wav2vec2-large-xls-r-300m-Indonesian / tokenizer_config.json
kingabzpro's picture
add LM preprocessing
aa94c13
raw
history blame
340 Bytes
{"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "special_tokens_map_file": "wav2vec2-xls-r-300m-indonesian/special_tokens_map.json", "tokenizer_file": null, "name_or_path": "wav2vec2-xls-r-300m-indonesian", "tokenizer_class": "Wav2Vec2CTCTokenizer"}