| { | |
| "tokenizer_class": "MorPieceTokenizer", | |
| "auto_map": { | |
| "AutoTokenizer": [ | |
| "morpiece_tokenizer.MorPieceTokenizer", | |
| null | |
| ] | |
| }, | |
| "bos_token": "<s>", | |
| "eos_token": "</s>", | |
| "unk_token": "<unk>", | |
| "pad_token": "<pad>", | |
| "mask_token": "<mask>", | |
| "sep_token": "<sep>", | |
| "cls_token": "<cls>", | |
| "model_max_length": 512, | |
| "padding_side": "left", | |
| "truncation_side": "right", | |
| "chat_template": null, | |
| "clean_up_tokenization_spaces": false, | |
| "split_special_tokens": false, | |
| "strip_accents": null, | |
| "add_prefix_space": true, | |
| "vocab_size": 29066, | |
| "min_frequency": 10, | |
| "cutoff": 100, | |
| "bf": 10, | |
| "use_tokenizers_lib": true | |
| } |