nb-wav2vec2-1b-nynorsk / tokenizer_config.json
versae's picture
Adding 5gram lang model bokmaal
2d2b4ab
raw
history blame
282 Bytes
{"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "special_tokens_map_file": null, "name_or_path": "./", "tokenizer_class": "Wav2Vec2CTCTokenizer", "processor_class": "Wav2Vec2ProcessorWithLM"}