GUARD-IT performs machine unlearning in LLMs via input-dependent activation steering at inference time, matching or exceeding gradient-based baselines on TOFU and MUSE while preserving utility and working under quantization.
Exploring Criteria of Loss Reweighting to Enhance
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
ETW uses predictive entropy as a proxy for token informativeness to improve selective unlearning in LLMs, achieving better forgetting with less utility loss than prior token-level methods.
citing papers explorer
-
Inference-Time Machine Unlearning via Gated Activation Redirection
GUARD-IT performs machine unlearning in LLMs via input-dependent activation steering at inference time, matching or exceeding gradient-based baselines on TOFU and MUSE while preserving utility and working under quantization.
-
Forget What Matters, Keep the Rest: Selective Unlearning of Informative Tokens
ETW uses predictive entropy as a proxy for token informativeness to improve selective unlearning in LLMs, achieving better forgetting with less utility loss than prior token-level methods.