paper-with-me

Papers

ALTER: Asymmetric LoRA for Token-Entropy-Guided Unlearning of LLMs

2026-03-02 · Xunlei Chen, Jinyu Guo, Yuang Li, Zhaokun Wang, Yi Gong, Jie Zou, Jiwei Wei, Wenhong Tian arxiv

Large language models (LLMs) have advanced to encompass extensive knowledge across diverse domains. Yet controlling what a LLMs should not know is important for ensuring alignment and thus safe use. However, effective unlearning in LLMs is difficult due to the fuzzy boundary between knowledge retention and forgetting. This challenge is exacerbated by entangled parameter spaces from continuous multi-domain training, often resulting in collateral damage, especially under aggressive unlearning strategies. Furthermore, the computational overhead required to optimize State-of-the-Art (SOTA) models with billions of parameters poses an additional barrier. In this work, we present ALTER, a lightweight unlearning framework for LLMs to address both the challenges of knowledge entanglement and unlearning efficiency. ALTER operates through two phases: (I) high entropy tokens are captured and learned via the shared A matrix in LoRA, followed by (II) an asymmetric LoRA architecture that achieves a specified forgetting objective by parameter isolation and unlearning tokens within the target subdomains. Serving as a new research direction for achieving unlearning via token-level isolation in the asymmetric framework. ALTER achieves SOTA performance on TOFU, WMDP, and MUSE benchmarks with over 95% forget quality and shows minimal side effects through preserving foundational tokens. By decoupling unlearning from LLMs' billion-scale parameters, this framework delivers excellent efficiency while preserving over 90% of model utility, exceeding baseline preservation rates of 47.8-83.6%.

📄 PDF Abstract BibTeX arXiv:2603.01792

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

ACPO: Asymmetric Credit Policy Optimization via Mode-Local Entropy Surrogate

2026-07-03 · Zijun Xie, Yuyang You, Yongzhi Li, Enlei Gong 외 arxiv

Outcome-supervised reinforcement learning scales to verifiable reasoning tasks, but trajectory-level rewards assign the same outcome signal to all sampled tokens, overlooking their unequal contributions to the reasoning …

Mathematical ReasoningReinforcement Learning

From Broad Exploration to Stable Synthesis: Entropy-Guided Optimization for Autoregressive Image Generation

2026-03-12 · Han Song, Yucheng Zhou, Jianbing Shen, Yu Cheng arxiv

Combining Chain-of-Thought (CoT) with Reinforcement Learning (RL) improves text-to-image (T2I) generation, yet the underlying interaction between CoT's exploration and RL's optimization remains unclear. We present a syst…

Reinforcement LearningImage Generation

Diversity or Precision? A Deep Dive into Next Token Prediction

2025-12-28 · Haoyuan Wu, Hai Wang, Jiajia Wu, Jinxiang Ou 외 arxiv

Recent advancements have shown that reinforcement learning (RL) can substantially improve the reasoning abilities of large language models (LLMs). The effectiveness of such RL training, however, depends critically on the…

Reinforcement Learning

Group Entropy-Controlled Policy Optimization

2026-07-18 · Guangran Cheng, Chengqi Lyu, Songyang Gao, Wenwei Zhang 외 hf

Entropy control has become an effective tool in reinforcement learning (RL) of large language models (LLMs), helping balance exploration-exploitation trade-off during alignment process. Such RL paradigm is often conducte…

Reinforcement LearningInstruction FollowingCode Generation

Heterogeneous Adaptive Policy Optimization: Tailoring Optimization to Every Token's Nature

2025-09-20 · Zheng Liu, Mengjie Liu, Siwei Wen, Mengzhang Cai 외 arxiv

Using entropy as a measure of heterogeneity to guide optimization has emerged as a crucial research direction in Reinforcement Learning for LLMs. However, existing methods typically treat it as a discrete filter or post-…

Reinforcement LearningMathematical Reasoning