./run_t5_mlm_flax.py \ | |
--output_dir="./" \ | |
--model_type="t5" \ | |
--config_name="./" \ | |
--tokenizer_name="./" \ | |
--train_file /mnt/disks/flaxdisk/NCC_nb_nn_balanced/nb_nn_balanced_shuffled_100k_train.json \ | |
--validation_file /mnt/disks/flaxdisk/NCC_nb_nn_balanced/nb_nn_balanced_shuffled_100k_validation.json \ | |
--max_seq_length="512" \ | |
--weight_decay="0.01" \ | |
--per_device_train_batch_size="32" \ | |
--per_device_eval_batch_size="32" \ | |
--learning_rate="8e-3" \ | |
--warmup_steps="2000" \ | |
--overwrite_output_dir \ | |
--cache_dir /mnt/disks/flaxdisk/cache/ \ | |
--num_train_epochs="15" \ | |
--adam_beta1="0.9" \ | |
--adam_beta2="0.98" \ | |
--logging_steps="500" \ | |
--save_steps="2500" \ | |
--eval_steps="2500" \ | |
--preprocessing_num_workers 96 \ | |
--adafactor \ | |
--push_to_hub | |