mlsanityData to 7 Oct 2026

Large Language Models Can Self-Correct with Key Condition Verification

arXiv 2405.14092 ↗First posted 23 May 2024cs.CLcited by 39

Not on the map. Why?

Abstract (arXiv)

Intrinsic self-correct was a method that instructed large language models (LLMs) to verify and correct their responses without external feedback. Unfortunately, the study concluded that the LLMs could not self-correct reasoning yet. We find that a simple yet effective verification method can unleash inherent capabilities of the LLMs. That is to mask a key condition in the question, add the current response to construct a verification question, and predict the condition to verify the response. The condition can be an entity in an open-domain question or a numeric value in a math question, which requires minimal effort (via prompting) to identify. We propose an iterative verify-then-correct framework to progressively identify and correct (probably) false responses, named ProCo. We conduct experiments on three reasoning tasks. On average, ProCo, with GPT-3.5-Turbo as the backend LLM, yields +6.8 exact match on four open-domain question answering datasets, +14.1 accuracy on three arithmetic reasoning datasets, and +9.6 accuracy on a commonsense reasoning dataset, compared to Self-Correct. Our implementation is made publicly available at https://wzy6642.github.io/proco.github.io/.

Key facts. Cited by 39 papers in the map (data through 7 Oct 2026). Builds on 28 of its 40 references that are in the map. First posted on arXiv 23 May 2024.

Builds on

28 of 40 references.

References · 28

  1. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  3. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  4. 2022Large Language Models are Zero-Shot Reasonerscited by 4,323
  5. 2018HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answeringcited by 3,686
  6. 2023Let's Verify Step by Stepcited by 3,598
  7. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
  8. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,895
  9. 2021SimCSE: Simple Contrastive Learning of Sentence Embeddingscited by 2,168
  10. 2020DeBERTa: Decoding-enhanced BERT with Disentangled Attentioncited by 2,157
  11. 2018CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledgecited by 1,793
  12. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 965
16 more
  1. 2023Teaching Large Language Models to Self-Debugcited by 963
  2. 2023CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquingcited by 751
  3. 2017Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problemscited by 642
  4. 2023Language Models can Solve Computer Taskscited by 371
  5. 2022Demonstrate-Search-Predict: Composing retrieval and language models for knowledge-intensive NLPcited by 316
  6. 2022Generate rather than Retrieve: Large Language Models are Strong Context Generatorscited by 296
  7. 2022Generating Sequences by Learning to Self-Correctcited by 271
  8. 2023FacTool: Factuality Detection in Generative AI -- A Tool Augmented Framework for Multi-Task and Multi-Domain Scenarioscited by 247
  9. 2022Making Large Language Models Better Reasoners with Step-Aware Verifiercited by 231
  10. 2023SelfCheck: Using LLMs to Zero-Shot Check Their Own Step-by-Step Reasoningcited by 157
  11. 2023Improving Language Models via Plug-and-Play Retrieval Feedbackcited by 87
  12. 2022Rainier: Reinforced Knowledge Introspector for Commonsense Question Answeringcited by 43
  13. 2023Self-prompted Chain-of-Thought on Large Language Models for Open-domain Multi-hop Reasoningcited by 21
  14. 2021Interscript: A dataset for interactive learning of scripts through error feedbackcited by 9
  15. 2021Validating Label Consistency in NER Data Annotationcited by 7
  16. 2023Get an A in Math: Progressive Rectification Promptingcited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 6; 2025: 19; 2026: 14.

39 papers.

Sort:

By area:

  1. 2024Mutual Reasoning Makes Smaller LLMs Stronger Problem-Solverscited by 137
  2. 2024Make Every Penny Count: Difficulty-Adaptive Self-Consistency for Cost-Efficient Reasoningcited by 61
  3. 2025Can Large Language Models Detect Errors in Long Chain-of-Thought Reasoning?cited by 43
  4. 2024Understanding the Dark Side of LLMs' Intrinsic Self-Correctioncited by 42
  5. 2024Confidence v.s. Critique: A Decomposition of Self-Correction Capability for LLMscited by 28
  6. 2024Critic-CoT: Boosting the reasoning abilities of large language model via Chain-of-thoughts Criticcited by 27
  7. 2025LEMMA: Learning from Errors for MatheMatical Advancement in LLMscited by 20
  8. 2025Zoom-Refine: Boosting High-Resolution Multimodal Understanding via Localized Zoom and Self-Refinementcited by 18
  9. 2025Beyond 'Aha!': Toward Systematic Meta-Abilities Alignment in Large Reasoning Modelscited by 17
  10. 2024Learn Beyond The Answer: Training Language Models with Reflection for Mathematical Reasoningcited by 15
  11. 2025ProgCo: Program Helps Self-Correction of Large Language Modelscited by 14
  12. 2025A Cognitive Writing Perspective for Constrained Long-Form Text Generationcited by 14
27 more
  1. 2025RealCritic: Towards Effectiveness-Driven Evaluation of Language Model Critiquescited by 12
  2. 2025Credible Plan-Driven RAG Method for Multi-Hop Question Answeringcited by 12
  3. 2025Failure Makes the Agent Stronger: Enhancing Accuracy through Structured Reflection for Reliable Tool Interactionscited by 12
  4. 2025Feedback Friction: LLMs Struggle to Fully Incorporate External Feedbackcited by 11
  5. 2025When Does Verification Pay Off? A Closer Look at LLMs as Solution Verifierscited by 11
  6. 2026Accurate Failure Prediction in Agents Does Not Imply Effective Failure Preventioncited by 11
  7. 2025VerifiAgent: a Unified Verification Agent in Language Model Reasoningcited by 10
  8. 2025Controlling Thinking Speed in Reasoning Modelscited by 10
  9. 2025A Probabilistic Inference Scaling Theory for LLM Self-Correctioncited by 9
  10. 2026SciCoQA: Quality Assurance for Scientific Paper--Code Alignmentcited by 9
  11. 2025AgentAsk: Multi-Agent Systems Need to Askcited by 6
  12. 2026ProMoral-Bench: Evaluating Prompting Strategies for Moral Reasoning and Safety in LLMscited by 4
  13. 2025From Misleading Queries to Accurate Answers: A Three-Stage Fine-Tuning Method for LLMscited by 2
  14. 2026Finding the Cracks: Improving LLMs Reasoning with Paraphrastic Probing and Consistency Verificationcited by 2
  15. 2025Beyond Prompting: An Efficient Embedding Framework for Open-Domain Question Answeringcited by 1
  16. 2025MM-SeR: Multimodal Self-Refinement for Lightweight Image Captioningcited by 1
  17. 2026Reflect: Transparent Principle-Guided Reasoning for Constitutional Alignment at Scalecited by 1
  18. 2026Who Flips? Self- and Cross-Model Counterarguments Reveal Answer Instability in LLMscited by 1
  19. 2026The Horizon Gap: Planning, Memory, Execution, Training, and Evaluation for Long-Horizon LLM Agentscited by 1
  20. 2025TaskEval: Synthesised Evaluation for Foundation-Model Taskscited by 0
  21. 2026Self-Correction Inside the Model: Leveraging Layer Attention to Mitigate Hallucinations in Large Vision Language Modelscited by 0
  22. 2026A Training-Free Regeneration Paradigm: Contrastive Reflection Memory Guided Self-Verification and Self-Improvementcited by 0
  23. 2026AutoVSR: Automatic Visual-to-Symbolic Reasoning for Symbolic Expression Generation from Circuit Schematiccited by 0
  24. 2026Accuracy Hides How Language Models Fail: Measuring Failure States Under Matched Output Budgetscited by 0
  25. 2026LLM-based Framework for Generating and Verifying Parallel DEVS Statechartscited by 0
  26. 2026The Error You See Is Not the Error You Made: Progression-aware Reasoning Origin for Reasoning Error Localizationcited by 0
  27. 2026How AI Agents Discover Scientific Equations: From Hydrotope Rediscovery to New Water-Wave Amplitudescited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Self-Verified Distillation: Your Language Model Is Secretly Its Own Synthetic Data Pipeline
  2. 2024Small Language Models Need Strong Verifiers to Self-Correct Reasoning
  3. 2022Large Language Models are Better Reasoners with Self-Verification
  4. 2025Incentivizing LLMs to Self-Verify Their Answers
  5. 2022Large Language Models Can Self-Improve
  6. 2025The Self-Improvement Paradox: Can Language Models Bootstrap Reasoning Capabilities without External Scaffolding?
  7. 2025Can LLMs Correct Themselves? A Benchmark of Self-Correction in LLMs
  8. 2024Embedding Self-Correction as an Inherent Ability in Large Language Models for Enhanced Mathematical Reasoning