xgml-564m-MoQlora-hindi-expert / tokenizer_config.json
sam2ai's picture
Upload folder using huggingface_hub
5af0d78
raw
history blame
2.49 kB
{
"added_tokens_decoder": {
"0": {
"content": "<s>",
"lstrip": true,
"normalized": false,
"rstrip": true,
"single_word": false,
"special": true
},
"1": {
"content": "<pad>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": false
},
"2": {
"content": "</s>",
"lstrip": true,
"normalized": false,
"rstrip": true,
"single_word": false,
"special": true
},
"3": {
"content": "<unk>",
"lstrip": true,
"normalized": false,
"rstrip": true,
"single_word": false,
"special": true
},
"256001": {
"content": "<madeupword0>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": false
},
"256002": {
"content": "<madeupword1>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": false
},
"256003": {
"content": "<madeupword2>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": false
},
"256004": {
"content": "<madeupword3>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": false
},
"256005": {
"content": "<madeupword4>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": false
},
"256006": {
"content": "<madeupword5>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": false
},
"256007": {
"content": "<madeupword6>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": false
}
},
"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",
"tokenizer_file": null,
"trust_remote_code": false,
"unk_token": "<unk>",
"use_fast": true
}