Saving weights and logs of step 1000

events.out.tfevents.1644155872.t1v-n-ccbf3e94-w-0.758395.3.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:129aa7fe54a781a63a3d2eda5cbc93f6cef8d55fb35128be130a9d0f624d8722
-size 21672797
+oid sha256:1e2f75854dcb7d407c3cfd31973c62ea38cc31c57a22379bde72c862bf66f37d
+size 21822009

events.out.tfevents.1644302791.t1v-n-ccbf3e94-w-0.915831.3.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:226c350e022b3511e53793006a0caac28582fede3ecc6565e3426daefbcffce8
+size 147136

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c293b90ab15441240d54c6c75a27de413989c9d9bf2c55370999d1a29158b8fb
+oid sha256:fa7660d133cd35c8785474afd6c2e37565627261fabcf6290363f7846ca43e16
 size 498796983

run_512.sh

@@ -12,7 +12,7 @@ python run_mlm_flax.py \
     --per_device_eval_batch_size="46" \
     --pad_to_max_length \
     --learning_rate="0.00015" \
-    --warmup_steps="10000" \
+    --warmup_steps="150000" \
     --overwrite_output_dir \
     --num_train_epochs="10" \
     --adam_beta1="0.9" \