Saving weights and logs of epoch 1

events.out.tfevents.1626000474.t1v-n-b95d739e-w-0.337760.3.v2 → events.out.tfevents.1626058987.t1v-n-b95d739e-w-0.391545.3.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d94790530d5a4895b4c0772b8b874e84d7b491a4b713d064413ee8f83beafea5
-size 6822986
+oid sha256:c490d811e1d2f0662ce85acb33aa3ecb8a8a1be68d7103f0cbe58564088c04a8
+size 3581817

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:04e73347e6b84dc12a46fb9439d340b613f9c16c1108345a4e95d1f1bd5e09b8
+oid sha256:0945c644f395e761506b26da253ba21489ee8e1acb5cf33c082f0644712cd504
 size 711588089

nohup.out

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aa7530b8d9514c6920b48143279239a3061ec753a0c3b8c180e80c8f361ada6f
-size 4002271
+oid sha256:a1a43800978026bb25dfe0a70cdb0c8a4129b57842af477a2a2f9e0a5801262a
+size 2037183

run.sh

@@ -11,7 +11,7 @@ python3 run_mlm_flax.py \
     --preprocessing_num_workers="64" \
     --per_device_train_batch_size="128" \
     --per_device_eval_batch_size="128" \
-    --learning_rate="1e-4" \
+    --learning_rate="8e-5" \
     --warmup_steps="1000" \
     --overwrite_output_dir \
     --num_train_epochs="10" \