arXiv subjects

Hao-Xiang Xu

Publications and source records attributed to Hao-Xiang Xu.

7 recordsLinked to original sources

arXiv2026

REINS: Refusal-Enhanced Inhibitory Steering with Sparse Autoencoder Features

Steering with Sparse Autoencoders (SAEs) offers a lightweight inference-time path for adapting the behavior of large language models without retraining. By exposing sparse and interpretable features, SAE steering provides a promising interface for safety control that guides harmful continuations toward refusal. However, we observe that complex wrappers can still undermine existing SAE steering methods on harmful prompts. To evaluate this failure mode systematically, we construct Generalized Undercover Instruction Safety Evaluation (GUISE), a dataset of harmful prompts with complex wrappers. Existing single direction SAE steering methods do not reliably produce refusals on harmful prompts, suggesting that refusal enhancement alone can be too weak when the harmful continuation path remains active. This motivates us to propose Refusal-Enhanced INhibitory Steering (REINS), which suppresses harmful continuation features and enhances safe refusal features in the same SAE feature space. Experiments on GUISE and other datasets show that prior methods either intervene too weakly or achieve only apparent safety through collapse, while REINS substantially reduces harmful responses, markedly improves safe refusals and largely preserves general capabilities.

cs.AI↗

arXiv2026

Evaluating the Reversal Curse in Model Editing

Large language models (LLMs) are prone to hallucinate unintended text due to false or outdated knowledge. Since retraining LLMs is resource intensive, there has been a growing interest in model editing. Despite the emergence of benchmarks and approaches, existing unidirectional editing and evaluation paradigms have failed to explore the reversal curse. In this paper, we study bidirectional language model editing, aiming to provide a rigorous evaluation to assess if edited LLMs can recall the editing knowledge bidirectionally. A metric of reverse generalization is introduced and a benchmark dubbed Bidirectional Assessment for Knowledge Editing (BAKE) is constructed to evaluate if post-edited models can recall the edited knowledge in the reverse direction of editing. We conduct extensive experiments using a variety of editing methods and LLMs. The results show that while most editing methods are able to accurately recall editing facts along the modification direction, they exhibit substantial systematic deficiencies when evaluating in the reverse direction. To further investigate the underlying causes of reversal curse and to explore potential strategies for mitigation, a detailed analysis is conducted from three perspectives. Our findings reveal that although In-Context Learning (ICL) can mitigate the reversal curse to a certain extent, it lacks continuity, is limited by the input length, and may introduce hallucinations. Therefore, combining the advantages of ICL and other editing methods is a promising direction for developing new editing paradigms.

cs.CL↗

arXiv2026

Constraining Sequential Model Editing with Editing Anchor Compression

Large language models (LLMs) struggle with hallucinations due to false or outdated knowledge. Given the high resource demands of retraining these models, there is an increasing focus on developing model editing. However, the general abilities of LLMs across downstream tasks are prone to significant degradation during sequential editing. This paper statistically observes that the parameter matrix after editing exhibits a significant deviation compared to its previous state as the number of edits increases. This serious deviation affects the original knowledge associations within LLMs and leads to the degradation of their general abilities. To this end, a framework termed Editing Anchor Compression (EAC) is proposed to constrain the deviation of the parameter matrix during sequential editing. It compresses the editing information by selecting editing anchors that are important in encoding new relations without deviating too much from the original matrix, thereby preserving the general abilities. Experiments of applying EAC to two popular editing methods on three LLMs across four tasks are conducted. Evaluation results show that EAC effectively minimizes unreasonable deviations caused by model editing, preserving over 70% of the general abilities while better retaining the editing knowledge compared to the original counterpart methods.

cs.CL↗

arXiv2026

GenesisFunc: Multi-Agent Data Generation for Accurate and Generalizable Function-Calling

Large Language Models (LLMs) extend their capabilities through function-calling (FC), which relies on training data with high quality, diversity, and broad coverage of scenario. However, obtaining and annotating real function-calling data is challenging, while synthetic data from existing pipelines often suffers from unreliable APIs, limited tool scalability, insufficient diversity, and weak quality control. To address these, we present GenesisFunc, an automated pipeline for generating FC training data. Starting from reliable tools in widely used public benchmarks, our GenesisFunc employs a multi-agent framework to support a dialogue generation system that produces conversations spanning diverse scenarios, while maintaining both diversity and quality throughout the process. The accuracy of the data is further reinforced through a multi-stage evaluation system. We fine-tune an 8B LLM on the synthetic dataset and show through extensive experiments that it outperforms similarly sized open-source models in in-domain FC performance and out-of-domain generalization, while reaching FC capabilities comparable to some of the latest API-based models. In addition, our method demonstrates strong potential to scale effectively across downstream tools, underscoring its real-world applicability.

cs.CL↗

arXiv2026

Multiplicative Orthogonal Sequential Editing for Language Models

Knowledge editing aims to efficiently modify the internal knowledge of large language models (LLMs) without compromising their other capabilities. The prevailing editing paradigm, which appends an update matrix to the original parameter matrix, has been shown by some studies to damage key numerical stability indicators (such as condition number and norm), thereby reducing editing performance and general abilities, especially in sequential editing scenario. Although subsequent methods have made some improvements, they remain within the additive framework and have not fundamentally addressed this limitation. To solve this problem, we analyze it from both statistical and mathematical perspectives and conclude that multiplying the original matrix by an orthogonal matrix does not change the numerical stability of the matrix. Inspired by this, different from the previous additive editing paradigm, a multiplicative editing paradigm termed Multiplicative Orthogonal Sequential Editing (MOSE) is proposed. Specifically, we first derive the matrix update in the multiplicative form, the new knowledge is then incorporated into an orthogonal matrix, which is multiplied by the original parameter matrix. In this way, the numerical stability of the edited matrix is unchanged, thereby maintaining editing performance and general abilities. We compared MOSE with several current knowledge editing methods, systematically evaluating their impact on both editing performance and the general abilities across three different LLMs. Experimental results show that MOSE effectively limits deviations in the edited parameter matrix and maintains its numerical stability. Compared to current methods, MOSE achieves a 12.08% improvement in sequential editing performance, while retaining 95.73% of general abilities across downstream tasks. The code is available at https://github.com/famoustourist/MOSE.

cs.LG↗

arXiv2025

Perturbation-Restrained Sequential Model Editing

Model editing is an emerging field that focuses on updating the knowledge embedded within large language models (LLMs) without extensive retraining. However, current model editing methods significantly compromise the general abilities of LLMs as the number of edits increases, and this trade-off poses a substantial challenge to the continual learning of LLMs. In this paper, we first theoretically analyze that the factor affecting the general abilities in sequential model editing lies in the condition number of the edited matrix. The condition number of a matrix represents its numerical sensitivity, and therefore can be used to indicate the extent to which the original knowledge associations stored in LLMs are perturbed after editing. Subsequently, statistical findings demonstrate that the value of this factor becomes larger as the number of edits increases, thereby exacerbating the deterioration of general abilities. To this end, a framework termed Perturbation Restraint on Upper bouNd for Editing (PRUNE) is proposed, which applies the condition number restraints in sequential editing. These restraints can lower the upper bound on perturbation to edited models, thus preserving the general abilities. Systematically, we conduct experiments employing three editing methods on three LLMs across four downstream tasks. The results show that PRUNE can preserve general abilities while maintaining the editing performance effectively in sequential model editing. The code are available at https://github.com/mjy1111/PRUNE.

cs.CL↗

arXiv2024

Model Editing Harms General Abilities of Large Language Models: Regularization to the Rescue

Model editing is a technique that edits the large language models (LLMs) with updated knowledge to alleviate hallucinations without resource-intensive retraining. While current model editing methods can effectively modify a model's behavior within a specific area of interest, they often overlook the potential unintended side effects on the general abilities of LLMs such as reasoning, natural language inference, and question answering. In this paper, we raise concerns that model editing's improvements on factuality may come at the cost of a significant degradation of the model's general abilities. We systematically analyze the side effects by evaluating four popular editing methods on three LLMs across eight representative tasks. Our extensive empirical experiments show that it is challenging for current editing methods to simultaneously improve factuality of LLMs and maintain their general abilities. Our analysis reveals that the side effects are caused by model editing altering the original model weights excessively, leading to overfitting to the edited facts. To mitigate this, a method named RECT is proposed to regularize the edit update weights by imposing constraints on their complexity based on the RElative Change in weighT. Evaluation results show that RECT can significantly mitigate the side effects of editing while still maintaining over 94% editing performance.

cs.CL↗