The paper introduces REDaCT, an inference-time unlearning framework for large reasoning models that removes forget-set information from both the reasoning trace and the final answer without causing reasoning collapse. REDaCT uses a binary leak detector to spot sentences that leak forget-set data and an embedding optimization module that injects a corrective embedding at the first detected leak, steering subsequent reasoning toward non-leaking sentences. Experiments demonstrate that REDaCT preserves coherent chain-of-thought reasoning, outperforms existing unlearning methods, and achieves perfect answer forgetting with near-perfect chain-of-thought forgetting.
By Joshua Maccoby, Wendy Hui Wang
arXiv:2610.14605v1 Announce Type: new
Abstract: Current evaluations of multimodal unlearning typically focus on whether a model stops outputting target information. This surface-level criterion, howe...
By Haodong Chen, Shengtao Wen, Dong Liang, Lingbing Guo, Xiang Chen
arXiv:2606. 05201v1 Announce Type: new Abstract: Reasoning language models do not distinguish tokens used for computation from tokens that constitute persistent state: once generated, all hidden thoughts remain in context and influence future predictions.
By Fei Ding, Yongkang Zhang, Runhao Liu, Yuhao Liao, Zijian Zeng, Huiming Yang
The paper identifies a specific issue in supervised fine‑tuning (SFT) of large language models called factual access failure, where models can recognize correct facts under constrained tests but fail to generate them in open‑ended settings. It demonstrates that SFT can cause both genuine wrong answers and expression‑level errors such as verbosity or formatting mismatches. To mitigate this, the authors propose Recall‑Anchored Distillation (RAD), a self‑distillation method that aligns the fine‑tuned model with the base model’s soft output distribution on unlabeled out‑of‑distribution text, thereby recovering lost factual recall without needing labeled data.
By Haodong Chen, Yadong Wang, Shengtao Wen, Dong Liang, Xiang Chen
arXiv:2607. 09236v1 Announce Type: new Abstract: Machine unlearning in LLMs is the targeted removal of specific knowledge while preserving all other capabilities, critical for privacy and safety.
By Amit Peleg, Naman Deep Singh, Naama Pearl, Bibhabasu Mohapatra, Matthias Hein
arXiv:2510. 21978v2 Announce Type: replace-cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has delivered impressive gains in mathematical and multimodal reasoning and has become a standard post-training paradigm for contemporary language and vision-language models.
By Hoang Phan, Xianjun Yang, Yuanshun Yao, Jingyu Zhang, Shengjie Bi, Xiaocheng Tang, Madian Khabsa, Lijuan Liu, Deren Lei
The paper investigates why large reasoning models (LRMs) lose safety alignment when faced with harmful queries. By analyzing token-level refusal dynamics, the authors identify a vulnerability called Onset Refusal Collapse (ORC), where the refusal signal drops sharply at the first generated token, leading to unsafe responses. They introduce SafeToken, a lightweight inference-time intervention that injects a learned safety anchor at reasoning onset, which mitigates ORC, improves safety on harmful-query benchmarks, and largely preserves reasoning utility.
By Yizheng Yang, Haining Yu, Yuechen Wang, Yikai Hou, Xing Fu, Jinbo Yang, Tianqing Zhu
SAVeR$^2$ is a safety alignment method for Large Reasoning Models that decomposes safety objectives into four fine‑grained verifiable reward signals and uses a difficulty‑aware data selection strategy to stabilize training. It introduces a reasoning‑preserving gradient projection mechanism that resolves conflicts between safety and reasoning gradients, eliminating the safety tax. Experiments on models from 1.5B to 671B show that SAVeR$^2$ improves safety performance and reduces over‑refusal rates without harming general reasoning abilities.
By Yichen Sun, Qirui Liu, Yan Wang, Jianan Lin, Linbo Jiang, Shiyu Wang, Zhibo Wang, Zhixuan Chu
Large Language Models (LLMs) can memorize and reproduce sensitive, copyrighted, or otherwise undesirable training content, creating privacy, safety, and regulatory concerns. Machine unlearning offers...
arXiv:2609.39346v1 Announce Type: new
Abstract: Large language models (LLMs) offer strong reasoning capabilities but are often costly to access through commercial APIs, while small language models (S...
By Bohan Zhang (Southeast University), Linan Yue (Southeast University), Weibo Gao (Hong Kong Polytechnic University), Pengyu Chen (Southeast University), Hong Guo (Southeast University), Yanqi Hao (ZTE Corporation)
The paper introduces Forgetting Only What Matters via Unlearning Layers (FOM-UL), a layer‑selective unlearning framework for large language models. FOM-UL uses a forget‑to‑retain significance score to identify transformer layers that strongly influence the forget set while being insensitive to the retain set, allowing targeted updates that preserve most of the model. Experiments on TOFU, KnowUnDo, and MUSE-style benchmarks show that FOM-UL reduces residual memorization and maintains utility better than several baselines, even after 8‑bit and 4‑bit post‑training quantization, and it also limits recovery of forgotten content in adversarial prompt tests.
By Ravi Ranjan, Olivera Kotevska, Agoritsa Polyzou
arXiv:2608. 07931v1 Announce Type: new Abstract: Large reasoning models (LRMs) are prone to hallucination, which undermines their reliability and poses challenges for safe deployment.
By Zhengze Huang, Luyang Yu, Di Hong, Xinzhe Huang, Wanyu Lin, Zhixuan Chu, Zhan Qin, Tianhang Zheng