| ./run_t5_mlm_flax.py \ | |
| --output_dir="./t5-small-2" \ | |
| --model_type="t5" \ | |
| --model_name_or_path="./t5-small-2" \ | |
| --config_name="./" \ | |
| --tokenizer_name="./" \ | |
| --dataset_name="large_spanish_corpus" \ | |
| --max_seq_length="512" \ | |
| --per_device_train_batch_size="32" \ | |
| --per_device_eval_batch_size="32" \ | |
| --adafactor \ | |
| --learning_rate="0.005" \ | |
| --weight_decay="0.001" \ | |
| --warmup_steps="2000" \ | |
| --overwrite_output_dir \ | |
| --num_train_epochs="10" \ | |
| --logging_steps="500" \ | |
| --save_steps="2500" \ | |
| --eval_steps="5000" \ | |
| --preprocessing_num_workers="64" | |