arXiv Computation and Language October 1, 2026 By Mehdi Jafari, Hao Xue, Flora Salim

MetaSteer: Context-Conditioned, nonlinear Steering via Attention-Projection Adaptation

Read the original on arXiv Computation and Language →

MetaSteer is a new method for steering large language models that learns nonlinear, context-dependent interventions applied to attention projection matrices. Unlike traditional linear, context-independent techniques, MetaSteer adapts its effects based on the input, requiring no linear concept-geometry assumption. Trained once on a pooled preference corpus, it transfers zero‑shot to unseen concepts and out‑of‑distribution contexts, matching or surpassing strong task‑specific baselines on multiple benchmarks and model families.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

arXiv AI

Aug 11

Learning Preference Adaptation for Large Language Model Personalization via Verbal Reinforcement Learning

arXiv:2608. 09507v1 Announce Type: cross Abstract: Natural language user preferences provide an interpretable interface for LLM personalization.

By Yuting Liu, Wei Wu, Jianzhe Zhao, Guibing Guo

llms rag agents reinforcement-learning

arXiv Machine Learning

Jun 2

Fine-Tuning Without Forgetting In-Context Learning: A Theoretical Analysis of Linear Attention Models

arXiv:2602. 23197v2 Announce Type: replace-cross Abstract: Transformer-based large language models exhibit in-context learning, enabling adaptation to downstream tasks via few-shot prompting with demonstrations.

By Chungpa Lee, Jy-yong Sohn, Kangwook Lee

llms fine-tuning

arXiv Machine Learning

Jun 9

Beyond Linear Activation Steering: Invertible Latent Transformations for Controlling LLM Behavior

arXiv:2606. 08454v1 Announce Type: new Abstract: Activation steering provides a lightweight inference-time mechanism for controlling large language models (LLMs) by modifying their internal activation vectors toward desired behaviors.

By Tuc Nguyen, Thai Le

llms diffusion benchmarks safety

arXiv Machine Learning

Jul 31

S-GRPO: Unified Post-Training for Large Vision-Language Models

arXiv:2604. 16557v2 Announce Type: replace Abstract: Current post-training methodologies for adapting Large Vision-Language Models (LVLMs) generally fall into two paradigms: Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL).

By Yuming Yan, Kai Tang, Sihong Chen, Ke Xu, Dan Hu, Qun Yu, Pengfei Hu

llms reinforcement-learning fine-tuning multimodal

arXiv AI

Jul 13

GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs

arXiv:2507. 18043v2 Announce Type: replace-cross Abstract: Inference-time steering methods offer a lightweight alternative to fine-tuning large language models (LLMs) and vision-language models (VLMs) by modifying internal activations at test time without updating model weights.

By Duy Nguyen, Archiki Prasad, Elias Stengel-Eskin, Mohit Bansal

llms fine-tuning multimodal safety

arXiv Computation and Language

5d ago

LocUS: Head Selection and Subspace Projection for Targeted Activation Steering

arXiv:2609.31122v1 Announce Type: new Abstract: Activation steering is a powerful training-free paradigm for controlling large language models at inference time. However, standard approaches estimate...

By Irene Tallini, Lorenzo Basile, Valentino Maiorca, Francesco Locatello, Alberto Cazzaniga

llms benchmarks safety

One email a morning, machine-written

Related stories

Learning Preference Adaptation for Large Language Model Personalization via Verbal Reinforcement Learning

Fine-Tuning Without Forgetting In-Context Learning: A Theoretical Analysis of Linear Attention Models

Beyond Linear Activation Steering: Invertible Latent Transformations for Controlling LLM Behavior

S-GRPO: Unified Post-Training for Large Vision-Language Models

GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs

LocUS: Head Selection and Subspace Projection for Targeted Activation Steering