reduce batch size

run.sh

@@ -9,9 +9,9 @@ python3 run_mlm_flax.py \
     --dataset_config_name="unshuffled_deduplicated_th" \
     --max_seq_length="128" \
     --preprocessing_num_workers="64" \
-    --per_device_train_batch_size="4" \
-    --per_device_eval_batch_size="4" \
-    --learning_rate="3e-4" \
+    --per_device_train_batch_size="64" \
+    --per_device_eval_batch_size="64" \
+    --learning_rate="2e-4" \
     --warmup_steps="1000" \
     --overwrite_output_dir \
     --num_train_epochs="8" \