| train: reflow-dataset | |
| valid: openwebtext-valid | |
| tokenizer_name_or_path: gpt2 | |
| cache_dir: /PATH/TO/TRAIN/DATASET | |
| wrap: True | |
| streaming: False | |
| insert_train_eos: True | |
| insert_valid_eos: True | |
| save_dir: /PATH/TO/SAVE/DATASET |
| train: reflow-dataset | |
| valid: openwebtext-valid | |
| tokenizer_name_or_path: gpt2 | |
| cache_dir: /PATH/TO/TRAIN/DATASET | |
| wrap: True | |
| streaming: False | |
| insert_train_eos: True | |
| insert_valid_eos: True | |
| save_dir: /PATH/TO/SAVE/DATASET |