tokenizer_config.json

{"target_lang": "eng", "source_lang": "tha"}