Transformers
MathFormer-16K-BPE / tokenizer_config.json
LastTransformer's picture
Deploy native SentencePiece-backed 16K Unigram tokenizer with verified space preservation.
6984b77 verified
Raw
History Blame Contribute Delete
248 Bytes
{
"add_prefix_space": true,
"backend": "tokenizers",
"bos_token": "<s>",
"eos_token": "</s>",
"model_max_length": 4096,
"pad_token": "<pad>",
"padding_side": "right",
"tokenizer_class": "TokenizersBackend",
"unk_token": "<unk>"
}