mlsanityData to 14 Oct 2026

Efficient Test-Time Scaling via Self-Calibration

arXiv 2503.00031 ↗First posted 25 Feb 2025cs.LG cs.AI cs.CLcited by 75

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

Abstract (arXiv)

Increasing test-time computation is a straightforward approach to enhancing the quality of responses in Large Language Models (LLMs). While Best-of-N sampling and Self-Consistency with majority voting are simple and effective, they require a fixed number of sampling responses for each query, regardless of its complexity. This could result in wasted computation for simpler questions and insufficient exploration for more challenging ones. In this work, we argue that model confidence of responses can be used for improving the efficiency of test-time scaling. Unfortunately, LLMs are known to be overconfident and provide unreliable confidence estimation. To address this limitation, we introduce Self-Calibration by distilling Self-Consistency-derived confidence into the model itself. This enables reliable confidence estimation at test time with one forward pass. We then design confidence-based efficient test-time scaling methods to handle queries of various difficulty, such as Early-Stopping for Best-of-N and Self-Consistency with calibrated confidence. Experiments on three LLMs across six datasets demonstrate the effectiveness of our approach. Specifically, applying confidence-based Early Stopping to Best-of-N improves MathQA accuracy from 81.0 to 83.6 with a sample budget of 16 responses, indicating the efficacy of confidence-based sampling strategy at inference time.

Key facts. Cited by 75 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 54 of its 63 references that are in the map. First posted on arXiv 25 Feb 2025.

Builds on

54 of 63 references.

Closest ancestors · 13

Same direction, newest first

  1. 2025s1: Simple test-time scaling31 Jan
  2. 2025SETS: Leveraging Self-Verification and Self-Correction for Improved Test-Time Scaling31 Jan
  3. 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs30 Dec
  4. 2024Forest-of-Thought: Scaling Test-Time Compute for Enhancing LLM Reasoning12 Dec
  5. 2024Reasoning Aware Self-Consistency: Leveraging Reasoning Paths for Efficient LLM Sampling30 Aug
  6. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters6 Aug
  7. 2024Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models1 Aug
  8. 2024A Theoretical Understanding of Self-Correction through In-context Alignment28 May
  9. 2024Escape Sky-high Cost: Early-stopping Self-Consistency for Multi-step Reasoning19 Jan
  10. 2023Let's Sample Step by Step: Adaptive-Consistency for Efficient Reasoning and Coding with LLMs19 May
  11. 2023Self-Refine: Iterative Refinement with Self-Feedback30 Mar
  12. 2022Large Language Models Can Self-Improve20 Oct
1 more
  1. 2022Language Models (Mostly) Know What They Know11 Jul

Other references · 41

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2024The Llama 3 Herd of Modelscited by 14,715
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  4. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  5. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  6. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  7. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,708
  8. 2017On Calibration of Modern Neural Networkscited by 4,960
  9. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  10. 2017Deep reinforcement learning from human preferencescited by 3,912
  11. 2023Let's Verify Step by Stepcited by 3,789
  12. 2016A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networkscited by 2,385
29 more
  1. 2018Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answeringcited by 2,164
  2. 2018CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledgecited by 1,822
  3. 2022Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Themcited by 1,807
  4. 2024Large Language Monkeys: Scaling Inference Compute with Repeated Samplingcited by 991
  5. 2023Teaching Large Language Models to Self-Debugcited by 988
  6. 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 974
  7. 2021Are NLP Models really able to Solve Simple Math Word Problems?cited by 878
  8. 2023Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedbackcited by 812
  9. 2017Crowdsourcing Multiple Choice Science Questionscited by 765
  10. 2022Teaching Models to Express Their Uncertainty in Wordscited by 751
  11. 2019MathQA: Towards Interpretable Math Word Problem Solving with Operation-Based Formalismscited by 662
  12. 2024Generative Verifiers: Reward Modeling as Next-Token Predictioncited by 480
  13. 2025The Lessons of Developing Process Reward Models in Mathematical Reasoningcited by 415
  14. 2022Generating Sequences by Learning to Self-Correctcited by 277
  15. 2020ReClor: A Reading Comprehension Dataset Requiring Logical Reasoningcited by 217
  16. 2020Retrieve, Program, Repeat: Complex Knowledge Base Question Answering via Alternate Meta-learningcited by 216
  17. 2019WinoGrande: An Adversarial Winograd Schema Challenge at Scalecited by 204
  18. 2022HyperTree Proof Search for Neural Theorem Provingcited by 197
  19. 2020Mix-n-Match: Ensemble and Compositional Methods for Uncertainty Calibration in Deep Learningcited by 190
  20. 2024Tree Search for Language Model Agentscited by 151
  21. 2024Taming Overconfidence in LLMs: Reward Calibration in RLHFcited by 112
  22. 2024SaySelf: Teaching LLMs to Express Confidence with Self-Reflective Rationalescited by 105
  23. 2024When to Trust LLMs: Aligning Confidence with Response Qualitycited by 43
  24. 2024AlphaVerus: Bootstrapping Formally Verified Code Generation through Self-Improving Translation and Treefinementcited by 35
  25. 2024Enhancing Confidence Expression in Large Language Models Through Learning from Past Experiencecited by 21
  26. 2024Self-Consistency Boosts Calibration for Math Reasoningcited by 13
  27. 2023Great Models Think Alike: Improving Model Reliability via Inter-Model Latent Agreementcited by 11
  28. 2024Reconfidencing LLMs from the Grouping Loss Perspectivecited by 11
  29. 2024Divide, Reweight, and Conquer: A Logit Arithmetic Approach for In-Context Learningcited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 51; 2026: 24.

75 papers.

Sort:

By area:

  1. 2025Stop Overthinking: A Survey on Efficient Reasoning for Large Language Modelscited by 464
  2. 2025Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Modelscited by 384
  3. 2025R-Zero: Self-Evolving Reasoning LLM from Zero Datacited by 228
  4. 2025A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?cited by 223
  5. 2025General-Reasoner: Advancing LLM Reasoning Across All Domainscited by 135
  6. 2025A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, and Beyondcited by 132
  7. 2025Efficient Reasoning Models: A Surveycited by 95
  8. 2025Parallel-R1: Towards Parallel Thinking via Reinforcement Learningcited by 61
  9. 2025VisPlay: Self-Evolving Vision-Language Models from Imagescited by 47
  10. 2025Long-Short Chain-of-Thought Mixture Supervised Fine-Tuning Eliciting Efficient Reasoning in Large Language Modelscited by 39
  11. 2025Trust, But Verify: A Self-Verification Approach to Reinforcement Learning with Verifiable Rewardscited by 38
  12. 2025Harnessing the Reasoning Economy: A Survey of Efficient Reasoning for Large Language Modelscited by 36
38 more
  1. 2025Retro-Search: Exploring Untaken Paths for Deeper and Efficient Reasoningcited by 34
  2. 2025UniGen: Enhanced Training & Test-Time Strategies for Unified Multimodal Understanding and Generationcited by 32
  3. 2025Reasoning on a Budget: A Survey of Adaptive and Controllable Test-Time Compute in LLMscited by 31
  4. 2025The LLM Already Knows: Estimating LLM-Perceived Question Difficulty via Hidden Representationscited by 29
  5. 2025A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Modelscited by 25
  6. 2025Thought calibration: Efficient and confident test-time scalingcited by 21
  7. 2025LoVeC: Reinforcement Learning for Better Verbalized Confidence in Long-Form Generationscited by 17
  8. 2025Generative AI Act II: Test Time Scaling Drives Cognition Engineeringcited by 15
  9. 2025Guiding Reasoning in Small Language Models with LLM Assistancecited by 14
  10. 2025Can LLMs Predict Their Own Failures? Self-Awareness via Internal Circuitscited by 14
  11. 2026Parallel-Probe: Towards Efficient Parallel Thinking via 2D Probingcited by 14
  12. 2025Efficient Test-Time Scaling for Small Vision-Language Modelscited by 12
  13. 2025AALC: Large Language Model Efficient Reasoning via Adaptive Accuracy-Length Controlcited by 11
  14. 2025CrossWordBench: Evaluating the Reasoning Capabilities of LLMs and LVLMs with Controllable Puzzle Generationcited by 10
  15. 2026LLMs Improving LLMs: Agentic Discovery for Test-Time Scalingcited by 10
  16. 2025Revisiting Test-Time Scaling: A Survey and a Diversity-Aware Method for Efficient Reasoningcited by 9
  17. 2025Optimal Self-Consistency for Efficient Reasoning with Large Language Modelscited by 9
  18. 2026VLA-ATTC: Adaptive Test-Time Compute for VLA Models with Relative Action Critic Modelcited by 9
  19. 2025Guided by Gut: Efficient Test-Time Scaling with Reinforced Intrinsic Confidencecited by 8
  20. 2025How Far Are We from Optimal Reasoning Efficiency?cited by 8
  21. 2025Better LLM Reasoning via Dual-Playcited by 8
  22. 2026RelayLLM: Efficient Reasoning via Collaborative Decodingcited by 8
  23. 2025AdaDecode: Accelerating LLM Decoding with Adaptive Layer Parallelismcited by 7
  24. 2025Optimal Stopping vs Best-of-N for Inference Time Optimizationcited by 7
  25. 2025The Era of Agentic Organization: Learning to Organize with Language Modelscited by 7
  26. 2025SSR: Socratic Self-Refine for Large Language Model Reasoningcited by 7
  27. 2025REA-RL: Reflection-Aware Online Reinforcement Learning for Efficient Reasoningcited by 6
  28. 2025CGES: Confidence-Guided Early Stopping for Efficient and Accurate Self-Consistencycited by 6
  29. 2025Zero-Overhead Introspection for Adaptive Test-Time Computecited by 6
  30. 2026NOVA: NOise-aware Verbal Confidence CAlibration for Robust Large Language Models in RAG Systemscited by 6
  31. 2025BrowseConf: Confidence-Guided Test-Time Scaling for Web Agentscited by 5
  32. 2025Test-time Verification via Optimal Transport: Coverage, ROC, & Sub-optimalitycited by 4
  33. 2025CritiCal: Can Critique Help LLM Uncertainty or Confidence Calibration?cited by 4
  34. 2026Online Reasoning Calibration: Test-Time Training Enables Generalizable Conformal LLM Reasoningcited by 4
  35. 2025Generalizable LLM Learning of Graph Synthetic Data with Post-training Alignmentcited by 3
  36. 2025GrACE: A Generative Approach to Better Confidence Elicitation and Efficient Test-Time Scaling in Large Language Modelscited by 3
  37. 2025ATTS: Asynchronous Test-Time Scaling via Conformal Predictioncited by 3
  38. 2025MARINE: Theoretical Optimization and Design for Multi-Agent Recursive IN-context Enhancementcited by 3

50 most cited of 75 All 75 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026Scaling with Confidence: Calibrating Confidence of LLMs for Adaptive Test Time Scaling
  2. 2024Calibrating Long-form Generations from Large Language Models
  3. 2026ConfidenceBench: Evaluating Confidence Calibration in Large Language Models
  4. 2025Sample Complexity and Representation Ability of Test-time Scaling Paradigms
  5. 2025A Theoretical Study on Bridging Internal Probability and Self-Consistency for LLM Reasoning
  6. 2024Calibrating the Confidence of Large Language Models by Eliciting Fidelity
  7. 2025Seer Self-Consistency: Advance Budget Estimation for Adaptive Test-Time Scaling
  8. 2025Is That Your Final Answer? Test-Time Scaling Improves Selective Question Answering