finetune for machine translation

[muhammed-saeed]

Hi guys, I am currently endeavoring to train a RWKV model on a dataset in either JSON or TSV format. The structure of the dataset is as follows: {"instruction": "Translate language 1 into language 2", "input": "raw text in language 1", "output": "Desired text in language 2"}. However, the same structure may have varied instructions, input, and output format.

Would anyone be able to direct me to a repository that provides a comprehensive guide on fine-tuning a pre-trained RWKV model on such data? Alternatively, if I were to utilize Hugging Face's resources, how could I go about fine-tuning the RWKV model for this particular task?