babylm-object-sp40-tokenizer / tokenizer_config.json
miguelcsx's picture
Publish corpus-derived object tournament artifacts
d951462 verified
Raw
History Blame Contribute Delete
198 Bytes
{
"bos_token": "<s>",
"eos_token": "</s>",
"mask_token": "<mask>",
"model_max_length": 1024,
"pad_token": "<pad>",
"tokenizer_class": "PreTrainedTokenizerFast",
"unk_token": "<unk>"
}