RISER: Orchestrating Latent Reasoning Skills for Adaptive Activation Steering

TL;DR

RISER is a flexible, reinforcement learning-based framework that adaptively guides large language models' reasoning by dynamically composing reusable latent vectors, leading to improved accuracy and efficiency across diverse tasks.

Contribution

It introduces RISER, a novel plug-and-play intervention method that adaptively steers LLM reasoning in activation space using a learned router and reasoning vectors.

Findings

Achieves 3.4-6.5% accuracy improvements on seven benchmarks.

Surpasses Chain-of-Thought reasoning with 2-3x token efficiency.

Produces interpretable, compositional control strategies.

Abstract

Recent work on domain-specific reasoning with large language models (LLMs) often relies on training-intensive approaches that require parameter updates. While activation steering has emerged as a parameter efficient alternative, existing methods apply static, manual interventions that fail to adapt to the dynamic nature of complex reasoning. To address this limitation, we propose RISER (Router-based Intervention for Steerable Enhancement of Reasoning), a plug-and-play intervention framework that adaptively steers LLM reasoning in activation space. RISER constructs a library of reusable reasoning vectors and employs a lightweight Router to dynamically compose them for each input. The Router is optimized via reinforcement learning under task-level rewards, activating latent cognitive primitives in an emergent and compositional manner. Across seven diverse benchmarks, RISER yields 3.4-6.5% average zero-shot accuracy improvements over the base model while surpassing CoT-style reasoning with 2-3x higher token efficiency and robust accuracy gains. Further analysis shows that RISER autonomously combines multiple vectors into interpretable, precise control strategies, pointing toward more controllable and efficient LLM reasoning.