Follow the Gradient: Tracing Safety Degradation Back to Training Data 待解读
新用户看这篇论文该怎么开始
- 先点「赞助解读」，AI 会把论文转成可直接执行的行动清单。
- 看完“可执行改进行动”后，可快速决定是否值得立项。
- 用上面的卡片内容直接发给团队，减少重复阅读。
摘要
Fine-tuning Large Language Models (LLMs) on instruction data can quietly erode their safety, even when no training example looks harmful. Because this safety-degrading data looks benign, standard content filters struggle to catch it. We tackle this problem by developing SimpleLoRIF, a scalable gradient embedding method that identifies fine-tuning examples whose gradients align with unsafe behavior. By simplifying a low-rank attribution method and restricting backpropagation to the final four transformer blocks, SimpleLoRIF can embed datasets on the order of a million examples, boosting gradient attribution from a small-scale analysis tool into a practical filter for full post-training datasets. We first validate it on emergent misalignment, where we find that gradients can connect insecure-code training data to the broadly harmful behaviors that appear after fine-tuning, with AUC as high as 0.865 in the gradient space of misaligned models. We then use SimpleLoRIF to build SafeTulu, a version of Tulu 3 with the 20% of examples most gradient-aligned with unsafe responses removed. Training on SafeTulu reduces GCG attack success rate by up to 22% across Qwen, Llama, and Olmo, outperforming both LLM-as-a-judge and prior gradient-based filtering, while largely preserving or even improving capability. Finally, we analyzed the removed data and identify possibly under-studied sources of safety degradation, such as tool-execution failures and narrative arithmetic. We release SafeTulu along with a Hugging Face Space that lets researchers build their own customized versions of SafeTulu to fine-tune on, such as by changing the removal fraction or filtering only certain subsets.
分析报告
暂无报告。点击“分析”开始生成。
个性化解读 与社区共享解读不同
用自己的话告诉 AI 你想要什么样的解读（比如"用大白话讲给非专业人士听"、"重点分析对我们团队 RAG 系统的可迁移性"），生成一份只属于你自己的版本；生成后也可以选择设为"愿意共享"，被更多人看到、点赞。