mlsanityData to 14 Oct 2026

TrimR: Verifier-based Training-Free Thinking Compression for Efficient Test-Time Scaling

arXiv 2505.17155 ↗First posted 22 May 2025cs.LG cs.AI cs.CLcited by 14

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

Abstract (arXiv)

Large Reasoning Models (LRMs) demonstrate exceptional capability in tackling complex mathematical, logical, and coding tasks by leveraging extended Chain-of-Thought (CoT) reasoning. Test-time scaling methods, such as prolonging CoT with explicit token-level exploration, can push LRMs' accuracy boundaries, but they incur significant decoding overhead. A key inefficiency source is LRMs often generate redundant thinking CoTs, which demonstrate clear structured overthinking and underthinking patterns. Inspired by human cognitive reasoning processes and numerical optimization theories, we propose TrimR, a verifier-based, training-free, efficient framework for dynamic CoT compression to trim reasoning and enhance test-time scaling, explicitly tailored for production-level deployment. Our method employs a lightweight, pretrained, instruction-tuned verifier to detect and truncate redundant intermediate thoughts of LRMs without any LRM or verifier fine-tuning. We present both the core algorithm and asynchronous online system engineered for high-throughput industrial applications. Empirical evaluations on Ascend NPUs and vLLM show that our framework delivers substantial gains in inference efficiency under large-batch workloads. In particular, on the four MATH500, AIME24, AIME25, and GPQA benchmarks, the reasoning runtime of Pangu Pro MoE, Pangu-R-38B, QwQ-32B, and DeepSeek-R1-Distill-Qwen-32B is improved by up to 70% with negligible impact on accuracy.

Key facts. Cited by 14 papers 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 31 of its 33 references that are in the map. First posted on arXiv 22 May 2025.

Builds on

31 of 33 references.

Closest ancestors · 14

Same direction, newest first

  1. 2025Efficient Reasoning for LLMs through Speculative Chain-of-Thought27 Apr
  2. 2025Dynamic Early Exit in Reasoning Models22 Apr
  3. 2025SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning10 Apr
  4. 2025ThinkPrune: Pruning Long Chain-of-Thought of LLMs via Reinforcement Learning2 Apr
  5. 2025InftyThink: Breaking the Length Limits of Long-Context Reasoning in Large Language Models9 Mar
  6. 2025Sampling-Efficient Test-Time Scaling: Self-Estimating the Best-of-N Sampling in Early Decoding3 Mar
  7. 2025Self-Training Elicits Concise Reasoning in Large Language Models27 Feb
  8. 2025LightThinker: Thinking Step-by-Step Compression21 Feb
  9. 2025TokenSkip: Controllable Chain-of-Thought Compression in LLMs17 Feb
  10. 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs30 Dec
  11. 2024Token-Budget-Aware LLM Reasoning24 Dec
  12. 2024Enhancing LLM Reasoning via Critique Models with Test-Time and Training-Time Supervision25 Nov
2 more
  1. 2024Adaptive Inference-Time Compute: LLMs Can Predict if They Can Do Better, Even Mid-Generation3 Oct
  2. 2024Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models1 Aug

Other references · 17

  1. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  3. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  4. 2023Let's Verify Step by Stepcited by 3,789
  5. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  6. 2024Qwen2 Technical Reportcited by 2,120
  7. 2025Kimi k1.5: Scaling Reinforcement Learning with LLMscited by 1,210
  8. 2023SGLang: Efficient Execution of Structured Language Model Programscited by 1,141
  9. 2023Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generationcited by 819
  10. 2024Training Large Language Models to Reason in a Continuous Latent Spacecited by 797
  11. 2025Chain of Draft: Thinking Faster by Writing Lesscited by 297
  12. 2025Training Language Models to Reason Efficientlycited by 247
5 more
  1. 2025Inference-Time Scaling for Generalist Reward Modelingcited by 239
  2. 2024Distilling System 2 into System 1cited by 159
  3. 2024Fast Best-of-N Decoding via Speculative Rejectioncited by 112
  4. 2025DeepSeek-R1 Thoughtology: Let's think about LLM Reasoningcited by 92
  5. 2025Pangu Ultra: Pushing the Limits of Dense Large Language Models on Ascend NPUscited by 13

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 4; 2026: 10.

14 papers.

Sort:

By area:

  1. 2026CRISP: Compressed Reasoning via Iterative Self-Policy Distillationcited by 58
  2. 2025Don't Overthink It: A Survey of Efficient R1-style Large Reasoning Modelscited by 37
  3. 2025The Markovian Thinker: Architecture-Agnostic Linear Scaling of Reasoningcited by 13
  4. 2026Neural Chain-of-Thought Search: Searching the Optimal Reasoning Path to Enhance Large Language Modelscited by 8
  5. 2026What Matters For Safety Alignment?cited by 7
  6. 2025Scaling Up, Speeding Up: A Benchmark of Speculative Decoding for Efficient LLM Test-Time Scalingcited by 6
  7. 2026DLLM Agent: See Farther, Run Fastercited by 6
  8. 2025Towards Efficient Agents: A Co-Design of Inference Architecture and Systemcited by 4
  9. 2026Efficient Reasoning with Balanced Thinkingcited by 4
  10. 2026SAT: Balancing Reasoning Accuracy and Efficiency with Stepwise Adaptive Thinkingcited by 2
  11. 2026Reasoning Models Know What's Important, and Encode It in Their Activationscited by 1
  12. 2026HyPER: Bridging Exploration and Exploitation for Scalable LLM Reasoning with Hypothesis Path Expansion and Reductioncited by 0
2 more
  1. 2026Breaking the Reward Barrier: Accelerating Tree-of-Thought Reasoning via Speculative Explorationcited by 0
  2. 2026DyCon: Dynamic Reasoning Control via Evolving Difficulty Modelingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025VeriThinker: Learning to Verify Makes Reasoning Model Efficient
  2. 2026Think Less, Know More: State-Aware Reasoning Compression with Knowledge Guidance for Efficient Reasoning
  3. 2025Optimizing Length Compression in Large Reasoning Models
  4. 2026CtrlCoT: Dual-Granularity Chain-of-Thought Compression for Controllable Reasoning
  5. 2025ConCISE: Confidence-guided Compression in Step-by-step Efficient Reasoning
  6. 2025Reasoning Efficiently Through Adaptive Chain-of-Thought Compression: A Self-Optimizing Framework
  7. 2025ThinkSwitcher: When to Think Hard, When to Think Fast
  8. 2026TRiMS: Real-Time Tracking of Minimal Sufficient Length for Efficient Reasoning via RL