tokenizer_config.json

{"target_lang": "de", "source_lang": "crs"}