| HUB_TOKEN=`cat $HOME/.huggingface/token` | |
| ./run_mlm_flax.py \ | |
| --output_dir="./" \ | |
| --model_type="roberta" \ | |
| --config_name="./" \ | |
| --tokenizer_name="./" \ | |
| --dataset_name="large_spanish_corpus" \ | |
| --dataset_config_name="combined" \ | |
| --max_seq_length="128" \ | |
| --pad_to_max_length \ | |
| --per_device_train_batch_size="128" \ | |
| --per_device_eval_batch_size="128" \ | |
| --learning_rate="3e-4" \ | |
| --save_strategy="steps" \ | |
| --save_steps="10000" \ | |
| --save_total_limit="5" \ | |
| --warmup_steps="1000" \ | |
| --overwrite_output_dir \ | |
| --num_train_epochs="8" \ | |
| --dtype="bfloat16" \ | |
| --push_to_hub_model_id="flax-community/bertin-roberta-large-spanish" \ | |
| --push_to_hub_token="$HUB_TOKEN" | |
| --push_to_hub 2>&1 | tee run.log | |