Saving weights and logs of step 10000

events.out.tfevents.1629925854.t1v-n-1ae8dadb-w-0.258038.0.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5cf2edebcf6ded37bb3bce8656df5a033466b7a787691ea4f4fda648a130517b
-size 7421353
+oid sha256:d3ba93cc73524ee91a324c9c60c9b1944661982351b153371e82ee189b547e08
+size 8166703

events.out.tfevents.1629954669.t1v-n-1ae8dadb-w-0.348909.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2754b0718aaa184806c065e66c1e4cd5de84a84f740d764fa0aa7b57e056c74
+size 1176479

events.out.tfevents.1629959691.t1v-n-1ae8dadb-w-0.364831.0.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f17a856d2fd8e07f6e5dfeff9ef002b11a709812bd05c57144322ca7fbd2b929
+size 1470757

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:73af60a78df8ccba5f3cdffe37da08484560c91bb8ad798654238986e7dfa131
+oid sha256:c4b00a73b13270205d21872705ba7bc03f609ecf3c5a6dc99ec19895a8184c9d
 size 711588089

start_train.sh

@@ -15,8 +15,9 @@ python3 run_mlm_flax.py \
     --adam_beta1="0.9" \
     --adam_beta2="0.98" \
     --adam_epsilon="1e-6" \
-    --learning_rate="4e-4" \
-    --warmup_steps="15000" \
+    --learning_rate="2e-4" \
+#    --weight_decay="0.01" \
+    --warmup_steps="1500" \
     --overwrite_output_dir \
     --num_train_epochs="3" \
     --save_strategy="steps" \