arXiv:2604.24693v2 Announce Type: replace Abstract: Linear activation steering is a powerful approach for controlling and specializing large language models by adding linear representations of concep...
arXiv:2609.07037v1 Announce Type: new Abstract: Activation steering has emerged as a lightweight, inference-time approach to control the behavior of Large Language Models (LLMs). However, traditional...
arXiv:2605. 05983v2 Announce Type: replace Abstract: Recently, steering vectors (SVs) have emerged as an effective and lightweight approach to steer behaviors of large language models (LLMs), among which fine-tuned SVs are more effective than optimization-free ones.
arXiv:2601. 03093v2 Announce Type: replace Abstract: Recent work on activation and latent steering has demonstrated that modifying internal representations can effectively guide large language models (LLMs) toward improved reasoning and efficiency without updating model parameters.
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.
The paper introduces Where to Steer (W2S), a framework that adaptively selects which layer of a large language model to intervene on based on the input. It argues that optimal steering layers vary across inputs and demonstrates that W2S outperforms fixed-layer steering methods on several LLMs and alignment tasks, both in‑distribution and out‑of‑distribution. The study highlights input‑dependent layer selection as a complementary design dimension to steering vector extraction.
arXiv:2502. 12446v3 Announce Type: replace-cross Abstract: Inference-time intervention (ITI) has emerged as a promising method for steering large language model (LLM) behavior in a particular direction (e.
arXiv:2602.09870v3 Announce Type: replace Abstract: Steering methods influence Large Language Model behavior by identifying semantic directions in hidden representations, but are typically realized t...
arXiv:2606. 11599v1 Announce Type: cross Abstract: Activation steering offers a lightweight approach to control language models' behavior at inference time, but whether it succeeds or fails heavily depends on the prompt, concept, model, and steering configuration.
The paper introduces EvoPrompt, a framework for adapting vision‑language models to new tasks with limited data while preventing catastrophic forgetting. EvoPrompt uses a Modality‑Shared Prompt Projector to create hierarchical prompts and an evolutionary training strategy that separates low‑rank updates into directional and magnitude components, preserving learned semantic directions. Experiments show that EvoPrompt achieves state‑of‑the‑art few‑shot performance while maintaining the original zero‑shot capabilities of the pre‑trained models.
arXiv:2603. 09493v2 Announce Type: replace-cross Abstract: The adaptation of large-scale vision-language models (VLMs) to downstream tasks with limited labeled data remains a significant challenge.
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.