mlsanityData to 7 Oct 2026

Confidence v.s. Critique: A Decomposition of Self-Correction Capability for LLMs

arXiv 2412.19513 ↗First posted 27 Dec 2024cs.CLcited by 28

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

Abstract (arXiv)

Large Language Models (LLMs) can correct their self-generated responses, but a decline in accuracy after self-correction is also witnessed. To have a deeper understanding of self-correction, we endeavor to decompose, evaluate, and analyze the self-correction behaviors of LLMs. By enumerating and analyzing answer correctness before and after self-correction, we decompose the self-correction capability into confidence (being confident to correct answers) and critique (turning wrong answers to correct) capabilities, and propose two metrics from a probabilistic perspective to measure these 2 capabilities, along with another metric for overall self-correction capability evaluation. Based on our decomposition and evaluation metrics, we conduct extensive experiments and draw some empirical conclusions. For example, we find different models can exhibit distinct behaviors: some models are confident while others are more critical. We also find the trade-off between the two capabilities (i.e. improving one can lead to a decline in the other) when manipulating model self-correction behavior by prompts or in-context learning. Further, we find a simple yet efficient strategy to improve self-correction capability by transforming Supervision Fine-Tuning (SFT) data format, and our strategy outperforms vanilla SFT in both capabilities and achieves much higher accuracy after self-correction. Our code will be publicly available on GitHub.

Key facts. Cited by 28 papers in the map (data through 7 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 31 of its 48 references that are in the map. First posted on arXiv 27 Dec 2024.

Builds on

31 of 48 references.

Closest ancestors · 7

Same direction, newest first

  1. 2024Training Language Models to Self-Correct via Reinforcement Learning19 Sep
  2. 2024S^3cMath: Spontaneous Step-level Self-correction Makes Large Language Models Better Mathematical Reasoners3 Sep
  3. 2024Large Language Models have Intrinsic Self-Correction Ability21 Jun
  4. 2024Small Language Models Need Strong Verifiers to Self-Correct Reasoning26 Apr
  5. 2024Self-Contrast: Better Reflection Through Inconsistent Solving Perspectives4 Jan
  6. 2023Large Language Models Cannot Self-Correct Reasoning Yet3 Oct
  7. 2023Self-Refine: Iterative Refinement with Self-Feedback30 Mar

Other references · 24

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2023GPT-4 Technical Reportcited by 18,132
  3. 2022Training language models to follow instructions with human feedbackcited by 15,712
  4. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  5. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  6. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  7. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  8. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  9. 2021Finetuned Language Models Are Zero-Shot Learnerscited by 3,747
  10. 2023Qwen Technical Reportcited by 3,273
  11. 2019BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questionscited by 2,140
  12. 2018CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledgecited by 1,793
12 more
  1. 2024ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Toolscited by 1,096
  2. 2023Teaching Large Language Models to Self-Debugcited by 963
  3. 2022Generating Sequences by Learning to Self-Correctcited by 271
  4. 2023A Survey of Confidence Estimation and Calibration in Large Language Modelscited by 253
  5. 2024When Can LLMs Actually Correct Their Own Mistakes? A Critical Survey of Self-Correction of LLMscited by 230
  6. 2023Can Large Language Models Really Improve by Self-critiquing Their Own Plans?cited by 107
  7. 2024Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedbackcited by 107
  8. 2023GPT-4 Doesn't Know It's Wrong: An Analysis of Iterative Prompting for Reasoning Problemscited by 88
  9. 2023Self-Polish: Enhance Reasoning in Large Language Models via Problem Refinementcited by 43
  10. 2024Large Language Models Can Self-Correct with Key Condition Verificationcited by 39
  11. 2024Small Language Model Can Self-correctcited by 23
  12. 2024Can Large Language Models Always Solve Easy Problems if They Can Solve Harder Ones?cited by 9

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 10; 2026: 18.

28 papers.

Sort:

By area:

  1. 2025Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Modelscited by 381
  2. 2025Table-Critic: A Multi-Agent Framework for Collaborative Criticism and Refinement in Table Reasoningcited by 32
  3. 2025LEMMA: Learning from Errors for MatheMatical Advancement in LLMscited by 20
  4. 2025Process or Result? Manipulated Ending Tokens Can Mislead Reasoning LLMs to Ignore the Correct Reasoning Stepscited by 9
  5. 2025A Probabilistic Inference Scaling Theory for LLM Self-Correctioncited by 9
  6. 2026Rethinking Table Pruning in TableQA: From Sequential Revisions to Gold Trajectory-Supervised Parallel Searchcited by 8
  7. 2025When Do LLMs Admit Their Mistakes? Understanding The Role Of Model Belief In Retractioncited by 6
  8. 2025Entropy-Based Adaptive Weighting for Self-Trainingcited by 5
  9. 2025CritiCal: Can Critique Help LLM Uncertainty or Confidence Calibration?cited by 4
  10. 2026When Small Models Are Right for Wrong Reasons: Process Verification for Trustworthy Agentscited by 3
  11. 2026How LLMs Detect and Correct Their Own Errors: The Role of Internal Confidence Signalscited by 3
  12. 2026When Does Intrinsic Self-Correction Help? A Task-Sensitive Analysiscited by 3
16 more
  1. 2025RETuning: Upgrading Inference-Time Scaling for Stock Movement Prediction with Large Language Modelscited by 1
  2. 2026iFlip: Iterative Feedback-driven Counterfactual Example Refinementcited by 1
  3. 2026Beyond In-Domain Detection: SpikeScore for Cross-Domain Hallucination Detectioncited by 1
  4. 2026Towards Better RL Training Data Utilization via Second-Order Rolloutcited by 1
  5. 2026LongSumEval: Question-Answering Based Evaluation and Feedback-Driven Refinement for Long Document Summarizationcited by 1
  6. 2026Precise but Uncoupled: Reviewer Precision Does Not Guarantee Critique Uptake in Multi-Agent Math Reasoningcited by 1
  7. 2025Error Notebook-Guided, Training-Free Part Retrieval in 3D CAD Assemblies via Vision-Language Modelscited by 0
  8. 2026Cognitive Loop of Thought: Reversible Hierarchical Markov Chain for Efficient Mathematical Reasoningcited by 0
  9. 2026Correction and Corruption: A Two-Rate View of Error Flow in LLM Protocolscited by 0
  10. 2026SPIN: Structural LLM Planning via Iterative Navigation for Industrial Taskscited by 0
  11. 2026Identifying High-Confidence Social Biases in LLMs for Trustworthy Conversational Tutoring Agentscited by 0
  12. 2026Confidence Composition for Multiagent Language Model Systemscited by 0
  13. 2026Recompute or Reuse? Diagnosing and Mitigating Textual Shortcuts in VLM Self-Reflectioncited by 0
  14. 2026Factorized Hypothesis Search for Evidence-to-Taxonomy Retrievalcited by 0
  15. 2026Prior Audit-Repair Context Shifts LLM Verifier Thresholds Toward Leniencycited by 0
  16. 2026When Should LLMs Trust Their Own Revisions? A Risk-Aware Study of Intrinsic Self-Correctioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Can LLMs Correct Themselves? A Benchmark of Self-Correction in LLMs
  2. 2024Understanding the Dark Side of LLMs' Intrinsic Self-Correction
  3. 2025High Accuracy, Less Talk (HALT): Reliable LLMs through Capability-Aligned Finetuning
  4. 2023Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategies
  5. 2026LLMs Should Express Uncertainty Explicitly
  6. 2024A Theoretical Understanding of Self-Correction through In-context Alignment
  7. 2023LLMs cannot find reasoning errors, but can correct them given the error location
  8. 2025Self-Correction Bench: Uncovering and Addressing the Self-Correction Blind Spot in Large Language Models