11 lines
262 B
JSON
11 lines
262 B
JSON
|
{
|
||
|
|
"add_prefix_space": false,
|
||
|
|
"bos_token": "<|startoftext|>",
|
||
|
|
"clean_up_tokenization_spaces": true,
|
||
|
|
"eos_token": "<|endoftext|>",
|
||
|
|
"model_max_length": 1024,
|
||
|
|
"pad_token": "<|pad|>",
|
||
|
|
"tokenizer_class": "GPT2Tokenizer",
|
||
|
|
"unk_token": "<|endoftext|>"
|
||
|
|
}
|