tokenizer_config.json

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