14 lines
342 B
JSON
14 lines
342 B
JSON
{
|
|
"add_prefix_space": false,
|
|
"bos_token": "<|endoftext|>",
|
|
"clean_up_tokenization_spaces": true,
|
|
"eos_token": "<|endoftext|>",
|
|
"max_length": 256,
|
|
"model_max_length": 1024,
|
|
"stride": 0,
|
|
"tokenizer_class": "GPT2Tokenizer",
|
|
"truncation_side": "right",
|
|
"truncation_strategy": "longest_first",
|
|
"unk_token": "<|endoftext|>"
|
|
}
|