Forget What Comes Next: LLM Unlearning via Statistical Independence
cs.LG
Submitted: 2026-11-02
Updated: 2026-11-02
Code: https://github.com/shkabrii/PINT
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Physics of Language Models: Part 3.1, Knowledge Storage and Extraction
- Quantifying Memorization Across Neural Language Models
- LOKI: Memory-Free Null-Space Constrained Lifelong Knowledge Editing
- Large sample analysis of the median heuristic
- Mechanistic Unlearning: Robust Knowledge Unlearning and Editing via Mechanistic Localization
- Towards Independence Criterion in Machine Unlearning of Features and Labels
- Measuring Massive Multitask Language Understanding
- The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning
- Unlearning in LLMs: Methods, Evaluation, and Open Challenges
- Decoupled Weight Decay Regularization
- An Adversarial Perspective on Machine Unlearning for AI Safety
- TOFU: A Task of Fictitious Unlearning for LLMs
- 2 OLMo 2 Furious
- In-context Pretraining: Language Modeling Beyond Document Boundaries
- FairSplit: Decomposing the Embedding Space for Fair Classification
- Llama 2: Open Foundation and Fine-Tuned Chat Models
- Zephyr: Direct Distillation of LM Alignment
- Unlearning Evaluation through Subset Statistical Independence
- Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis