Skip to content

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

Oct 12 · Monday

All papers

Found 7 papers
Filter3

Attack type

Extraction and theft7

Defense type

Model hardening7Inference-time intervention5Instruction–data separation3

Impact

Harmful content23Data and privacy leaks7

System type

Models and APIs7

Attack surface / entry point

User input7Training data3

Layer

Training and data layer5
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 265papers have no tags yet and are not included in filtered results.
Defense type：Model hardeningImpact：Data and privacy leaksAttack surface / entry point：User inputClear all

Another 265papers do not have category tags yet and are not included in filtered results.

  1. DefensesarXiv 2609.40286

    研究者提出 COVER：在语言预算下选择源语言实现跨语言遗忘

    提出 COVER，在语言预算下选择源语言实现跨语言知识遗忘

    arXiv
    2609.40286
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Tiny Aya Global, Qwen3-4B-Instruct-2507, Llama-3.2-3B-Instruct
    AbstractCOVER 在语言预算下选源语言，以降低未监督语言上相对 oracle 的残余访问。

    语言预算下的多语言遗忘：遗忘监督只放在 K 种源语言上，目标是降低其余语言对同一事实的残余访问，而不是把每种评测语言都译成监督数据。

    Paper analysisRead analysis
  3. DefensesarXiv 2609.34591

    TULIP：按输入定位层级的定向大模型遗忘方法

    提出 TULIP，按输入定位形成层并去除目标 token 对齐以实现定向遗忘

    arXiv
    2609.34591
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Llama-3.1-8B, Llama-3.2-3B, Llama-3.2-1B and others7
    AbstractTULIP 在形成结束的层去掉遗忘目标对齐，多数设置下 FQ 更高，4-bit 下 FQ 下降。

    TULIP 是一种逐输入选择层的表示级遗忘方法，目标是去掉遗忘答案的形成、留下读出。输出级方法可能只压低特定回答的似然。RMU、LUNAR 等则对全部遗忘样本共用一层。

    Paper analysisRead analysis
  4. Evaluation and benchmarksarXiv 2609.33481

    IDUnlearn-Bench：面向视觉语言模型的个体级遗忘基准

    提出 IDUnlearn-Bench，评测 VLM 个体级遗忘后的身份残留

    arXiv
    2609.33481
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3-VL-2B, Qwen3-VL-4B, Qwen3-VL-8B and others6
    AbstractIDUnlearn-Bench 用四族探针显示：压低直接属性访问后，身份仍常能被反向查出、绑定或重构。

    IDUnlearn-Bench 评测 VLM 是否忘掉一个可跨模态识别的人，而不只是答不出某条监督事实。

    Paper analysisRead analysis
  7. DefensesarXiv 2602.24210

    研究者提出 Staged Decoding 与 SFT 数据集，减少大推理模型推理轨迹中的隐私泄露

    提出 Staged Decoding 与指令跟随训练，减少推理轨迹隐私泄露

    arXiv
    2602.24210
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    Qwen 3 1.7B, Qwen 3 4B, Qwen 3 8B and others6
    Abstract用一般指令跟随训练加分段解码，可降低轨迹隐私泄露，但数学效用下降。

    本文把大推理模型的隐私问题做成轨迹可控性：隐私指令是指令，轨迹不跟随就会把敏感上下文写进可被提示注入导出的文本。

    Paper analysisRead analysis
You have reached the end