asr-wav2vec2-ctc-french / tokenizer_config.json
bofenghuang's picture
initial commit
b9072de
raw history blame
No virus
431 Bytes
{
"bos_token": "<s>",
"do_lower_case": false,
"eos_token": "</s>",
"model_max_length": 1000000000000000019884624838656,
"name_or_path": "outputs/big/wav2vec2-FR-7K-large-ft",
"pad_token": "[PAD]",
"processor_class": "Wav2Vec2ProcessorWithLM",
"replace_word_delimiter_char": " ",
"special_tokens_map_file": null,
"tokenizer_class": "Wav2Vec2CTCTokenizer",
"unk_token": "[UNK]",
"word_delimiter_token": "|"
}