Guardrail Baselines for Unlearning in LLMs
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:MVBRHB7Lrecord.jsonopen to challenge →
read the original abstract
Recent work has demonstrated that finetuning is a promising approach to 'unlearn' concepts from large language models. However, finetuning can be expensive, as it requires both generating a set of examples and running iterations of finetuning to update the model. In this work, we show that simple guardrail-based approaches such as prompting and filtering can achieve unlearning results comparable to finetuning. We recommend that researchers investigate these lightweight baselines when evaluating the performance of more computationally intensive finetuning methods. While we do not claim that methods such as prompting or filtering are universal solutions to the problem of unlearning, our work suggests the need for evaluation metrics that can better separate the power of guardrails vs. finetuning, and highlights scenarios where guardrails expose possible unintended behavior in existing metrics and benchmarks.
This paper has not been read by Pith yet.
Forward citations
Cited by 15 Pith papers
-
Can VLMs Truly Forget? Benchmarking Training-Free Visual Concept Unlearning
VLM-UnBench demonstrates that prompt-based training-free unlearning in VLMs leaves forget accuracy near the no-instruction baseline except under oracle conditions that reveal the target concept.
-
Improving LLM Unlearning Robustness via Random Perturbations
LLM unlearning is reframed as inadvertently installing backdoor triggers on forget-tokens; Random Noise Augmentation is introduced as a defense that improves robustness with theoretical guarantees.
-
Fast Unlearning at Scale via Margin Self-Correction
MASC achieves competitive forget-retain trade-offs in language model unlearning at lower computational cost via margin self-correction and an online stopping criterion on TOFU, MUSE News, and MUSE Books.
-
Distinguishable Deletion: Unifying Knowledge Erasure and Refusal for Large Language Model Unlearning
Distinguishable Deletion unifies knowledge erasure and refusal for LLM unlearning via an energy index that enforces boundaries during training and enables refusal at inference.
-
Robust LLM Unlearning Against Relearning Attacks: The Minor Components in Representations Matter
Targeting minor components in LLM representations during unlearning yields substantially better resistance to relearning attacks than prior methods.
-
CAP: Controllable Alignment Prompting for Unlearning in LLMs
CAP is a reinforcement-learning-driven prompt optimization framework that suppresses target knowledge in LLMs while preserving general capabilities, enabling reversible unlearning without any parameter updates.
-
CAP: Controllable Alignment Prompting for Unlearning in LLMs
CAP optimizes prompts via reinforcement learning to selectively unlearn target knowledge in LLMs while preserving general capabilities, without any parameter updates and with reversible revocation.
-
CAP: Controllable Alignment Prompting for Unlearning in LLMs
CAP enables reversible unlearning of targeted knowledge in LLMs through optimized prompts generated via reinforcement learning, without any parameter updates.
-
Representation-Guided Parameter-Efficient LLM Unlearning
REGLU guides LoRA-based unlearning via representation subspaces and orthogonal regularization to outperform prior methods on forget-retain trade-off in LLM benchmarks.
-
CURaTE: Continual Unlearning in Real Time with Ensured Preservation of LLM Knowledge
CURaTE performs continual unlearning in LLMs in real time by using sentence embeddings to detect and refuse forget requests without changing model parameters, achieving effective forgetting and perfect knowledge preservation.
-
Downgrade to Upgrade: Optimizer Simplification Enhances Robustness in LLM Unlearning
Downgrading optimizers to lower-information variants during LLM unlearning yields more robust forgetting on MUSE and WMDP benchmarks by converging to harder-to-perturb loss basins.
-
Short paper: Models in the dark -- Rectification and erasure under GDPR in ML supply chains
Survey identifying technical and supply-chain barriers to GDPR data subject rights in ML, with new framing of 'models in the dark' for downstream opacity.
-
Runtime-Structured Task Decomposition for Agentic Coding Systems
Runtime-structured task decomposition reduces retry costs in agentic coding systems by up to 51.7% versus monolithic prompts by rerunning only failed subtasks on two software engineering workloads.
-
AI as a Tool for Simulation-Based Experiments in Literary Studies
Proposes AI-driven simulations for literary-historical experiments and reports preliminary text-generation results claiming the first limited in-distribution outputs matching human novels.
-
On the Hidden Costs of Counterfactual Knowledge Training in LLM Unlearning
Counterfactual tuning for LLM unlearning induces knowledge conflict and hallucination spillover, diagnosed via the RWKU+ benchmark.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.