Saving weights and logs of step 1000

events.out.tfevents.1635881280.t1v-n-f6f5b6cc-w-0.918853.0.v2

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:23924ecfccda20b364d62c01624662b25de5099ea50e43300192d5f6ba0dddf7
-size 3251345

events.out.tfevents.1635877753.t1v-n-f6f5b6cc-w-0.912674.0.v2 → events.out.tfevents.1635885577.t1v-n-f6f5b6cc-w-0.927806.0.v2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:afe1b8df5b6d8c3e0b0c5e437fe5ea534eb15c751f20081863e9a34c7b8fc2cd
-size 1472727
+oid sha256:c9deea6fae194a5e82fb314e234a1910c60fd5ee6ff52aae4d172ae30beb7bd2
+size 147207

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69368bc2cba2e7f9866a48643413163512720990b5d8cc8fcdbc86ef6dc9eed6
+oid sha256:517b0984754a2093b404bd907a4446ed02d7865b73ff35f1ae029a32b2916ddf
 size 497764120

run.sh

@@ -11,11 +11,11 @@ python run_clm_flax.py \
     --block_size="512" \
     --per_device_train_batch_size="8" \
     --per_device_eval_batch_size="8" \
-    --learning_rate="3e-5" \
-    --warmup_steps="500" \
+    --learning_rate="4e-5" \
+    --warmup_steps="1000" \
     --adam_beta1="0.9" --adam_beta2="0.98" --weight_decay="0.01" \
     --overwrite_output_dir \
-    --num_train_epochs="30" \
+    --num_train_epochs="10" \
     --logging_steps="500" \
     --save_steps="1000" \
     --eval_steps="1000" \