urdumodel / tokenizer_config.json
Talha's picture
URDU-ASR-LM
e3a805a
raw
history blame contribute delete
381 Bytes
{
"bos_token": "<s>",
"do_lower_case": false,
"eos_token": "</s>",
"name_or_path": "urdumodel",
"pad_token": "[PAD]",
"processor_class": "Wav2Vec2ProcessorWithLM",
"replace_word_delimiter_char": " ",
"special_tokens_map_file": "urdumodel/special_tokens_map.json",
"tokenizer_class": "Wav2Vec2CTCTokenizer",
"unk_token": "[UNK]",
"word_delimiter_token": "|"
}