deepspeed run_mlm_local.py \
    --model_config_id "bert-base-uncased" \
    --dataset_id "chaoyan/processed_bert_dataset" \
    --tokenizer_id "chaoyan/bert-base-uncased-cat_tokenizer" \
    --repository_id "bert-base-uncased-cat_model" \
    --max_steps 1000_000 \
    --per_device_train_batch_size 64 \
    --learning_rate 5e-5 \
    --deepspeed ds_config_zero3.json