sansar-20m / tokenizer_config.json
ks2812's picture
v0.1.0: tokv3_v01_d20m (ex-Gita 0.7177)
329a89b verified
Raw History Blame Contribute Delete
335 Bytes
{
"tokenizer_class": "SansarTokenizer",
"auto_map": {
"AutoTokenizer": [
"tokenization_sansar.SansarTokenizer",
null
]
},
"unk_token": "<unk>",
"bos_token": "<s>",
"eos_token": "</s>",
"pad_token": "</s>",
"model_max_length": 512,
"padding_side": "left",
"clean_up_tokenization_spaces": false
}