Saving weights and logs of epoch 1

events.out.tfevents.1625825107.t1v-n-b95d739e-w-0.172623.3.v2 → events.out.tfevents.1625848981.t1v-n-b95d739e-w-0.195447.3.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0e20a3b8b97c289a397b20b2c4d4300223d194b7dd9967272c6f575bc756a67d
+oid sha256:553197e40e1ee37f960da354350ed70147d500b9887c5c46dc7caa0d5d15f635
 size 3581817

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3903ba16f2702cb02d4cbc39d63a80d5264c9a4389fefcf48c05986133f500f3
+oid sha256:4e4a15d3a3656c720be8f016191292ed4fd5ec92a7b80f9226fad84210702158
 size 711588089

nohup.out

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f6d2a0bcc862bb1cff548f19aa1ce5d5314ed010bc327bda0876e0f0df804c03
-size 2037187
+oid sha256:5372b784c18a06c3f8582bd4711dc0b69cff87f87abb4355c4e10163b99f38ab
+size 4112389

run.sh

@@ -12,7 +12,7 @@ python3 run_mlm_flax.py \
     --preprocessing_num_workers="64" \
     --per_device_train_batch_size="128" \
     --per_device_eval_batch_size="128" \
-    --learning_rate="2e-4" \
+    --learning_rate="2e-5" \
     --warmup_steps="1000" \
     --overwrite_output_dir \
     --num_train_epochs="10" \