analiza aiqbench
This paper introduces REDaCT, an inference-time unlearning framework for large reasoning models that successfully prevents "reasoning collapse" while ensuring complete removal of sensitive information. By combining a binary leak detector with an offline corrective embedding module, the method preserves coherent chain-of-thought generation better than existing state-of-the-art techniques. This advancement is critical for deploying reliable AI systems where data privacy and model integrity must coexist without degrading reasoning capabilities.
REDaCT: Collapse-Free Unlearning in Large Reasoning Models
arXiv:2610.12769v1 Announce Type: cross Abstract: Unlearning in large reasoning models (LRMs) requires removing forget-set information from both the reasoning trace and the final answer. Existing approaches achieve this objective at the cost of reasoning collapse, where the model fails to produce a coherent chain-of-thought (CoT) and does not properly terminate its reasoning process. To address this challenge, we propose REDaCT (REgenerate via Detection and Corrective Tuning), an inference-time unlearning framework for LRMs. REDaCT consists of two key components: (1) a binary leak detector that identifies sentences in the reasoning trace that leak forget-set information, and (2) an embedding optimization module that learns a corrective embedding offline, such that when injected at the first detected leak during inference, it guides subsequent reasoning towards sentences that do not leak forget-set information. Experimental results show that REDaCT completely avoids reasoning collapse, substantially improving reasoning preservation compared to state-of-the-art unlearning methods while maintaining perfect answer forgetting and near-perfect CoT forgetting. Our code is available at: https://github.com/jmaccoby/redact.
Abstract:Unlearning in large reasoning models (LRMs) requires removing forget-set information from both the reasoning trace and the final answer. Existing approaches achieve this objective at the cost of reasoning collapse, where the model fails to produce a coherent chain-of-thought (CoT) and does not properly terminate its reasoning process. To address this challenge, we propose REDaCT (REgenerate via Detection and Corrective Tuning), an inference-time unlearning framework for LRMs. REDaCT consists of two key components: (1) a binary leak detector that identifies sentences in the reasoning trace that leak forget-set information, and (2) an embedding optimization module that learns a corrective embedding offline, such that when injected at the first detected leak during inference, it guides subsequent reasoning towards sentences that do not leak forget-set information. Experimental results show that REDaCT completely avoids reasoning collapse, substantially improving reasoning preservation compared to state-of-the-art unlearning methods while maintaining perfect answer forgetting and near-perfect CoT forgetting. Our code is available at: this https URL.
Accepted to Findings of EMNLP 2026. 19 pages, 7 figures, 6 tables. Copyright 2026 Association for Computational Linguistics (ACL). Licensed under CC BY 4.0
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
arXiv:2610.12769 [cs.LG]
(or arXiv:2610.12769v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.12769 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 208 cuvinte
Subiecte conexe
Links #5: 2026/07
LessWrong · 31 iul. 2026
Concrete Evaluations to Investigate the OpenAI Model That Hacked Hugging Face
AI Alignment Forum · 3 aug. 2026
Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident
METR · 26 aug. 2026
[AINews] High Taste Testers Ep. 1
Latent Space · 17 oct. 2026