Saving weights and logs of step 400

events.out.tfevents.1626356336.t1v-n-1926f308-w-0.93622.3.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0e236e573933e78a338ba55589b3035b720199a69c0aef2a8e5ab36f5d5bab3b
-size 60409
+oid sha256:037f21c4c9c7d03064fbcff6dd96669b8be250f23b9678cd795971bc9a83c0b1
+size 113336

events.out.tfevents.1626357469.t1v-n-1926f308-w-0.95268.3.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65d6d947b3da9df3dfc09cc3a657592d85fc8ba3783af91c196fbab5f0846dd3
+size 60409

events.out.tfevents.1626357733.t1v-n-1926f308-w-0.96764.3.v2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb1f4267c9667e83a6c35e620806d8776ce4dc66d1d56559186813567ba5443e
+size 40

run.sh

@@ -11,11 +11,11 @@ python3 run_clm_flax.py \
     --block_size="512" \
     --per_device_train_batch_size="64" \
     --per_device_eval_batch_size="32" \
-    --learning_rate="5e-4"  \
+    --learning_rate="5e-5"  \
     --adam_beta1="0.9" --adam_beta2="0.98" --weight_decay="0.01" \
     --overwrite_output_dir \
     --num_train_epochs="100" \
     --logging_steps="50" \
-    --save_steps="400" \
+    --save_steps="750" \
     --eval_steps="50" \
     --push_to_hub