|
python run_clm_mp.py \ |
|
--model_name_or_path /mnt/disks/flaxdisk/norwegian-gptneo-red/ \ |
|
--tokenizer_name /mnt/disks/flaxdisk/norwegian-gptneo-red/ \ |
|
--train_file /mnt/disks/flaxdisk/corpus/social_train.json \ |
|
--validation_file /mnt/disks/flaxdisk/corpus/social_validation.json \ |
|
--do_train \ |
|
--do_eval \ |
|
--block_size 1024 \ |
|
--num_train_epochs 5 \ |
|
--learning_rate 4e-6 \ |
|
--per_device_train_batch_size 3 \ |
|
--per_device_eval_batch_size 3 \ |
|
--overwrite_output_dir \ |
|
--output_dir /mnt/disks/flaxdisk/output/ \ |
|
--cache_dir /mnt/disks/flaxdisk/cache/ \ |
|
--dtype bfloat16 \ |
|
--logging_steps 97 \ |
|
--eval_steps 96\ |
|
--push_to_hub |
|
|