mlsanityData to 14 Oct 2026

Adaptive Activation Steering for Efficient LLM Reasoning via Closed-Loop PID Control

arXiv 2506.18831 ↗First posted 23 Jun 2025cs.CLcited by 1

On the map: LLM reasoning → Test-time compute and efficient thinking

also in 1 more
Abstract (arXiv)

Reasoning LLMs trained with long chain-of-thought often overthink: they spend tokens on redundant reflection and transitions that inflate cost without improving accuracy. Static activation steering (e.g. SEAL) suppresses such content with a fixed vector, but applies the same strength regardless of how redundant the current chunk actually is. We describe PID-steering, a training-free, decoding-time method that modulates the steering strength with a PID controller driven by a lightweight chunk-level redundancy classifier. On a subset of GSM8K with DeepSeek-R1-Distill-Qwen-1.5B, the method improves accuracy from 85.7% to 89.6% (+3.9 pp) while cutting average output length from 1026 to 790 tokens (-23%). We report it as a small-scale proof of concept rather than a benchmark result.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 3 of its 4 references that are in the map. First posted on arXiv 23 Jun 2025.

Builds on

3 of 4 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025SEAL: Steerable Reasoning Calibration of Large Language Models for Free7 Apr
  2. 2025Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models20 Mar

Other references · 1

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Closing the Loop: PID Feedback Control for Interpretable Activation Steering in Symbolic Music Generationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026ATLAS: Verifier-Guided Adaptive Latent Activation Steering for Efficient LLM Reasoning
  2. 2025Steering LLM Reasoning Through Bias-Only Adaptation
  3. 2025Prototype-Based Dynamic Steering for Large Language Models
  4. 2025Learning to Ponder: Adaptive Reasoning in Latent Space
  5. 2026Can We Break LLMs Out of Self-Loops? Fine-Grained Reasoning Control with Activation Steering
  6. 2026LoopUS: Recasting Pretrained LLMs into Looped Latent Refinement Models
  7. 2025Small Vectors, Big Effects: A Mechanistic Study of RL-Induced Reasoning via Steering Vectors
  8. 2025The Rogue Scalpel: Activation Steering Compromises LLM Safety