I have not tried fine-tuning, but I would recommend following OpenAI's guidebook and considering fine-tuning only if prompting is not yielding satisfactory results and high-quality data is available. Besides, if you are considering fine-tuning an open-source model, you may refer to Permutation Distillation and Instruction Distillation.
I have not tried fine-tuning, but I would recommend following OpenAI's guidebook and considering fine-tuning only if prompting is not yielding satisfactory results and high-quality data is available. Besides, if you are considering fine-tuning an open-source model, you may refer to Permutation Distillation and Instruction Distillation.