tokenizer_config.json

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