Minor SFT loss for LLM fine-tune to increase performance and reduce model deviation

Open in new window