Model Editing

Follow this topic Ask about this topic

Momentum

8 papers in the last four weeks, against 1 the four weeks before. 0.1% of all new papers.

Jul 13Week of Sep 28

Latest papers 53

Oct 4, 2026cs.CL

FORGE: Verification-Gated Behavioral Repair for Generative Language Models

Generative large language models (LLMs) inherit undesirable behaviors from pre-training, including demographic bias and toxic generation, that often emerge only after deployment and affect a small subset of inputs. A repair should eliminate the identified defect, preserve the model's overall functionality and, ideally, provide correctness guarantees. Existing approaches address this only partially: gradient-based fine-tuning lacks per-instance guarantees and becomes unstable with few defect samples; model editing assumes explicit knowledge replacement rather than behavioral correction; and constraint-based repair is largely restricted to discriminative models with unique target outputs. We present FORGE, a framework for targeted behavioral repair of generative language models that separates defect localization, weight editing, and behavioral verification into independent stages. Its core is a repair abstraction that converts localized defective generation into explicit optimization objectives, enabling verification-oriented repair techniques to operate on autoregressive generation. FORGE is editing-mechanism agnostic: we instantiate it with (1) a constraint-based quadratic optimization method that provides per-sample repair certificates and (2) a null-space projection editor that minimizes interference with the original model distribution, both under the same localization and verification protocol. On five open-source LLMs, FORGE consistently achieves larger reductions in bias and toxicity than gradient-based fine-tuning with minor perplexity degradation. The two backends exhibit complementary performance across architectures, which a lightweight causal probe traces to where toxicity-related signals concentrate. FORGE also remains effective with only a handful of defective examples, where conventional fine-tuning often oscillates or fails to converge.

Hsin-Ling Hsu, Min-Yu Chen, Nai-Chia Chen +3

Department of Management Information Systems National Chengchi University Taipei, Taiwan

Social Bias in Language Models LLM Safety Model Editing

Oct 4, 2026cs.LG

Operational Abstractions of Neural Network Concepts via Topological Representations

Concept-based methods provide a semantic level for interpreting and manipulating learned representations, but existing editing approaches are typically specialized to particular interventions and do not provide a common and editable representation of concept organization. To achieve this, we introduce Topological Concept Representations (TCR), a post-hoc operational abstraction that jointly characterizes the concepts encoded in a learned representation and their relationships. TCR constructs an intermediate concept space from concept recoverability and interaction scores, and compactly encodes its organization through topology. Interventions are expressed through modifications of this abstraction that are propagated back to the underlying learned representation. This allows different concept-level operations to share the same optimization framework and separates the desired concept organization from the mechanism to achieve it. We establish stability and reparameterization-invariance properties of TCR and its connections to existing concept-editing formulations. We use TCR to disentangle concepts as a preprocessing step for existing erasure methods, improving worst-group accuracy by 21.89 on average at comparable concept leakage. We further use TCR to transfer concepts from teacher to student models, improving concept recoverability by up to 5.54 while also improving or maintaining competitive test top-1 accuracy.

Mathieu Pont, Christoph Garth

RPTU Kaiserslautern-Landau RPTU Kaiserslautern-Landau Kaiserslautern, Germany

Disentangled Representation Learning Model Editing Topological Representation Learning

Oct 1, 2026cs.LG

Repurposing Obsolete Representations for Post-Deployment Adaptation

Deep neural networks are increasingly deployed in long-lived systems, where task requirements may change after training. In such settings, part of the original output space may become obsolete: a class, prediction region, or learned behaviour may no longer be valid. Existing approaches either leave the obsolete behaviour intact or require fine-tuning, which can be expensive. We propose Deep Repurposing (DR), a post-hoc framework for adapting models under task obsolescence. DR estimates the latent geometry of obsolete and retained regions, removes obsolete-supporting components, and reallocates retained-compatible evidence through an analytic repair map without gradient updates. This yields repaired predictions and representations in which obsolete regions no longer act as valid outputs, while useful obsolete structure can support the retained task. Across multiple task settings, DR removes obsolete behaviour while preserving retained utility. More importantly, across classification benchmarks, DR matches or exceeds competing unlearning and editing baselines in retained accuracy, eliminates obsolete predictions, and adapts up to

60\times

faster than competing unlearning methods.

Daniel Bethell, Charmaine Barker, Simos Gerasimou

Department of Computer Science, University of York, UK · Department of Elect. Eng., and Computer Science and Eng., Cyprus University of Technology, Cyprus

Model Editing Machine Unlearning

Oct 1, 2026cs.CV

STAGE: Subspace-Targeted Affine Generative Erasure for Text-to-3D Models

Concept erasure suppresses a target concept while preserving behavior on unrelated inputs. Existing closed-form methods were designed for 2D image diffusion and assume a single generative pathway, so one edit must cover geometry and texture at once. Native 3D generators, which synthesize structured 3D representations directly rather than by lifting 2D samples, violate this assumption. We show that shape and object concepts must be erased in the structural stage of the pipeline and material concepts in the appearance stage. We therefore formulate erasure in native text-to-3D as a stage-aware editing problem and introduce STAGE, a training-free, closed-form framework. STAGE confines each edit to the low-dimensional subspace spanned by the differences between erase and anchor embeddings, and relaxes the norm-preserving (orthogonal) constraint of prior editors into a least-squares affine correction that maps target activations onto safe anchors subject to a penalty on the displacement of retained prompts. The correction applies to the structural stage, the appearance stage, or both. We find that the stage an edit must reach is determined by concept type. On TRELLIS, the standard open native 3D generator, across 15 shape, material, and object concepts, STAGE reaches 66.7 on a composite score that balances forgetting the target concept against preserving everything else, aggregating CLIP-based semantic and physical metrics, versus 53.2 for the strongest adapted baseline. Code: https://github.com/gmum/STAGE/ Project Page https://gmum.github.io/STAGE/

Karol Dziekan, Przemysław Spurek, Dawid Malarz

Jagiellonian University · IDEAS Research Institute

3D Asset Generation Text-to-3D Generation Model Editing

Oct 1, 2026cs.LG

Platonic Task Arithmetic

Models specialized for the same task converge to similar behavior, yet the parameter updates that produce it share no common coordinate system, so weight-space task arithmetic stays confined to a single model and cannot cross architectures without a structural correspondence. Drawing on Plato's allegory of the cave, we hypothesize that these model-specific updates are shadows of one shared, model-agnostic object, which we call the platonic task vector. To make it operational for models that pair an image or audio encoder with a text encoder, we introduce Universal Task Descriptors: matrices whose shape is independent of architecture and embedding dimension, which record a task's functional effect and support addition and negation as matrix operations. Transferring a descriptor into a target means editing the target until it reproduces the descriptor on the task's unlabeled probe images and class-name prompts, requiring no per-image labels. We realize this edit in two ways. First, the descriptor factorizes into a shift field on image embeddings, so a single least-squares solve yields a linear operator that folds into the target's last layer as a weight edit; by linearity, a bank of such operators admits any composition at any strength as a signed sum. Second, a low-rank adapter trained on the same objective reaches every layer and fits compositions jointly, at the cost of one optimization per edit. Heterogeneous models share this object only partially, with a model-specific residual comparable in norm to the shared component, yet cross-model transfer still retains 74-80 percent of the gain of the target's own descriptors. Experiments across six model families, eight classification tasks, and an audio-text setting show that task knowledge transfers and composes across heterogeneous models under both realizations.

Junghwan Park, Woojin Cho

TelePIX

VLM Adaptation Task Arithmetic Transfer Learning Model Editing

Sep 30, 2026cs.CL

LLM Persona Unlearning

Pre-training equips large language models (LLMs) with a broad repertoire of behavioral patterns associated with roles, styles, values, and goals. Post-training teaches conditional enactment and makes a helpful Assistant the default, but it does not erase alternative modes from the weights; explicit prompts can therefore elicit personas that repeatedly shape judgment, language, and action. In open-weight settings, runtime controls can be removed, motivating persona unlearning: a weight-level edit that makes a designated persona difficult to elicit and enact on unseen contexts. We introduce PersonaUnlearnBench, a model-specific paired benchmark spanning six LLMs from three families and five personas, with aligned forget/retain sets, held-out instruction paraphrases, and four-axis evaluation. The benchmark shows that standard unlearning methods cannot reliably erase the target persona without sacrificing meaningful generation or general utility. We therefore propose PaCE, which compares target and desirable responses to the same questions to locate an internal behavior direction, then trains target-prompt states away from the target mode and toward the matched desirable response. Experiments show that PaCE consistently suppresses target personas with high response quality and useful counterpart behavior, at moderate utility cost. These results establish persona unlearning as a distinct behavior-level editing problem and a practical route toward persistent control of latent LLM response policies.

Kemou Li, Zhuan Shi, Qizhou Wang +4

State Key Laboratory of Internet of Things for Smart City, University of Macau · Mila – Québec AI Institute · McGill University +3

Model Editing Persona-Conditioned Generation Machine Unlearning LLM Unlearning

Sep 28, 2026cs.AI

Imprint Reader: From Weight-Update Readout to Behavioral Intervention

As language models take a growing role in AI development, a natural aspiration is for them to reflect on their own learning process, as humans do, and use that reflection to improve themselves. At the same time, these models have an advantage that human learners lack, since training leaves parameter-level traces that can, in principle, be inspected directly. However, current models cannot decode these traces into an explicit account of what they have learned. To this end, we introduce the \textit{Imprint Reader}, a model trained with \textit{Semantic Mount-and-Read Tuning} (SaRT) to describe frozen weight updates. SMaRT mounts each update onto the Reader and uses an anchor-free meta-query to elicit a natural-language description, while no-change and random-perturbation controls discourage unsupported claims. On held-out updates, the joint Reader reaches judge-based Pass@100 of

2\%

for knowledge and

16\%

for behavior. These results demonstrate the feasibility of natural-language readout while pointing to reliability across updates as the next step. Beyond free-form generation, the Reader provides a differentiable proxy for the gap between a specified target behavior and a candidate weight update. Its coordinate-aligned gradients support intervention through MetaEdit. At a

0.5\%

pruning rate, Reader-guided selection raises measured harmful-prompt refusal from

57.9\%

to

64.1\%

under a safety-maintenance target. Using behavior descriptions without target-task training data, MetaEdit increases the frequency of backtracking and sub-goal expressions in mathematical reasoning traces and raises BFCL Overall from

41.69\%

to

44.60\%

.

Guanxu Chen, Qihao Lin, Jing Shao

Shanghai Artificial Intelligence Laboratory · Shanghai Jiao Tong University

LLM Interpretability Neural Network Interpretability Model Editing

Sep 14, 2026cs.LG

Decoy Direction Optimization: A Post-Hoc Defense Against LLM Abliteration

Safety guardrails in open-weight language models can be readily bypassed using Refusal Feature Ablation (RFA), a technique that identifies and projects out a linear refusal direction from the residual stream, often achieving a high attack success rate (ASR) while preserving model capability. Defending against these attacks typically requires computationally expensive safety finetuning for every new checkpoint. We introduce Decoy Direction Optimization (DDO), a fast, post-hoc weight-editing defense that requires no base-model finetuning. Our approach is based on a simple mechanistic insight: ablation attacks rely on contrastive estimators to find the refusal direction. Rather than trying to hide the true refusal circuitry, DDO actively injects a high-magnitude, nonlinear decoy signal into the network's MLP neurons. When an attacker attempts to locate the refusal direction, the decoy corrupts their estimator, tricking them into ablating a harmless orthogonal feature while the actual safety mechanism remains intact. We prove a spectral bound formalizing this effect and evaluate DDO across six model families, achieving <10% ASR under standard RFA. On Llama-3-8B-Instruct, DDO remains comparable to trained defenses under adaptive multi-phase attacks (65% vs. 58% worst-case ASR) and reduces Heretic weight-level attack ASR from 88.7% to 18%, all at 30 to 450 times lower optimization cost per configuration than the trained baselines.

Aashiq Muhamed, Mona T. Diab, Virginia Smith

Carnegie Mellon University

Adversarial Attacks on LLMs LLM Safety Model Editing

Aug 10, 2026cs.CV

Beyond Global Editing: Per-Instance Disentangled Subspaces for Training-Free Hallucination Mitigation in LVLMs

Recent advances in large vision-language models (LVLMs) have enabled powerful multimodal reasoning by integrating visual encoders with large language models (LLMs). However, their reliability is frequently undermined by hallucinations, where generated text inaccurately describes the visual input. Although fine-tuning can mitigate this problem, it is computationally expensive and requires large, curated datasets, making training-free alternatives attractive. Among these, model editing is more promising than decoding-based approaches: decoding methods adapt outputs per input but introduce computational overhead and instability, whereas model editing modifies internal representations offline, providing a more efficient and stable solution. However, existing model-editing techniques typically rely on a single global subspace to correct hallucinations, treating all test samples identically and failing to capture diverse hallucination modes across inputs. To address this limitation, we propose a training-free hallucination mitigation framework for dynamic, per-instance suppression at test time. Our method first constructs a set of Disentangled Hallucination Subspaces, each isolating a distinct hallucination mode. During inference, the model adaptively calculates weights reflecting each input's relationship to these subspaces, guiding a dynamically combined projection that selectively suppresses the most probable hallucination directions while preserving image-grounded semantics. Extensive experiments across multiple vision-language benchmarks and LVLM families demonstrate consistent improvements, highlighting the robustness, generalizability, and efficiency of our approach.

Ali Cheraghian, Hamidreza Dastmalchi, Hamed Barzamini +4

Macquarie University, Australia · Australian National University, Australia · York University, Canada +4

Large Vision-Language Models Model Editing VLM Hallucination Mitigation

Jul 13, 2026cs.LG

PRISM Edit: One Vector for All Temporal Answers

Model editing keeps large language models (LLMs) up to date without retraining, but temporal facts expose a limitation of the prevailing locate-and-edit paradigm: an update is not always a replacement. When a fact changes, the new answer should become current while the old answer may remain correct in historical time contexts. Building on this insight, we use causal tracing to show that LLMs already support this distinction via a two-stage internal computation: early MLP layers retrieve a time-agnostic subject representation, and later layers modulate it with temporal context to yield the time-correct answer. Motivated by this finding, we introduce PRISM Edit, which optimizes a single polysemous representation across temporal contexts and leverages the model's inherent modulation pathway to route it to temporally correct predictions without requiring any architectural modification. We evaluate on TimeConflict, a newly introduced temporal editing benchmark, and on temporally augmented CounterFact. PRISM Edit improves multiple core metrics over the best baseline, most notably +23.3 Temporal Consistency (TC) and +33.7 Current Relative-time Score (CRS) on LLaMA-3, while being more than 2x faster. Code and data are publicly available at https://github.com/CheerCHuang/PRISM-Edit.

Chen Huang, Qi Zheng, Ruiqin Zheng +2

Tsinghua University · ByteDance

Temporal Reasoning in Language Models Model Editing Temporal Consistency Language Model Editing Temporal QA

Jul 6, 2026cs.AI

Evaluating and Understanding Model Editing for Medical Vision Language Models

Model editing promises a fast, targeted way to correct post-deployment mistakes in medical vision-language models (VLMs) without costly retraining. However, existing multimodal model editing benchmarks focus on general-purpose tasks and do not reflect realistic clinical domain requirements and variability. To address this, we introduce M3Bench, a clinically grounded benchmark for multimodal model editing that evaluates whether an edit remains reliable, precise, and generalizable under the challenges of image and text variation, modality and protocol shifts, clinical knowledge composition, and temporal progression. M3Bench contains 16,276 questions spanning diverse anatomy, modalities, and specialties, and supports both single and sequential edits. By evaluating 4 representative editors across 6 medical and general VLMs, we find that no method excels across all criteria. Gradient-based editors achieve strong transfer but suffer from catastrophic locality violations, whereas memory-based methods preserve locality but lack compositional generality and exhibit high backbone-dependent hyperparameter sensitivity. We further attribute these failures to the latent space geometry of VLMs and how different editing methods shift its landscape. Overall, M3Bench establishes a rigorous clinical stress test for multimodal model editing and offers actionable guidance for safer post-deployment adaptation. The benchmark is publicly available at https://github.com/BioMed-AI-Lab-U-Michgan/M3Bench .

Guli Zhu, Chenwei Wu, Liyue Shen

EECS, University of Michigan, Ann Arbor, MI 48105, USA

Knowledge Editing Medical VLMs Model Editing Multimodal Model Evaluation Medical Image Benchmarks

Jul 2, 2026cs.AI

Multimodal Knowledge Edit-Scoped Generalization for Online Recursive MLLM Editing

Online multimodal knowledge editing requires injecting a continual stream of visual-textual corrections into multimodal large language models (MLLMs) with bounded overhead and minimal disruption to unrelated behaviors. Existing editors mainly emphasize edit reliability and long-horizon stability, but rarely control the semantic boundary of each edit. Our pilot analyses of post-edit behaviors and internal neuronal activities reveal a scope gap behind reliable edits: instance-level success neither guarantees transfer to valid cross-modal variants nor prevents leakage to unrelated inputs, while edit-related cross-modal responses concentrate in deeper semantic layers. Therefore, we formulate Edit-Scoped Generalization, reframing online MLLM editing from merely correcting an instance to controlling the propagation boundary of each edit. To this end, we propose ScopeEdit, a scope-aware online editor that decomposes each update into a modality-local absorption branch and an evidence-gated shared generalization branch. The local branch supports stable edit absorption, whereas the shared branch enables cross-modal propagation only when visual and textual evidence are sufficiently aligned. Both branches perform scope-separated write geometries in orthogonal low-rank spaces and maintain branch-wise preconditioners via Sherman--Morrison recursions, yielding constant per-edit overhead. Extensive experiments across diverse benchmarks, long-horizon edit streams, MLLM backbones, real-world VLKEB scenarios, and complex vision-language architectures show that ScopeEdit consistently improves the trade-off between in-scope cross-modal transfer and out-of-scope locality, while preserving edit reliability, stability and online efficiency. Our code is available at https://github.com/lab-klc/ScopeEdit.

Siyuan Li, Youyuan Zhang, Ruitong Liu +2

Harbin Institute of Technology, Shenzhen, China · Peng Cheng Laboratory · Peking University +1

Knowledge Editing Multimodal Large Language Models Model Editing

Jul 1, 2026cs.CL

KnowledgeDebugger -- an Exploration Tool for Knowledge Localization and Editing in Transformers

Recent research has increasingly focused on understanding how Transformers store and process knowledge, as well as how this knowledge can be edited. Research work in this area is often conducted in two phases: first, phenomena are explored on individual samples. Then, when results appear promising, more statistically robust experiments follow. To support the first phase, we propose KnowledgeDebugger, a GUI-based exploration tool for knowledge localization and editing in Transformers. Our tool - inspired by LM-Debugger - offers no-code access to the methods in EasyEdit, a widely used library of state-of-the-art Knowledge Editing approaches. We demonstrate the tool's effectiveness through case studies of recent findings in this field.

Eric Benz, Lennart Stöpler, Nikolai Bolik +1

Heidelberg University

Knowledge Editing Transformer Interpretability Model Editing

Jun 29, 2026cs.LG

Revocable Learned State via Process Sidecars

Language models are often adapted in stages: a public skill phase, a private memory phase, and a later safety phase that learns to refuse outputs tied to the remembered entities. Revoking the memory after the safety phase is not the same problem as subtracting the memory update: the later safety optimizer has transported the memory direction. We introduce process sidecars, a two-coefficient edit family

\hatθ(λ,γ)=θ_{\mathrm{AMS}}-λΔ_{\mathrm{M}}-γ\hat{R}_{\mathrm{S}\leftarrow\mathrm{M}}

, with

\hat{R}_{\mathrm{S}\leftarrow\mathrm{M}}=\hat{J}_{\mathrm{S},\varepsilon}(Δ_{\mathrm{M}})-Δ_{\mathrm{M}}

, where

\hat{J}_{\mathrm{S},\varepsilon}

is a centered secant through the realized future AdamW safety-training process. The implementation uses

\varepsilon=1

at the natural memory-edit scale; it reuses

θ_{\mathrm{AMS}}

as the positive endpoint and computes one additional safety trace at

θ_{\mathrm{A}}-Δ_{\mathrm{M}}

. We prove two things. First, the exact sidecar, using the true transported direction

R_{\mathrm{S}\leftarrow\mathrm{M}}

rather than the secant estimate, at

(λ,γ)=(1,1)

recovers the counterfactual safety-only oracle

θ_{\mathrm{AS}}

up to second order; the proof treats AdamW as an augmented-state map over parameters, first moments, and second moments. Second, this process information is necessary: whenever future safety training bends the memory direction, every scalar task-arithmetic edit leaves first-order counterfactual error, while the process-sidecar edit is second-order accurate. Across three models, the validation-selected 2D edit improves held-out refusal closure over naive task arithmetic in all trials, and over the

γ=λ

process-JVP subfamily, the diagonal slice of the cached 2D grid, in all paired trials.

John Sweeney

Sideplane AI

LLM Safety Alignment Task Arithmetic Model Editing LLM Unlearning

Jun 25, 2026cs.LG

Reproducibility Study of "AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models"

Fang et al. (2025) introduced a null-space constrained projection, named AlphaEdit, for locate-then-edit knowledge editing methods, theoretically guaranteeing that edits do not disrupt previously preserved knowledge, and reports substantial gains over existing editing methods on LLaMA3, GPT2-XL, and GPT-J. In this work, we present a reproducibility study of AlphaEdit, reproducing its reported results under the original experimental setup and extending the evaluation along three axes: new model architectures, additional downstream benchmarks, and substantially longer sequential editing horizons. We successfully reproduce AlphaEdit's reported metrics across the original models, though we identify a discrepancy in the reported fluency and consistency metric. Extending AlphaEdit to newer model families, we find that its advantage does not generalize uniformly, which we trace to architectural assumptions in the locate-then-edit paradigm that are violated by these newer models. We further stress-test AlphaEdit's central sequential-editing claim by extending the number of edits well beyond those evaluated in the original paper, and find that performance, which is stable at the originally reported scale, degrades as edits reach a much higher count, indicating that the null-space projection's protection against catastrophic forgetting is bounded rather than unconditional. Finally, we extend evaluation of edited models on three extra benchmarks, namely, BoolQ, HellaSwag, and XSTest, and we find that large-scale sequential editing degrades both general downstream task competence and safety-relevant refusal behavior. Our results confirm that AlphaEdit performs as reported within its original scope, while showing that its core theoretical guarantees are sensitive to model architecture and editing scale in ways that have practical implications for its deployment.

Ananth K Suresh, Arya Hariharan

Independent

Knowledge Editing Model Editing Language Model Editing Catastrophic Forgetting

Jun 21, 2026cs.CL

Orthogonal Representation Editing: Decoupling Semantic Entanglement in Batch Knowledge Editing of LLMs

Knowledge editing aims to efficiently update factual information in Large Language Models (LLMs) without full retraining. However, existing methods still suffer from performance degradation in batch knowledge editing. We identify that semantic representation entanglement, such as overlapping concepts and shared syntactic patterns, accumulates interference in the representation space and reduces editing precision. To bridge this gap, in this paper, we propose Orthogonal Representation Editing (ORE), which performs edits in the hidden representation space of LLMs by constructing a general semantic subspace and enforcing orthogonal constraints on edit vectors, effectively decoupling semantic entanglement. Furthermore, we introduce a gated non-linear representation head to enable adaptive learning of editing locations and precise control over knowledge injection. Extensive experiments show that ORE outperforms existing methods and achieves superior performance in cross-lingual knowledge editing scenarios. We release our code at https://github.com/YVVH/ORE.

Wenhao Yu, Zhicong Lu, Bo Lv +4

School of Computer Science and Technology, Tianjin University · Kexin Technology · University of Chinese Academy of Sciences +2

Knowledge Editing Model Editing Language Model Editing Representation Geometry in Language Models

Jun 18, 2026cs.AI

BIM-Edit: Benchmarking Large Language Models for IFC-Based Building Information Modeling

Large language models (LLMs) are increasingly applied to computer-aided design (CAD) to generate design artifacts from textual instructions. In engineering practice, this requires more than creating new geometry, models must also understand existing scenes, edit them correctly, and preserve semantics and relations. However, many CAD benchmarks focus on creating new models rather than editing existing ones, and mostly evaluate geometric correctness. We introduce BIM-Edit, a benchmark for evaluating LLMs on natural-language editing of Building Information Models (BIM) represented in the Industry Foundation Classes (IFC) format. BIM provides a challenging testbed because building models encode geometry together with semantic and relational structure. BIM-Edit contains 324 editing tasks spanning 11 realistic building models and 36 synthetic scenes. Tasks are expressed using three instruction categories - direct, spatial, and topological - covering both explicit and scene-grounded edits. We evaluate outputs along three dimensions: geometric accuracy, semantic validity, and topological consistency. Across evaluated LLMs, the best-performing model achieves only 49.5% average score across the three metrics, and no model fully solves more than 3.4% of tasks. These results demonstrate a substantial gap between current LLM capabilities and the requirements of structured engineering design workflows.

Bharathi Kannan Nithyanantham, Clemens Kujat, Tobias Sesterhenn +5

University of Rostock · Clausthal University of Technology

LLM Evaluation Building Information Modeling Text-to-CAD Generation Model Editing

Jun 18, 2026cs.LG

LOKI: Memory-Free Null-Space Constrained Lifelong Knowledge Editing

Lifelong knowledge editing aims to efficiently and sequentially update language models over time, as new knowledge becomes available or when the model makes mistakes, while preserving acceptable performance on past knowledge. One unresolved challenge is that existing methods modify a fixed set of layers for all new knowledge samples, reducing flexibility and increasing catastrophic forgetting. Another is requiring access to previous knowledge and extensive pre-processing to obtain data statistics. To address these challenges, we introduce LOKI, a novel approach that uses dynamic layer selection based on the Hilbert-Schmidt Independence Criterion and projects gradient updates onto the null-space of the model weights, bypassing the requirement for previous knowledge access. We show that LOKI achieves superior performance to existing approaches across a wide variety of experiments, achieving up to a 14% improvement in average accuracy.

Masih Eskandar, Miquel Sirera Perelló, Stratis Ioannidis +1

Department of Electrical and Computer Engineering Northeastern University Boston, MA, 02115

Knowledge Editing Continual Learning for LLMs Model Editing Language Model Editing Catastrophic Forgetting

Jun 17, 2026cs.AI

Repeated Shared Access Enables Grokking, but Edit Propagation Depends on an Addressable Memory

We study factual edit propagation in a controlled synthetic knowledge-graph QA setting using a 2x2 grid that crosses loop recurrence with shared-memory access: a dense transformer (Dense), a looped transformer (Loop), a dense backbone with shared memory (Dense+Mem), and a looped backbone with shared memory (loop-memory coupling, LMC). The two factors dissociate. For learning, both routes to repeated shared access -- looped recomputation and repeated memory rereading -- cross the out-of-distribution (OOD) grokking barrier that Dense fails, so repeated shared access is the behavioral regularity, not a specific architecture. For editing, the substrates split along a different axis: applying a single localized factual edit (conditioned on direct success) and measuring 2-hop propagation on a shared pre-edit-correct set, the edit propagates strongly in both memory-bearing cells (LMC 0.78-0.92, Dense+Mem 0.71-0.96) and only weakly in the memory-free ones (Loop 0.04-0.30, Dense 0.00-0.03). The split is along the memory axis, not the loop axis: every memory-bearing seed exceeds every memory-free seed, with no detectable difference between the two memory cells. Crucially Dense+Mem has no recurrence, so the propagating ingredient is an addressable site that an edit can write to and later computation rereads, not loop recomputation; Loop is at best a partial intermediate. The affordance survives coarsening the store (N=128 to N=13): propagation attenuates but the memory/no-memory split persists, so fine granularity buys precision rather than the affordance itself. These results dissociate learning competence from editing affordance -- repeated shared access suffices to grok, but edit propagation depends on whether the substrate exposes an addressable memory that the forward computation can write to and later reread, an affordance that loop recurrence provides only partially.

Yanan Niu

Kuaishou Technology

Memory-Augmented Neural Networks Knowledge Editing Recurrent Transformers Model Editing Grokking

Jun 17, 2026cs.AI

User as Engram: Internalizing Per-User Memory as Local Parametric Edits

Personal memory in a language model is two problems: content and reasoning skill. The brain keeps the two apart (a sparse, local engram in the hippocampus for each episode, a slow neocortex for the shared skills that interpret it), so a new fact need not overwrite everything else. Most personalization today keeps a user's facts outside the weights, in a natural-language memory file or a retrieval index. When facts are written into the model instead, the standard recipe is the per-user LoRA adapter, which does the opposite of the brain, folding content and skill into one global weight delta. Writing a user's facts as a LoRA contaminates text unrelated to them; writing the same facts as local Engram rows leaves it mathematically untouched, resulting in a roughly 33,000x smaller memory footprint. We therefore propose User as Engram: store a user's content as surgical edits to the hash-keyed memory table of an Engram model, and carry the reasoning skill in one shared adapter. This layered design matches per-user LoRA's direct recall while delivering 5.6x higher indirect-reasoning accuracy on average, and never makes a single user worse at reasoning than the untouched base. The edit is a glass box: writing a fact switches on its lookup at exactly the trigger, adds the value the answer needs, leaves every other position unchanged to the last bit, and fails if written into the wrong layer. Because different users' facts land in disjoint hash slots, their edits compose: many users live in one shared table at once, stacking additively and losslessly, where a per-user LoRA, a single global weight delta, admits only one. Upon retrieval, a per-user Engram table does not grow with the population the retriever must search, so past ~100 facts it overtakes a retrieval pipeline on a 2.5x larger model.

Bojie Li

Pine AI

Knowledge Editing LLM Personalization Memory-Augmented Language Models Persistent Memory for Language Models LLM Memory Personalized Language Models+2

Jun 12, 2026cs.LG

When to Write and When to Suppress: Route-Specialized Dual Adapters for Memory-Assisted Knowledge Editing

Knowledge editing systems must update selected facts while preserving nearby but irrelevant behavior. This paper studies this problem in a memory-assisted setting where an edit memory is retrieved at inference time and a parameter-efficient adapter corrects the model's object preference. We argue that the central design question is not only how to write an edit, but also when to suppress it. We introduce RRDA, a route-specialized dual-adapter editor. A relevance router first decides whether a prompt should receive an edit memory. Routed prompts use an edit adapter trained to prefer the new object over the original object; unrouted non-direct prompts use a separate locality adapter trained to preserve or restore the original-object preference. We evaluate RRDA on three 1,000-case protocols, CounterFact, ZsRE, and MQuAKE-CF, under the same memory protocol and two 7B/8B base models. On Llama-3.1-8B-Instruct, RRDA obtains the best overall probability-preference accuracy on all three benchmarks: 0.8180 on CounterFact, 0.8946 on ZsRE, and 0.9922 on MQuAKE-CF. The same trend holds on Qwen3-8B. Router ablations show that the relevant memory boundary differs across datasets: a lexical neural router is safest on CounterFact, while BGE embedding routing is better on ZsRE and MQuAKE-CF. Memory, component, and module ablations show that explicit memory supplies the largest edit gain, while route-specialized adapters improve the final reliability-locality balance rather than simply increasing LoRA capacity.

Baijia Zhang, Yining Huang

Xinjiang University · Meta Emergence Laboratory

Knowledge Editing Factual Knowledge in Language Models LLM Routing Model Editing

Jun 11, 2026cs.LG

Reinforcement Learning for Neural Model Editing

Editing pretrained neural networks requires specialized algorithms tailored to specific objectives. Designing such algorithms is often time-consuming and demands significant effort. We present an exploratory framework that formulates neural model editing as a reinforcement learning problem, where agents modify models using reward feedback. We introduce two environments: MaskWorld, where agents scale weights multiplicatively, and ShiftWorld, where agents apply additive weight updates. The reward function combines a utility-preservation objective with a task-specific editing objective, enabling agents to learn targeted modifications while maintaining overall model performance. We evaluate the framework on bias mitigation in text classification and machine unlearning in image classification, both of which traditionally rely on specialized algorithms. Our results show that the learned policies reduce forget set accuracy to nearly 0% while preserving over 90% retain set accuracy on the unlearning task. In the bias mitigation setting, the learned policies improve bias-related performance by more than 5% while maintaining general classification utility. Our findings show that neural model editing can be cast as a reinforcement learning problem, allowing editing policies to be learned from reward feedback rather than manually engineered for each task.

Shaivi Malik

Reinforcement Learning Model Editing Machine Unlearning

Jun 11, 2026cs.LG

TimeROME-DLM: Temporal Causal Tracing and Low-Rank Inference-Time Knowledge Editing for Masked Diffusion Language Models

Masked diffusion language models (MDLMs) such as LLaDA now rival autoregressive (AR) LLMs, but every existing knowledge-editing and unlearning method (ROME, MEMIT, etc.) targets AR transformers and either makes assumptions that fail under iterative denoising, or requires gradient updates whose backward-pass activations cost tens of GB of extra VRAM and which collapse MDLMs at standard learning rates. We introduce TimeROME-DLM, the first training-free, gradient-free, inference-time knowledge-editing framework for MDLMs. It couples two components: a Temporal Indirect Effect (TIE) causal-tracing protocol that identifies, for each fact, the coordinate whose intervention most strongly drives the object prediction at later denoising steps; and a closed-form, low-rank residual edit memory that aggregates subject keys and target deltas across all forget facts and applies a single ridge-regularised update at that coordinate at every diffusion forward, with sparsification to limit utility spillover. Backbone weights stay frozen; only three hyperparameters (alpha, lambda, q) are tuned on a small validation split. On TOFU forget01 with TOFU-finetuned LLaDA-8B-Base, TimeROME-DLM cuts forget-set log-probability by roughly 83 nats. The same configuration transfers to LLaDA-8B-Instruct, Dream-7B, MMaDA-8B, DiffuLLaMA-7B, and LLaDA-MoE-1.4B. It keeps retain-set log-probability nearly flat (within ~1 nat at the utility-safe operating point) across 50 sequentially inserted facts, delivers a four- to fourteen-fold wall-clock speedup with zero additional VRAM over the strongest converged training-time baseline, and scales sub-linearly to 400 facts. TimeROME-DLM closes the locate-then-edit gap between AR LLMs and MDLMs at a fraction of the computational cost.

Zhengtao Yao, Liuyang Song, Hongbo Zhang +4

University of Southern California, USA · Peking University, China · Stevens Institute of Technology, USA +1

Knowledge Editing Masked Diffusion Models Model Editing Diffusion Language Models Language Model Editing Causal Interventions in Language Models+1

Jun 10, 2026stat.ML

LoMC: Localized Multidirectional Correction for Refusal Suppression in Routed Foundation Models

We study controlled post-training refusal suppression in routed MoE and hybrid-MoE foundation models, aiming to increase non-refusal target-response behavior while preserving general capability under a compact intervention footprint. Existing broad direction-based edits can perturb general-purpose computation, whereas support-only expert edits often lack sufficient capacity to correct heterogeneous refusal representations. To address this limitation, we introduce Localized Multidirectional Correction (LoMC), a support-gated intervention framework that follows a support-then-correction execution order: it first identifies a compact edit support, then aggregates prototype correction directions into layer-wise correction directions, and finally applies rank-one layer-wise correction only within the selected support. By using the edit support as a structural gating constraint, LoMC increases correction capacity without expanding the intervention scope. Experiments on text-only and multimodal safety benchmarks across four routed backbones show that LoMC substantially improves non-refusal target-response behavior while maintaining general capability under a compact intervention footprint.

Yan Hong, Kedong Xiu, Wei Li +6

Ant Group · Zhejiang University · Shanghai Jiao Tong University

Mixture-of-Experts Language Models LLM Refusal Behavior LLM Safety Model Editing Refusal Behavior in Language Models

Jun 9, 2026cs.LG

Can Editing 1 Neuron Fix Repetition Loops in LLMs?

Yes. Can it cure doom loops? Probably not. The Gemma 4 instruction-tuned models share a reproducible failure: on long factual enumeration prompts, such as listing every episode of a TV series, the 88 IAU constellations, or the 151 original Pokemon, they collapse into repetition, either a tight verbatim loop or a list whose entries decay onto a single answer. These loops occur at rates as high as 95% and survive prompt rewording, inference-engine changes, and most sampling adjustments. In this paper we explore whether this behavior is localized enough to remove by weight edits. To localize the cause, we use per-layer ablation and per-neuron attribution, then confirm the strongest candidates with full-generation sweeps. The loops trace to a small set of MLP neurons (or, in the 26B-A4B Mixture-of-Experts model, a few routed experts) which we suppress with static weight edits. These "surgeries" can be as small as a single sign-inverted neuron (in the E2B model). The size of the effective edits grows with model scale, but in all cases, the loop patterns can be addressed at normal generation budgets while preserving general-purpose benchmark scores. However, the edits do not solve everything: we also study longer thinking budgets, where the two larger models most visibly enter doom looping, i.e. a non-convergent regime in which the model self-corrects in circles over a fact it cannot recall, exhausting the budget without committing to a final answer. We show this residual failure is reduced but not eliminated by the same edits, and argue it is fundamentally a knowledge-precision problem rather than a removable circuit; weight surgery can delete a loop, but it cannot supply a missing fact. Our results are both a feasibility demonstration, that is, evidence that a concrete generation pathology can be localized to a few parameters and edited out, and a delineation of where that approach stops.

Aristotelis Lazaridis, Aman Sharma, Dylan Bates +3

Edgerunner AI

LLM Self-Correction Mechanistic Interpretability Model Editing Language Model Editing Causal Interventions in Language Models Data Repetition in Language Model Training

Jun 8, 2026cs.CV

CRANE: Knowledge Editing for Reasoning MLLMs

The emergence of reasoning multimodal large language models (MLLMs), which generate explicit chain-of-thought (CoT) reasoning before producing answers, has introduced a new challenge for knowledge editing: methods that appear successful under traditional metrics (teacher-forcing accuracy up to 100%) can fail severely when the model's reasoning process is examined (Grounded Success as low as 0%). We identify three failure modes: (1) Structural Collapse, where weight-modifying methods destroy the CoT format; (2) Cognitive Dissonance, where the model's reasoning chain actively rejects the injected edit fact based on visual evidence; and (3) Shallow Internalization, where methods succeed on exact queries but fail on rephrase or multi-hop variants. On reasoning MLLMs, these modes interact: methods that generalize (FT, LoRA) trigger format collapse, while methods without deep modification cannot generalize. To expose these failures, we propose a CoT-aware evaluation protocol and construct ReasonEdit-Bench, with conflict stratification, multi-level probes, and multi-hop portability tests. We propose CRANE, a retrieval-augmented framework that requires no per-edit parameter modification. CRANE combines a modality-aware dual-library retrieval system with a two-phase training strategy: Supervised Fine-Tuning (SFT) for structural initialization, followed by GRPO with a Cognitive Routing Reward that trains the model to arbitrate between visual priors and injected edit facts. On ReasonEdit-Bench, CRANE achieves 96.9% Grounded Success on conflict scenarios and 96.9% intermediate entity usage in multi-hop chains, with 97.6% text-locality and 68.1% image-locality Edit Independence. On the out-of-distribution MMEVOKE benchmark, CRANE reaches 87.0% under gold retrieval.

Han Huang, Hao Wang, Mengqi Zhang +3

University of Chinese Academy of Sciences, Beijing 101408, China, and also with the New Laboratory of Pattern Recognition (NLPR), CASIA, Beijing 100045, China · Harbin Institute of Technology, Weihai 264209, China · School of Computer Science and Technology, Shandong University, Qingdao 266237, China

Knowledge Editing Knowledge Conflicts in Language Models Multimodal Large Language Models Multimodal CoT Reasoning Model Editing

Jun 2, 2026cs.CL

Knowledge Editing for Masked Diffusion Language Models

Knowledge editing aims to update or correct factual knowledge in a language model. A widely used approach, locate-then-edit, first localizes a fact within the model and then edits the weights there. To date, such methods have been developed exclusively for autoregressive models (ARMs). Whether they work for masked diffusion models (MDMs), which model text bidirectionally and generate by iterative denoising rather than next-token prediction, remains an open question. We address it by transferring locate-then-edit to MDMs and comparing multiple MDMs with their matched ARMs. Our central finding has two parts. First, where an edit should be applied transfers between them: the same early-to-mid-layer MLP at the last subject token is most effective for both. Second, this shared location does not guarantee a shared outcome. Single-token edits succeed in both, but as targets grow longer, editing degrades far more sharply in the MDMs than in the ARMs. The failure stems from how the edited fact is generated: producing a multi-token target passes through intermediate states in which the target is partially unmasked, for which the edit was never optimized. Guided by this diagnosis, we introduce a simple correction that optimizes the edit including such states, substantially restoring multi-token performance. Our code is available at https://github.com/holi-lab/MDM-KE.

Haewon Park, Yohan Jo

Graduate School of Data Science, Seoul National University

Knowledge Editing Masked Diffusion Models Masked Language Modeling Model Editing Diffusion Language Models Language Model Editing

Jun 1, 2026cs.AI

Revisiting Ripple Effects in Knowledge Editing through Pressure-Aware Joint Neighborhood Optimization

Single-edit updates in large language models can trigger ripple effects across local knowledge neighborhoods: desirable propagation to related facts and unintended perturbation of preserved ones. Existing methods address these two effects separately, without explicitly modeling their coupling. We challenge this separation through an analysis of ripple responses across typical baselines, identifying two coupled design pressures: editable-side coordination and preserved-side leakage. We propose Joint Neighborhood Optimization (JNO), a new knowledge-editing framework to formalize and jointly address both pressures at the target-planning stage. JNO instantiates this principle through Pressure-Aware Coordination (PAC), which jointly optimizes neighborhood target representations under coupled constraints, and a semantic pre-execution gate that rejects high-risk target plans before parameter execution. Experiments on RippleEdits show JNO improves propagation and preservation metrics by at least 7.0% while preserving cross-backbone editing stability.

Haoben Huang, Shuxin Liu, Ou Wu +1

Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences, Hangzhou, China · College of Information Engineering, Zhejiang University of Technology, Hangzhou, China

Knowledge Editing Model Editing

May 30, 2026cs.CL

Revisiting Parameter-Based Knowledge Editing in Large Language Models: Theoretical Limits and Empirical Evidence

Parameter-based knowledge editing updates the internal knowledge of large language models (LLMs) via localized weight modifications and has attracted significant attention. However, most existing methods overlook fundamental theoretical limitations and are rarely evaluated under realistic, practice-oriented settings. In this paper, we first present a theoretical analysis based on the dimensional Collapse Hypothesis, explaining how localized parameter edits can propagate along fragile directions in the representation space, inducing global interference and ultimately causing reasoning collapse. Building on this insight, we conduct a comprehensive empirical evaluation by systematically varying knowledge complexity, number of edits, evaluation dimensions, and baseline methods. Our results show that parameter-based editing methods consistently damage core LLM capabilities. In contrast, a simple retrieval-based baseline achieves consistently stronger performance than all parameter-editing methods across all evaluated conditions. These findings highlight that preserving the fundamental capabilities of LLMs after knowledge editing should be a central concern for future research.

Wanying Ren, Xin Song, Futing Wang +2

East China Normal University, Shanghai, China · Westlake University, Hangzhou, China · Nanyang Technological University, Singapore

Model Editing Language Model Editing Catastrophic Forgetting

May 30, 2026cs.CL

Do Text Edits Generalize to Visual Generation? Benchmarking Cross-Modal Knowledge Editing in UMMs

Unified multimodal models (UMMs) have emerged as a promising paradigm for general-purpose multimodal intelligence. As they are deployed in real-world applications, effectively updating internal knowledge becomes critical. While knowledge editing has matured for text-only models, it remains unclear whether edits that successfully modify textual outputs also transfer to image generation in UMMs. To study this question, we introduce UniKE, the first benchmark for cross-modality knowledge editing in UMMs, comprising 2,971 edit subjects spanning attribute and relation edits. Using VQA-based visual verification, we reveal a striking modality gap: text-side efficacy can reach approximately 92%, whereas the best overall VQA accuracy under direct image generation is only 18.5%. We further propose Reasoning-augmented Parameter Editing, which explicitly activates edited knowledge before generation and improves overall VQA accuracy for all evaluated model-editor pairs, with gains up to 18.6 percentage points. Mechanistic analysis shows that this gap is associated with partial alignment between edited textual representations and the conditioning pathways for visual generation, where edits sufficient for text outputs may remain too weak or misaligned to steer image synthesis. These findings show that textual knowledge edits do not guarantee reliable cross-modality transfer and motivate modality-aware editing methods. Our code and data are available at https://github.com/gxx27/UniKE.

Xin Gao, Cheng Yang, Chufan Shi +1

University of California San Diego · University of Southern California

Knowledge Editing Unified Multimodal Models Multimodal Generation Model Editing Multimodal Model Evaluation

May 27, 2026cs.AI

Orthogonal Concept Erasure for Diffusion Models

Concept erasure has emerged as a promising approach to mitigate undesired or unsafe content in diffusion models, yet existing methods still face significant limitations. While training-based methods are effective, their high computational cost limits scalability. Editing-based methods are more efficient and deployment-friendly, yet they struggle to simultaneously achieve precise concept erasure and preserve overall generative capacity. We identify this core limitation of the editing-based methods as reliance on additive parameter updates. Our empirical analysis reveals that concept semantics primarily depend on neuron direction rather than neuron magnitude, while overall generative capacity relies on the angular geometry of neurons. As additive updates inherently entangle direction, magnitude, and angular geometry, they inevitably introduce unintended interference between concept erasure and overall generation performance. To address this, we propose Orthogonal Concept Erasure (OCE), which reformulates editing-based erasure as multiplicative parameter updates from a geometric perspective. Specifically, OCE applies layer-wise orthogonal transformations derived from a closed-form solution to the parameters, enabling precise concept erasure while preserving the neuron magnitude and angular geometry. Furthermore, to address conflicting constraints in multi-concept erasure, OCE introduces a subspace-level objective with structured subspace manipulation, yielding a more effective and scalable erasure. Extensive experiments on single- and multi-concept erasure demonstrate that OCE outperforms existing methods in concept erasure and non-target preservation, erasing up to 100 concepts in 4.3 s. Code: https://github.com/HansSunY/OCE.

Yuhao Sun, Lingyun Yu, Haoxiang Xu +3

University of Science and Technology of China · Ant Group

Model Editing Concept Erasure Diffusion Models

May 27, 2026cs.LG

Interpretability-Guided Layer Selection over Subspace Projection: SAEs as Stethoscopes, Not Scalpels, for Raw Task Vector Model Editing

LLMs increasingly require surgical model editing to enhance domain-specific capabilities without incurring the computational cost or catastrophic forgetting associated with full fine-tuning. Sparse Autoencoders (SAEs) have emerged as a promising tool in this setting, in principle allowing for feature-level identification of where to intervene. In this work, we rigorously evaluate an SAE-guided editing pipeline for mathematical reasoning on Gemma-3-4B-IT and uncover a fundamental failure mode: the intuitively appealing approach of projecting task vectors onto SAE feature subspaces acts as an information bottleneck that discards approximately 97% of the modification energy, yielding no statistically significant improvements across seven math subjects. We show that this failure stems from a geometric misalignment between activation-space SAE directions and weight-space task vectors. We then propose a shift in perspective: SAE as a Stethoscope, Not a Scalpel, where SAEs are used for layer-level diagnosis rather than intervention-level filtering. By injecting unfiltered raw task vectors only into layers identified by an SAE-derived specificity score, we improve Number Theory accuracy from 29.6% to 39.4% (z=+3.41, p=0.0007) on the Minerva Math benchmark; 5 of 7 math subjects significantly improved and none significantly degraded. Our method is fully deterministic, requires no additional inference cost, and provides a principled framework for interpretability-guided model editing.

Li Lei, Madalina Ciobanu, Qingqing Mao +1

Incept Labs Houston, TX · Incept Labs, Houston, TX · Titan Holdings, San Francisco, CA

Task Vectors LLM Interpretability Sparse Autoencoders Model Editing Language Model Editing

May 27, 2026cs.AI

From Fact Overwriting to Knowledge Evolution: Causal Editing via On-Policy Self-Distillation

While Knowledge Editing (KE) enables efficient updates, its dominant Static Fact Overwriting paradigm treats LLMs as discrete databases, forcibly injecting isolated facts. Fracturing pre-trained logical topologies, this triggers Epistemic Dissonance -- a pathology where un-evolved legacy priors force the model to explicitly negate the injected update. Idealized interventions reveal that this is an inherent structural flaw rather than mere algorithmic noise, with a zero-distortion proxy yielding a catastrophic 95.6% self-refutation rate. Given the causally driven nature of real-world knowledge, grounding updates in explicit causal narratives effectively collapses this conflict rate to just 6.6%, underscoring the imperative for a paradigm shift toward Causal Editing. To internalize this evolution, we propose CODE (Causal On-policy Distillation for Editing). By coupling causal bootstrapping with asymmetric on-policy distillation, CODE engraves causal transition logic directly into parametric memory. Experiments on LLaMA-3.1 and Qwen-2.5 show CODE drastically suppresses self-refutation to 1.8% while securing robust multi-hop accuracy (up to 83.5%), seamlessly transforming discrete fact injection into coherent knowledge evolution. Code is available at https://github.com/CrashBugger/CODE.

Shuaike Li, Kai Zhang, Xianquan Wang +2

State Key Laboratory of Cognitive Intelligence, University of Science and Technology of China

Knowledge Editing Knowledge Conflicts in Language Models Model Editing Causal Interventions in Language Models

May 27, 2026cs.CL

Where Does Toxicity Live? Mechanistic Localization and Targeted Suppression in Language Models

Large language models frequently generate toxic, hateful, or harmful content, yet existing mitigation methods rely on costly retraining or output-level filtering with no mechanistic insight into where toxicity originates internally. We introduce Meow2X and TRNE, two complementary retraining-free frameworks that localize toxicity to specific layers and neurons by analyzing activation differentials between toxic and neutral prompts, then suppress them via inference-time scaling or minimal rank-one weight edits -- without any gradient descent. Evaluations across five LMs, two benchmarks, and 90 configurations using dual safety evaluators demonstrate consistent toxicity reduction while preserving language modeling quality. Our analysis reveals that toxicity is disproportionately encoded in early MLP layers, varies across architectures, and is systematically underestimated by single-evaluator setups -- underscoring the need for multi-evaluator safety assessment. By bridging mechanistic interpretability with practical detoxification, our framework offers a principled path toward safer, more transparent language models.

Himanshu Beniwal, Mayank Singh

Indian Institute of Technology Gandhinagar

Language Model Safety Evaluation LLM Interpretability LLM Safety Model Editing Causal Interventions in Language Models

May 26, 2026cs.CL

The Labyrinth and the Thread: Rethinking Regularizations in Sequential Knowledge Editing for Large Language Models

Sequential editing of structured knowledge in large language models allows targeted factual updates without retraining, yet existing methods often rely on complex regularization or constraint mechanisms whose necessity remains unclear. In this work, we systematically investigate the mechanisms underlying effective and stable sequential editing. Specifically, we first analyze the empirical success of AlphaEdit and establish, via a rigorous optimization analysis, the formal equivalence between one-time and sequential editing. Building on this insight, we generalize the equivalence to a broader class of editing objectives, demonstrating that stability emerges naturally from properly accounting for accumulated editing constraints, rather than from specialized regularization or null-space operations. We empirically confirm that many commonly used regularization strategies are unnecessary for reliable sequential updates. Furthermore, we extend our framework to handle conflicting edits, ensuring robust and consistent behavior under contradictory updates. Ultimately, our work provides Ariadne's thread through the labyrinth of sequential editing, charting a path toward simpler, more interpretable, and dependable knowledge updates. Our code is available at https://github.com/Wangzzzzzzzz/OTE-SE-Alignment.

Zheng Wang, Kaixuan Zhang, Wanfang Chen +2

Bosch Center for Artificial Intelligence (BCAI) · Bosch (China) Investment Ltd. · School of Statistics, East China Normal University.

Knowledge Editing Factual Knowledge in Language Models Model Editing

May 24, 2026cs.CV

X-Edit: Exact, Explicit, and Explainable Null-Space Editing for Medical Vision Transformers

Pre-trained Vision Transformers (ViTs) are increasingly deployed for medical image classification. However, correcting their inevitable failure cases in dynamic clinical scenarios poses a critical challenge. Conventional fine-tuning approaches inherently suffer from catastrophic forgetting, severely degrading previously acquired diagnostic capabilities. Such instability fundamentally compromises clinical safety. Addressing this vulnerability requires an active, controllable, and reliable intervention mechanism that is both theoretically grounded and inherently interpretable. To this end, we propose X-Edit (eXact, eXplicit, and eXplainable Editing), an efficient null-space model editing framework. X-Edit transitions the editing process from iterative gradient-based optimization to a theoretically grounded, closed-form solution. Specifically, we first explicitly localize the influential layers via causal tracing governing the erroneous prediction. Subsequently, we construct an orthogonal null-space projection matrix from a curated anchor set. By geometrically constraining the exact parameter update strictly within this null space, we provide mathematical guarantees that the intervention rectifies targeted errors without perturbing established diagnostic representations. Extensive evaluations on six medical imaging benchmarks demonstrate that X-Edit comprehensively suppresses catastrophic forgetting while achieving superior edit success rates. Our code is available at https://github.com/HenryLau7/X-Edit.

Yuanye Liu, Siyuan Zhou, Ke Zhang +3

Fudan University, Shanghai, China · Johns Hopkins University, Baltimore, USA · National University of Singapore, Singapore +1

Vision Transformer Medical Image Classification Explainable Medical Image Analysis Model Editing Catastrophic Forgetting

May 22, 2026cs.AI

Beyond Binary Edits Robust Multimodal Knowledge Editing with Adversarial Subspace Alignment

Multimodal large language models (MLLMs) need efficient mechanisms to update knowledge without degrading existing capabilities. While intrinsic multimodal knowledge editing achieves strong reliability and locality, it often exhibits limited generality, failing to propagate edits across semantically equivalent visual and linguistic variations. This issue arises from the lack of explicit semantic supervision, rigid editing scopes, and biased anchoring to individual samples in high-dimensional multimodal spaces. We address robust intrinsic multimodal knowledge editing by explicitly targeting generalization. We formalize robustness through knowledge units that group semantically equivalent multimodal inputs and define generality as consistent predictions within each unit. To expose fragile semantic regions, we introduce Latent Adversarial Robustification (LAR), which generates adversarial yet semantically coherent variants in the joint latent space. We further propose Rank-Constrained Subspace Learning (RCSL), enforcing low-rank alignment of adversarial representations at the edit layer via a singular value-based objective. Extensive analysis demonstrates the effectiveness of ASAM empirically.

Haoyuan Wang, Xiaohao Liu, Jiajie Su +2

Zhejiang University · National University of Singapore · Jiangxi Normal University

Multimodal Robustness Adversarial Robustness Multimodal Large Language Models Model Editing

May 19, 2026cs.LG

Modality-Decoupled Online Recursive Editing

Online model editing for multimodal large language models (MLLMs) requires assimilating a stream of corrections under tight compute and memory budgets. Yet editors developed for text-only LLMs often degrade on MLLMs: visually dominant activations skew the statistics that shape updates, causing cross-modal conflict, while sequential writes become entangled in a shared edit space and amplify long-horizon interference, causing inter-edit interference. To address these, we propose M-ORE, a modality-decoupled online recursive editor for lifelong MLLM adaptation. M-ORE is derived from a unified proximal-projection formulation and admits a closed-form update with a Sherman-Morrison recursion, yielding constant per-edit overhead. It maintains module-wise locality statistics for the text stack and the visual projector to avoid visually dominated update shaping and performs continual updates in a fixed orthogonal low-rank edit subspace via a Sherman-Morrison recursion to mitigate long-horizon interference. Experiments on multiple MLLM backbones and online editing benchmarks show that our M-ORE method consistently improves reliability, generality, and locality over strong baselines, while achieving favorable quality-efficiency scaling. Our code is publicly available at https://github.com/lab-klc/M-ORE.

Siyuan Li, Youyuan Zhang, Fangming Liu +1

Harbin Institute of Technology, Shenzhen, China. · Peng Cheng Laboratory, China. · Huazhong University of Science and Technology, China.

Continual Learning for LLMs Multimodal Large Language Models Model Editing Language Model Editing

May 18, 2026cs.LG

Residual Paving: Diagnosing the Routing Bottleneck in Selective Refusal Editing

We study selective refusal editing as a three-way control problem: induce non-refusal on designated edit prompts while preserving benign behavior and harmful refusals outside the edit set. We introduce Residual Paving, a routed residual editing method for frozen instruction-tuned transformers that separates route selectivity, whether to intervene, from residual-edit capacity, what edit to apply. An early-layer router predicts a scalar gate and expert mixture; when active, prompt-conditioned bottleneck residual experts apply later-layer residual updates while leaving the backbone unchanged. This decomposition supports an oracle-routing diagnostic where only the learned scalar gate is replaced with the held-out edit/keep label, leaving the residual editor and frozen backbone fixed. On the primary Gemma-3-4B-IT held-out split, learned Residual Paving reduces edit refusal from 88.6% to 4.0%, with 95.5% benign distribution preservation and 87.3% harmful distribution preservation. Same-protocol one-direction steering controls are much weaker on edit success, leaving edit refusal at 86.8% for Edit-target ActAdd and 78.9% for DIM-style refusal steering. The remaining failure is off-target harmful-keep degradation: harmful refusal remains below the frozen-base rate, 65.3% vs. 81.6%. Across six backbones, oracle routing improves the keep-side diagnostic score on every reported row, with median gain +12.9 pp, supporting the interpretation that learned route selectivity is the main observed bottleneck. Trajectory diagnostics on two backbones further suggest directed movement toward edit-target continuations rather than generic refusal suppression.

Bryce Hinkley, Peyman Najafirad

University of Texas at San Antonio

Residual Learning LLM Refusal Behavior Model Editing

May 16, 2026cs.LG

ZeroUnlearn: Few-Shot Knowledge Unlearning in Large Language Models

Large language models inevitably retain sensitive information, defined as inputs that may induce harmful generations, due to training on massive web corpora, raising concerns for privacy and safety. Existing machine unlearning methods primarily rely on retraining or aggressive fine-tuning, which are either computationally expensive or prone to degrading related knowledge and overall model utility. In this work, we reformulate machine unlearning as a precise knowledge re-mapping problem via model editing. We propose ZeroUnlearn, a few-shot unlearning framework. It overwrites sensitive inputs by mapping them to a neutral target state and removing their original representations. ZeroUnlearn enforces representational orthogonality through a multiplicative parameter update with a closed-form solution, enabling efficient and targeted unlearning. We further extend ZeroUnlearn to a gradient-based variant for multi-sample unlearning. Experiments demonstrate that our approach outperforms existing baselines while preserving general model utility. Our code is available at the github: https://github.com/XMUDeepLIT/ZeroUnlearn.

Yujie Lin, Chengyi Yang, Zhishang Xiang +2

School of Informatics, Xiamen University, China · Key Laboratory of Digital Protection and Intelligent Processing of Intangible Cultural Heritage of Fujian and Taiwan (Xiamen University), Ministry of Culture and Tourism, China · School of Film, Xiamen University, China +2

Privacy Leakage in Language Models Model Editing Machine Unlearning LLM Unlearning

May 15, 2026cs.LG

Scalable Knowledge Editing for Mixture-of-Experts LLMs via Tensor-Structured Updates

Knowledge editing (KE) provides a lightweight alternative to repeated fine-tuning of LLMs. However, most existing KE methods target dense feed-forward layers, while modern LLMs increasingly adopt Mixture-of-Experts (MoE) architectures for their superior memory footprint and inference efficiency. This mismatch leaves a growing class of production models without principled editing tools. We propose a MEMIT-like framework for knowledge editing in MoE-based LLMs. Our method exploits the tensor structure of MoE layers to formulate the editing objective faithfully at the per expert level, and applies the Woodbury matrix identity to avoid materializing or inverting the full stacked matrix of expert weights. The resulting update reduces to inversions of fixed low-rank matrices and requires no additional backward passes. Empirically, our approach matches the editing quality of strong baselines on the main KE metrics while accelerating the editing procedure by up to 6x, owing to the batched MEMIT-style formulation and the low-dimensional inversions enabled by the Woodbury identity. These results show that closed-form, parameter-modifying KE can be extended efficiently beyond dense layers, opening a path toward scalable knowledge editing in modern sparse LLM architectures.

Roman Maksimov, Vladimir Aletov, Dmitry Bylinkin +3

1Moscow Independent Research Institute of Artificial Intelligence · 2Basic Research of Artificial Intelligence Laboratory · 3Innopolis University

Knowledge Editing Mixture-of-Experts Language Models Model Editing Language Model Editing

May 14, 2026cs.CR

EVA: Editing for Versatile Alignment against Jailbreaks

Large Language Models (LLMs) and Vision Language Models (VLMs) have demonstrated impressive capabilities but remain vulnerable to jailbreaking attacks, where adversaries exploit textual or visual triggers to bypass safety guardrails. Recent defenses typically rely on safety fine-tuning or external filters to reduce the model's likelihood of producing harmful content. While effective to some extent, these methods often incur significant computational overheads and suffer from the safety utility trade-off, degrading the model's performance on benign tasks. To address these challenges, we propose EVA (Editing for Versatile Alignment against Jailbreaks), a novel framework that pioneers the application of direct model editing for safety alignment. EVA reframes safety alignment as a precise knowledge correction task. Instead of retraining massive parameters, EVA identifies and surgically edits specific neurons responsible for the model's susceptibility to harmful instructions, while leaving the vast majority of the model unchanged. By localizing the updates, EVA effectively neutralizes harmful behaviors without compromising the model's general reasoning capabilities. Extensive experiments demonstrate that EVA outperforms baselines in mitigating jailbreaks across both LLMs and VLMs, offering a precise and efficient solution for post-deployment safety alignment.

Yi Wang, Hongye Qiu, Yue Xu +4

VLM Robustness LLM Safety Alignment LLM Jailbreak Attacks Model Editing

May 13, 2026cs.CL

Merging Methods for Multilingual Knowledge Editing for Large Language Models: An Empirical Odyssey

Multilingual knowledge editing (MKE) remains challenging because language-specific edits interfere with one another, even when locate-then-edit methods work well in monolingual settings. This paper focuses on three issues: the effectiveness of vector merging methods for MKE, the extent to which Task Singular Vectors for Merging (TSVM) can reduce multilingual interference, and the influence of the weight scaling factor and rank compression ratio on performance. We evaluate six merging variants with two popular backbone large language models, two base knowledge editing methods, and 12 languages on the MzsRE benchmark under a large-scale batch-editing setting. Our results show that vector summation with shared covariance is the most reliable overall strategy, whereas simple summation without shared covariance performs poorly. TSVM improves performance in some settings, but its ability to mitigate multilingual interference is limited. We also find that performance is sensitive to both weight scale and rank ratio, with larger-than-default scaling and relatively low rank often yielding better results. These findings clarify the practical strengths and limits of current vector merging methods for MKE and provide guidance for future multilingual knowledge editing research.

Kunil Lee, Ki-Young Shin, Jong-Hyeok Lee +1

Department of Computer Science and Engineering, POSTECH · Designovel Co., Ltd. · LLSOLLU +1

Knowledge Editing Multilingual Language Models Task Vector Merging Model Editing Language Model Merging

May 12, 2026cs.LG

More Edits, More Stable: Understanding the Lifelong Normalization in Sequential Model Editing

Lifelong Model Editing aims to continuously update evolving facts in Large Language Models while preserving unrelated knowledge and general capabilities, yet it remains plagued by catastrophic forgetting and model collapse. Empirically, we find that recent editors resilient over long horizons share the same core strategy: Lifelong Normalization (LN), which normalizes value gradients using running statistics. Removing LN causes immediate performance collapse, and we observe a counter-intuitive positive cumulative effect where early edits can promote the success of future edits. Yet the mechanism of LN remains a "black box", leaving its precise role in lifelong stability poorly understood. In this work, we provide the first theoretical account of LN in the lifelong regime. Our analysis reveals a self-reinforcing stability loop and proves that, when combined with ridge-regularized regression, LN yields parameter updates with asymptotic orthogonality and bounded norms, directly mitigating forgetting and systemic collapse. Based on these insights, we derive StableEdit, which strengthens this stability loop via an explicit warm-up stage and full whitening, improving long-horizon stability at minimal overhead. Extensive experiments validate our theory and demonstrate competitive performance. Our code is available at https://github.com/MINE-USTC/StableEdit.

Xin Ma, Wei Chen, Qi Liu +4

State Key Laboratory of Cognitive Intelligence, University of Science and Technology of China, Anhui, China.

Knowledge Editing Continual Learning Model Editing Catastrophic Forgetting

May 10, 2026cs.CL

BetaEdit: Null-Space Constrained Sequential Model Editing

Null-space-based methods have garnered considerable attention in model editing by constraining updates to the null space of the pre-existing knowledge representation, thereby preserving the model's original behavior. However, in practice these methods rely on an approximate null space--leading to knowledge leakage--and further suffer from severe performance degradation during sequential editing. Recent work shows that history-aware editing strategies can empirically mitigate this decline, yet the underlying reason remains unclear. In this paper, we first expose the knowledge leakage inherent in existing null-space approaches and then analyze why history-aware updates effectively preserve both editing performance and general capabilities during long-horizon editing. Building on these insights, we propose BetaEdit, a refined framework that effectively controls the knowledge leakage and integrates history-aware updates into the null-space paradigm. Extensive experiments on three large language models across two standard benchmarks show that BetaEdit consistently outperforms prior methods in the challenging regime of massive-scale sequential editing. Code is available at: https://github.com/lbq8942/BetaEdit.

Bingqing Liu, Wei Liu, Yuhua Li

Huazhong University of Science and Technology, Wuhan, China

Knowledge Editing Model Editing Language Model Editing

May 7, 2026cs.CL

Uncovering Entity Identity Confusion in Multimodal Knowledge Editing

Multimodal knowledge editing (MKE) aims to correct the internal knowledge of large vision-language models after deployment, yet the behavioral patterns of post-edit models remain underexplored. In this paper, we identify a systemic failure mode in edited models, termed Entity Identity Confusion (EIC): edited models exhibit an absurd behavior where text-only queries about the original entity's identity unexpectedly return information about the new entity. To rigorously investigate EIC, we construct EC-Bench, a diagnostic benchmark that directly probes how image-entity bindings shift before and after editing. Our analysis reveals that EIC stems from existing methods failing to distinguish between Image-Entity (I-E) binding and Entity-Entity (E-E) relational knowledge in the model, causing models to overfit E-E associations as a shortcut: the image is still perceived as the original entity, with the new entity's name serving only as a spurious identity label. We further explore potential mitigation strategies, showing that constraining edits to the model's I-E processing stage encourages edits to act more faithfully on I-E binding, thereby substantially reducing EIC. Based on these findings, we discuss principled desiderata for faithful MKE and provide methodological guidance for future research.

Shu Wu, Xiaotian Ye, Xinyu Mou +3

New Laboratory of Pattern Recognition (NLPR) State Key Laboratory of Multimodal Artificial Intelligence Systems (MAIS) Institute of Automation, Chinese Academy of Sciences · Beijing University of Posts and Telecommunications · School of Artificial Intelligence, University of Chinese Academy of Sciences +3

Multimodal Grounding Multimodal Large Language Models Model Editing Multimodal Model Evaluation

May 2, 2026cs.LG

HoReN: Normalized Hopfield Retrieval for Large-Scale Sequential Model Editing

Large language models encode vast factual knowledge that can become outdated or incorrect after deployment, yet retraining is prohibitively costly. This motivates lifelong model editing, which updates targeted behavior while preserving the rest of the model. Existing editors, both parameter-modifying and parameter-preserving, degrade severely as edits accumulate and struggle to generalize across paraphrases. We propose HoReN, a codebook-based parameter-preserving editor that wraps a single MLP layer with a discrete key-value memory. HoReN treats each codebook entry as both a knowledge key and a Hopfield stored pattern, retrieves edits by angular similarity on the unit hypersphere, and refines queries through damped Hopfield dynamics so paraphrases converge to the correct memory basin while unrelated inputs remain stable. HoReN achieves strong editing performance with consistent gains across diverse benchmarks spanning standard ZsRE, structured WikiBigEdit, and unstructured UnKE evaluations. Moreover, HoReN scales to 50K sequential edits on ZsRE with stable overall performance above 0.93, while prior editors collapse or degrade severely before reaching 10K. Our code is available at https://github.com/ha11ucin8/HoReN.

Yuan Fang, Yi Xie, Xuming Ran

1IXL Learning, Inc · 2Technical University of Munich · 3National University of Singapore

Knowledge Editing Memory-Augmented Language Models Hopfield Networks Model Editing Language Model Editing

May 1, 2026cs.CL

From Backward Spreading to Forward Replay: Revisiting Target Construction in LLM Parameter Editing

LLM parameter editing methods commonly rely on computing an ideal target hidden-state at a target layer (referred as anchor point) and distributing the target vector to multiple preceding layers (commonly known as backward spreading) for cooperative editing. Although widely used for a long time, its underlying basis have not been systematically investigated. In this paper, we first conduct a systematic study of its foundations, which helps clarify its capability boundaries, practical considerations, and potential failure modes. Then, we propose a simple and elegant alternative that replaces backward spreading with forward-propagation. Instead of optimizing the target at the last editing layer, we optimize the anchor point at the first editing layer, and then propagate it forward to obtain accurate and mutually compatible target hidden-states for all subsequent editing layers. This approach achieves the same computational complexity as existing methods while producing more accurate layer-wise targets. Our method is simple, without interfering with either the computation of the initial target hidden state or any other components of the subsequent editing pipeline, and thus constituting a benefit for a wide range of LLM parameter editing methods.

Wei Liu, Hongkai Liu, Zhiying Deng +2

National University of Singapore · Laboratory for Artificial Intelligence and New Forms of Education, National Engineering Research Center for Educational Big Data, Faculty of Artificial Intelligence in Education, Central China Normal University · University of Oxford

Knowledge Editing Model Editing Language Model Editing

Apr 20, 2026cs.LG

Correct When Paired, Wrong When Split: Decoupling and Editing Modality-Specific Neurons in MLLMs

Although Knowledge Editing provides an efficient mechanism for updating the knowledge of Multimodal Large Language Models (MLLMs), we find that current paradigms still suffer from an important yet remain underexplored issue : editing decoupling failure, where entity-related knowledge can be updated when the model is triggered by multimodal inputs (text--image query pairs), however, it often reverts to outdated pre-edit facts when the paired inputs are split into unimodal ones. Our in-depth empirical analysis reveals that the entity knowledge in MLLMs is not stored as a unified representation, but is instead distributed across disentangled modality-specific pathways. As a result, updates biased toward multimodal queries fail to propagate effectively to unimodal circuits. To bridge this gap, we propose DECODE, which explicitly disentangles and localizes modality-specific neuron groups for targeted knowledge. Extensive experiments demonstrate that DECODE consistently achieves effective knowledge updates under different modality triggers, thereby mitigating editing decoupling failures.

Tingchao Fu, Wenkai Wang, Fanxiao Li +6

School of Information Science and Engineering, Yunnan University · School of Software, Yunnan University · 3National University of Singapore +1

Knowledge Editing Multimodal Large Language Models Model Editing Multimodal Foundation Models

Apr 18, 2026cs.LG

One Mask to Rule Them All: On Hidden Facts after Editing and How to Find Them

Knowledge editing methods such as ROME and MEMIT update factual associations in transformer models by modifying MLP weights. While evaluated mainly by output behavior, their internal mechanism remains underexplored. We investigate whether edits rely on a common mechanism, regardless of which fact is modified. Despite fact-specific weight changes, we argue that ROME and MEMIT target the same subset of weights critical for maintaining edits. To isolate this subset, we train a compact binary mask over the edited weights. The mask reverses 80% of edits on the training set and over 70% on the test set, confirming that diverse edits share a common functional structure. Our analysis reveals that the mask reverses edits by eliminating overattention in later layers. Additionally, we show that injecting the mask during editing drops editing success from 98% to 38%, demonstrating that this mechanism is necessary for edits to succeed. Our finding that edits suppress rather than overwrite knowledge explains why ROME and MEMIT fail to propagate changes to related facts. The identified common functional subspace informs detection and defense against unwanted edits.

Ali Holmov, Paul Youssef, Nandi Schoots +1

Technical University of Munich · Marburg University · University of Oxford

Knowledge Editing Transformer Interpretability LLM Auditing Factual Knowledge in Language Models Model Editing Language Model Editing+1

Nov 8, 2025cs.CL

Can Fine-Tuning Erase Your Edits? On the Fragile Coexistence of Knowledge Editing and Adaptation

Knowledge editing (KE) offers a lightweight alternative to retraining for updating large language models (LLMs). Meanwhile, fine-tuning remains the default operation for adapting LLMs to new domains and tasks. Despite their widespread adoption, these two post-training interventions have been studied in isolation, leaving open a crucial question: if we fine-tune an edited model, do the edits survive? This question is motivated by practical objectives: removing covert or malicious edits, and preserving beneficial edits. If fine-tuning impairs edits (Fig.1), current KE methods become less efficient, as a newly fine-tuned model requires re-editing; if edits persist, fine-tuned models risk propagating hidden malicious edits, raising serious safety concerns. To this end, we systematically quantify edit decay after fine-tuning across 254 experimental configurations. Our results show that in general, edits decay substantially after subsequent fine-tuning. AlphaEdit exhibits the greatest decay on the zsRE benchmark when applied to GPT-J, where 25.27% of previously successful edits become unsuccessful after fine-tuning. We further find that fine-tuning only the edited layers is sufficient to effectively remove edits, while incurring only modest degradation in downstream performance. Surprisingly, fine-tuning non-edited layers leads to greater edit decay than all-layer fine-tuning. Besides, our activation space analysis reveals that fine-tuning produces a larger and more coherent representational shift, both in magnitude and direction, than KE. Overall, our study underscores the necessity of evaluating KE within the broader LLM application pipeline.

Yinjie Cheng, Paul Youssef, Christin Seifert +2

School of Computer Science University of Sheffield Sheffield, UK · Mathematics and Computer Science Marburg University Marburg, Germany

Knowledge Editing LLM Fine-Tuning Model Editing Language Model Editing

Aug 28, 2025cs.CL

Turning the Spell Around: Lightweight Alignment Amplification via Rank-One Safety Injection

Safety alignment in Large Language Models (LLMs) often involves mediating internal representations to refuse harmful requests. Recent research has demonstrated that these safety mechanisms can be bypassed by ablating or removing specific representational directions within the model. In this paper, we propose the opposite approach: Rank-One Safety Injection (ROSI), a white-box method that amplifies a model's safety alignment by permanently steering its activations toward the refusal-mediating subspace. ROSI operates as a simple, fine-tuning-free rank-one weight modification applied to all residual stream write matrices. The required safety direction can be computed from a small set of harmful and harmless instruction pairs. We show that ROSI consistently increases safety refusal rates - as evaluated by Llama Guard 3 - while preserving the utility of the model on standard benchmarks such as MMLU, HellaSwag, and Arc. Furthermore, we show that ROSI can also re-align 'uncensored' models by amplifying their own latent safety directions, demonstrating its utility as an effective last-mile safety procedure. Our results suggest that targeted, interpretable weight steering is a cheap and potent mechanism to improve LLM safety, complementing more resource-intensive fine-tuning paradigms.

Harethah Abu Shairah, Hasan Abed Al Kader Hammoud, George Turkiyyah +1

King Abdullah University of Science and Technology (KAUST)

LLM Guardrails LLM Safety Alignment Model Editing Activation Steering

May 28, 2025cs.CL

InComeS: Integrating Compression and Selection Mechanisms into LLMs for Efficient Model Editing

Although existing model editing methods perform well in recalling exact edit facts, they often struggle in complex scenarios that require deeper semantic understanding rather than mere knowledge regurgitation. Leveraging the strong contextual reasoning abilities of large language models (LLMs), in-context learning (ICL) becomes a promising editing method by comprehending edit information through context encoding. However, this method is constrained by the limited context window of LLMs, leading to degraded performance and efficiency as the number of edits increases. To overcome this limitation, we propose InComeS, a flexible framework that enhances LLMs' ability to process editing contexts through explicit compression and selection mechanisms. Specifically, InComeS compresses each editing context into the key-value (KV) cache of a special gist token, enabling efficient handling of multiple edits without being restricted by the model's context window. Furthermore, specialized cross-attention modules are added to dynamically select the most relevant information from the gist pools, enabling adaptive and effective utilization of edit information. We conduct experiments on diverse model editing benchmarks with various editing formats, and the results demonstrate the effectiveness and efficiency of our method.

Shuaiyi Li, Zhisong Zhang, Yang Deng +6

The Chinese University of Hong Kong · City University of Hong Kong · Singapore Management University +1

In-Context Learning Model Editing LLM Context Management Context Compression