19 lines
494 B
JSON
19 lines
494 B
JSON
|
{
|
||
|
|
"add_bos_token": true,
|
||
|
|
"add_eos_token": false,
|
||
|
|
"unk_token": "<unk>",
|
||
|
|
"bos_token": "<s>",
|
||
|
|
"eos_token": "</s>",
|
||
|
|
"pad_token": "<PAD|LLM-jp>",
|
||
|
|
"cls_token": "<CLS|LLM-jp>",
|
||
|
|
"sep_token": "<SEP|LLM-jp>",
|
||
|
|
"eod_token": "</s>",
|
||
|
|
"mask_token": "<MASK|LLM-jp>",
|
||
|
|
"extra_ids": 0,
|
||
|
|
"sp_model_kwargs": {},
|
||
|
|
"model_max_length": 1000000000000000019884624838656,
|
||
|
|
"clean_up_tokenization_spaces": false,
|
||
|
|
"special_tokens_map_file": null,
|
||
|
|
"tokenizer_class": "PreTrainedTokenizerFast"
|
||
|
|
}
|