KarveeSaimai / tokenizer_config.json
Thanravee's picture
Training in progress, step 1
18d5230
raw
history blame
456 Bytes
{
"additional_special_tokens": [
"<madeupword0>",
"<madeupword1>",
"<madeupword2>",
"<madeupword3>",
"<madeupword4>",
"<madeupword5>",
"<madeupword6>"
],
"bos_token": "<s>",
"clean_up_tokenization_spaces": true,
"cls_token": "<s>",
"eos_token": "</s>",
"model_max_length": 2048,
"pad_token": "<pad>",
"sep_token": "</s>",
"sp_model_kwargs": {},
"tokenizer_class": "XGLMTokenizer",
"unk_token": "<unk>"
}