Saving weights and logs of step 10000

events.out.tfevents.1644662890.t1v-n-ccbf3e94-w-0.1307490.3.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94f1dd0065038d65f57c3b0d41e411f83804552e6852555e2cd0ef5842c1be0b
+size 1470136

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:559249c8e42a5a372de69c99340b0c4ab3da15cac6c4ff253cb1adafc42889ba
+oid sha256:3f2523dae16815c083d982b608a29065d97c475203bab4289bcf0657159c4770
 size 498796983

run_512.sh

@@ -11,16 +11,16 @@ python run_mlm_flax.py \
     --per_device_train_batch_size="46" \
     --per_device_eval_batch_size="46" \
     --pad_to_max_length \
-    --learning_rate="0.00015" \
+    --learning_rate="0.00005" \
     --warmup_steps="150000" \
     --overwrite_output_dir \
     --num_train_epochs="10" \
     --adam_beta1="0.9" \
     --adam_beta2="0.98" \
     --adam_epsilon="1e-6" \
-    --logging_steps="1000" \
-    --save_steps="1000" \
-    --eval_steps="1000" \
+    --logging_steps="10000" \
+    --save_steps="10000" \
+    --eval_steps="10000" \
     --auth_token="True" \
     --do_train \
     --do_eval \