Adaptive Overclocking: Dynamic Control of Thinking Path Length via Real-Time Reasoning Signals
On the map: LLM reasoning → Test-time compute and efficient thinking
Abstract (arXiv)
Large Reasoning Models (LRMs) often suffer from computational inefficiency due to overthinking, where a fixed reasoning budget fails to match the varying complexity of tasks. To address this issue, we propose Adaptive Overclocking, a method that makes the overclocking hyperparameter α dynamic and context-aware. Our method adjusts reasoning speed in real time through two complementary signals: (1) token-level model uncertainty for fine-grained step-wise control, and (2) input complexity estimation for informed initialization. We implement this approach with three strategies: Uncertainty-Aware Alpha Scheduling (UA-αS), Complexity-Guided Alpha Initialization (CG-αI), and a Hybrid Adaptive Control (HAC) that combines both. Experiments on GSM8K, MATH, and SVAMP show that HAC achieves superior accuracy-latency trade-offs, reducing unnecessary computation on simple problems while allocating more resources to challenging ones. By mitigating overthinking, Adaptive Overclocking enhances both efficiency and overall reasoning performance.
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 19 of its 24 references that are in the map. First posted on arXiv 21 Sep 2025.
Builds on
19 of 24 references.
Closest ancestors · 9
Same direction, newest first
- 2025Overclocking LLM Reasoning: Monitoring and Controlling Thinking Path Lengths in LLMs8 Jun
- 2025Between Underthinking and Overthinking: An Empirical Study of Reasoning Length and correctness in LLMs30 Apr
- 2025Effectively Controlling Reasoning Models through Thinking Intervention31 Mar
- 2025Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models20 Mar
- 2025L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning6 Mar
- 2025The Danger of Overthinking: Examining the Reasoning-Action Dilemma in Agentic Tasks12 Feb
- 2025Thoughts Are All Over the Place: On the Underthinking of o1-Like LLMs30 Jan
- 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs30 Dec
- 2024Escape Sky-high Cost: Early-stopping Self-Consistency for Multi-step Reasoning19 Jan
Other references · 10
- 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
- 2021Training Verifiers to Solve Math Word Problemscited by 10,109
- 2025Qwen3 Technical Reportcited by 8,820
- 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
- 2022Large Language Models are Zero-Shot Reasonerscited by 4,374
- 2023Let's Verify Step by Stepcited by 3,789
- 2021Are NLP Models really able to Solve Simple Math Word Problems?cited by 878
- 2025Training Language Models to Reason Efficientlycited by 247
- 2025When More is Less: Understanding Chain-of-Thought Length in LLMscited by 183
- 2024The Impact of Reasoning Step Length on Large Language Modelscited by 153
Built on by
Citing papers per year: 2026: 1.
1 paper.
Similar, not cited
Close in wording, no citation either way
- 2025Controlling Thinking Speed in Reasoning Models
- 2026When2Think: Learning Difficulty-Aware Length Control for Efficient Hybrid Reasoning Models
- 2026Mitigating Overthinking in Large Reasoning Language Models via Reasoning Path Deviation Monitoring
- 2025ARS: Adaptive Reasoning Suppression for Efficient Large Reasoning Language Models
- 2026CODA: Difficulty-Aware Compute Allocation for Adaptive Reasoning
- 2026DyCon: Dynamic Reasoning Control via Evolving Difficulty Modeling
- 2026Thinking Traps in Long Chain-of-Thought: A Measurable Study and Trap-Aware Adaptive Restart
- 2026ROM: Real-time Overthinking Mitigation via Streaming Detection and Intervention