Skip to content

Translation in progress Some entries still include text in the original language.

Oct 12 · Monday

Defenses and guardrails · Papers

Found 1 paper
Filter2

Attack technique

Conversation structure manipulation1

Defense type

Inference-time intervention1
Select any value within a category; different categories must all match. Counts include the other filters you have selected.
Attack technique：Conversation structure manipulationDefense type：Inference-time interventionTopic: Defenses and guardrailsClear all
  1. InterpretabilityarXiv 2603.08234

    续写与拒答之争：大语言模型中续写触发越狱的机制分析

    分析续写触发越狱的注意力头机制并提出 HCS

    arXiv
    2603.08234
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    LLaMA-2-7B-Chat, Qwen2.5-7B-Instruct
    Abstract作者称越狱来自两类头竞争，HCS 加蒸馏可把该竞争用于推理时干预。

    作者用注意力头机制分析 continuation-triggered jailbreak：同一有害指令下，把 continuation 后缀从指令后面挪到提示词边界外，模型更容易顺着后缀继续生成。

    Paper analysisRead analysis
You have reached the end