tokenizer_config.json

{"target_lang": "rnd", "source_lang": "en"}