wav2vec2-1b-nst / tokenizer_config.json
versae's picture
Add 5-gram lang model and eval code
e0cc13d
raw
history blame
343 Bytes
{
"bos_token": "<s>",
"do_lower_case": false,
"eos_token": "</s>",
"name_or_path": "./",
"pad_token": "[PAD]",
"processor_class": "Wav2Vec2ProcessorWithLM",
"replace_word_delimiter_char": " ",
"special_tokens_map_file": null,
"tokenizer_class": "Wav2Vec2CTCTokenizer",
"unk_token": "[UNK]",
"word_delimiter_token": "|"
}