gpt2-xl-japanese / tokenizer_config.json
schnell's picture
train 10 epochs
47ae502
raw
history blame
336 Bytes
{
"bos_token": "<s>",
"eos_token": "</s>",
"model_max_length": 1024,
"name_or_path": "juman-bpe-wiki-cc100-50000",
"padding_side": "right",
"special_tokens_map_file": "juman-bpe-wiki-cc100-50000/special_tokens_map.json",
"tokenizer_class": "PreTrainedTokenizerFast",
"truncation_side": "right",
"unk_token": "<unk>"
}