all_results.json

{
    "achieved_tflops_per_gpu": 0.0036400978834181994,
    "achieved_tflops_per_gpu_theoretical": 918.1221365250786,
    "epoch": 7.0,
    "loss_nan_ranks": 0,
    "loss_rank_avg": 0.05594291538000107,
    "mfu_percent": 0.0002572507338104735,
    "mfu_percent_theoretical": 64.88495664488188,
    "total_flos": 1429285307547648.0,
    "train_loss": 0.14208543798229403,
    "train_runtime": 24540.64,
    "train_samples_per_second": 4.338,
    "train_steps_per_second": 0.271,
    "valid_targets_mean": 2843.0,
    "valid_targets_min": 791
}