spe_unigram_10m_50_3k_datamixC / tokenizer_config.json
kartikmosaicml's picture
Upload tokenizer
04a0564
{
"bos_token": "<s>",
"clean_up_tokenization_spaces": false,
"eos_token": "</s>",
"model_max_length": 1000000000000000019884624838656,
"tokenizer_class": "PreTrainedTokenizerFast",
"unk_token": "<unk>",
"vocab_file": "kartik-dbx-spe-dmixC.model"
}