| { | |
| "added_tokens_decoder": { | |
| "1": { | |
| "content": "AA", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "2": { | |
| "content": "AA1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "3": { | |
| "content": "AE", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "4": { | |
| "content": "AE1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "5": { | |
| "content": "AH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "6": { | |
| "content": "AH1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "7": { | |
| "content": "AO", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "8": { | |
| "content": "AO1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "9": { | |
| "content": "AW", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "10": { | |
| "content": "AW1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "11": { | |
| "content": "AX-H1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "12": { | |
| "content": "AY", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "13": { | |
| "content": "AY1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "15": { | |
| "content": "CH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "17": { | |
| "content": "DH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "18": { | |
| "content": "EH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "19": { | |
| "content": "EH1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "20": { | |
| "content": "ER", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "21": { | |
| "content": "ER1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "22": { | |
| "content": "EY", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "23": { | |
| "content": "EY1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "26": { | |
| "content": "HH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "27": { | |
| "content": "IX", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "28": { | |
| "content": "IX1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "29": { | |
| "content": "IY", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "30": { | |
| "content": "IY1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "31": { | |
| "content": "JH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "35": { | |
| "content": "NG", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "36": { | |
| "content": "NX", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "37": { | |
| "content": "OW", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "38": { | |
| "content": "OW1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "39": { | |
| "content": "OY", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "40": { | |
| "content": "OY1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "44": { | |
| "content": "S1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "45": { | |
| "content": "SH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "47": { | |
| "content": "T1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "48": { | |
| "content": "TH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "49": { | |
| "content": "UH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "50": { | |
| "content": "UH1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "51": { | |
| "content": "UW", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "52": { | |
| "content": "UW1", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "57": { | |
| "content": "ZH", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "58": { | |
| "content": "[UNK]", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "59": { | |
| "content": "[PAD]", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": false | |
| }, | |
| "60": { | |
| "content": "<s>", | |
| "lstrip": false, | |
| "normalized": false, | |
| "rstrip": false, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "61": { | |
| "content": "</s>", | |
| "lstrip": false, | |
| "normalized": false, | |
| "rstrip": false, | |
| "single_word": false, | |
| "special": true | |
| } | |
| }, | |
| "bos_token": "<s>", | |
| "clean_up_tokenization_spaces": false, | |
| "do_lower_case": false, | |
| "eos_token": "</s>", | |
| "extra_special_tokens": {}, | |
| "model_max_length": 1000000000000000019884624838656, | |
| "pad_token": "[PAD]", | |
| "replace_word_delimiter_char": " ", | |
| "target_lang": null, | |
| "tokenizer_class": "Wav2Vec2CTCTokenizer", | |
| "unk_token": "[UNK]", | |
| "word_delimiter_token": "|" | |
| } | |