tokenizer_config.json

{
    "tokenizer_class": "SentencePieceTokenizerFast",
    "model_file": "spm_model.model"
}