wav2vec2-large-asr-th-2 / tokenizer_config.json
DylanonWic's picture
Create tokenizer_config.json
840a504
raw
history blame
547 Bytes
{
"bos_token": "<s>",
"do_lower_case": false,
"eos_token": "</s>",
"model_max_length": 1000000000000000019884624838656,
"name_or_path": "airesearch/wav2vec2-large-xlsr-53-th",
"num_proc": 4,
"pad_token": "[PAD]",
"replace_word_delimiter_char": " ",
"special_tokens_map_file": "/root/.cache/huggingface/hub/models--airesearch--wav2vec2-large-xlsr-53-th/snapshots/3155938c549b23eee16b1d4b55dcb161b7fe4bcf/special_tokens_map.json",
"tokenizer_class": "Wav2Vec2CTCTokenizer",
"unk_token": "[UNK]",
"word_delimiter_token": "|"
}