pith. sign in

arxiv: 2410.12299 · v2 · pith:BU5CLHHPnew · submitted 2024-10-16 · 💻 cs.CL

Semantics-Adaptive Activation Intervention for LLMs via Dynamic Steering Vectors

classification 💻 cs.CL
keywords interventionsadiactivationactivationsdynamicllmsmodelsteering
0
0 comments X
read the original abstract

Large language models (LLMs) have achieved remarkable performance across many tasks, yet aligning them with desired behaviors remains challenging. Activation intervention has emerged as an effective and economical method to modify the behavior of LLMs. Despite considerable interest in this area, current intervention methods exclusively employ a fixed steering vector to modify model activations, lacking adaptability to diverse input semantics. To address this limitation, we propose Semantics-Adaptive Dynamic Intervention (SADI), a novel method that constructs a dynamic steering vector to intervene model activations at inference time. More specifically, SADI utilizes activation differences in contrastive pairs to precisely identify critical elements of an LLM (i.e., attention heads, hidden states, and neurons) for targeted intervention. During inference, SADI dynamically steers model behavior by scaling element-wise activations based on the directions of input semantics. Experimental results show that SADI outperforms established baselines by substantial margins, improving task performance without training. SADI's cost-effectiveness and generalizability across various LLM backbones and tasks highlight its potential as a versatile alignment technique.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Decodable Is Not Grounded: A Vision-Ablation Arbiter for VLM Spatial Reasoning

    cs.CV 2026-06 unverdicted novelty 8.0

    A blank-image ablation test reveals that high probe accuracy on VLM spatial reasoning frequently reflects priors or inverted signs rather than image grounding, with horizontal grounded, vertical prior, and depth inverted.

  2. Subliminal Clocks: Latent Time Modelling in Diffusion Language Models

    cs.AI 2026-07 unverdicted novelty 6.0

    DLMs encode a decodable latent timestep signal in residual activations that can be steered to predictably change model confidence and entropy.

  3. Latent Reward Steering: An Adaptive Inference-Time Framework that Implicitly Promotes Cognitive Behaviors in Reasoning LLMs

    cs.AI 2026-05 unverdicted novelty 6.0

    LRS trains a latent reward model on final-answer correctness to steer SAE states during inference, improving reasoning performance and implicitly encouraging better cognitive behaviors.

  4. Activation Steering for Synthetic Data Generation: The Role of Diversity in Downstream Safety Detection

    cs.LG 2026-05 unverdicted novelty 6.0

    Activation steering produces synthetic safety-violating data that improves downstream classifiers over prompting on most tested concepts when a harmonic mean of alignment, coherence, and diversity is optimized.

  5. Hallucination as an Anomaly: Dynamic Intervention via Probabilistic Circuits

    cs.CL 2026-05 unverdicted novelty 6.0

    Probabilistic circuits detect LLM hallucinations as residual-stream anomalies with up to 99% AUROC and enable dynamic correction that raises truthfulness scores while cutting unnecessary output corruption.

  6. Chain of Risk: Safety Failures in Large Reasoning Models and Mitigation via Adaptive Multi-Principle Steering

    cs.AI 2026-05 unverdicted novelty 6.0

    Reasoning traces in large reasoning models expose safety failures missed by final-answer checks, and adaptive multi-principle steering reduces unsafe content in both traces and answers while preserving task performance.

  7. CreativityNeuro: Steering Language Model Weights to Improve Divergent Thinking and Reduce Mode Collapse

    cs.AI 2026-07 unverdicted novelty 5.0

    CreativityNeuro applies contrastive weight steering to LLMs, yielding up to 14 percentile gains on the Divergent Association Task and improved originality in human-rated tests while reducing mode collapse.