mlsanityData to 14 Oct 2026

Calibrated Reasoning: An Explanatory Verifier for Dynamic and Efficient Problem-Solving

arXiv 2509.19681 ↗First posted 24 Sep 2025cs.AIcited by 1

On the map: LLM reasoning → Reinforcement learning for reasoning

Abstract (arXiv)

Advanced test-time computing strategies are essential for scaling reasoning models, but their effectiveness is capped by the models' poor self-evaluation. We propose a pairwise Explanatory Verifier, trained via reinforcement learning (GRPO), that produces calibrated confidence scores and associated natural language reasoning for generated solutions. Our verifier improves the accuracy and efficiency of test-time strategies like best-of-n and self-reflection. Crucially, it excels at identifying challenging failure modes, such as when both candidate solutions are identically incorrect, succeeding where standard methods like majority voting fail.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Reinforcement learning for reasoning, one of 17 directions in LLM reasoning. Builds on 20 of its 26 references that are in the map. First posted on arXiv 24 Sep 2025.

Builds on

20 of 26 references.

Closest ancestors · 7

Same direction, newest first

  1. 2025VerifyBench: A Systematic Benchmark for Evaluating Reasoning Verifiers Across Domains14 Jul
  2. 2025AceReason-Nemotron: Advancing Math and Code Reasoning through Reinforcement Learning22 May
  3. 2025TinyV: Reducing False Negatives in Verification Improves RL for LLM Reasoning20 May
  4. 2025Putting the Value Back in RL: Better Test-Time Scaling by Unifying LLM Reasoners With Verifiers7 May
  5. 2025Reinforcement Learning for Reasoning in Large Language Models with One Training Example29 Apr
  6. 2025Inference-Time Scaling for Generalist Reward Modeling3 Apr
  7. 2025Process Reinforcement through Implicit Rewards3 Feb

Other references · 13

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  3. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,442
  4. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,986
  5. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 1,017
  6. 2024Quiet-STaR: Language Models Can Teach Themselves to Think Before Speakingcited by 301
  7. 2025Skywork Open Reasoner 1 Technical Reportcited by 165
  8. 2023TACO: Topics in Algorithmic COde generation datasetcited by 160
  9. 2024RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learningcited by 151
  10. 2025LeetCodeDataset: A Temporal Dataset for Robust Evaluation and Efficient Training of Code LLMscited by 70
  11. 2025xVerify: Efficient Answer Verifier for Reasoning Model Evaluationscited by 55
  12. 2024Search, Verify and Feedback: Towards Next Generation Post-training Paradigm of Foundation Models via Verifier Engineeringcited by 16
1 more
  1. 2025Scaling Flaws of Verifier-Guided Search in Mathematical Reasoningcited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026C-Instrument: Automating RL Data Generation and Hillclimbing with a Constitution-Grid Instrumentcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Decoupling Reasoning and Confidence: Resurrecting Calibration in Reinforcement Learning from Verifiable Rewards
  2. 2025SHARP: Synthesizing High-quality Aligned Reasoning Problems for Large Reasoning Models Reinforcement Learning
  3. 2026Learning to Correct: Calibrated Reinforcement Learning for Multi-Attempt Chain-of-Thought
  4. 2025Critique-GRPO: Advancing LLM Reasoning with Natural Language and Numerical Feedback
  5. 2026The Easy, the Hard, and the Learnable: Confidence and Difficulty-Adaptive Policy Optimization for LLM Reasoning
  6. 2025ConciseRL: Conciseness-Guided Reinforcement Learning for Efficient Reasoning Models
  7. 2025Reinforcing General Reasoning without Verifiers
  8. 2025From Solving to Verifying: A Unified Objective for Robust Reasoning in LLMs