Semantic Loss Guided Data Efficient Supervised Fine Tuning for Safe Responses in LLMs

Open in new window