tokenizer_config.json

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