On the Role of Attention Heads in Large Language Model Safety

• URL: http://arxiv.org/abs/2410.13708v1
• Date: Thu, 17 Oct 2024 16:08:06 GMT
• Title: On the Role of Attention Heads in Large Language Model Safety
• Authors: Zhenhong Zhou, Haiyang Yu, Xinghua Zhang, Rongwu Xu, Fei Huang, Kun Wang, Yang Liu, Junfeng Fang, Yongbin Li,
• Abstract summary: Large language models (LLMs) achieve state-of-the-art performance on multiple language tasks, yet their safety guardrails can be circumvented. We propose a novel metric which tailored for multi-head attention, the Safety Head ImPortant Score (Ships) to assess the individual heads' contributions to model safety.
• Score: 64.51534137177491
• License:
• Abstract: Large language models (LLMs) achieve state-of-the-art performance on multiple language tasks, yet their safety guardrails can be circumvented, leading to harmful generations. In light of this, recent research on safety mechanisms has emerged, revealing that when safety representations or component are suppressed, the safety capability of LLMs are compromised. However, existing research tends to overlook the safety impact of multi-head attention mechanisms, despite their crucial role in various model functionalities. Hence, in this paper, we aim to explore the connection between standard attention mechanisms and safety capability to fill this gap in the safety-related mechanistic interpretability. We propose a novel metric which tailored for multi-head attention, the Safety Head ImPortant Score (Ships), to assess the individual heads' contributions to model safety. Based on this, we generalize Ships to the dataset level and further introduce the Safety Attention Head AttRibution Algorithm (Sahara) to attribute the critical safety attention heads inside the model. Our findings show that the special attention head has a significant impact on safety. Ablating a single safety head allows aligned model (e.g., Llama-2-7b-chat) to respond to 16 times more harmful queries, while only modifying 0.006% of the parameters, in contrast to the ~ 5% modification required in previous studies. More importantly, we demonstrate that attention heads primarily function as feature extractors for safety and models fine-tuned from the same base model exhibit overlapping safety heads through comprehensive experiments. Together, our attribution approach and findings provide a novel perspective for unpacking the black box of safety mechanisms within large models.

Related papers

• The Hidden Risks of Large Reasoning Models: A Safety Assessment of R1 [70.94607997570729]
  We present a comprehensive safety assessment of OpenAI-o3 and DeepSeek-R1 reasoning models. We investigate their susceptibility to adversarial attacks, such as jailbreaking and prompt injection, to assess their robustness in real-world applications.
  arXiv  Detail & Related papers  (2025-02-18T09:06:07Z)
• Vulnerability Mitigation for Safety-Aligned Language Models via Debiasing [12.986006070964772]
  Safety alignment is an essential research topic for real-world AI applications. Our study first identified the difficulty of eliminating such vulnerabilities without sacrificing the model's helpfulness. Our method could enhance the model's helpfulness while maintaining safety, thus improving the trade-off-front.
  arXiv  Detail & Related papers  (2025-02-04T09:31:54Z)
• Internal Activation as the Polar Star for Steering Unsafe LLM Behavior [50.463399903987245]
  We introduce SafeSwitch, a framework that dynamically regulates unsafe outputs by monitoring and utilizing the model's internal states. Our empirical results show that SafeSwitch reduces harmful outputs by over 80% on safety benchmarks while maintaining strong utility.
  arXiv  Detail & Related papers  (2025-02-03T04:23:33Z)
• Spot Risks Before Speaking! Unraveling Safety Attention Heads in Large Vision-Language Models [9.318094073527563]
  Internal activations of large vision-language models (LVLMs) can identify malicious prompts across different attacks. This inherent safety perception is governed by sparse attention heads, which we term safety heads" By locating these safety heads and concatenating their activations, we construct a straightforward but powerful malicious prompt detector.
  arXiv  Detail & Related papers  (2025-01-03T07:01:15Z)
• Multimodal Situational Safety [73.63981779844916]
  We present the first evaluation and analysis of a novel safety challenge termed Multimodal Situational Safety. For an MLLM to respond safely, whether through language or action, it often needs to assess the safety implications of a language query within its corresponding visual context. We develop the Multimodal Situational Safety benchmark (MSSBench) to assess the situational safety performance of current MLLMs.
  arXiv  Detail & Related papers  (2024-10-08T16:16:07Z)
• Superficial Safety Alignment Hypothesis [8.297367440457508]
  We propose the Superficial Safety Alignment Hypothesis (SSAH), which posits that safety alignment should teach an otherwise unsafe model to choose the correct reasoning direction. We identify four types of attribute-critical components in safety-aligned large language models (LLMs) Our findings show that freezing certain safety-critical components 7.5% during fine-tuning allows the model to retain its safety attributes while adapting to new tasks.
  arXiv  Detail & Related papers  (2024-10-07T19:53:35Z)
• What Makes and Breaks Safety Fine-tuning? A Mechanistic Study [64.9691741899956]
  Safety fine-tuning helps align Large Language Models (LLMs) with human preferences for their safe deployment. We design a synthetic data generation framework that captures salient aspects of an unsafe input. Using this, we investigate three well-known safety fine-tuning methods.
  arXiv  Detail & Related papers  (2024-07-14T16:12:57Z)
• Navigating the Safety Landscape: Measuring Risks in Finetuning Large Language Models [65.06446825020578]
  Safety alignment is crucial to ensure that large language models (LLMs) behave in ways that align with human preferences and prevent harmful actions during inference. We aim to measure the risks in finetuning LLMs through navigating the LLM safety landscape.
  arXiv  Detail & Related papers  (2024-05-27T17:31:56Z)
• USC: Uncompromising Spatial Constraints for Safety-Oriented 3D Object Detectors in Autonomous Driving [7.355977594790584]
  We consider the safety-oriented performance of 3D object detectors in autonomous driving contexts. We present uncompromising spatial constraints (USC), which characterize a simple yet important localization requirement. We incorporate the quantitative measures into common loss functions to enable safety-oriented fine-tuning for existing models.
  arXiv  Detail & Related papers  (2022-09-21T14:03:08Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.