radiogroup-crits's picture
Update tokenizer_config.json
5cde451
raw
history blame contribute delete
423 Bytes
{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "<pad>", "do_lower_case": false, "word_delimiter_token": "|", "replace_word_delimiter_char": " ", "special_tokens_map_file": null, "tokenizer_file": null, "name_or_path": "radiogroup-crits/wav2vec2-xls-r-1b-italian-doc4lm-5gram", "processor_class": "Wav2Vec2ProcessorWithLM", "trust_remote_code": false, "tokenizer_class": "Wav2Vec2CTCTokenizer"}