umberto-wikipedia-uncased-v1 / tokenizer_config.json
Narsil's picture
Narsil HF staff
Adding precomputed fast-tokenizer.
483eca5
raw
history blame contribute delete
309 Bytes
{
"bos_token": "<s>",
"eos_token": "</s>",
"sep_token": "</s>",
"cls_token": "<s>",
"unk_token": "<unk>",
"pad_token": "<pad>",
"mask_token": "<mask>",
"additional_special_tokens": ["<s>NOTUSED", "</s>NOTUSED"],
"do_lower_case": true,
"special_tokens_map_file": null
}