| { | |
| "add_prefix_space": false, | |
| "auto_map": { | |
| "AutoTokenizer": [ | |
| "tokenization_milli.MiLLiTokenizer", | |
| null | |
| ] | |
| }, | |
| "clean_up_tokenization_spaces": true, | |
| "model_max_length": 512, | |
| "tokenizer_class": "MiLLiTokenizer", | |
| "unk_token": "<UNK>", | |
| "pad_token": "<PAD>", | |
| "eos_token": "<EOS>", | |
| "mask_token": "<mask]", | |
| "additional_special_tokens": ["<UPPER>"] | |
| } |