss": 0.0045,
      "step": 13303
    },
    {
      "epoch": 24.5466297322253,
      "grad_norm": 0.16021010279655457,
      "learning_rate": 2.8899666666666667e-05,
      "loss": 0.004,
      "step": 13304
    },
    {
      "epoch": 24.548476454293628,
      "grad_norm": 0.39383822679519653,
      "learning_rate": 2.8899333333333333e-05,
      "loss": 0.0203,
      "step": 13305
    },
    {
      "epoch": 24.550323176361957,
      "grad_norm": 0.6963154077529907,
      "learning_rate": 2.8899000000000002e-05,
      "loss": 0.0139,
      "step": 13306
    },
    {
      "epoch": 24.552169898430286,
      "grad_norm": 0.4663408100605011,
      "learning_rate": 2.8898666666666668e-05,
      "loss": 0.0112,
      "step": 13307
    },
    {
      "epoch": 24.554016620498615,
      "grad_norm": 0.6084351539611816,
      "learning_rate": 2.8898333333333334e-05,
      "loss": 0.0091,
      "step": 13308
    },
    {
      "epoch": 24.555863342566944,
      "grad_norm": 0.4496295154094696,
      "learning_rate": 2.8898000000000004e-05,
      "loss": 0.0855,
      "step": 13309
    },
    {
      "epoch": 24.557710064635273,
      "grad_norm": 0.5155513882637024,
      "learning_rate": 2.8897666666666666e-05,
      "loss": 0.0998,
      "step": 13310
    },
    {
      "epoch": 24.559556786703602,
      "grad_norm": 0.6382702589035034,
      "learning_rate": 2.8897333333333335e-05,
      "loss": 0.1049,
      "step": 13311
    },
    {
      "epoch": 24.56140350877193,
      "grad_norm": 0.8368784785270691,
      "learning_rate": 2.8897e-05,
      "loss": 0.0648,
      "step": 13312
    },
    {
      "epoch": 24.563250230840257,
      "grad_norm": 0.5362563729286194,
      "learning_rate": 2.8896666666666667e-05,
      "loss": 0.0595,
      "step": 13313
    },
    {
      "epoch": 24.565096952908586,
      "grad_norm": 0.42777219414711,
      "learning_rate": 2.8896333333333333e-05,
      "loss": 0.0354,
      "step": 13314
    },
    {
      "epoch": 24.566943674976915,
      "grad_norm": 0.5480246543884277,
      "learning_rate": 2.8896e-05,
      "loss": 0.0547,
      "step": 13315
    },
    {
      "epoch": 24.568790397045245,
      "grad_norm": 0.5275306105613708,
      "learning_rate": 2.889566666666667e-05,
      "loss": 0.0548,
      "step": 13316
    },
    {
      "epoch": 24.570637119113574,
      "grad_norm": 0.4799412786960602,
      "learning_rate": 2.8895333333333334e-05,
      "loss": 0.0294,
      "step": 13317
    },
    {
      "epoch": 24.572483841181903,
      "grad_norm": 0.3169337213039398,
      "learning_rate": 2.8895e-05,
      "loss": 0.0237,
      "step": 13318
    },
    {
      "epoch": 24.574330563250232,
      "grad_norm": 0.3254132866859436,
      "learning_rate": 2.8894666666666666e-05,
      "loss": 0.018,
      "step": 13319
    },
    {
      "epoch": 24.57617728531856,
      "grad_norm": 0.1958315074443817,
      "learning_rate": 2.8894333333333336e-05,
      "loss": 0.0166,
      "step": 13320
    },
    {
      "epoch": 24.578024007386887,
      "grad_norm": 0.3534824848175049,
      "learning_rate": 2.8893999999999998e-05,
      "loss": 0.0379,
      "step": 13321
    },
    {
      "epoch": 24.579870729455216,
      "grad_norm": 0.2735038101673126,
      "learning_rate": 2.8893666666666667e-05,
      "loss": 0.0295,
      "step": 13322
    },
    {
      "epoch": 24.581717451523545,
      "grad_norm": 0.2168884128332138,
      "learning_rate": 2.8893333333333333e-05,
      "loss": 0.0195,
      "step": 13323
    },
    {
      "epoch": 24.583564173591874,
      "grad_norm": 0.3940054774284363,
      "learning_rate": 2.8893e-05,
      "loss": 0.0138,
      "step": 13324
    },
    {
      "epoch": 24.585410895660203,
      "grad_norm": 0.10811630636453629,
      "learning_rate": 2.889266666666667e-05,
      "loss": 0.0043,
      "step": 13325
    },
    {
      "epoch": 24.587257617728532,
      "grad_norm": 0.27094316482543945,
      "learning_rate": 2.8892333333333335e-05,
      "loss": 0.0207,
      "step": 13326
    },
    {
      "epoch": 24.58910433979686,
      "grad_norm": 0.2474210113286972,
      "learning_rate": 2.8892e-05,
      "loss": 0.0125,
      "step": 13327
    },
    {
      "epoch": 24.59095106186519,
      "grad_norm": 0.24952836334705353,
      "learning_rate": 2.8891666666666666e-05,
      "loss": 0.038,
      "step": 13328
    },
    {
      "epoch": 24.592797783933516,
      "grad_norm": 0.19649440050125122,
      "learning_rate": 2.8891333333333336e-05,
      "loss": 0.0074,
      "step": 13329
    },
    {
      "epoch": 24.594644506001845,
      "grad_norm": 0.290304571390152,
      "learning_rate": 2.8891e-05,
      "loss": 0.009,
      "step": 13330
    },
    {
      "epoch": 24.596491228070175,
      "grad_norm": 0.28430837392807007,
      "learning_rate": 2.8890666666666668e-05,
      "loss": 0.0062,
      "step": 13331
    },
    {
      "epoch": 24.598337950138504,
      "grad_norm": 0.14937494695186615,
      "learning_rate": 2.8890333333333334e-05,
      "loss": 0.0045,
      "step": 13332
    },
    {
      "epoch": 24.600184672206833,
      "grad_norm": 0.17749227583408356,
      "learning_rate": 2.889e-05,
      "loss": 0.0039,
      "step": 13333
    },
    {
      "epoch": 24.602031394275162,
      "grad_norm": 0.16499114036560059,
      "learning_rate": 2.888966666666667e-05,
      "loss": 0.0054,
      "step": 13334
    },
    {
      "epoch": 24.60387811634349,
      "grad_norm": 0.38722649216651917,
      "learning_rate": 2.8889333333333335e-05,
      "loss": 0.0109,
      "step": 13335
    },
    {
      "epoch": 24.60572483841182,
      "grad_norm": 0.5073453187942505,
      "learning_rate": 2.8889e-05,
      "loss": 0.0095,
      "step": 13336
    },
    {
      "epoch": 24.607571560480146,
      "grad_norm": 0.1586906909942627,
      "learning_rate": 2.8888666666666667e-05,
      "loss": 0.0034,
      "step": 13337
    },
    {
      "epoch": 24.609418282548475,
      "grad_norm": 0.15562497079372406,
      "learning_rate": 2.8888333333333336e-05,
      "loss": 0.0036,
      "step": 13338
    },
    {
      "epoch": 24.611265004616804,
      "grad_norm": 0.23065587878227234,
      "learning_rate": 2.8888e-05,
      "loss": 0.0048,
      "step": 13339
    },
    {
      "epoch": 24.613111726685133,
      "grad_norm": 0.5878494381904602,
      "learning_rate": 2.8887666666666668e-05,
      "loss": 0.0104,
      "step": 13340
    },
    {
      "epoch": 24.614958448753463,
      "grad_norm": 0.2063795030117035,
      "learning_rate": 2.8887333333333337e-05,
      "loss": 0.0055,
      "step": 13341
    },
    {
      "epoch": 24.61680517082179,
      "grad_norm": 0.1259927600622177,
      "learning_rate": 2.8887e-05,
      "loss": 0.003,
      "step": 13342
    },
    {
      "epoch": 24.61865189289012,
      "grad_norm": 0.15318113565444946,
      "learning_rate": 2.888666666666667e-05,
      "loss": 0.0035,
      "step": 13343
    },
    {
      "epoch": 24.62049861495845,
      "grad_norm": 0.2823616862297058,
      "learning_rate": 2.8886333333333335e-05,
      "loss": 0.0066,
      "step": 13344
    },
    {
      "epoch": 24.62234533702678,
      "grad_norm": 0.23812097311019897,
      "learning_rate": 2.8886e-05,
      "loss": 0.0325,
      "step": 13345
    },
    {
      "epoch": 24.624192059095105,
      "grad_norm": 0.14409127831459045,
      "learning_rate": 2.8885666666666667e-05,
      "loss": 0.0033,
      "step": 13346
    },
    {
      "epoch": 24.626038781163434,
      "grad_norm": 0.0659145936369896,
      "learning_rate": 2.8885333333333333e-05,
      "loss": 0.0012,
      "step": 13347
    },
    {
      "epoch": 24.627885503231763,
      "grad_norm": 0.3200194239616394,
      "learning_rate": 2.8885e-05,
      "loss": 0.0105,
      "step": 13348
    },
    {
      "epoch": 24.629732225300092,
      "grad_norm": 0.26891952753067017,
      "learning_rate": 2.8884666666666668e-05,
      "loss": 0.0052,
      "step": 13349
    },
    {
      "epoch": 24.63157894736842,
      "grad_norm": 0.328961580991745,
      "learning_rate": 2.8884333333333334e-05,
      "loss": 0.0074,
      "step": 13350
    },
    {
      "epoch": 24.63342566943675,
      "grad_norm": 0.1385984867811203,
      "learning_rate": 2.8884e-05,
      "loss": 0.0094,
      "step": 13351
    },
    {
      "epoch": 24.63527239150508,
      "grad_norm": 0.15883958339691162,
      "learning_rate": 2.888366666666667e-05,
      "loss": 0.0046,
      "step": 13352
    },
    {
      "epoch": 24.63711911357341,
      "grad_norm": 0.21640440821647644,
      "learning_rate": 2.8883333333333332e-05,
      "loss": 0.003,
      "step": 13353
    },
    {
      "epoch": 24.638965835641734,
      "grad_norm": 0.18219715356826782,
      "learning_rate": 2.8883e-05,
      "loss": 0.0043,
      "step": 13354
    },
    {
      "epoch": 24.640812557710063,
      "grad_norm": 0.23579628765583038,
      "learning_rate": 2.8882666666666667e-05,
      "loss": 0.0042,
      "step": 13355
    },
    {
      "epoch": 24.642659279778393,
      "grad_norm": 0.3087804913520813,
      "learning_rate": 2.8882333333333333e-05,
      "loss": 0.0094,
      "step": 13356
    },
    {
      "epoch": 24.64450600184672,
      "grad_norm": 0.3579120635986328,
      "learning_rate": 2.8882000000000002e-05,
      "loss": 0.0137,
      "step": 13357
    },
    {
      "epoch": 24.64635272391505,
      "grad_norm": 0.3803880512714386,
      "learning_rate": 2.8881666666666668e-05,
      "loss": 0.0114,
      "step": 13358
    },
    {
      "epoch": 24.64819944598338,
      "grad_norm": 0.5040902495384216,
      "learning_rate": 2.8881333333333334e-05,
      "loss": 0.1614,
      "step": 13359
    },
    {
      "epoch": 24.65004616805171,
      "grad_norm": 0.5189815163612366,
      "learning_rate": 2.8881e-05,
      "loss": 0.0932,
      "step": 13360
    },
    {
      "epoch": 24.65189289012004,
      "grad_norm": 0.3309284448623657,
      "learning_rate": 2.888066666666667e-05,
      "loss": 0.1499,
      "step": 13361
    },
    {
      "epoch": 24.653739612188367,
      "grad_norm": 0.4587146043777466,
      "learning_rate": 2.8880333333333332e-05,
      "loss": 0.0822,
      "step": 13362
    },
    {
      "epoch": 24.655586334256693,
      "grad_norm": 0.34885290265083313,
      "learning_rate": 2.888e-05,
      "loss": 0.0672,
      "step": 13363
    },
    {
      "epoch": 24.657433056325022,
      "grad_norm": 0.38283026218414307,
      "learning_rate": 2.8879666666666667e-05,
      "loss": 0.0574,
      "step": 13364
    },
    {
      "epoch": 24.65927977839335,
      "grad_norm": 0.38761594891548157,
      "learning_rate": 2.8879333333333333e-05,
      "loss": 0.0586,
      "step": 13365
    },
    {
      "epoch": 24.66112650046168,
      "grad_norm": 0.3336412012577057,
      "learning_rate": 2.8879000000000003e-05,
      "loss": 0.0365,
      "step": 13366
    },
    {
      "epoch": 24.66297322253001,
      "grad_norm": 0.3232431411743164,
      "learning_rate": 2.887866666666667e-05,
      "loss": 0.03,
      "step": 13367
    },
    {
      "epoch": 24.66481994459834,
      "grad_norm": 0.5756363868713379,
      "learning_rate": 2.8878333333333334e-05,
      "loss": 0.0777,
      "step": 13368
    },
    {
      "epoch": 24.666666666666668,
      "grad_norm": 0.5154371857643127,
      "learning_rate": 2.8878e-05,
      "loss": 0.0313,
      "step": 13369
    },
    {
      "epoch": 24.668513388734997,
      "grad_norm": 0.35584941506385803,
      "learning_rate": 2.887766666666667e-05,
      "loss": 0.02,
      "step": 13370
    },
    {
      "epoch": 24.670360110803323,
      "grad_norm": 0.5981518030166626,
      "learning_rate": 2.8877333333333332e-05,
      "loss": 0.0728,
      "step": 13371
    },
    {
      "epoch": 24.67220683287165,
      "grad_norm": 0.4004424810409546,
      "learning_rate": 2.8877e-05,
      "loss": 0.066,
      "step": 13372
    },
    {
      "epoch": 24.67405355493998,
      "grad_norm": 0.27876365184783936,
      "learning_rate": 2.8876666666666667e-05,
      "loss": 0.0089,
      "step": 13373
    },
    {
      "epoch": 24.67590027700831,
      "grad_norm": 0.22904369235038757,
      "learning_rate": 2.8876333333333333e-05,
      "loss": 0.0088,
      "step": 13374
    },
    {
      "epoch": 24.67774699907664,
      "grad_norm": 0.2941538691520691,
      "learning_rate": 2.8876000000000003e-05,
      "loss": 0.0203,
      "step": 13375
    },
    {
      "epoch": 24.67959372114497,
      "grad_norm": 0.30584147572517395,
      "learning_rate": 2.8875666666666665e-05,
      "loss": 0.0091,
      "step": 13376
    },
    {
      "epoch": 24.681440443213297,
      "grad_norm": 0.31425368785858154,
      "learning_rate": 2.8875333333333335e-05,
      "loss": 0.0103,
      "step": 13377
    },
    {
      "epoch": 24.683287165281627,
      "grad_norm": 0.07397730648517609,
      "learning_rate": 2.8875e-05,
      "loss": 0.0021,
      "step": 13378
    },
    {
      "epoch": 24.685133887349952,
      "grad_norm": 0.19247840344905853,
      "learning_rate": 2.8874666666666666e-05,
      "loss": 0.005,
      "step": 13379
    },
    {
      "epoch": 24.68698060941828,
      "grad_norm": 0.30215951800346375,
      "learning_rate": 2.8874333333333332e-05,
      "loss": 0.0075,
      "step": 13380
    },
    {
      "epoch": 24.68882733148661,
      "grad_norm": 0.34927937388420105,
      "learning_rate": 2.8874000000000002e-05,
      "loss": 0.0176,
      "step": 13381
    },
    {
      "epoch": 24.69067405355494,
      "grad_norm": 0.17897507548332214,
      "learning_rate": 2.8873666666666668e-05,
      "loss": 0.0047,
      "step": 13382
    },
    {
      "epoch": 24.69252077562327,
      "grad_norm": 0.7408897280693054,
      "learning_rate": 2.8873333333333334e-05,
      "loss": 0.0087,
      "step": 13383
    },
    {
      "epoch": 24.694367497691598,
      "grad_norm": 0.16339300572872162,
      "learning_rate": 2.8873000000000003e-05,
      "loss": 0.0054,
      "step": 13384
    },
    {
      "epoch": 24.696214219759927,
      "grad_norm": 0.11997639387845993,
      "learning_rate": 2.8872666666666665e-05,
      "loss": 0.0025,
      "step": 13385
    },
    {
      "epoch": 24.698060941828256,
      "grad_norm": 0.618619978427887,
      "learning_rate": 2.8872333333333335e-05,
      "loss": 0.0377,
      "step": 13386
    },
    {
      "epoch": 24.69990766389658,
      "grad_norm": 0.08947763592004776,
      "learning_rate": 2.8872e-05,
      "loss": 0.0018,
      "step": 13387
    },
    {
      "epoch": 24.70175438596491,
      "grad_norm": 0.16694006323814392,
      "learning_rate": 2.8871666666666667e-05,
      "loss": 0.0054,
      "step": 13388
    },
    {
      "epoch": 24.70360110803324,
      "grad_norm": 0.44772008061408997,
      "learning_rate": 2.8871333333333333e-05,
      "loss": 0.0081,
      "step": 13389
    },
    {
      "epoch": 24.70544783010157,
      "grad_norm": 0.3581324815750122,
      "learning_rate": 2.8871000000000002e-05,
      "loss": 0.0362,
      "step": 13390
    },
    {
      "epoch": 24.7072945521699,
      "grad_norm": 0.17675131559371948,
      "learning_rate": 2.8870666666666668e-05,
      "loss": 0.0092,
      "step": 13391
    },
    {
      "epoch": 24.709141274238227,
      "grad_norm": 0.2986243665218353,
      "learning_rate": 2.8870333333333334e-05,
      "loss": 0.006,
      "step": 13392
    },
    {
      "epoch": 24.710987996306557,
      "grad_norm": 0.14071765542030334,
      "learning_rate": 2.8870000000000003e-05,
      "loss": 0.0032,
      "step": 13393
    },
    {
      "epoch": 24.712834718374886,
      "grad_norm": 0.3196142911911011,
      "learning_rate": 2.8869666666666666e-05,
      "loss": 0.0084,
      "step": 13394
    },
    {
      "epoch": 24.714681440443215,
      "grad_norm": 0.5192102789878845,
      "learning_rate": 2.8869333333333335e-05,
      "loss": 0.0095,
      "step": 13395
    },
    {
      "epoch": 24.71652816251154,
      "grad_norm": 0.21429091691970825,
      "learning_rate": 2.8869e-05,
      "loss": 0.0118,
      "step": 13396
    },
    {
      "epoch": 24.71837488457987,
      "grad_norm": 0.309095561504364,
      "learning_rate": 2.8868666666666667e-05,
      "loss": 0.0055,
      "step": 13397
    },
    {
      "epoch": 24.7202216066482,
      "grad_norm": 0.1888907253742218,
      "learning_rate": 2.8868333333333333e-05,
      "loss": 0.0069,
      "step": 13398
    },
    {
      "epoch": 24.722068328716528,
      "grad_norm": 0.3549118638038635,
      "learning_rate": 2.8868000000000002e-05,
      "loss": 0.0056,
      "step": 13399
    },
    {
      "epoch": 24.723915050784857,
      "grad_norm": 0.15123812854290009,
      "learning_rate": 2.8867666666666668e-05,
      "loss": 0.0048,
      "step": 13400
    },
    {
      "epoch": 24.725761772853186,
      "grad_norm": 0.18749882280826569,
      "learning_rate": 2.8867333333333334e-05,
      "loss": 0.0049,
      "step": 13401
    },
    {
      "epoch": 24.727608494921515,
      "grad_norm": 0.14293302595615387,
      "learning_rate": 2.8867000000000003e-05,
      "loss": 0.0025,
      "step": 13402
    },
    {
      "epoch": 24.729455216989845,
      "grad_norm": 0.14083927869796753,
      "learning_rate": 2.8866666666666666e-05,
      "loss": 0.0048,
      "step": 13403
    },
    {
      "epoch": 24.73130193905817,
      "grad_norm": 0.3582073450088501,
      "learning_rate": 2.8866333333333335e-05,
      "loss": 0.0081,
      "step": 13404
    },
    {
      "epoch": 24.7331486611265,
      "grad_norm": 0.32324060797691345,
      "learning_rate": 2.8866e-05,
      "loss": 0.0052,
      "step": 13405
    },
    {
      "epoch": 24.73499538319483,
      "grad_norm": 0.42408451437950134,
      "learning_rate": 2.8865666666666667e-05,
      "loss": 0.0073,
      "step": 13406
    },
    {
      "epoch": 24.736842105263158,
      "grad_norm": 0.7171687483787537,
      "learning_rate": 2.8865333333333336e-05,
      "loss": 0.0139,
      "step": 13407
    },
    {
      "epoch": 24.738688827331487,
      "grad_norm": 0.43229997158050537,
      "learning_rate": 2.8865e-05,
      "loss": 0.0085,
      "step": 13408
    },
    {
      "epoch": 24.740535549399816,
      "grad_norm": 0.7140260338783264,
      "learning_rate": 2.8864666666666668e-05,
      "loss": 0.2258,
      "step": 13409
    },
    {
      "epoch": 24.742382271468145,
      "grad_norm": 0.34567001461982727,
      "learning_rate": 2.8864333333333334e-05,
      "loss": 0.0694,
      "step": 13410
    },
    {
      "epoch": 24.744228993536474,
      "grad_norm": 0.6018345355987549,
      "learning_rate": 2.8864e-05,
      "loss": 0.1287,
      "step": 13411
    },
    {
      "epoch": 24.746075715604803,
      "grad_norm": 0.5588078498840332,
      "learning_rate": 2.8863666666666666e-05,
      "loss": 0.0671,
      "step": 13412
    },
    {
      "epoch": 24.74792243767313,
      "grad_norm": 0.4431043863296509,
      "learning_rate": 2.8863333333333335e-05,
      "loss": 0.0574,
      "step": 13413
    },
    {
      "epoch": 24.749769159741458,
      "grad_norm": 0.3815719187259674,
      "learning_rate": 2.8862999999999998e-05,
      "loss": 0.0652,
      "step": 13414
    },
    {
      "epoch": 24.751615881809787,
      "grad_norm": 0.4633784592151642,
      "learning_rate": 2.8862666666666667e-05,
      "loss": 0.0853,
      "step": 13415
    },
    {
      "epoch": 24.753462603878116,
      "grad_norm": 0.30097484588623047,
      "learning_rate": 2.8862333333333337e-05,
      "loss": 0.0583,
      "step": 13416
    },
    {
      "epoch": 24.755309325946445,
      "grad_norm": 0.3656047284603119,
      "learning_rate": 2.8862e-05,
      "loss": 0.0418,
      "step": 13417
    },
    {
      "epoch": 24.757156048014775,
      "grad_norm": 0.32067063450813293,
      "learning_rate": 2.886166666666667e-05,
      "loss": 0.0238,
      "step": 13418
    },
    {
      "epoch": 24.759002770083104,
      "grad_norm": 0.2664825916290283,
      "learning_rate": 2.8861333333333334e-05,
      "loss": 0.0244,
      "step": 13419
    },
    {
      "epoch": 24.760849492151433,
      "grad_norm": 0.261997789144516,
      "learning_rate": 2.8861e-05,
      "loss": 0.0197,
      "step": 13420
    },
    {
      "epoch": 24.76269621421976,
      "grad_norm": 0.3810383677482605,
      "learning_rate": 2.8860666666666666e-05,
      "loss": 0.035,
      "step": 13421
    },
    {
      "epoch": 24.764542936288088,
      "grad_norm": 0.26574939489364624,
      "learning_rate": 2.8860333333333336e-05,
      "loss": 0.0398,
      "step": 13422
    },
    {
      "epoch": 24.766389658356417,
      "grad_norm": 0.33361557126045227,
      "learning_rate": 2.8859999999999998e-05,
      "loss": 0.0297,
      "step": 13423
    },
    {
      "epoch": 24.768236380424746,
      "grad_norm": 0.19486159086227417,
      "learning_rate": 2.8859666666666667e-05,
      "loss": 0.0255,
      "step": 13424
    },
    {
      "epoch": 24.770083102493075,
      "grad_norm": 0.23205748200416565,
      "learning_rate": 2.8859333333333337e-05,
      "loss": 0.0089,
      "step": 13425
    },
    {
      "epoch": 24.771929824561404,
      "grad_norm": 0.28048232197761536,
      "learning_rate": 2.8859e-05,
      "loss": 0.0266,
      "step": 13426
    },
    {
      "epoch": 24.773776546629733,
      "grad_norm": 0.2901442348957062,
      "learning_rate": 2.885866666666667e-05,
      "loss": 0.0424,
      "step": 13427
    },
    {
      "epoch": 24.775623268698062,
      "grad_norm": 0.19676855206489563,
      "learning_rate": 2.8858333333333335e-05,
      "loss": 0.0249,
      "step": 13428
    },
    {
      "epoch": 24.777469990766388,
      "grad_norm": 0.1431822031736374,
      "learning_rate": 2.8858e-05,
      "loss": 0.0037,
      "step": 13429
    },
    {
      "epoch": 24.779316712834717,
      "grad_norm": 0.17462551593780518,
      "learning_rate": 2.8857666666666666e-05,
      "loss": 0.0206,
      "step": 13430
    },
    {
      "epoch": 24.781163434903046,
      "grad_norm": 0.130209818482399,
      "learning_rate": 2.8857333333333336e-05,
      "loss": 0.0037,
      "step": 13431
    },
    {
      "epoch": 24.783010156971375,
      "grad_norm": 0.22776542603969574,
      "learning_rate": 2.8857000000000002e-05,
      "loss": 0.0085,
      "step": 13432
    },
    {
      "epoch": 24.784856879039705,
      "grad_norm": 0.22143611311912537,
      "learning_rate": 2.8856666666666668e-05,
      "loss": 0.0063,
      "step": 13433
    },
    {
      "epoch": 24.786703601108034,
      "grad_norm": 0.15124769508838654,
      "learning_rate": 2.8856333333333337e-05,
      "loss": 0.0047,
      "step": 13434
    },
    {
      "epoch": 24.788550323176363,
      "grad_norm": 0.2723243832588196,
      "learning_rate": 2.8856e-05,
      "loss": 0.004,
      "step": 13435
    },
    {
      "epoch": 24.790397045244692,
      "grad_norm": 0.41200390458106995,
      "learning_rate": 2.885566666666667e-05,
      "loss": 0.0171,
      "step": 13436
    },
    {
      "epoch": 24.792243767313018,
      "grad_norm": 0.17938579618930817,
      "learning_rate": 2.885533333333333e-05,
      "loss": 0.0081,
      "step": 13437
    },
    {
      "epoch": 24.794090489381347,
      "grad_norm": 0.12062520533800125,
      "learning_rate": 2.8855e-05,
      "loss": 0.0021,
      "step": 13438
    },
    {
      "epoch": 24.795937211449676,
      "grad_norm": 0.5112106800079346,
      "learning_rate": 2.8854666666666667e-05,
      "loss": 0.014,
      "step": 13439
    },
    {
      "epoch": 24.797783933518005,
      "grad_norm": 0.21921005845069885,
      "learning_rate": 2.8854333333333333e-05,
      "loss": 0.0039,
      "step": 13440
    },
    {
      "epoch": 24.799630655586334,
      "grad_norm": 0.4647342562675476,
      "learning_rate": 2.8854000000000002e-05,
      "loss": 0.0102,
      "step": 13441
    },
    {
      "epoch": 24.801477377654663,
      "grad_norm": 0.17171858251094818,
      "learning_rate": 2.8853666666666668e-05,
      "loss": 0.0029,
      "step": 13442
    },
    {
      "epoch": 24.803324099722992,
      "grad_norm": 0.39785921573638916,
      "learning_rate": 2.8853333333333334e-05,
      "loss": 0.0057,
      "step": 13443
    },
    {
      "epoch": 24.80517082179132,
      "grad_norm": 0.1778843104839325,
      "learning_rate": 2.8853e-05,
      "loss": 0.0053,
      "step": 13444
    },
    {
      "epoch": 24.80701754385965,
      "grad_norm": 0.26482951641082764,
      "learning_rate": 2.885266666666667e-05,
      "loss": 0.0084,
      "step": 13445
    },
    {
      "epoch": 24.808864265927976,
      "grad_norm": 0.5930157899856567,
      "learning_rate": 2.885233333333333e-05,
      "loss": 0.0062,
      "step": 13446
    },
    {
      "epoch": 24.810710987996305,
      "grad_norm": 0.7362170815467834,
      "learning_rate": 2.8852e-05,
      "loss": 0.008,
      "step": 13447
    },
    {
      "epoch": 24.812557710064635,
      "grad_norm": 0.2874982953071594,
      "learning_rate": 2.8851666666666667e-05,
      "loss": 0.01,
      "step": 13448
    },
    {
      "epoch": 24.814404432132964,
      "grad_norm": 0.2321842610836029,
      "learning_rate": 2.8851333333333333e-05,
      "loss": 0.0048,
      "step": 13449
    },
    {
      "epoch": 24.816251154201293,
      "grad_norm": 0.550155520439148,
      "learning_rate": 2.8851000000000002e-05,
      "loss": 0.0074,
      "step": 13450
    },
    {
      "epoch": 24.818097876269622,
      "grad_norm": 0.4127471148967743,
      "learning_rate": 2.8850666666666668e-05,
      "loss": 0.007,
      "step": 13451
    },
    {
      "epoch": 24.81994459833795,
      "grad_norm": 0.33397412300109863,
      "learning_rate": 2.8850333333333334e-05,
      "loss": 0.0131,
      "step": 13452
    },
    {
      "epoch": 24.82179132040628,
      "grad_norm": 0.25564906001091003,
      "learning_rate": 2.885e-05,
      "loss": 0.0039,
      "step": 13453
    },
    {
      "epoch": 24.823638042474606,
      "grad_norm": 0.0578516349196434,
      "learning_rate": 2.884966666666667e-05,
      "loss": 0.0013,
      "step": 13454
    },
    {
      "epoch": 24.825484764542935,
      "grad_norm": 0.14229914546012878,
      "learning_rate": 2.8849333333333332e-05,
      "loss": 0.0038,
      "step": 13455
    },
    {
      "epoch": 24.827331486611264,
      "grad_norm": 0.5849806070327759,
      "learning_rate": 2.8849e-05,
      "loss": 0.0081,
      "step": 13456
    },
    {
      "epoch": 24.829178208679593,
      "grad_norm": 0.2880188524723053,
      "learning_rate": 2.884866666666667e-05,
      "loss": 0.0052,
      "step": 13457
    },
    {
      "epoch": 24.831024930747922,
      "grad_norm": 0.32646384835243225,
      "learning_rate": 2.8848333333333333e-05,
      "loss": 0.0126,
      "step": 13458
    },
    {
      "epoch": 24.83287165281625,
      "grad_norm": 0.651218593120575,
      "learning_rate": 2.8848000000000002e-05,
      "loss": 0.1134,
      "step": 13459
    },
    {
      "epoch": 24.83471837488458,
      "grad_norm": 0.3713066577911377,
      "learning_rate": 2.8847666666666668e-05,
      "loss": 0.0834,
      "step": 13460
    },
    {
      "epoch": 24.83656509695291,
      "grad_norm": 0.4958324730396271,
      "learning_rate": 2.8847333333333334e-05,
      "loss": 0.061,
      "step": 13461
    },
    {
      "epoch": 24.83841181902124,
      "grad_norm": 0.5096185803413391,
      "learning_rate": 2.8847e-05,
      "loss": 0.0809,
      "step": 13462
    },
    {
      "epoch": 24.840258541089565,
      "grad_norm": 0.2968968451023102,
      "learning_rate": 2.884666666666667e-05,
      "loss": 0.0972,
      "step": 13463
    },
    {
      "epoch": 24.842105263157894,
      "grad_norm": 0.3534603416919708,
      "learning_rate": 2.8846333333333332e-05,
      "loss": 0.0804,
      "step": 13464
    },
    {
      "epoch": 24.843951985226223,
      "grad_norm": 0.3805117607116699,
      "learning_rate": 2.8846e-05,
      "loss": 0.0425,
      "step": 13465
    },
    {
      "epoch": 24.845798707294552,
      "grad_norm": 0.2468799501657486,
      "learning_rate": 2.8845666666666667e-05,
      "loss": 0.0254,
      "step": 13466
    },
    {
      "epoch": 24.84764542936288,
      "grad_norm": 0.4495117962360382,
      "learning_rate": 2.8845333333333333e-05,
      "loss": 0.0336,
      "step": 13467
    },
    {
      "epoch": 24.84949215143121,
      "grad_norm": 0.38449251651763916,
      "learning_rate": 2.8845000000000003e-05,
      "loss": 0.0221,
      "step": 13468
    },
    {
      "epoch": 24.85133887349954,
      "grad_norm": 0.38495227694511414,
      "learning_rate": 2.8844666666666665e-05,
      "loss": 0.0217,
      "step": 13469
    },
    {
      "epoch": 24.85318559556787,
      "grad_norm": 0.5466144680976868,
      "learning_rate": 2.8844333333333334e-05,
      "loss": 0.0275,
      "step": 13470
    },
    {
      "epoch": 24.855032317636194,
      "grad_norm": 0.4924391508102417,
      "learning_rate": 2.8844e-05,
      "loss": 0.0384,
      "step": 13471
    },
    {
      "epoch": 24.856879039704523,
      "grad_norm": 0.33810392022132874,
      "learning_rate": 2.8843666666666666e-05,
      "loss": 0.0243,
      "step": 13472
    },
    {
      "epoch": 24.858725761772853,
      "grad_norm": 0.17373476922512054,
      "learning_rate": 2.8843333333333332e-05,
      "loss": 0.0079,
      "step": 13473
    },
    {
      "epoch": 24.86057248384118,
      "grad_norm": 0.5740019083023071,
      "learning_rate": 2.8843e-05,
      "loss": 0.0169,
      "step": 13474
    },
    {
      "epoch": 24.86241920590951,
      "grad_norm": 0.1319696605205536,
      "learning_rate": 2.8842666666666667e-05,
      "loss": 0.0051,
      "step": 13475
    },
    {
      "epoch": 24.86426592797784,
      "grad_norm": 0.46193742752075195,
      "learning_rate": 2.8842333333333333e-05,
      "loss": 0.0219,
      "step": 13476
    },
    {
      "epoch": 24.86611265004617,
      "grad_norm": 0.13610051572322845,
      "learning_rate": 2.8842000000000003e-05,
      "loss": 0.0043,
      "step": 13477
    },
    {
      "epoch": 24.8679593721145,
      "grad_norm": 0.2926333248615265,
      "learning_rate": 2.8841666666666665e-05,
      "loss": 0.0078,
      "step": 13478
    },
    {
      "epoch": 24.869806094182824,
      "grad_norm": 0.35700613260269165,
      "learning_rate": 2.8841333333333335e-05,
      "loss": 0.0068,
      "step": 13479
    },
    {
      "epoch": 24.871652816251153,
      "grad_norm": 0.4237408936023712,
      "learning_rate": 2.8841e-05,
      "loss": 0.0059,
      "step": 13480
    },
    {
      "epoch": 24.873499538319482,
      "grad_norm": 0.26153576374053955,
      "learning_rate": 2.8840666666666666e-05,
      "loss": 0.0112,
      "step": 13481
    },
    {
      "epoch": 24.87534626038781,
      "grad_norm": 0.22728842496871948,
      "learning_rate": 2.8840333333333336e-05,
      "loss": 0.0069,
      "step": 13482
    },
    {
      "epoch": 24.87719298245614,
      "grad_norm": 0.22643160820007324,
      "learning_rate": 2.8840000000000002e-05,
      "loss": 0.0142,
      "step": 13483
    },
    {
      "epoch": 24.87903970452447,
      "grad_norm": 0.2977308928966522,
      "learning_rate": 2.8839666666666668e-05,
      "loss": 0.0188,
      "step": 13484
    },
    {
      "epoch": 24.8808864265928,
      "grad_norm": 0.17739634215831757,
      "learning_rate": 2.8839333333333334e-05,
      "loss": 0.0048,
      "step": 13485
    },
    {
      "epoch": 24.882733148661128,
      "grad_norm": 0.1741904318332672,
      "learning_rate": 2.8839000000000003e-05,
      "loss": 0.0024,
      "step": 13486
    },
    {
      "epoch": 24.884579870729453,
      "grad_norm": 0.2034701704978943,
      "learning_rate": 2.8838666666666665e-05,
      "loss": 0.0073,
      "step": 13487
    },
    {
      "epoch": 24.886426592797783,
      "grad_norm": 0.24229569733142853,
      "learning_rate": 2.8838333333333335e-05,
      "loss": 0.0707,
      "step": 13488
    },
    {
      "epoch": 24.88827331486611,
      "grad_norm": 0.49798405170440674,
      "learning_rate": 2.8838e-05,
      "loss": 0.0124,
      "step": 13489
    },
    {
      "epoch": 24.89012003693444,
      "grad_norm": 0.29836997389793396,
      "learning_rate": 2.8837666666666667e-05,
      "loss": 0.0045,
      "step": 13490
    },
    {
      "epoch": 24.89196675900277,
      "grad_norm": 0.2838921546936035,
      "learning_rate": 2.8837333333333336e-05,
      "loss": 0.0053,
      "step": 13491
    },
    {
      "epoch": 24.8938134810711,
      "grad_norm": 0.2837573289871216,
      "learning_rate": 2.8837000000000002e-05,
      "loss": 0.0093,
      "step": 13492
    },
    {
      "epoch": 24.89566020313943,
      "grad_norm": 0.12989038228988647,
      "learning_rate": 2.8836666666666668e-05,
      "loss": 0.0037,
      "step": 13493
    },
    {
      "epoch": 24.897506925207757,
      "grad_norm": 0.22318127751350403,
      "learning_rate": 2.8836333333333334e-05,
      "loss": 0.0057,
      "step": 13494
    },
    {
      "epoch": 24.899353647276087,
      "grad_norm": 0.30415669083595276,
      "learning_rate": 2.8836000000000003e-05,
      "loss": 0.01,
      "step": 13495
    },
    {
      "epoch": 24.901200369344412,
      "grad_norm": 0.11953115463256836,
      "learning_rate": 2.8835666666666666e-05,
      "loss": 0.0038,
      "step": 13496
    },
    {
      "epoch": 24.90304709141274,
      "grad_norm": 0.3473064601421356,
      "learning_rate": 2.8835333333333335e-05,
      "loss": 0.0108,
      "step": 13497
    },
    {
      "epoch": 24.90489381348107,
      "grad_norm": 0.30331888794898987,
      "learning_rate": 2.8834999999999998e-05,
      "loss": 0.0097,
      "step": 13498
    },
    {
      "epoch": 24.9067405355494,
      "grad_norm": 0.4642525315284729,
      "learning_rate": 2.8834666666666667e-05,
      "loss": 0.013,
      "step": 13499
    },
    {
      "epoch": 24.90858725761773,
      "grad_norm": 0.23262901604175568,
      "learning_rate": 2.8834333333333336e-05,
      "loss": 0.0106,
      "step": 13500
    },
    {
      "epoch": 24.910433979686058,
      "grad_norm": 0.22102048993110657,
      "learning_rate": 2.8834e-05,
      "loss": 0.0037,
      "step": 13501
    },
    {
      "epoch": 24.912280701754387,
      "grad_norm": 0.44870537519454956,
      "learning_rate": 2.8833666666666668e-05,
      "loss": 0.0084,
      "step": 13502
    },
    {
      "epoch": 24.914127423822716,
      "grad_norm": 0.1932738572359085,
      "learning_rate": 2.8833333333333334e-05,
      "loss": 0.0049,
      "step": 13503
    },
    {
      "epoch": 24.91597414589104,
      "grad_norm": 0.3363226056098938,
      "learning_rate": 2.8833e-05,
      "loss": 0.0096,
      "step": 13504
    },
    {
      "epoch": 24.91782086795937,
      "grad_norm": 0.2664887309074402,
      "learning_rate": 2.8832666666666666e-05,
      "loss": 0.0088,
      "step": 13505
    },
    {
      "epoch": 24.9196675900277,
      "grad_norm": 0.34887585043907166,
      "learning_rate": 2.8832333333333335e-05,
      "loss": 0.0173,
      "step": 13506
    },
    {
      "epoch": 24.92151431209603,
      "grad_norm": 0.1695641726255417,
      "learning_rate": 2.8832e-05,
      "loss": 0.0062,
      "step": 13507
    },
    {
      "epoch": 24.92336103416436,
      "grad_norm": 0.09887981414794922,
      "learning_rate": 2.8831666666666667e-05,
      "loss": 0.0023,
      "step": 13508
    },
    {
      "epoch": 24.925207756232687,
      "grad_norm": 0.6092426180839539,
      "learning_rate": 2.8831333333333336e-05,
      "loss": 0.1358,
      "step": 13509
    },
    {
      "epoch": 24.927054478301017,
      "grad_norm": 0.3662014603614807,
      "learning_rate": 2.8831e-05,
      "loss": 0.0617,
      "step": 13510
    },
    {
      "epoch": 24.928901200369346,
      "grad_norm": 0.38646215200424194,
      "learning_rate": 2.8830666666666668e-05,
      "loss": 0.0661,
      "step": 13511
    },
    {
      "epoch": 24.930747922437675,
      "grad_norm": 0.6269811987876892,
      "learning_rate": 2.8830333333333334e-05,
      "loss": 0.1526,
      "step": 13512
    },
    {
      "epoch": 24.932594644506,
      "grad_norm": 0.3474038243293762,
      "learning_rate": 2.883e-05,
      "loss": 0.0508,
      "step": 13513
    },
    {
      "epoch": 24.93444136657433,
      "grad_norm": 0.3669436573982239,
      "learning_rate": 2.8829666666666666e-05,
      "loss": 0.0335,
      "step": 13514
    },
    {
      "epoch": 24.93628808864266,
      "grad_norm": 0.3087623119354248,
      "learning_rate": 2.8829333333333335e-05,
      "loss": 0.0312,
      "step": 13515
    },
    {
      "epoch": 24.938134810710988,
      "grad_norm": 0.4781001806259155,
      "learning_rate": 2.8829e-05,
      "loss": 0.0507,
      "step": 13516
    },
    {
      "epoch": 24.939981532779317,
      "grad_norm": 0.26677459478378296,
      "learning_rate": 2.8828666666666667e-05,
      "loss": 0.0321,
      "step": 13517
    },
    {
      "epoch": 24.941828254847646,
      "grad_norm": 0.3729875981807709,
      "learning_rate": 2.8828333333333337e-05,
      "loss": 0.0253,
      "step": 13518
    },
    {
      "epoch": 24.943674976915975,
      "grad_norm": 0.3847418427467346,
      "learning_rate": 2.8828e-05,
      "loss": 0.0217,
      "step": 13519
    },
    {
      "epoch": 24.945521698984304,
      "grad_norm": 0.3041211664676666,
      "learning_rate": 2.882766666666667e-05,
      "loss": 0.0143,
      "step": 13520
    },
    {
      "epoch": 24.94736842105263,
      "grad_norm": 0.6895516514778137,
      "learning_rate": 2.8827333333333334e-05,
      "loss": 0.0477,
      "step": 13521
    },
    {
      "epoch": 24.94921514312096,
      "grad_norm": 0.6736093163490295,
      "learning_rate": 2.8827e-05,
      "loss": 0.0163,
      "step": 13522
    },
    {
      "epoch": 24.95106186518929,
      "grad_norm": 0.15927226841449738,
      "learning_rate": 2.8826666666666666e-05,
      "loss": 0.0078,
      "step": 13523
    },
    {
      "epoch": 24.952908587257618,
      "grad_norm": 0.2903159558773041,
      "learning_rate": 2.8826333333333336e-05,
      "loss": 0.0118,
      "step": 13524
    },
    {
      "epoch": 24.954755309325947,
      "grad_norm": 0.23991265892982483,
      "learning_rate": 2.8826e-05,
      "loss": 0.0081,
      "step": 13525
    },
    {
      "epoch": 24.956602031394276,
      "grad_norm": 0.17719998955726624,
      "learning_rate": 2.8825666666666667e-05,
      "loss": 0.008,
      "step": 13526
    },
    {
      "epoch": 24.958448753462605,
      "grad_norm": 0.266058087348938,
      "learning_rate": 2.8825333333333333e-05,
      "loss": 0.0066,
      "step": 13527
    },
    {
      "epoch": 24.960295475530934,
      "grad_norm": 0.16486336290836334,
      "learning_rate": 2.8825e-05,
      "loss": 0.007,
      "step": 13528
    },
    {
      "epoch": 24.96214219759926,
      "grad_norm": 0.10596636682748795,
      "learning_rate": 2.882466666666667e-05,
      "loss": 0.0033,
      "step": 13529
    },
    {
      "epoch": 24.96398891966759,
      "grad_norm": 0.43525850772857666,
      "learning_rate": 2.882433333333333e-05,
      "loss": 0.0086,
      "step": 13530
    },
    {
      "epoch": 24.965835641735918,
      "grad_norm": 0.17290306091308594,
      "learning_rate": 2.8824e-05,
      "loss": 0.008,
      "step": 13531
    },
    {
      "epoch": 24.967682363804247,
      "grad_norm": 0.3244077265262604,
      "learning_rate": 2.882366666666667e-05,
      "loss": 0.0079,
      "step": 13532
    },
    {
      "epoch": 24.969529085872576,
      "grad_norm": 0.16322195529937744,
      "learning_rate": 2.8823333333333332e-05,
      "loss": 0.0047,
      "step": 13533
    },
    {
      "epoch": 24.971375807940905,
      "grad_norm": 0.09850552678108215,
      "learning_rate": 2.8823000000000002e-05,
      "loss": 0.0026,
      "step": 13534
    },
    {
      "epoch": 24.973222530009235,
      "grad_norm": 0.17957203090190887,
      "learning_rate": 2.8822666666666668e-05,
      "loss": 0.008,
      "step": 13535
    },
    {
      "epoch": 24.975069252077564,
      "grad_norm": 0.42165297269821167,
      "learning_rate": 2.8822333333333334e-05,
      "loss": 0.0103,
      "step": 13536
    },
    {
      "epoch": 24.97691597414589,
      "grad_norm": 0.21143174171447754,
      "learning_rate": 2.8822e-05,
      "loss": 0.0082,
      "step": 13537
    },
    {
      "epoch": 24.97876269621422,
      "grad_norm": 0.1594984382390976,
      "learning_rate": 2.882166666666667e-05,
      "loss": 0.0063,
      "step": 13538
    },
    {
      "epoch": 24.980609418282548,
      "grad_norm": 0.13408170640468597,
      "learning_rate": 2.882133333333333e-05,
      "loss": 0.0041,
      "step": 13539
    },
    {
      "epoch": 24.982456140350877,
      "grad_norm": 0.19753006100654602,
      "learning_rate": 2.8821e-05,
      "loss": 0.0056,
      "step": 13540
    },
    {
      "epoch": 24.984302862419206,
      "grad_norm": 0.21510076522827148,
      "learning_rate": 2.882066666666667e-05,
      "loss": 0.0076,
      "step": 13541
    },
    {
      "epoch": 24.986149584487535,
      "grad_norm": 0.10175257921218872,
      "learning_rate": 2.8820333333333333e-05,
      "loss": 0.003,
      "step": 13542
    },
    {
      "epoch": 24.987996306555864,
      "grad_norm": 0.10908100754022598,
      "learning_rate": 2.8820000000000002e-05,
      "loss": 0.0031,
      "step": 13543
    },
    {
      "epoch": 24.989843028624193,
      "grad_norm": 0.37116262316703796,
      "learning_rate": 2.8819666666666668e-05,
      "loss": 0.0077,
      "step": 13544
    },
    {
      "epoch": 24.991689750692522,
      "grad_norm": 0.13784131407737732,
      "learning_rate": 2.8819333333333334e-05,
      "loss": 0.0027,
      "step": 13545
    },
    {
      "epoch": 24.993536472760848,
      "grad_norm": 0.1495315283536911,
      "learning_rate": 2.8819e-05,
      "loss": 0.0063,
      "step": 13546
    },
    {
      "epoch": 24.995383194829177,
      "grad_norm": 0.17947818338871002,
      "learning_rate": 2.881866666666667e-05,
      "loss": 0.004,
      "step": 13547
    },
    {
      "epoch": 24.997229916897506,
      "grad_norm": 0.22032637894153595,
      "learning_rate": 2.881833333333333e-05,
      "loss": 0.0035,
      "step": 13548
    },
    {
      "epoch": 24.999076638965835,
      "grad_norm": 0.08717142790555954,
      "learning_rate": 2.8818e-05,
      "loss": 0.002,
      "step": 13549
    },
    {
      "epoch": 25.0,
      "grad_norm": 0.766228973865509,
      "learning_rate": 2.881766666666667e-05,
      "loss": 0.0297,
      "step": 13550
    },
    {
      "epoch": 25.00184672206833,
      "grad_norm": 0.502040445804596,
      "learning_rate": 2.8817333333333333e-05,
      "loss": 0.1689,
      "step": 13551
    },
    {
      "epoch": 25.00369344413666,
      "grad_norm": 0.41068142652511597,
      "learning_rate": 2.8817000000000002e-05,
      "loss": 0.0909,
      "step": 13552
    },
    {
      "epoch": 25.005540166204987,
      "grad_norm": 0.3879607319831848,
      "learning_rate": 2.8816666666666668e-05,
      "loss": 0.0564,
      "step": 13553
    },
    {
      "epoch": 25.007386888273317,
      "grad_norm": 0.31146150827407837,
      "learning_rate": 2.8816333333333334e-05,
      "loss": 0.0549,
      "step": 13554
    },
    {
      "epoch": 25.009233610341642,
      "grad_norm": 0.515511691570282,
      "learning_rate": 2.8816e-05,
      "loss": 0.0529,
      "step": 13555
    },
    {
      "epoch": 25.01108033240997,
      "grad_norm": 0.349189817905426,
      "learning_rate": 2.881566666666667e-05,
      "loss": 0.0903,
      "step": 13556
    },
    {
      "epoch": 25.0129270544783,
      "grad_norm": 0.5013755559921265,
      "learning_rate": 2.8815333333333335e-05,
      "loss": 0.0776,
      "step": 13557
    },
    {
      "epoch": 25.01477377654663,
      "grad_norm": 0.42771899700164795,
      "learning_rate": 2.8815e-05,
      "loss": 0.0319,
      "step": 13558
    },
    {
      "epoch": 25.01662049861496,
      "grad_norm": 0.22613854706287384,
      "learning_rate": 2.8814666666666667e-05,
      "loss": 0.0173,
      "step": 13559
    },
    {
      "epoch": 25.018467220683288,
      "grad_norm": 0.34730762243270874,
      "learning_rate": 2.8814333333333333e-05,
      "loss": 0.0725,
      "step": 13560
    },
    {
      "epoch": 25.020313942751617,
      "grad_norm": 0.24891577661037445,
      "learning_rate": 2.8814000000000002e-05,
      "loss": 0.0161,
      "step": 13561
    },
    {
      "epoch": 25.022160664819946,
      "grad_norm": 0.4423004388809204,
      "learning_rate": 2.8813666666666665e-05,
      "loss": 0.0184,
      "step": 13562
    },
    {
      "epoch": 25.02400738688827,
      "grad_norm": 0.2701072096824646,
      "learning_rate": 2.8813333333333334e-05,
      "loss": 0.0303,
      "step": 13563
    },
    {
      "epoch": 25.0258541089566,
      "grad_norm": 0.29532355070114136,
      "learning_rate": 2.8813e-05,
      "loss": 0.0128,
      "step": 13564
    },
    {
      "epoch": 25.02770083102493,
      "grad_norm": 0.27251678705215454,
      "learning_rate": 2.8812666666666666e-05,
      "loss": 0.0256,
      "step": 13565
    },
    {
      "epoch": 25.02954755309326,
      "grad_norm": 0.2066493034362793,
      "learning_rate": 2.8812333333333335e-05,
      "loss": 0.0235,
      "step": 13566
    },
    {
      "epoch": 25.03139427516159,
      "grad_norm": 0.12427642196416855,
      "learning_rate": 2.8812e-05,
      "loss": 0.003,
      "step": 13567
    },
    {
      "epoch": 25.033240997229917,
      "grad_norm": 0.3608299493789673,
      "learning_rate": 2.8811666666666667e-05,
      "loss": 0.0099,
      "step": 13568
    },
    {
      "epoch": 25.035087719298247,
      "grad_norm": 0.1806800216436386,
      "learning_rate": 2.8811333333333333e-05,
      "loss": 0.0179,
      "step": 13569
    },
    {
      "epoch": 25.036934441366576,
      "grad_norm": 0.5715934634208679,
      "learning_rate": 2.8811000000000002e-05,
      "loss": 0.0304,
      "step": 13570
    },
    {
      "epoch": 25.0387811634349,
      "grad_norm": 0.10905469208955765,
      "learning_rate": 2.8810666666666665e-05,
      "loss": 0.0012,
      "step": 13571
    },
    {
      "epoch": 25.04062788550323,
      "grad_norm": 0.15964914858341217,
      "learning_rate": 2.8810333333333334e-05,
      "loss": 0.0043,
      "step": 13572
    },
    {
      "epoch": 25.04247460757156,
      "grad_norm": 0.07477544993162155,
      "learning_rate": 2.881e-05,
      "loss": 0.002,
      "step": 13573
    },
    {
      "epoch": 25.04432132963989,
      "grad_norm": 0.11960326135158539,
      "learning_rate": 2.8809666666666666e-05,
      "loss": 0.0009,
      "step": 13574
    },
    {
      "epoch": 25.046168051708218,
      "grad_norm": 0.17168718576431274,
      "learning_rate": 2.8809333333333336e-05,
      "loss": 0.0036,
      "step": 13575
    },
    {
      "epoch": 25.048014773776547,
      "grad_norm": 0.14043472707271576,
      "learning_rate": 2.8809e-05,
      "loss": 0.0041,
      "step": 13576
    },
    {
      "epoch": 25.049861495844876,
      "grad_norm": 0.18809548020362854,
      "learning_rate": 2.8808666666666667e-05,
      "loss": 0.0031,
      "step": 13577
    },
    {
      "epoch": 25.051708217913205,
      "grad_norm": 0.16893936693668365,
      "learning_rate": 2.8808333333333333e-05,
      "loss": 0.0043,
      "step": 13578
    },
    {
      "epoch": 25.053554939981534,
      "grad_norm": 0.5226263403892517,
      "learning_rate": 2.8808000000000003e-05,
      "loss": 0.0116,
      "step": 13579
    },
    {
      "epoch": 25.05540166204986,
      "grad_norm": 0.7630420923233032,
      "learning_rate": 2.8807666666666665e-05,
      "loss": 0.0162,
      "step": 13580
    },
    {
      "epoch": 25.05724838411819,
      "grad_norm": 0.17915871739387512,
      "learning_rate": 2.8807333333333335e-05,
      "loss": 0.0222,
      "step": 13581
    },
    {
      "epoch": 25.05909510618652,
      "grad_norm": 0.30622854828834534,
      "learning_rate": 2.8807000000000004e-05,
      "loss": 0.0036,
      "step": 13582
    },
    {
      "epoch": 25.060941828254848,
      "grad_norm": 0.16288454830646515,
      "learning_rate": 2.8806666666666666e-05,
      "loss": 0.0027,
      "step": 13583
    },
    {
      "epoch": 25.062788550323177,
      "grad_norm": 0.41513580083847046,
      "learning_rate": 2.8806333333333336e-05,
      "loss": 0.0124,
      "step": 13584
    },
    {
      "epoch": 25.064635272391506,
      "grad_norm": 0.28535616397857666,
      "learning_rate": 2.8806e-05,
      "loss": 0.0056,
      "step": 13585
    },
    {
      "epoch": 25.066481994459835,
      "grad_norm": 0.3623763620853424,
      "learning_rate": 2.8805666666666668e-05,
      "loss": 0.0071,
      "step": 13586
    },
    {
      "epoch": 25.068328716528164,
      "grad_norm": 0.14072564244270325,
      "learning_rate": 2.8805333333333334e-05,
      "loss": 0.0022,
      "step": 13587
    },
    {
      "epoch": 25.07017543859649,
      "grad_norm": 0.3874894082546234,
      "learning_rate": 2.8805e-05,
      "loss": 0.0103,
      "step": 13588
    },
    {
      "epoch": 25.07202216066482,
      "grad_norm": 0.08016061782836914,
      "learning_rate": 2.8804666666666665e-05,
      "loss": 0.0011,
      "step": 13589
    },
    {
      "epoch": 25.073868882733148,
      "grad_norm": 0.22445538640022278,
      "learning_rate": 2.8804333333333335e-05,
      "loss": 0.0054,
      "step": 13590
    },
    {
      "epoch": 25.075715604801477,
      "grad_norm": 0.11151864379644394,
      "learning_rate": 2.8804e-05,
      "loss": 0.0022,
      "step": 13591
    },
    {
      "epoch": 25.077562326869806,
      "grad_norm": 0.2736489176750183,
      "learning_rate": 2.8803666666666667e-05,
      "loss": 0.0089,
      "step": 13592
    },
    {
      "epoch": 25.079409048938135,
      "grad_norm": 0.14825448393821716,
      "learning_rate": 2.8803333333333336e-05,
      "loss": 0.0024,
      "step": 13593
    },
    {
      "epoch": 25.081255771006465,
      "grad_norm": 0.2685433626174927,
      "learning_rate": 2.8803e-05,
      "loss": 0.0032,
      "step": 13594
    },
    {
      "epoch": 25.083102493074794,
      "grad_norm": 0.17419521510601044,
      "learning_rate": 2.8802666666666668e-05,
      "loss": 0.0026,
      "step": 13595
    },
    {
      "epoch": 25.08494921514312,
      "grad_norm": 0.2433479279279709,
      "learning_rate": 2.8802333333333334e-05,
      "loss": 0.0058,
      "step": 13596
    },
    {
      "epoch": 25.08679593721145,
      "grad_norm": 0.15277212858200073,
      "learning_rate": 2.8802e-05,
      "loss": 0.0035,
      "step": 13597
    },
    {
      "epoch": 25.088642659279778,
      "grad_norm": 0.20970101654529572,
      "learning_rate": 2.8801666666666666e-05,
      "loss": 0.0017,
      "step": 13598
    },
    {
      "epoch": 25.090489381348107,
      "grad_norm": 0.12431564927101135,
      "learning_rate": 2.8801333333333335e-05,
      "loss": 0.0021,
      "step": 13599
    },
    {
      "epoch": 25.092336103416436,
      "grad_norm": 0.5341439843177795,
      "learning_rate": 2.8801e-05,
      "loss": 0.0074,
      "step": 13600
    },
    {
      "epoch": 25.094182825484765,
      "grad_norm": 0.7216334342956543,
      "learning_rate": 2.8800666666666667e-05,
      "loss": 0.1557,
      "step": 13601
    },
    {
      "epoch": 25.096029547553094,
      "grad_norm": 0.6455937027931213,
      "learning_rate": 2.8800333333333336e-05,
      "loss": 0.0688,
      "step": 13602
    },
    {
      "epoch": 25.097876269621423,
      "grad_norm": 0.36682724952697754,
      "learning_rate": 2.88e-05,
      "loss": 0.0601,
      "step": 13603
    },
    {
      "epoch": 25.099722991689752,
      "grad_norm": 0.42346009612083435,
      "learning_rate": 2.8799666666666668e-05,
      "loss": 0.0536,
      "step": 13604
    },
    {
      "epoch": 25.101569713758078,
      "grad_norm": 0.6256864666938782,
      "learning_rate": 2.8799333333333334e-05,
      "loss": 0.0467,
      "step": 13605
    },
    {
      "epoch": 25.103416435826407,
      "grad_norm": 0.35584160685539246,
      "learning_rate": 2.8799e-05,
      "loss": 0.0529,
      "step": 13606
    },
    {
      "epoch": 25.105263157894736,
      "grad_norm": 0.29889115691185,
      "learning_rate": 2.879866666666667e-05,
      "loss": 0.0317,
      "step": 13607
    },
    {
      "epoch": 25.107109879963065,
      "grad_norm": 0.37145760655403137,
      "learning_rate": 2.8798333333333335e-05,
      "loss": 0.0296,
      "step": 13608
    },
    {
      "epoch": 25.108956602031395,
      "grad_norm": 0.4296289086341858,
      "learning_rate": 2.8798e-05,
      "loss": 0.0252,
      "step": 13609
    },
    {
      "epoch": 25.110803324099724,
      "grad_norm": 0.3164059817790985,
      "learning_rate": 2.8797666666666667e-05,
      "loss": 0.0168,
      "step": 13610
    },
    {
      "epoch": 25.112650046168053,
      "grad_norm": 0.31333017349243164,
      "learning_rate": 2.8797333333333336e-05,
      "loss": 0.0176,
      "step": 13611
    },
    {
      "epoch": 25.114496768236382,
      "grad_norm": 0.31346407532691956,
      "learning_rate": 2.8797e-05,
      "loss": 0.0344,
      "step": 13612
    },
    {
      "epoch": 25.116343490304708,
      "grad_norm": 0.43452003598213196,
      "learning_rate": 2.8796666666666668e-05,
      "loss": 0.0455,
      "step": 13613
    },
    {
      "epoch": 25.118190212373037,
      "grad_norm": 0.3308866322040558,
      "learning_rate": 2.8796333333333334e-05,
      "loss": 0.0211,
      "step": 13614
    },
    {
      "epoch": 25.120036934441366,
      "grad_norm": 0.1883377730846405,
      "learning_rate": 2.8796e-05,
      "loss": 0.0078,
      "step": 13615
    },
    {
      "epoch": 25.121883656509695,
      "grad_norm": 0.15821512043476105,
      "learning_rate": 2.879566666666667e-05,
      "loss": 0.0049,
      "step": 13616
    },
    {
      "epoch": 25.123730378578024,
      "grad_norm": 0.21383026242256165,
      "learning_rate": 2.8795333333333332e-05,
      "loss": 0.0052,
      "step": 13617
    },
    {
      "epoch": 25.125577100646353,
      "grad_norm": 0.2946529984474182,
      "learning_rate": 2.8795e-05,
      "loss": 0.0084,
      "step": 13618
    },
    {
      "epoch": 25.127423822714682,
      "grad_norm": 0.5541951656341553,
      "learning_rate": 2.8794666666666667e-05,
      "loss": 0.0091,
      "step": 13619
    },
    {
      "epoch": 25.12927054478301,
      "grad_norm": 0.1361600160598755,
      "learning_rate": 2.8794333333333333e-05,
      "loss": 0.0047,
      "step": 13620
    },
    {
      "epoch": 25.131117266851337,
      "grad_norm": 0.18995541334152222,
      "learning_rate": 2.8794e-05,
      "loss": 0.0068,
      "step": 13621
    },
    {
      "epoch": 25.132963988919666,
      "grad_norm": 0.15388409793376923,
      "learning_rate": 2.879366666666667e-05,
      "loss": 0.0061,
      "step": 13622
    },
    {
      "epoch": 25.134810710987995,
      "grad_norm": 0.29169967770576477,
      "learning_rate": 2.879333333333333e-05,
      "loss": 0.0126,
      "step": 13623
    },
    {
      "epoch": 25.136657433056325,
      "grad_norm": 0.8038426041603088,
      "learning_rate": 2.8793e-05,
      "loss": 0.0102,
      "step": 13624
    },
    {
      "epoch": 25.138504155124654,
      "grad_norm": 0.24247054755687714,
      "learning_rate": 2.879266666666667e-05,
      "loss": 0.0061,
      "step": 13625
    },
    {
      "epoch": 25.140350877192983,
      "grad_norm": 1.249650001525879,
      "learning_rate": 2.8792333333333332e-05,
      "loss": 0.0122,
      "step": 13626
    },
    {
      "epoch": 25.142197599261312,
      "grad_norm": 0.27134811878204346,
      "learning_rate": 2.8792e-05,
      "loss": 0.0089,
      "step": 13627
    },
    {
      "epoch": 25.14404432132964,
      "grad_norm": 0.33459794521331787,
      "learning_rate": 2.8791666666666667e-05,
      "loss": 0.0147,
      "step": 13628
    },
    {
      "epoch": 25.14589104339797,
      "grad_norm": 0.42010146379470825,
      "learning_rate": 2.8791333333333333e-05,
      "loss": 0.0053,
      "step": 13629
    },
    {
      "epoch": 25.147737765466296,
      "grad_norm": 0.16870661079883575,
      "learning_rate": 2.8791e-05,
      "loss": 0.0063,
      "step": 13630
    },
    {
      "epoch": 25.149584487534625,
      "grad_norm": 0.2020006626844406,
      "learning_rate": 2.879066666666667e-05,
      "loss": 0.0045,
      "step": 13631
    },
    {
      "epoch": 25.151431209602954,
      "grad_norm": 0.273642361164093,
      "learning_rate": 2.8790333333333335e-05,
      "loss": 0.0271,
      "step": 13632
    },
    {
      "epoch": 25.153277931671283,
      "grad_norm": 0.32549455761909485,
      "learning_rate": 2.879e-05,
      "loss": 0.0083,
      "step": 13633
    },
    {
      "epoch": 25.155124653739612,
      "grad_norm": 0.3531877398490906,
      "learning_rate": 2.878966666666667e-05,
      "loss": 0.0069,
      "step": 13634
    },
    {
      "epoch": 25.15697137580794,
      "grad_norm": 0.16902898252010345,
      "learning_rate": 2.8789333333333332e-05,
      "loss": 0.0033,
      "step": 13635
    },
    {
      "epoch": 25.15881809787627,
      "grad_norm": 0.20311951637268066,
      "learning_rate": 2.8789e-05,
      "loss": 0.0038,
      "step": 13636
    },
    {
      "epoch": 25.1606648199446,
      "grad_norm": 0.15935593843460083,
      "learning_rate": 2.8788666666666668e-05,
      "loss": 0.0042,
      "step": 13637
    },
    {
      "epoch": 25.162511542012926,
      "grad_norm": 0.642253041267395,
      "learning_rate": 2.8788333333333334e-05,
      "loss": 0.0058,
      "step": 13638
    },
    {
      "epoch": 25.164358264081255,
      "grad_norm": 0.20698927342891693,
      "learning_rate": 2.8788e-05,
      "loss": 0.0045,
      "step": 13639
    },
    {
      "epoch": 25.166204986149584,
      "grad_norm": 0.16226263344287872,
      "learning_rate": 2.878766666666667e-05,
      "loss": 0.0027,
      "step": 13640
    },
    {
      "epoch": 25.168051708217913,
      "grad_norm": 0.19789494574069977,
      "learning_rate": 2.8787333333333335e-05,
      "loss": 0.0033,
      "step": 13641
    },
    {
      "epoch": 25.169898430286242,
      "grad_norm": 0.4175894856452942,
      "learning_rate": 2.8787e-05,
      "loss": 0.0059,
      "step": 13642
    },
    {
      "epoch": 25.17174515235457,
      "grad_norm": 0.25816839933395386,
      "learning_rate": 2.878666666666667e-05,
      "loss": 0.0048,
      "step": 13643
    },
    {
      "epoch": 25.1735918744229,
      "grad_norm": 0.24436330795288086,
      "learning_rate": 2.8786333333333333e-05,
      "loss": 0.006,
      "step": 13644
    },
    {
      "epoch": 25.17543859649123,
      "grad_norm": 0.26902177929878235,
      "learning_rate": 2.8786000000000002e-05,
      "loss": 0.0044,
      "step": 13645
    },
    {
      "epoch": 25.177285318559555,
      "grad_norm": 0.6639602184295654,
      "learning_rate": 2.8785666666666668e-05,
      "loss": 0.0166,
      "step": 13646
    },
    {
      "epoch": 25.179132040627884,
      "grad_norm": 0.05455506592988968,
      "learning_rate": 2.8785333333333334e-05,
      "loss": 0.0012,
      "step": 13647
    },
    {
      "epoch": 25.180978762696213,
      "grad_norm": 0.29231777787208557,
      "learning_rate": 2.8785e-05,
      "loss": 0.0068,
      "step": 13648
    },
    {
      "epoch": 25.182825484764543,
      "grad_norm": 0.08994735032320023,
      "learning_rate": 2.8784666666666666e-05,
      "loss": 0.0021,
      "step": 13649
    },
    {
      "epoch": 25.18467220683287,
      "grad_norm": 0.4141843318939209,
      "learning_rate": 2.8784333333333335e-05,
      "loss": 0.0105,
      "step": 13650
    },
    {
      "epoch": 25.1865189289012,
      "grad_norm": 0.45748546719551086,
      "learning_rate": 2.8784e-05,
      "loss": 0.1008,
      "step": 13651
    },
    {
      "epoch": 25.18836565096953,
      "grad_norm": 0.4383915662765503,
      "learning_rate": 2.8783666666666667e-05,
      "loss": 0.1024,
      "step": 13652
    },
    {
      "epoch": 25.19021237303786,
      "grad_norm": 0.5043680667877197,
      "learning_rate": 2.8783333333333333e-05,
      "loss": 0.1288,
      "step": 13653
    },
    {
      "epoch": 25.19205909510619,
      "grad_norm": 0.44734859466552734,
      "learning_rate": 2.8783000000000002e-05,
      "loss": 0.0724,
      "step": 13654
    },
    {
      "epoch": 25.193905817174514,
      "grad_norm": 0.7525545358657837,
      "learning_rate": 2.8782666666666665e-05,
      "loss": 0.0877,
      "step": 13655
    },
    {
      "epoch": 25.195752539242843,
      "grad_norm": 0.43058881163597107,
      "learning_rate": 2.8782333333333334e-05,
      "loss": 0.0608,
      "step": 13656
    },
    {
      "epoch": 25.197599261311172,
      "grad_norm": 0.6278600096702576,
      "learning_rate": 2.8782000000000003e-05,
      "loss": 0.0535,
      "step": 13657
    },
    {
      "epoch": 25.1994459833795,
      "grad_norm": 0.40241897106170654,
      "learning_rate": 2.8781666666666666e-05,
      "loss": 0.0471,
      "step": 13658
    },
    {
      "epoch": 25.20129270544783,
      "grad_norm": 0.35044482350349426,
      "learning_rate": 2.8781333333333335e-05,
      "loss": 0.0289,
      "step": 13659
    },
    {
      "epoch": 25.20313942751616,
      "grad_norm": 0.555040717124939,
      "learning_rate": 2.8781e-05,
      "loss": 0.0781,
      "step": 13660
    },
    {
      "epoch": 25.20498614958449,
      "grad_norm": 0.5007652640342712,
      "learning_rate": 2.8780666666666667e-05,
      "loss": 0.0326,
      "step": 13661
    },
    {
      "epoch": 25.206832871652818,
      "grad_norm": 0.4233347475528717,
      "learning_rate": 2.8780333333333333e-05,
      "loss": 0.033,
      "step": 13662
    },
    {
      "epoch": 25.208679593721143,
      "grad_norm": 0.4547087252140045,
      "learning_rate": 2.8780000000000002e-05,
      "loss": 0.0214,
      "step": 13663
    },
    {
      "epoch": 25.210526315789473,
      "grad_norm": 0.31385135650634766,
      "learning_rate": 2.8779666666666665e-05,
      "loss": 0.016,
      "step": 13664
    },
    {
      "epoch": 25.2123730378578,
      "grad_norm": 0.7495867609977722,
      "learning_rate": 2.8779333333333334e-05,
      "loss": 0.0244,
      "step": 13665
    },
    {
      "epoch": 25.21421975992613,
      "grad_norm": 0.4695708453655243,
      "learning_rate": 2.8779000000000003e-05,
      "loss": 0.013,
      "step": 13666
    },
    {
      "epoch": 25.21606648199446,
      "grad_norm": 0.16245493292808533,
      "learning_rate": 2.8778666666666666e-05,
      "loss": 0.0057,
      "step": 13667
    },
    {
      "epoch": 25.21791320406279,
      "grad_norm": 0.29763901233673096,
      "learning_rate": 2.8778333333333335e-05,
      "loss": 0.0103,
      "step": 13668
    },
    {
      "epoch": 25.21975992613112,
      "grad_norm": 0.1941731572151184,
      "learning_rate": 2.8778e-05,
      "loss": 0.0107,
      "step": 13669
    },
    {
      "epoch": 25.221606648199447,
      "grad_norm": 0.3459717929363251,
      "learning_rate": 2.8777666666666667e-05,
      "loss": 0.0147,
      "step": 13670
    },
    {
      "epoch": 25.223453370267773,
      "grad_norm": 0.375107079744339,
      "learning_rate": 2.8777333333333333e-05,
      "loss": 0.0133,
      "step": 13671
    },
    {
      "epoch": 25.225300092336102,
      "grad_norm": 0.3817448616027832,
      "learning_rate": 2.8777000000000002e-05,
      "loss": 0.0231,
      "step": 13672
    },
    {
      "epoch": 25.22714681440443,
      "grad_norm": 0.2049887627363205,
      "learning_rate": 2.8776666666666665e-05,
      "loss": 0.002,
      "step": 13673
    },
    {
      "epoch": 25.22899353647276,
      "grad_norm": 0.3521776497364044,
      "learning_rate": 2.8776333333333334e-05,
      "loss": 0.0086,
      "step": 13674
    },
    {
      "epoch": 25.23084025854109,
      "grad_norm": 0.04273456335067749,
      "learning_rate": 2.8776000000000004e-05,
      "loss": 0.0011,
      "step": 13675
    },
    {
      "epoch": 25.23268698060942,
      "grad_norm": 0.2826165556907654,
      "learning_rate": 2.8775666666666666e-05,
      "loss": 0.0071,
      "step": 13676
    },
    {
      "epoch": 25.234533702677748,
      "grad_norm": 0.17444173991680145,
      "learning_rate": 2.8775333333333336e-05,
      "loss": 0.0072,
      "step": 13677
    },
    {
      "epoch": 25.236380424746077,
      "grad_norm": 0.27987492084503174,
      "learning_rate": 2.8774999999999998e-05,
      "loss": 0.0048,
      "step": 13678
    },
    {
      "epoch": 25.238227146814406,
      "grad_norm": 0.35382506251335144,
      "learning_rate": 2.8774666666666667e-05,
      "loss": 0.0044,
      "step": 13679
    },
    {
      "epoch": 25.24007386888273,
      "grad_norm": 0.22350679337978363,
      "learning_rate": 2.8774333333333333e-05,
      "loss": 0.0063,
      "step": 13680
    },
    {
      "epoch": 25.24192059095106,
      "grad_norm": 0.2280522584915161,
      "learning_rate": 2.8774e-05,
      "loss": 0.0049,
      "step": 13681
    },
    {
      "epoch": 25.24376731301939,
      "grad_norm": 0.22186324000358582,
      "learning_rate": 2.877366666666667e-05,
      "loss": 0.0064,
      "step": 13682
    },
    {
      "epoch": 25.24561403508772,
      "grad_norm": 0.30040767788887024,
      "learning_rate": 2.8773333333333335e-05,
      "loss": 0.0096,
      "step": 13683
    },
    {
      "epoch": 25.24746075715605,
      "grad_norm": 0.12468577921390533,
      "learning_rate": 2.8773e-05,
      "loss": 0.0029,
      "step": 13684
    },
    {
      "epoch": 25.249307479224377,
      "grad_norm": 0.2888202667236328,
      "learning_rate": 2.8772666666666666e-05,
      "loss": 0.0055,
      "step": 13685
    },
    {
      "epoch": 25.251154201292707,
      "grad_norm": 0.49866822361946106,
      "learning_rate": 2.8772333333333336e-05,
      "loss": 0.0086,
      "step": 13686
    },
    {
      "epoch": 25.253000923361036,
      "grad_norm": 0.31418851017951965,
      "learning_rate": 2.8771999999999998e-05,
      "loss": 0.0069,
      "step": 13687
    },
    {
      "epoch": 25.25484764542936,
      "grad_norm": 0.49223172664642334,
      "learning_rate": 2.8771666666666668e-05,
      "loss": 0.0102,
      "step": 13688
    },
    {
      "epoch": 25.25669436749769,
      "grad_norm": 0.16645334661006927,
      "learning_rate": 2.8771333333333334e-05,
      "loss": 0.0056,
      "step": 13689
    },
    {
      "epoch": 25.25854108956602,
      "grad_norm": 0.23253782093524933,
      "learning_rate": 2.8771e-05,
      "loss": 0.0042,
      "step": 13690
    },
    {
      "epoch": 25.26038781163435,
      "grad_norm": 0.17528776824474335,
      "learning_rate": 2.877066666666667e-05,
      "loss": 0.005,
      "step": 13691
    },
    {
      "epoch": 25.262234533702678,
      "grad_norm": 0.1108769103884697,
      "learning_rate": 2.8770333333333335e-05,
      "loss": 0.0033,
      "step": 13692
    },
    {
      "epoch": 25.264081255771007,
      "grad_norm": 0.32757946848869324,
      "learning_rate": 2.877e-05,
      "loss": 0.0039,
      "step": 13693
    },
    {
      "epoch": 25.265927977839336,
      "grad_norm": 0.29985570907592773,
      "learning_rate": 2.8769666666666667e-05,
      "loss": 0.0072,
      "step": 13694
    },
    {
      "epoch": 25.267774699907665,
      "grad_norm": 0.16443781554698944,
      "learning_rate": 2.8769333333333336e-05,
      "loss": 0.0048,
      "step": 13695
    },
    {
      "epoch": 25.26962142197599,
      "grad_norm": 0.15562738478183746,
      "learning_rate": 2.8769e-05,
      "loss": 0.003,
      "step": 13696
    },
    {
      "epoch": 25.27146814404432,
      "grad_norm": 0.29432448744773865,
      "learning_rate": 2.8768666666666668e-05,
      "loss": 0.0039,
      "step": 13697
    },
    {
      "epoch": 25.27331486611265,
      "grad_norm": 0.265778511762619,
      "learning_rate": 2.8768333333333334e-05,
      "loss": 0.0178,
      "step": 13698
    },
    {
      "epoch": 25.27516158818098,
      "grad_norm": 0.32704463601112366,
      "learning_rate": 2.8768e-05,
      "loss": 0.0085,
      "step": 13699
    },
    {
      "epoch": 25.277008310249307,
      "grad_norm": 0.0786413848400116,
      "learning_rate": 2.876766666666667e-05,
      "loss": 0.0009,
      "step": 13700
    },
    {
      "epoch": 25.278855032317637,
      "grad_norm": 0.43745529651641846,
      "learning_rate": 2.8767333333333335e-05,
      "loss": 0.1118,
      "step": 13701
    },
    {
      "epoch": 25.280701754385966,
      "grad_norm": 0.6298274993896484,
      "learning_rate": 2.8767e-05,
      "loss": 0.0922,
      "step": 13702
    },
    {
      "epoch": 25.282548476454295,
      "grad_norm": 0.3927988111972809,
      "learning_rate": 2.8766666666666667e-05,
      "loss": 0.076,
      "step": 13703
    },
    {
      "epoch": 25.284395198522624,
      "grad_norm": 0.34882792830467224,
      "learning_rate": 2.8766333333333336e-05,
      "loss": 0.055,
      "step": 13704
    },
    {
      "epoch": 25.28624192059095,
      "grad_norm": 0.31682804226875305,
      "learning_rate": 2.8766e-05,
      "loss": 0.0384,
      "step": 13705
    },
    {
      "epoch": 25.28808864265928,
      "grad_norm": 0.4535244405269623,
      "learning_rate": 2.8765666666666668e-05,
      "loss": 0.0295,
      "step": 13706
    },
    {
      "epoch": 25.289935364727608,
      "grad_norm": 0.4537026584148407,
      "learning_rate": 2.8765333333333337e-05,
      "loss": 0.0274,
      "step": 13707
    },
    {
      "epoch": 25.291782086795937,
      "grad_norm": 0.30101144313812256,
      "learning_rate": 2.8765e-05,
      "loss": 0.0326,
      "step": 13708
    },
    {
      "epoch": 25.293628808864266,
      "grad_norm": 0.6969102621078491,
      "learning_rate": 2.876466666666667e-05,
      "loss": 0.0912,
      "step": 13709
    },
    {
      "epoch": 25.295475530932595,
      "grad_norm": 0.3315484821796417,
      "learning_rate": 2.8764333333333332e-05,
      "loss": 0.0395,
      "step": 13710
    },
    {
      "epoch": 25.297322253000925,
      "grad_norm": 0.2811791002750397,
      "learning_rate": 2.8764e-05,
      "loss": 0.0426,
      "step": 13711
    },
    {
      "epoch": 25.299168975069254,
      "grad_norm": 0.4676089584827423,
      "learning_rate": 2.8763666666666667e-05,
      "loss": 0.0153,
      "step": 13712
    },
    {
      "epoch": 25.30101569713758,
      "grad_norm": 0.5364266633987427,
      "learning_rate": 2.8763333333333333e-05,
      "loss": 0.0456,
      "step": 13713
    },
    {
      "epoch": 25.30286241920591,
      "grad_norm": 0.25764209032058716,
      "learning_rate": 2.8763e-05,
      "loss": 0.0225,
      "step": 13714
    },
    {
      "epoch": 25.304709141274238,
      "grad_norm": 0.4249553680419922,
      "learning_rate": 2.8762666666666668e-05,
      "loss": 0.0166,
      "step": 13715
    },
    {
      "epoch": 25.306555863342567,
      "grad_norm": 0.3083512783050537,
      "learning_rate": 2.8762333333333334e-05,
      "loss": 0.0352,
      "step": 13716
    },
    {
      "epoch": 25.308402585410896,
      "grad_norm": 0.2641586363315582,
      "learning_rate": 2.8762e-05,
      "loss": 0.0079,
      "step": 13717
    },
    {
      "epoch": 25.310249307479225,
      "grad_norm": 0.2892804443836212,
      "learning_rate": 2.876166666666667e-05,
      "loss": 0.0073,
      "step": 13718
    },
    {
      "epoch": 25.312096029547554,
      "grad_norm": 0.3115205466747284,
      "learning_rate": 2.8761333333333332e-05,
      "loss": 0.0058,
      "step": 13719
    },
    {
      "epoch": 25.313942751615883,
      "grad_norm": 0.6394883990287781,
      "learning_rate": 2.8761e-05,
      "loss": 0.0246,
      "step": 13720
    },
    {
      "epoch": 25.31578947368421,
      "grad_norm": 0.29537880420684814,
      "learning_rate": 2.8760666666666667e-05,
      "loss": 0.0052,
      "step": 13721
    },
    {
      "epoch": 25.317636195752538,
      "grad_norm": 0.39419668912887573,
      "learning_rate": 2.8760333333333333e-05,
      "loss": 0.0063,
      "step": 13722
    },
    {
      "epoch": 25.319482917820867,
      "grad_norm": 0.11710794270038605,
      "learning_rate": 2.876e-05,
      "loss": 0.0033,
      "step": 13723
    },
    {
      "epoch": 25.321329639889196,
      "grad_norm": 0.20329396426677704,
      "learning_rate": 2.875966666666667e-05,
      "loss": 0.0032,
      "step": 13724
    },
    {
      "epoch": 25.323176361957525,
      "grad_norm": 0.1527843326330185,
      "learning_rate": 2.8759333333333334e-05,
      "loss": 0.0025,
      "step": 13725
    },
    {
      "epoch": 25.325023084025855,
      "grad_norm": 0.16141752898693085,
      "learning_rate": 2.8759e-05,
      "loss": 0.0057,
      "step": 13726
    },
    {
      "epoch": 25.326869806094184,
      "grad_norm": 0.2185315191745758,
      "learning_rate": 2.875866666666667e-05,
      "loss": 0.008,
      "step": 13727
    },
    {
      "epoch": 25.328716528162513,
      "grad_norm": 0.23512014746665955,
      "learning_rate": 2.8758333333333332e-05,
      "loss": 0.0059,
      "step": 13728
    },
    {
      "epoch": 25.330563250230842,
      "grad_norm": 0.18767587840557098,
      "learning_rate": 2.8758e-05,
      "loss": 0.0064,
      "step": 13729
    },
    {
      "epoch": 25.332409972299168,
      "grad_norm": 0.24704614281654358,
      "learning_rate": 2.8757666666666667e-05,
      "loss": 0.0064,
      "step": 13730
    },
    {
      "epoch": 25.334256694367497,
      "grad_norm": 0.7131418585777283,
      "learning_rate": 2.8757333333333333e-05,
      "loss": 0.0059,
      "step": 13731
    },
    {
      "epoch": 25.336103416435826,
      "grad_norm": 0.20158667862415314,
      "learning_rate": 2.8757000000000003e-05,
      "loss": 0.0052,
      "step": 13732
    },
    {
      "epoch": 25.337950138504155,
      "grad_norm": 0.23540963232517242,
      "learning_rate": 2.875666666666667e-05,
      "loss": 0.0063,
      "step": 13733
    },
    {
      "epoch": 25.339796860572484,
      "grad_norm": 0.15179704129695892,
      "learning_rate": 2.8756333333333335e-05,
      "loss": 0.004,
      "step": 13734
    },
    {
      "epoch": 25.341643582640813,
      "grad_norm": 0.3278792202472687,
      "learning_rate": 2.8756e-05,
      "loss": 0.0057,
      "step": 13735
    },
    {
      "epoch": 25.343490304709142,
      "grad_norm": 0.2591707408428192,
      "learning_rate": 2.875566666666667e-05,
      "loss": 0.0204,
      "step": 13736
    },
    {
      "epoch": 25.34533702677747,
      "grad_norm": 0.1418246179819107,
      "learning_rate": 2.8755333333333332e-05,
      "loss": 0.0036,
      "step": 13737
    },
    {
      "epoch": 25.347183748845797,
      "grad_norm": 0.16631825268268585,
      "learning_rate": 2.8755e-05,
      "loss": 0.0046,
      "step": 13738
    },
    {
      "epoch": 25.349030470914126,
      "grad_norm": 0.06749407947063446,
      "learning_rate": 2.8754666666666664e-05,
      "loss": 0.0022,
      "step": 13739
    },
    {
      "epoch": 25.350877192982455,
      "grad_norm": 0.07478956133127213,
      "learning_rate": 2.8754333333333334e-05,
      "loss": 0.0022,
      "step": 13740
    },
    {
      "epoch": 25.352723915050785,
      "grad_norm": 0.5097907185554504,
      "learning_rate": 2.8754000000000003e-05,
      "loss": 0.0154,
      "step": 13741
    },
    {
      "epoch": 25.354570637119114,
      "grad_norm": 0.1690087914466858,
      "learning_rate": 2.8753666666666665e-05,
      "loss": 0.0044,
      "step": 13742
    },
    {
      "epoch": 25.356417359187443,
      "grad_norm": 0.2194225937128067,
      "learning_rate": 2.8753333333333335e-05,
      "loss": 0.007,
      "step": 13743
    },
    {
      "epoch": 25.358264081255772,
      "grad_norm": 0.14810651540756226,
      "learning_rate": 2.8753e-05,
      "loss": 0.0057,
      "step": 13744
    },
    {
      "epoch": 25.3601108033241,
      "grad_norm": 0.21244044601917267,
      "learning_rate": 2.8752666666666667e-05,
      "loss": 0.0078,
      "step": 13745
    },
    {
      "epoch": 25.361957525392427,
      "grad_norm": 1.2006323337554932,
      "learning_rate": 2.8752333333333333e-05,
      "loss": 0.0092,
      "step": 13746
    },
    {
      "epoch": 25.363804247460756,
      "grad_norm": 0.177195206284523,
      "learning_rate": 2.8752000000000002e-05,
      "loss": 0.0025,
      "step": 13747
    },
    {
      "epoch": 25.365650969529085,
      "grad_norm": 0.3909979462623596,
      "learning_rate": 2.8751666666666664e-05,
      "loss": 0.0049,
      "step": 13748
    },
    {
      "epoch": 25.367497691597414,
      "grad_norm": 1.1116374731063843,
      "learning_rate": 2.8751333333333334e-05,
      "loss": 0.017,
      "step": 13749
    },
    {
      "epoch": 25.369344413665743,
      "grad_norm": 0.2975039780139923,
      "learning_rate": 2.8751000000000003e-05,
      "loss": 0.017,
      "step": 13750
    },
    {
      "epoch": 25.371191135734072,
      "grad_norm": NaN,
      "learning_rate": 2.8751000000000003e-05,
      "loss": 0.1609,
      "step": 13751
    },
    {
      "epoch": 25.3730378578024,
      "grad_norm": 0.5848435759544373,
      "learning_rate": 2.8750666666666666e-05,
      "loss": 0.1283,
      "step": 13752
    },
    {
      "epoch": 25.37488457987073,
      "grad_norm": 0.435305118560791,
      "learning_rate": 2.8750333333333335e-05,
      "loss": 0.0762,
      "step": 13753
    },
    {
      "epoch": 25.37673130193906,
      "grad_norm": 1.2225451469421387,
      "learning_rate": 2.875e-05,
      "loss": 0.0796,
      "step": 13754
    },
    {
      "epoch": 25.378578024007385,
      "grad_norm": 0.7846620678901672,
      "learning_rate": 2.8749666666666667e-05,
      "loss": 0.0573,
      "step": 13755
    },
    {
      "epoch": 25.380424746075715,
      "grad_norm": 0.41929861903190613,
      "learning_rate": 2.8749333333333333e-05,
      "loss": 0.055,
      "step": 13756
    },
    {
      "epoch": 25.382271468144044,
      "grad_norm": 0.5668908357620239,
      "learning_rate": 2.8749000000000002e-05,
      "loss": 0.0698,
      "step": 13757
    },
    {
      "epoch": 25.384118190212373,
      "grad_norm": 0.4807840883731842,
      "learning_rate": 2.8748666666666668e-05,
      "loss": 0.0487,
      "step": 13758
    },
    {
      "epoch": 25.385964912280702,
      "grad_norm": 0.503312885761261,
      "learning_rate": 2.8748333333333334e-05,
      "loss": 0.0667,
      "step": 13759
    },
    {
      "epoch": 25.38781163434903,
      "grad_norm": 0.6881386041641235,
      "learning_rate": 2.8748000000000003e-05,
      "loss": 0.0217,
      "step": 13760
    },
    {
      "epoch": 25.38965835641736,
      "grad_norm": 0.3284233510494232,
      "learning_rate": 2.8747666666666666e-05,
      "loss": 0.0201,
      "step": 13761
    },
    {
      "epoch": 25.39150507848569,
      "grad_norm": 0.49530842900276184,
      "learning_rate": 2.8747333333333335e-05,
      "loss": 0.0245,
      "step": 13762
    },
    {
      "epoch": 25.393351800554015,
      "grad_norm": 0.3061735928058624,
      "learning_rate": 2.8747e-05,
      "loss": 0.0183,
      "step": 13763
    },
    {
      "epoch": 25.395198522622344,
      "grad_norm": 0.29596859216690063,
      "learning_rate": 2.8746666666666667e-05,
      "loss": 0.03,
      "step": 13764
    },
    {
      "epoch": 25.397045244690673,
      "grad_norm": 0.27488064765930176,
      "learning_rate": 2.8746333333333333e-05,
      "loss": 0.0253,
      "step": 13765
    },
    {
      "epoch": 25.398891966759003,
      "grad_norm": 0.3477145731449127,
      "learning_rate": 2.8746000000000002e-05,
      "loss": 0.0144,
      "step": 13766
    },
    {
      "epoch": 25.40073868882733,
      "grad_norm": 0.319806843996048,
      "learning_rate": 2.8745666666666668e-05,
      "loss": 0.0111,
      "step": 13767
    },
    {
      "epoch": 25.40258541089566,
      "grad_norm": 0.3649977147579193,
      "learning_rate": 2.8745333333333334e-05,
      "loss": 0.022,
      "step": 13768
    },
    {
      "epoch": 25.40443213296399,
      "grad_norm": 0.11577984690666199,
      "learning_rate": 2.8745000000000003e-05,
      "loss": 0.0039,
      "step": 13769
    },
    {
      "epoch": 25.40627885503232,
      "grad_norm": 0.1643013060092926,
      "learning_rate": 2.8744666666666666e-05,
      "loss": 0.0048,
      "step": 13770
    },
    {
      "epoch": 25.408125577100645,
      "grad_norm": 0.25984930992126465,
      "learning_rate": 2.8744333333333335e-05,
      "loss": 0.0058,
      "step": 13771
    },
    {
      "epoch": 25.409972299168974,
      "grad_norm": 0.37410005927085876,
      "learning_rate": 2.8743999999999998e-05,
      "loss": 0.0093,
      "step": 13772
    },
    {
      "epoch": 25.411819021237303,
      "grad_norm": 0.5347874164581299,
      "learning_rate": 2.8743666666666667e-05,
      "loss": 0.0103,
      "step": 13773
    },
    {
      "epoch": 25.413665743305632,
      "grad_norm": 0.30911314487457275,
      "learning_rate": 2.8743333333333333e-05,
      "loss": 0.0097,
      "step": 13774
    },
    {
      "epoch": 25.41551246537396,
      "grad_norm": 0.2698220908641815,
      "learning_rate": 2.8743e-05,
      "loss": 0.006,
      "step": 13775
    },
    {
      "epoch": 25.41735918744229,
      "grad_norm": 0.3670751750469208,
      "learning_rate": 2.874266666666667e-05,
      "loss": 0.0166,
      "step": 13776
    },
    {
      "epoch": 25.41920590951062,
      "grad_norm": 0.31863951683044434,
      "learning_rate": 2.8742333333333334e-05,
      "loss": 0.0046,
      "step": 13777
    },
    {
      "epoch": 25.42105263157895,
      "grad_norm": 0.3697512447834015,
      "learning_rate": 2.8742e-05,
      "loss": 0.0069,
      "step": 13778
    },
    {
      "epoch": 25.422899353647278,
      "grad_norm": 0.1975865662097931,
      "learning_rate": 2.8741666666666666e-05,
      "loss": 0.0191,
      "step": 13779
    },
    {
      "epoch": 25.424746075715603,
      "grad_norm": 0.189303457736969,
      "learning_rate": 2.8741333333333336e-05,
      "loss": 0.003,
      "step": 13780
    },
    {
      "epoch": 25.426592797783933,
      "grad_norm": 0.747004508972168,
      "learning_rate": 2.8740999999999998e-05,
      "loss": 0.0108,
      "step": 13781
    },
    {
      "epoch": 25.42843951985226,
      "grad_norm": 0.07769419997930527,
      "learning_rate": 2.8740666666666667e-05,
      "loss": 0.0022,
      "step": 13782
    },
    {
      "epoch": 25.43028624192059,
      "grad_norm": 0.15571707487106323,
      "learning_rate": 2.8740333333333337e-05,
      "loss": 0.0038,
      "step": 13783
    },
    {
      "epoch": 25.43213296398892,
      "grad_norm": 0.06311183422803879,
      "learning_rate": 2.874e-05,
      "loss": 0.0021,
      "step": 13784
    },
    {
      "epoch": 25.43397968605725,
      "grad_norm": 0.12324243038892746,
      "learning_rate": 2.873966666666667e-05,
      "loss": 0.0031,
      "step": 13785
    },
    {
      "epoch": 25.43582640812558,
      "grad_norm": 1.1392844915390015,
      "learning_rate": 2.8739333333333335e-05,
      "loss": 0.0103,
      "step": 13786
    },
    {
      "epoch": 25.437673130193907,
      "grad_norm": 0.33735352754592896,
      "learning_rate": 2.8739e-05,
      "loss": 0.0333,
      "step": 13787
    },
    {
      "epoch": 25.439519852262233,
      "grad_norm": 0.18859614431858063,
      "learning_rate": 2.8738666666666666e-05,
      "loss": 0.0037,
      "step": 13788
    },
    {
      "epoch": 25.441366574330562,
      "grad_norm": 0.4658453166484833,
      "learning_rate": 2.8738333333333336e-05,
      "loss": 0.0044,
      "step": 13789
    },
    {
      "epoch": 25.44321329639889,
      "grad_norm": 0.29348111152648926,
      "learning_rate": 2.8737999999999998e-05,
      "loss": 0.0084,
      "step": 13790
    },
    {
      "epoch": 25.44506001846722,
      "grad_norm": 0.17754070460796356,
      "learning_rate": 2.8737666666666668e-05,
      "loss": 0.0055,
      "step": 13791
    },
    {
      "epoch": 25.44690674053555,
      "grad_norm": 0.2359544187784195,
      "learning_rate": 2.8737333333333337e-05,
      "loss": 0.0039,
      "step": 13792
    },
    {
      "epoch": 25.44875346260388,
      "grad_norm": 0.11587945371866226,
      "learning_rate": 2.8737e-05,
      "loss": 0.0019,
      "step": 13793
    },
    {
      "epoch": 25.450600184672208,
      "grad_norm": 0.3465985357761383,
      "learning_rate": 2.873666666666667e-05,
      "loss": 0.0096,
      "step": 13794
    },
    {
      "epoch": 25.452446906740537,
      "grad_norm": 0.31631413102149963,
      "learning_rate": 2.8736333333333335e-05,
      "loss": 0.0063,
      "step": 13795
    },
    {
      "epoch": 25.454293628808863,
      "grad_norm": 0.4622824788093567,
      "learning_rate": 2.8736e-05,
      "loss": 0.0081,
      "step": 13796
    },
    {
      "epoch": 25.45614035087719,
      "grad_norm": 0.1349441111087799,
      "learning_rate": 2.8735666666666667e-05,
      "loss": 0.0029,
      "step": 13797
    },
    {
      "epoch": 25.45798707294552,
      "grad_norm": 0.5047300457954407,
      "learning_rate": 2.8735333333333336e-05,
      "loss": 0.0045,
      "step": 13798
    },
    {
      "epoch": 25.45983379501385,
      "grad_norm": 0.4179021716117859,
      "learning_rate": 2.8735e-05,
      "loss": 0.0098,
      "step": 13799
    },
    {
      "epoch": 25.46168051708218,
      "grad_norm": 0.23024718463420868,
      "learning_rate": 2.8734666666666668e-05,
      "loss": 0.0035,
      "step": 13800
    },
    {
      "epoch": 25.46352723915051,
      "grad_norm": 0.7658423781394958,
      "learning_rate": 2.8734333333333334e-05,
      "loss": 0.1122,
      "step": 13801
    },
    {
      "epoch": 25.465373961218837,
      "grad_norm": 0.4380490183830261,
      "learning_rate": 2.8734e-05,
      "loss": 0.1327,
      "step": 13802
    },
    {
      "epoch": 25.467220683287167,
      "grad_norm": 0.5566225647926331,
      "learning_rate": 2.873366666666667e-05,
      "loss": 0.0584,
      "step": 13803
    },
    {
      "epoch": 25.469067405355496,
      "grad_norm": 1.2810173034667969,
      "learning_rate": 2.873333333333333e-05,
      "loss": 0.102,
      "step": 13804
    },
    {
      "epoch": 25.47091412742382,
      "grad_norm": 0.5459073781967163,
      "learning_rate": 2.8733e-05,
      "loss": 0.0893,
      "step": 13805
    },
    {
      "epoch": 25.47276084949215,
      "grad_norm": 0.4217475652694702,
      "learning_rate": 2.8732666666666667e-05,
      "loss": 0.0529,
      "step": 13806
    },
    {
      "epoch": 25.47460757156048,
      "grad_norm": 0.3302052617073059,
      "learning_rate": 2.8732333333333333e-05,
      "loss": 0.0353,
      "step": 13807
    },
    {
      "epoch": 25.47645429362881,
      "grad_norm": 0.39502203464508057,
      "learning_rate": 2.8732000000000002e-05,
      "loss": 0.053,
      "step": 13808
    },
    {
      "epoch": 25.478301015697138,
      "grad_norm": 0.3716045618057251,
      "learning_rate": 2.8731666666666668e-05,
      "loss": 0.0433,
      "step": 13809
    },
    {
      "epoch": 25.480147737765467,
      "grad_norm": 0.30203694105148315,
      "learning_rate": 2.8731333333333334e-05,
      "loss": 0.025,
      "step": 13810
    },
    {
      "epoch": 25.481994459833796,
      "grad_norm": 0.621515691280365,
      "learning_rate": 2.8731e-05,
      "loss": 0.0464,
      "step": 13811
    },
    {
      "epoch": 25.483841181902125,
      "grad_norm": 0.2422146201133728,
      "learning_rate": 2.873066666666667e-05,
      "loss": 0.013,
      "step": 13812
    },
    {
      "epoch": 25.48568790397045,
      "grad_norm": 0.3242398500442505,
      "learning_rate": 2.8730333333333332e-05,
      "loss": 0.0166,
      "step": 13813
    },
    {
      "epoch": 25.48753462603878,
      "grad_norm": 0.5535221099853516,
      "learning_rate": 2.873e-05,
      "loss": 0.0324,
      "step": 13814
    },
    {
      "epoch": 25.48938134810711,
      "grad_norm": 0.19889982044696808,
      "learning_rate": 2.8729666666666667e-05,
      "loss": 0.0244,
      "step": 13815
    },
    {
      "epoch": 25.49122807017544,
      "grad_norm": 0.14758522808551788,
      "learning_rate": 2.8729333333333333e-05,
      "loss": 0.0065,
      "step": 13816
    },
    {
      "epoch": 25.493074792243767,
      "grad_norm": 0.207974374294281,
      "learning_rate": 2.8729000000000002e-05,
      "loss": 0.0082,
      "step": 13817
    },
    {
      "epoch": 25.494921514312097,
      "grad_norm": 0.2960265278816223,
      "learning_rate": 2.8728666666666668e-05,
      "loss": 0.0108,
      "step": 13818
    },
    {
      "epoch": 25.496768236380426,
      "grad_norm": 0.48034244775772095,
      "learning_rate": 2.8728333333333334e-05,
      "loss": 0.0098,
      "step": 13819
    },
    {
      "epoch": 25.498614958448755,
      "grad_norm": 0.2875449061393738,
      "learning_rate": 2.8728e-05,
      "loss": 0.0097,
      "step": 13820
    },
    {
      "epoch": 25.50046168051708,
      "grad_norm": 0.2575296461582184,
      "learning_rate": 2.872766666666667e-05,
      "loss": 0.0059,
      "step": 13821
    },
    {
      "epoch": 25.50230840258541,
      "grad_norm": 0.19673795998096466,
      "learning_rate": 2.8727333333333332e-05,
      "loss": 0.0048,
      "step": 13822
    },
    {
      "epoch": 25.50415512465374,
      "grad_norm": 0.6091803908348083,
      "learning_rate": 2.8727e-05,
      "loss": 0.005,
      "step": 13823
    },
    {
      "epoch": 25.506001846722068,
      "grad_norm": 0.26289331912994385,
      "learning_rate": 2.8726666666666667e-05,
      "loss": 0.0073,
      "step": 13824
    },
    {
      "epoch": 25.507848568790397,
      "grad_norm": 0.2928977310657501,
      "learning_rate": 2.8726333333333333e-05,
      "loss": 0.0194,
      "step": 13825
    },
    {
      "epoch": 25.509695290858726,
      "grad_norm": 0.1658514142036438,
      "learning_rate": 2.8726000000000002e-05,
      "loss": 0.0038,
      "step": 13826
    },
    {
      "epoch": 25.511542012927055,
      "grad_norm": 0.5831472277641296,
      "learning_rate": 2.872566666666667e-05,
      "loss": 0.011,
      "step": 13827
    },
    {
      "epoch": 25.513388734995385,
      "grad_norm": 0.22511175274848938,
      "learning_rate": 2.8725333333333334e-05,
      "loss": 0.0051,
      "step": 13828
    },
    {
      "epoch": 25.51523545706371,
      "grad_norm": 0.2004740834236145,
      "learning_rate": 2.8725e-05,
      "loss": 0.0034,
      "step": 13829
    },
    {
      "epoch": 25.51708217913204,
      "grad_norm": 0.4446840286254883,
      "learning_rate": 2.8724666666666666e-05,
      "loss": 0.0075,
      "step": 13830
    },
    {
      "epoch": 25.51892890120037,
      "grad_norm": 0.45033150911331177,
      "learning_rate": 2.8724333333333332e-05,
      "loss": 0.0082,
      "step": 13831
    },
    {
      "epoch": 25.520775623268698,
      "grad_norm": 0.17718258500099182,
      "learning_rate": 2.8724e-05,
      "loss": 0.0063,
      "step": 13832
    },
    {
      "epoch": 25.522622345337027,
      "grad_norm": 0.17210236191749573,
      "learning_rate": 2.8723666666666667e-05,
      "loss": 0.0042,
      "step": 13833
    },
    {
      "epoch": 25.524469067405356,
      "grad_norm": 0.43359464406967163,
      "learning_rate": 2.8723333333333333e-05,
      "loss": 0.0059,
      "step": 13834
    },
    {
      "epoch": 25.526315789473685,
      "grad_norm": 0.3382089138031006,
      "learning_rate": 2.8723000000000003e-05,
      "loss": 0.0094,
      "step": 13835
    },
    {
      "epoch": 25.528162511542014,
      "grad_norm": 0.7305532693862915,
      "learning_rate": 2.8722666666666665e-05,
      "loss": 0.0186,
      "step": 13836
    },
    {
      "epoch": 25.530009233610343,
      "grad_norm": 0.691562831401825,
      "learning_rate": 2.8722333333333335e-05,
      "loss": 0.0353,
      "step": 13837
    },
    {
      "epoch": 25.53185595567867,
      "grad_norm": 0.22980351746082306,
      "learning_rate": 2.8722e-05,
      "loss": 0.0069,
      "step": 13838
    },
    {
      "epoch": 25.533702677746998,
      "grad_norm": 1.1313012838363647,
      "learning_rate": 2.8721666666666666e-05,
      "loss": 0.0122,
      "step": 13839
    },
    {
      "epoch": 25.535549399815327,
      "grad_norm": 0.2683294713497162,
      "learning_rate": 2.8721333333333332e-05,
      "loss": 0.0052,
      "step": 13840
    },
    {
      "epoch": 25.537396121883656,
      "grad_norm": 0.4816579520702362,
      "learning_rate": 2.8721e-05,
      "loss": 0.0124,
      "step": 13841
    },
    {
      "epoch": 25.539242843951985,
      "grad_norm": 0.2540108859539032,
      "learning_rate": 2.8720666666666668e-05,
      "loss": 0.0043,
      "step": 13842
    },
    {
      "epoch": 25.541089566020315,
      "grad_norm": 0.16645418107509613,
      "learning_rate": 2.8720333333333334e-05,
      "loss": 0.0051,
      "step": 13843
    },
    {
      "epoch": 25.542936288088644,
      "grad_norm": 0.447847843170166,
      "learning_rate": 2.8720000000000003e-05,
      "loss": 0.0087,
      "step": 13844
    },
    {
      "epoch": 25.544783010156973,
      "grad_norm": 0.4923630952835083,
      "learning_rate": 2.8719666666666665e-05,
      "loss": 0.0082,
      "step": 13845
    },
    {
      "epoch": 25.5466297322253,
      "grad_norm": 0.19476549327373505,
      "learning_rate": 2.8719333333333335e-05,
      "loss": 0.0048,
      "step": 13846
    },
    {
      "epoch": 25.548476454293628,
      "grad_norm": 0.5130172371864319,
      "learning_rate": 2.8719e-05,
      "loss": 0.0075,
      "step": 13847
    },
    {
      "epoch": 25.550323176361957,
      "grad_norm": 0.8781978487968445,
      "learning_rate": 2.8718666666666667e-05,
      "loss": 0.0066,
      "step": 13848
    },
    {
      "epoch": 25.552169898430286,
      "grad_norm": 0.5883736610412598,
      "learning_rate": 2.8718333333333333e-05,
      "loss": 0.0145,
      "step": 13849
    },
    {
      "epoch": 25.554016620498615,
      "grad_norm": 0.3041222393512726,
      "learning_rate": 2.8718000000000002e-05,
      "loss": 0.0065,
      "step": 13850
    },
    {
      "epoch": 25.555863342566944,
      "grad_norm": 0.588570773601532,
      "learning_rate": 2.8717666666666668e-05,
      "loss": 0.1257,
      "step": 13851
    },
    {
      "epoch": 25.557710064635273,
      "grad_norm": 0.49004557728767395,
      "learning_rate": 2.8717333333333334e-05,
      "loss": 0.1183,
      "step": 13852
    },
    {
      "epoch": 25.559556786703602,
      "grad_norm": 0.5519627332687378,
      "learning_rate": 2.8717000000000003e-05,
      "loss": 0.0888,
      "step": 13853
    },
    {
      "epoch": 25.56140350877193,
      "grad_norm": 0.6542931795120239,
      "learning_rate": 2.8716666666666666e-05,
      "loss": 0.0951,
      "step": 13854
    },
    {
      "epoch": 25.563250230840257,
      "grad_norm": 0.5098896026611328,
      "learning_rate": 2.8716333333333335e-05,
      "loss": 0.0795,
      "step": 13855
    },
    {
      "epoch": 25.565096952908586,
      "grad_norm": 0.561345636844635,
      "learning_rate": 2.8716e-05,
      "loss": 0.0592,
      "step": 13856
    },
    {
      "epoch": 25.566943674976915,
      "grad_norm": 0.38035711646080017,
      "learning_rate": 2.8715666666666667e-05,
      "loss": 0.0378,
      "step": 13857
    },
    {
      "epoch": 25.568790397045245,
      "grad_norm": 0.4322569966316223,
      "learning_rate": 2.8715333333333336e-05,
      "loss": 0.0544,
      "step": 13858
    },
    {
      "epoch": 25.570637119113574,
      "grad_norm": 0.6671208143234253,
      "learning_rate": 2.8715000000000002e-05,
      "loss": 0.0436,
      "step": 13859
    },
    {
      "epoch": 25.572483841181903,
      "grad_norm": 0.3293718695640564,
      "learning_rate": 2.8714666666666668e-05,
      "loss": 0.0358,
      "step": 13860
    },
    {
      "epoch": 25.574330563250232,
      "grad_norm": 0.2848909795284271,
      "learning_rate": 2.8714333333333334e-05,
      "loss": 0.0175,
      "step": 13861
    },
    {
      "epoch": 25.57617728531856,
      "grad_norm": 0.2711280286312103,
      "learning_rate": 2.8714e-05,
      "loss": 0.0316,
      "step": 13862
    },
    {
      "epoch": 25.578024007386887,
      "grad_norm": 0.2882595360279083,
      "learning_rate": 2.8713666666666666e-05,
      "loss": 0.0186,
      "step": 13863
    },
    {
      "epoch": 25.579870729455216,
      "grad_norm": 0.5483098030090332,
      "learning_rate": 2.8713333333333335e-05,
      "loss": 0.0553,
      "step": 13864
    },
    {
      "epoch": 25.581717451523545,
      "grad_norm": 1.0351965427398682,
      "learning_rate": 2.8712999999999998e-05,
      "loss": 0.0106,
      "step": 13865
    },
    {
      "epoch": 25.583564173591874,
      "grad_norm": 0.8314506411552429,
      "learning_rate": 2.8712666666666667e-05,
      "loss": 0.0213,
      "step": 13866
    },
    {
      "epoch": 25.585410895660203,
      "grad_norm": 0.5011735558509827,
      "learning_rate": 2.8712333333333336e-05,
      "loss": 0.01,
      "step": 13867
    },
    {
      "epoch": 25.587257617728532,
      "grad_norm": 0.19918778538703918,
      "learning_rate": 2.8712e-05,
      "loss": 0.0073,
      "step": 13868
    },
    {
      "epoch": 25.58910433979686,
      "grad_norm": 0.44274699687957764,
      "learning_rate": 2.8711666666666668e-05,
      "loss": 0.0138,
      "step": 13869
    },
    {
      "epoch": 25.59095106186519,
      "grad_norm": 0.2641814947128296,
      "learning_rate": 2.8711333333333334e-05,
      "loss": 0.0344,
      "step": 13870
    },
    {
      "epoch": 25.592797783933516,
      "grad_norm": 0.2949162721633911,
      "learning_rate": 2.8711e-05,
      "loss": 0.0069,
      "step": 13871
    },
    {
      "epoch": 25.594644506001845,
      "grad_norm": 0.4576517641544342,
      "learning_rate": 2.8710666666666666e-05,
      "loss": 0.0179,
      "step": 13872
    },
    {
      "epoch": 25.596491228070175,
      "grad_norm": 0.3257240653038025,
      "learning_rate": 2.8710333333333335e-05,
      "loss": 0.0097,
      "step": 13873
    },
    {
      "epoch": 25.598337950138504,
      "grad_norm": 0.266026109457016,
      "learning_rate": 2.871e-05,
      "loss": 0.0045,
      "step": 13874
    },
    {
      "epoch": 25.600184672206833,
      "grad_norm": 0.28286173939704895,
      "learning_rate": 2.8709666666666667e-05,
      "loss": 0.0074,
      "step": 13875
    },
    {
      "epoch": 25.602031394275162,
      "grad_norm": 0.5088093280792236,
      "learning_rate": 2.8709333333333337e-05,
      "loss": 0.0041,
      "step": 13876
    },
    {
      "epoch": 25.60387811634349,
      "grad_norm": 0.25021037459373474,
      "learning_rate": 2.8709e-05,
      "loss": 0.005,
      "step": 13877
    },
    {
      "epoch": 25.60572483841182,
      "grad_norm": 0.24160067737102509,
      "learning_rate": 2.870866666666667e-05,
      "loss": 0.0035,
      "step": 13878
    },
    {
      "epoch": 25.607571560480146,
      "grad_norm": 0.34590643644332886,
      "learning_rate": 2.8708333333333334e-05,
      "loss": 0.0074,
      "step": 13879
    },
    {
      "epoch": 25.609418282548475,
      "grad_norm": 0.5404072403907776,
      "learning_rate": 2.8708e-05,
      "loss": 0.0101,
      "step": 13880
    },
    {
      "epoch": 25.611265004616804,
      "grad_norm": 0.1285506635904312,
      "learning_rate": 2.8707666666666666e-05,
      "loss": 0.0035,
      "step": 13881
    },
    {
      "epoch": 25.613111726685133,
      "grad_norm": 0.23897305130958557,
      "learning_rate": 2.8707333333333336e-05,
      "loss": 0.0072,
      "step": 13882
    },
    {
      "epoch": 25.614958448753463,
      "grad_norm": 0.29718905687332153,
      "learning_rate": 2.8707e-05,
      "loss": 0.006,
      "step": 13883
    },
    {
      "epoch": 25.61680517082179,
      "grad_norm": 0.08984941989183426,
      "learning_rate": 2.8706666666666667e-05,
      "loss": 0.0024,
      "step": 13884
    },
    {
      "epoch": 25.61865189289012,
      "grad_norm": 0.223371222615242,
      "learning_rate": 2.8706333333333337e-05,
      "loss": 0.0074,
      "step": 13885
    },
    {
      "epoch": 25.62049861495845,
      "grad_norm": 0.16034305095672607,
      "learning_rate": 2.8706e-05,
      "loss": 0.0079,
      "step": 13886
    },
    {
      "epoch": 25.62234533702678,
      "grad_norm": 0.2069733887910843,
      "learning_rate": 2.870566666666667e-05,
      "loss": 0.0039,
      "step": 13887
    },
    {
      "epoch": 25.624192059095105,
      "grad_norm": 0.3699100613594055,
      "learning_rate": 2.8705333333333335e-05,
      "loss": 0.0145,
      "step": 13888
    },
    {
      "epoch": 25.626038781163434,
      "grad_norm": 0.5772353410720825,
      "learning_rate": 2.8705e-05,
      "loss": 0.0065,
      "step": 13889
    },
    {
      "epoch": 25.627885503231763,
      "grad_norm": 0.0805864930152893,
      "learning_rate": 2.8704666666666666e-05,
      "loss": 0.002,
      "step": 13890
    },
    {
      "epoch": 25.629732225300092,
      "grad_norm": 0.19519758224487305,
      "learning_rate": 2.8704333333333332e-05,
      "loss": 0.005,
      "step": 13891
    },
    {
      "epoch": 25.63157894736842,
      "grad_norm": 0.21675550937652588,
      "learning_rate": 2.8704e-05,
      "loss": 0.0061,
      "step": 13892
    },
    {
      "epoch": 25.63342566943675,
      "grad_norm": 0.15496018528938293,
      "learning_rate": 2.8703666666666668e-05,
      "loss": 0.0043,
      "step": 13893
    },
    {
      "epoch": 25.63527239150508,
      "grad_norm": 0.5796242356300354,
      "learning_rate": 2.8703333333333334e-05,
      "loss": 0.0093,
      "step": 13894
    },
    {
      "epoch": 25.63711911357341,
      "grad_norm": 0.26933422684669495,
      "learning_rate": 2.8703e-05,
      "loss": 0.004,
      "step": 13895
    },
    {
      "epoch": 25.638965835641734,
      "grad_norm": 1.446235179901123,
      "learning_rate": 2.870266666666667e-05,
      "loss": 0.0082,
      "step": 13896
    },
    {
      "epoch": 25.640812557710063,
      "grad_norm": 1.0523089170455933,
      "learning_rate": 2.870233333333333e-05,
      "loss": 0.015,
      "step": 13897
    },
    {
      "epoch": 25.642659279778393,
      "grad_norm": 0.45611003041267395,
      "learning_rate": 2.8702e-05,
      "loss": 0.0084,
      "step": 13898
    },
    {
      "epoch": 25.64450600184672,
      "grad_norm": 0.15992005169391632,
      "learning_rate": 2.8701666666666667e-05,
      "loss": 0.0025,
      "step": 13899
    },
    {
      "epoch": 25.64635272391505,
      "grad_norm": 0.08912533521652222,
      "learning_rate": 2.8701333333333333e-05,
      "loss": 0.0017,
      "step": 13900
    },
    {
      "epoch": 25.64819944598338,
      "grad_norm": 1.0898728370666504,
      "learning_rate": 2.8701000000000002e-05,
      "loss": 0.1265,
      "step": 13901
    },
    {
      "epoch": 25.65004616805171,
      "grad_norm": 0.5395174622535706,
      "learning_rate": 2.8700666666666668e-05,
      "loss": 0.1012,
      "step": 13902
    },
    {
      "epoch": 25.65189289012004,
      "grad_norm": 0.4160155951976776,
      "learning_rate": 2.8700333333333334e-05,
      "loss": 0.0628,
      "step": 13903
    },
    {
      "epoch": 25.653739612188367,
      "grad_norm": 0.5603948831558228,
      "learning_rate": 2.87e-05,
      "loss": 0.0745,
      "step": 13904
    },
    {
      "epoch": 25.655586334256693,
      "grad_norm": 0.4057348370552063,
      "learning_rate": 2.869966666666667e-05,
      "loss": 0.0692,
      "step": 13905
    },
    {
      "epoch": 25.657433056325022,
      "grad_norm": 0.379995733499527,
      "learning_rate": 2.869933333333333e-05,
      "loss": 0.0394,
      "step": 13906
    },
    {
      "epoch": 25.65927977839335,
      "grad_norm": 0.5244229435920715,
      "learning_rate": 2.8699e-05,
      "loss": 0.0504,
      "step": 13907
    },
    {
      "epoch": 25.66112650046168,
      "grad_norm": 0.34626778960227966,
      "learning_rate": 2.869866666666667e-05,
      "loss": 0.0344,
      "step": 13908
    },
    {
      "epoch": 25.66297322253001,
      "grad_norm": 0.5717699527740479,
      "learning_rate": 2.8698333333333333e-05,
      "loss": 0.0491,
      "step": 13909
    },
    {
      "epoch": 25.66481994459834,
      "grad_norm": 0.33867087960243225,
      "learning_rate": 2.8698000000000002e-05,
      "loss": 0.0252,
      "step": 13910
    },
    {
      "epoch": 25.666666666666668,
      "grad_norm": 0.7220029830932617,
      "learning_rate": 2.8697666666666668e-05,
      "loss": 0.0261,
      "step": 13911
    },
    {
      "epoch": 25.668513388734997,
      "grad_norm": 0.48679837584495544,
      "learning_rate": 2.8697333333333334e-05,
      "loss": 0.0953,
      "step": 13912
    },
    {
      "epoch": 25.670360110803323,
      "grad_norm": 0.45722731947898865,
      "learning_rate": 2.8697e-05,
      "loss": 0.0296,
      "step": 13913
    },
    {
      "epoch": 25.67220683287165,
      "grad_norm": 0.27095258235931396,
      "learning_rate": 2.869666666666667e-05,
      "loss": 0.033,
      "step": 13914
    },
    {
      "epoch": 25.67405355493998,
      "grad_norm": 0.12416177988052368,
      "learning_rate": 2.8696333333333332e-05,
      "loss": 0.004,
      "step": 13915
    },
    {
      "epoch": 25.67590027700831,
      "grad_norm": 0.27076831459999084,
      "learning_rate": 2.8696e-05,
      "loss": 0.0085,
      "step": 13916
    },
    {
      "epoch": 25.67774699907664,
      "grad_norm": 0.4575108289718628,
      "learning_rate": 2.869566666666667e-05,
      "loss": 0.0153,
      "step": 13917
    },
    {
      "epoch": 25.67959372114497,
      "grad_norm": 0.33490633964538574,
      "learning_rate": 2.8695333333333333e-05,
      "loss": 0.0392,
      "step": 13918
    },
    {
      "epoch": 25.681440443213297,
      "grad_norm": 0.4233061969280243,
      "learning_rate": 2.8695000000000002e-05,
      "loss": 0.0212,
      "step": 13919
    },
    {
      "epoch": 25.683287165281627,
      "grad_norm": 0.37904930114746094,
      "learning_rate": 2.8694666666666668e-05,
      "loss": 0.0341,
      "step": 13920
    },
    {
      "epoch": 25.685133887349952,
      "grad_norm": 0.3937544524669647,
      "learning_rate": 2.8694333333333334e-05,
      "loss": 0.0148,
      "step": 13921
    },
    {
      "epoch": 25.68698060941828,
      "grad_norm": 0.21121932566165924,
      "learning_rate": 2.8694e-05,
      "loss": 0.0095,
      "step": 13922
    },
    {
      "epoch": 25.68882733148661,
      "grad_norm": 0.17242980003356934,
      "learning_rate": 2.8693666666666666e-05,
      "loss": 0.0067,
      "step": 13923
    },
    {
      "epoch": 25.69067405355494,
      "grad_norm": 0.6399073600769043,
      "learning_rate": 2.8693333333333335e-05,
      "loss": 0.006,
      "step": 13924
    },
    {
      "epoch": 25.69252077562327,
      "grad_norm": 0.4123702943325043,
      "learning_rate": 2.8693e-05,
      "loss": 0.0123,
      "step": 13925
    },
    {
      "epoch": 25.694367497691598,
      "grad_norm": 0.2438940405845642,
      "learning_rate": 2.8692666666666667e-05,
      "loss": 0.0168,
      "step": 13926
    },
    {
      "epoch": 25.696214219759927,
      "grad_norm": 0.18913382291793823,
      "learning_rate": 2.8692333333333333e-05,
      "loss": 0.0063,
      "step": 13927
    },
    {
      "epoch": 25.698060941828256,
      "grad_norm": 0.2355855107307434,
      "learning_rate": 2.8692000000000002e-05,
      "loss": 0.0047,
      "step": 13928
    },
    {
      "epoch": 25.69990766389658,
      "grad_norm": 0.16520513594150543,
      "learning_rate": 2.8691666666666665e-05,
      "loss": 0.0062,
      "step": 13929
    },
    {
      "epoch": 25.70175438596491,
      "grad_norm": 0.50859135389328,
      "learning_rate": 2.8691333333333334e-05,
      "loss": 0.0138,
      "step": 13930
    },
    {
      "epoch": 25.70360110803324,
      "grad_norm": 0.3660358488559723,
      "learning_rate": 2.8691e-05,
      "loss": 0.0085,
      "step": 13931
    },
    {
      "epoch": 25.70544783010157,
      "grad_norm": 0.2390100210905075,
      "learning_rate": 2.8690666666666666e-05,
      "loss": 0.0045,
      "step": 13932
    },
    {
      "epoch": 25.7072945521699,
      "grad_norm": 0.2729160785675049,
      "learning_rate": 2.8690333333333336e-05,
      "loss": 0.011,
      "step": 13933
    },
    {
      "epoch": 25.709141274238227,
      "grad_norm": 0.2517271041870117,
      "learning_rate": 2.869e-05,
      "loss": 0.0084,
      "step": 13934
    },
    {
      "epoch": 25.710987996306557,
      "grad_norm": 0.291607141494751,
      "learning_rate": 2.8689666666666667e-05,
      "loss": 0.0109,
      "step": 13935
    },
    {
      "epoch": 25.712834718374886,
      "grad_norm": 0.43152952194213867,
      "learning_rate": 2.8689333333333333e-05,
      "loss": 0.0105,
      "step": 13936
    },
    {
      "epoch": 25.714681440443215,
      "grad_norm": 0.21754711866378784,
      "learning_rate": 2.8689000000000003e-05,
      "loss": 0.0107,
      "step": 13937
    },
    {
      "epoch": 25.71652816251154,
      "grad_norm": 0.3606095016002655,
      "learning_rate": 2.8688666666666665e-05,
      "loss": 0.0068,
      "step": 13938
    },
    {
      "epoch": 25.71837488457987,
      "grad_norm": 0.3872680962085724,
      "learning_rate": 2.8688333333333335e-05,
      "loss": 0.0066,
      "step": 13939
    },
    {
      "epoch": 25.7202216066482,
      "grad_norm": 0.24474932253360748,
      "learning_rate": 2.8688e-05,
      "loss": 0.0074,
      "step": 13940
    },
    {
      "epoch": 25.722068328716528,
      "grad_norm": 0.0998922809958458,
      "learning_rate": 2.8687666666666666e-05,
      "loss": 0.003,
      "step": 13941
    },
    {
      "epoch": 25.723915050784857,
      "grad_norm": 0.36174479126930237,
      "learning_rate": 2.8687333333333336e-05,
      "loss": 0.0061,
      "step": 13942
    },
    {
      "epoch": 25.725761772853186,
      "grad_norm": 0.3098700940608978,
      "learning_rate": 2.8687e-05,
      "loss": 0.0055,
      "step": 13943
    },
    {
      "epoch": 25.727608494921515,
      "grad_norm": 0.638459324836731,
      "learning_rate": 2.8686666666666668e-05,
      "loss": 0.0109,
      "step": 13944
    },
    {
      "epoch": 25.729455216989845,
      "grad_norm": 0.35461121797561646,
      "learning_rate": 2.8686333333333334e-05,
      "loss": 0.0093,
      "step": 13945
    },
    {
      "epoch": 25.73130193905817,
      "grad_norm": 0.14713731408119202,
      "learning_rate": 2.8686000000000003e-05,
      "loss": 0.0042,
      "step": 13946
    },
    {
      "epoch": 25.7331486611265,
      "grad_norm": 0.21553030610084534,
      "learning_rate": 2.8685666666666665e-05,
      "loss": 0.0061,
      "step": 13947
    },
    {
      "epoch": 25.73499538319483,
      "grad_norm": 0.24078437685966492,
      "learning_rate": 2.8685333333333335e-05,
      "loss": 0.0054,
      "step": 13948
    },
    {
      "epoch": 25.736842105263158,
      "grad_norm": 0.4704957604408264,
      "learning_rate": 2.8685e-05,
      "loss": 0.0136,
      "step": 13949
    },
    {
      "epoch": 25.738688827331487,
      "grad_norm": 1.031306505203247,
      "learning_rate": 2.8684666666666667e-05,
      "loss": 0.0218,
      "step": 13950
    },
    {
      "epoch": 25.740535549399816,
      "grad_norm": 0.7354793548583984,
      "learning_rate": 2.8684333333333336e-05,
      "loss": 0.1239,
      "step": 13951
    },
    {
      "epoch": 25.742382271468145,
      "grad_norm": 0.43336817622184753,
      "learning_rate": 2.8684e-05,
      "loss": 0.0813,
      "step": 13952
    },
    {
      "epoch": 25.744228993536474,
      "grad_norm": 0.48846545815467834,
      "learning_rate": 2.8683666666666668e-05,
      "loss": 0.111,
      "step": 13953
    },
    {
      "epoch": 25.746075715604803,
      "grad_norm": 0.48932769894599915,
      "learning_rate": 2.8683333333333334e-05,
      "loss": 0.0533,
      "step": 13954
    },
    {
      "epoch": 25.74792243767313,
      "grad_norm": 0.5916051864624023,
      "learning_rate": 2.8683e-05,
      "loss": 0.0647,
      "step": 13955
    },
    {
      "epoch": 25.749769159741458,
      "grad_norm": 0.5129830837249756,
      "learning_rate": 2.8682666666666666e-05,
      "loss": 0.0752,
      "step": 13956
    },
    {
      "epoch": 25.751615881809787,
      "grad_norm": 0.4459969699382782,
      "learning_rate": 2.8682333333333335e-05,
      "loss": 0.0473,
      "step": 13957
    },
    {
      "epoch": 25.753462603878116,
      "grad_norm": 0.8263006210327148,
      "learning_rate": 2.8682e-05,
      "loss": 0.0351,
      "step": 13958
    },
    {
      "epoch": 25.755309325946445,
      "grad_norm": 0.6295078992843628,
      "learning_rate": 2.8681666666666667e-05,
      "loss": 0.042,
      "step": 13959
    },
    {
      "epoch": 25.757156048014775,
      "grad_norm": 0.3544279932975769,
      "learning_rate": 2.8681333333333336e-05,
      "loss": 0.0261,
      "step": 13960
    },
    {
      "epoch": 25.759002770083104,
      "grad_norm": 0.2778114676475525,
      "learning_rate": 2.8681e-05,
      "loss": 0.0233,
      "step": 13961
    },
    {
      "epoch": 25.760849492151433,
      "grad_norm": 0.26822900772094727,
      "learning_rate": 2.8680666666666668e-05,
      "loss": 0.0319,
      "step": 13962
    },
    {
      "epoch": 25.76269621421976,
      "grad_norm": 0.7484460473060608,
      "learning_rate": 2.8680333333333334e-05,
      "loss": 0.0482,
      "step": 13963
    },
    {
      "epoch": 25.764542936288088,
      "grad_norm": 0.1940113604068756,
      "learning_rate": 2.868e-05,
      "loss": 0.0103,
      "step": 13964
    },
    {
      "epoch": 25.766389658356417,
      "grad_norm": 0.3328947126865387,
      "learning_rate": 2.8679666666666666e-05,
      "loss": 0.0206,
      "step": 13965
    },
    {
      "epoch": 25.768236380424746,
      "grad_norm": 0.3248881697654724,
      "learning_rate": 2.8679333333333335e-05,
      "loss": 0.0189,
      "step": 13966
    },
    {
      "epoch": 25.770083102493075,
      "grad_norm": 0.17101606726646423,
      "learning_rate": 2.8679e-05,
      "loss": 0.0071,
      "step": 13967
    },
    {
      "epoch": 25.771929824561404,
      "grad_norm": 0.18649211525917053,
      "learning_rate": 2.8678666666666667e-05,
      "loss": 0.0083,
      "step": 13968
    },
    {
      "epoch": 25.773776546629733,
      "grad_norm": 0.2769390940666199,
      "learning_rate": 2.8678333333333336e-05,
      "loss": 0.0091,
      "step": 13969
    },
    {
      "epoch": 25.775623268698062,
      "grad_norm": 0.2639617323875427,
      "learning_rate": 2.8678e-05,
      "loss": 0.0082,
      "step": 13970
    },
    {
      "epoch": 25.777469990766388,
      "grad_norm": 0.27369749546051025,
      "learning_rate": 2.8677666666666668e-05,
      "loss": 0.0112,
      "step": 13971
    },
    {
      "epoch": 25.779316712834717,
      "grad_norm": 0.2493075728416443,
      "learning_rate": 2.8677333333333334e-05,
      "loss": 0.0053,
      "step": 13972
    },
    {
      "epoch": 25.781163434903046,
      "grad_norm": 0.35896432399749756,
      "learning_rate": 2.8677e-05,
      "loss": 0.0094,
      "step": 13973
    },
    {
      "epoch": 25.783010156971375,
      "grad_norm": 0.6993874311447144,
      "learning_rate": 2.867666666666667e-05,
      "loss": 0.0189,
      "step": 13974
    },
    {
      "epoch": 25.784856879039705,
      "grad_norm": 0.27760475873947144,
      "learning_rate": 2.8676333333333335e-05,
      "loss": 0.0064,
      "step": 13975
    },
    {
      "epoch": 25.786703601108034,
      "grad_norm": 0.32611775398254395,
      "learning_rate": 2.8676e-05,
      "loss": 0.0074,
      "step": 13976
    },
    {
      "epoch": 25.788550323176363,
      "grad_norm": 0.17413999140262604,
      "learning_rate": 2.8675666666666667e-05,
      "loss": 0.0061,
      "step": 13977
    },
    {
      "epoch": 25.790397045244692,
      "grad_norm": 0.1638621687889099,
      "learning_rate": 2.8675333333333336e-05,
      "loss": 0.008,
      "step": 13978
    },
    {
      "epoch": 25.792243767313018,
      "grad_norm": 0.4006885588169098,
      "learning_rate": 2.8675e-05,
      "loss": 0.0101,
      "step": 13979
    },
    {
      "epoch": 25.794090489381347,
      "grad_norm": 0.29481589794158936,
      "learning_rate": 2.867466666666667e-05,
      "loss": 0.009,
      "step": 13980
    },
    {
      "epoch": 25.795937211449676,
      "grad_norm": 0.10241249948740005,
      "learning_rate": 2.867433333333333e-05,
      "loss": 0.0025,
      "step": 13981
    },
    {
      "epoch": 25.797783933518005,
      "grad_norm": 0.198036789894104,
      "learning_rate": 2.8674e-05,
      "loss": 0.0068,
      "step": 13982
    },
    {
      "epoch": 25.799630655586334,
      "grad_norm": 0.14961473643779755,
      "learning_rate": 2.867366666666667e-05,
      "loss": 0.0051,
      "step": 13983
    },
    {
      "epoch": 25.801477377654663,
      "grad_norm": 0.11135370284318924,
      "learning_rate": 2.8673333333333332e-05,
      "loss": 0.0036,
      "step": 13984
    },
    {
      "epoch": 25.803324099722992,
      "grad_norm": 0.38078421354293823,
      "learning_rate": 2.8673e-05,
      "loss": 0.0071,
      "step": 13985
    },
    {
      "epoch": 25.80517082179132,
      "grad_norm": 0.15546701848506927,
      "learning_rate": 2.8672666666666667e-05,
      "loss": 0.0037,
      "step": 13986
    },
    {
      "epoch": 25.80701754385965,
      "grad_norm": 0.3690662980079651,
      "learning_rate": 2.8672333333333333e-05,
      "loss": 0.0069,
      "step": 13987
    },
    {
      "epoch": 25.808864265927976,
      "grad_norm": 0.21701663732528687,
      "learning_rate": 2.8672e-05,
      "loss": 0.0119,
      "step": 13988
    },
    {
      "epoch": 25.810710987996305,
      "grad_norm": 0.5214126706123352,
      "learning_rate": 2.867166666666667e-05,
      "loss": 0.0079,
      "step": 13989
    },
    {
      "epoch": 25.812557710064635,
      "grad_norm": 0.24381795525550842,
      "learning_rate": 2.867133333333333e-05,
      "loss": 0.0042,
      "step": 13990
    },
    {
      "epoch": 25.814404432132964,
      "grad_norm": 0.36028149724006653,
      "learning_rate": 2.8671e-05,
      "loss": 0.0048,
      "step": 13991
    },
    {
      "epoch": 25.816251154201293,
      "grad_norm": 0.2194744199514389,
      "learning_rate": 2.867066666666667e-05,
      "loss": 0.0033,
      "step": 13992
    },
    {
      "epoch": 25.818097876269622,
      "grad_norm": 0.3237725496292114,
      "learning_rate": 2.8670333333333332e-05,
      "loss": 0.0069,
      "step": 13993
    },
    {
      "epoch": 25.81994459833795,
      "grad_norm": 0.4830024540424347,
      "learning_rate": 2.867e-05,
      "loss": 0.0158,
      "step": 13994
    },
    {
      "epoch": 25.82179132040628,
      "grad_norm": 0.8800148367881775,
      "learning_rate": 2.8669666666666668e-05,
      "loss": 0.0311,
      "step": 13995
    },
    {
      "epoch": 25.823638042474606,
      "grad_norm": 0.18014296889305115,
      "learning_rate": 2.8669333333333334e-05,
      "loss": 0.0028,
      "step": 13996
    },
    {
      "epoch": 25.825484764542935,
      "grad_norm": 0.25388187170028687,
      "learning_rate": 2.8669e-05,
      "loss": 0.0049,
      "step": 13997
    },
    {
      "epoch": 25.827331486611264,
      "grad_norm": 0.5490654706954956,
      "learning_rate": 2.866866666666667e-05,
      "loss": 0.0101,
      "step": 13998
    },
    {
      "epoch": 25.829178208679593,
      "grad_norm": 0.4083966910839081,
      "learning_rate": 2.8668333333333335e-05,
      "loss": 0.008,
      "step": 13999
    },
    {
      "epoch": 25.831024930747922,
      "grad_norm": 0.3681202232837677,
      "learning_rate": 2.8668e-05,
      "loss": 0.0142,
      "step": 14000
    },
    {
      "epoch": 25.831024930747922,
      "eval_cer": 0.11683127251989528,
      "eval_loss": 0.417955607175827,
      "eval_runtime": 16.0102,
      "eval_samples_per_second": 18.988,
      "eval_steps_per_second": 0.625,
      "eval_wer": 0.4121258633921719,
      "step": 14000
    },
    {
      "epoch": 25.831024930747922,
      "step": 14000,
      "total_flos": 1.8828835451512226e+20,
      "train_loss": 0.3484456565128265,
      "train_runtime": 27570.5866,
      "train_samples_per_second": 232.131,
      "train_steps_per_second": 3.627
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 185,
  "save_steps": 1000,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 5
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.8828835451512226e+20,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}