Saving weights and logs of step 2500

config.json

@@ -9,14 +9,14 @@
   "eos_token_id": 2,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
-  "hidden_size": 768,
+  "hidden_size": 1024,
   "initializer_range": 0.02,
-  "intermediate_size": 3072,
+  "intermediate_size": 4096,
   "layer_norm_eps": 1e-05,
   "max_position_embeddings": 514,
   "model_type": "roberta",
-  "num_attention_heads": 12,
-  "num_hidden_layers": 12,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
   "pad_token_id": 1,
   "position_embedding_type": "absolute",
   "transformers_version": "4.47.0",

events.out.tfevents.1733906256.t1v-n-53cd541d-w-35.909229.0.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eed9bfb6af731f00d239909198c68d3e9fad81727b8ca4aa8e406015ba75a520
-size 735935
+oid sha256:89d27471d60c3cb5cb254f9c1ba89e1af7dede5278e4b0a16d6abe9b787895ed
+size 883073

events.out.tfevents.1733907637.t1v-n-53cd541d-w-35.912013.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18647887c2a0849db306f9fe96fecdcd6767d7ae6fb34691cdb5eab7bbcb86fb
+size 78

events.out.tfevents.1733907869.t1v-n-53cd541d-w-35.913593.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d6daeca2897dae66c89f6beaad87f0b73759a3670013ca028398143a71d4637
+size 367950

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:064c3f63bea271a2bb660ef51d5dc8d1df8a3346dc3e2d22f85e6e55ea49929e
-size 498796983
+oid sha256:fec6d5cf91d71d588c62a1e8945778185c81e4158a1a3e119ac93fc82fc1648a
+size 1421662309

run_train.sh

@@ -8,8 +8,8 @@ python run_mlm_flax.py \
     --dataset_config_name="unshuffled_deduplicated_no" \
     --max_seq_length="128" \
     --weight_decay="0.01" \
-    --per_device_train_batch_size="2" \
-    --per_device_eval_batch_size="2" \
+    --per_device_train_batch_size="1" \
+    --per_device_eval_batch_size="1" \
     --learning_rate="3e-4" \
     --warmup_steps="1000" \
     --overwrite_output_dir \