Training in progress, step 126

config.json

@@ -23,7 +23,7 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.48.1",
-  "use_cache": true,
+  "use_cache": false,
   "use_sliding_window": false,
   "vocab_size": 152064
 }

runs/Mar27_03-28-34_fdse/events.out.tfevents.1743046230.fdse.1748328.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac2f3108a02d1f9d6bd9f5871b7ca91df34c579c5ab73db4d196c5a19fbebc3c
+size 6302

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:33155a996014a45c187f44946a989f6a5e1fc1fe6fe66873f688ff833920fbb2
+oid sha256:abdd55fe8e4a7f4763c1f4b9e13238187adf363b1ad29bc64de1a26aafc4682c
 size 6840