tokenizer_config.json

{
  "clean_up_tokenization_spaces": true,  
  "tokenizer_class": "SentencePieceJA",
  "auto_map": {
    "AutoTokenizer": "sentencepiece_ja.SentencePieceJA"
  }
}