tokenizer_config.json

{"target_lang": "msa", "source_lang": "jpn"}