wav2vec2-ru / tokenizer_config.json
Alikhan Urumov
add tokenizer
f4403d2
raw
history blame
415 Bytes
{
"bos_token": "<s>",
"do_lower_case": false,
"eos_token": "</s>",
"name_or_path": "/content/gdrive/MyDrive/NeuroNet/SST/wav2vec_model",
"pad_token": "[PAD]",
"replace_word_delimiter_char": " ",
"special_tokens_map_file": "/content/gdrive/MyDrive/NeuroNet/SST/wav2vec_model/special_tokens_map.json",
"tokenizer_class": "Wav2Vec2CTCTokenizer",
"unk_token": "[UNK]",
"word_delimiter_token": "|"
}