LLM Self-Correction

LLM: Large Language Model

Follow this topic Ask about this topic

Momentum

8 papers in the last four weeks, up 167% on the four weeks before. 0.1% of all new papers.

Jul 13Week of Sep 28

Latest papers 105

All topics

Structured Feedback Improves Repair in an LLM Agent Loop
Jul 15, 2026Jaideep Ray, Ankit GoyalLLM Self-Correction LLM Agent Evaluation
Independent Researcher
Form, Not Content? A Preregistered, Placebo-Controlled Evaluation of Learned Error-Conditioned Self-Repair Through Prompts and Weights in Frozen Small Code Models
Jul 14, 2026Mehmet IscanLLM Self-Correction Automated Program Repair
PythaLab, Yıldız Technical University, Istanbul, Turkey
Self-Improving AI Coding Agents Through Accumulated Behavioral Rules: A Closed-Loop Framework
Jul 13, 2026Aditya Aggarwal, Nahid Farhady GhalatyLLM Self-Correction Continual Learning for LLM Agents
Microsoft
Spectral Origins of the Self-Correction Blind Spot in Autoregressive Generation
Jul 9, 2026Ingrid Petrova, Luan VejsiuLLM Self-Correction RL for Language Models
European University of Tirana
Search, Fail, Recover: A Training Framework for Correction-Aware Reasoning
Jul 8, 2026Dmitry Beresnev, Vladimir Makharev, Roman Khalikov +2LLM Self-Correction Inference-Time Search
Innopolis University, Innopolis, Russia · AXXX, Moscow, Russia · Lomonosov Moscow State University, Moscow, Russia
When Does In-Context Search Help? A Sampling-Complexity Theory of Reflection-Driven Reasoning
Jul 7, 2026Yotam Wolf, Noam Wies, Amnon ShashuaLLM Self-Correction Inference-Time Search
The Hebrew University
Reason, Reward, Refine: Step-Level Errors Corrections with Structured Feedback for Physics Reasoning in Small Language Models
Jul 6, 2026Raj Jaiswal, Dhruv Jain, Rishabh Dhawan +4LLM Self-Correction Physical Reasoning
IIIT Delhi · IIT BHU · NII Tokyo +2
Rethinking On-Policy Self-Distillation for Thinking Models
Jul 6, 2026Simran Kaur, Narutatsu Ri, Yinghui He +2LLM Self-Correction On-Policy Self-Distillation
Princeton Language and Intelligence, Princeton University
Obey, Diverge, Collapse: Blind Obedience to Incorrect Instructions Drives Code LLMs to Irrecoverable Code Semantic Collapse
Jul 5, 2026Raj Jaiswal, Anany Singh Divy, Savar Bhasin +3LLM Self-Correction Automated Program Repair
IIIT Delhi · Microsoft Research India · IIT Kanpur
ESC: Emotional Self-Correction for Reliable Vision-Language Models
Jul 1, 2026Tien-Huy Nguyen, Minh-Nhat Nguyen, Nguyen Nhat Huy +9LLM Self-Correction VLM Reasoning
GenAI4E Lab · University of Information Technology, Ho Chi Minh City, Vietnam · Vietnam National University, Ho Chi Minh City, Vietnam +10
Know When to Stop: Segment-Level Credit Assignment for Reducing Overthinking
Jul 1, 2026Chia-Hsuan Lee, Sihui Dai, Mingyang Zhou +4LLM Self-Correction Overthinking in Language Models
Capital One
Falsification, Not Exposure: An Internally Preregistered Placebo-Controlled Decomposition of Self-Repair Feedback in Frozen Small Code Models
Jun 30, 2026Mehmet IscanLLM Self-Correction Automated Program Repair
PythaLab, Yıldız Technical University, Istanbul, Turkey
One Reflection Is Not Enough: Self-Correcting Autonomous Research via Multi-Hypothesis Failure Attribution
Jun 30, 2026Jie Ma, Binfei Chu, Jie Gao +6Agent Failure Analysis LLM Self-Correction
Xiamen University · Hello Inc. · Xiaohongshu Inc.
How Far Do On-Prem Open LLMs Get on Text-to-SQL? A Cross-Family Size x Technique Frontier on BIRD
Jun 29, 2026Vladimir BeskorovainyiLLM Self-Correction LLM Evaluation
Besk Tech · Moscow Institute of Physics and Technology (MIPT)
The Signal-Coverage Matrix: Stratifying Type and Semantic Errors in Statement Autoformalization
Jun 26, 2026Chengxiao Dai, Zhaokun Yan, Zhanhui LinLLM Self-Correction Lean Theorem Proving
School of Computer Science, University of Sydney, Sydney, Australia · China Academy of Information and Communications Technology, China
Reflect-R1: Evidence-Driven Reflection for Self-Correction in Long Video Understanding
Jun 26, 2026Shuimu Chen, Yuteng Chen, Yuanshen Guan +7LLM Self-Correction Long-Video Understanding
Tsinghua University · Nanyang Technological University · University of Science and Technology of China +4
When Does Intrinsic Self-Correction Help? A Task-Sensitive Analysis
Jun 22, 2026Elroy Stav, Dvir Berlowitz, Maayan Orner +1LLM Self-Correction LLM Evaluation
Department of Computer Science and Artificial Intelligence, Bar-Ilan University · Independent Researcher
Denoising Iterative Self-Correction: Structured Verification Loops for Reliable LLM Reasoning
Jun 19, 2026Shen Yin, David Ken, Joel StremmelLLM Self-Correction LLM Self-Refinement
Thomson Reuters Labs
Self-Preference Is Weak or Absent in Verifiable Instruction-Following Revision: A Four-Model Test Under Genuine Authorship
Jun 18, 2026William Guey, Pierrick BougaultLLM Self-Correction Language Model Bias Evaluation
Department of Industrial Engineering, Tsinghua University, Beijing, China
CREDENCE: Claim Reduction for Decomposition & Enhanced Credibility -- Semantic Metrics and Convergence Analysis
Jun 18, 2026Phuong Huu Vu Tran, Thuan Duc Mai, Bach Xuan LeLLM Self-Correction Automated Fact-Checking
Vietnamese-German University, Ho Chi Minh City, Vietnam · Ho Chi Minh University of Technology, Ho Chi Minh City, Vietnam
Emergent Alignment
Jun 17, 2026Martin KolářLLM Self-Correction LLM Alignment
CIIRC, Czech Technical University in Prague, Prague, Czech Republic.
Correct Yourself, Keep My Trust: How Self-Correction and Social Connection Shape Credibility in Social Chatbots
Jun 17, 2026Biswadeep Sen, Yi-Chieh LeeLLM Self-Correction Human-AI Interaction
School of Computing National University of Singapore Singapore, Singapore
REVES: REvision and VErification--Augmented Training for Test-Time Scaling
Jun 17, 2026Yuanxin Liu, Ruida Zhou, Xinyan Zhao +6LLM Self-Correction LLM Answer Verification
Northwestern University · Amazon AGI · Qualcomm AI Research +1
Learning from Your Own Mistakes: Constructing Learnable Micro-Reflective Trajectories for Self-Distillation
Jun 17, 2026Zhilin Huang, Hang Gao, Ziqiang Dong +6LLM Self-Correction CoT Distillation
Qwen Business Unit of Alibaba · Tsinghua University · Peking University
Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs
Jun 16, 2026Ziliang Wang, Kang An, Faqiang Qian +5LLM Self-Correction Reinforcement Learning
1SenseTime · 2Shanghai Jiao Tong University
Unlocking LLM Code Correction with Iterative Feedback Loops
Jun 16, 2026Le Zhang, Suresh KothariLLM Self-Correction Execution-Guided Code Generation
Iowa State University, Ames Iowa 50010, USA
Self-Generated Error Training for Token Editing in Diffusion Language Models
Jun 15, 2026Lin YaoLLM Self-Correction Diffusion Model Fine-Tuning
School of Computer Science, Shanghai Jiao Tong University, Shanghai, 200240, China · Zhongguancun Academy, Beijing, 100097, China
Can Editing 1 Neuron Fix Repetition Loops in LLMs?
Jun 9, 2026Aristotelis Lazaridis, Aman Sharma, Dylan Bates +3LLM Self-Correction Mechanistic Interpretability
Edgerunner AI
More Yap Less Meaning: Uncovering Self-Improvement Behavior in SLMs
Jun 7, 2026Marina Igitkhanian, Erik ArakelyanLLM Self-Correction Test-Time Scaling
American University of Armenia · NVIDIA
The Self-Correction Illusion: LLMs Correct Others but Not Themselves
Jun 4, 2026Kuan-Yen Chen, Fang-Yi Su, Jung-Hsien ChiangLLM Self-Correction LLM Prompting
National Cheng Kung University

Jul 15, 2026, cs.SEJ/K move · Enter open · S save

Structured Feedback Improves Repair in an LLM Agent Loop

Jaideep Ray, Ankit Goyal

Independent Researcher

LLM Self-Correction LLM Agent Evaluation LLM Agent Workflow Optimization LLM Agent Verification LLM Agent Harnesses LLM Agent Reliability

Open paper AskPDF arXiv

LLM agents often retry after external validation rejects a candidate, but the interface between validation and the next model call remains underspecified. We introduce VeriHarness, a code-controlled agent loop in which models generate candidates while external validators control acceptance, budgets, and traces. We use it to compare raw diagnostics with feedback that identifies the failure location, observed value, and admissible alternatives. Across 50 paired TextWorld games under a four-call cap, feedback containing all three fields raises terminal success from 14/50 to 36/50 for Qwen2.5-Coder-14B (+44 percentage points) and from 8/50 to 29/50 for Llama-3.1-8B (+42 points). Ablations locate most of the gain in the admissible alternatives: feedback containing only the location and observed value remains near the raw diagnostic baseline. Presenting the complete repair information in prose instead of a keyed JSON record yields nearly the same success, providing no evidence that JSON syntax itself improves repair. The ordering persists across the tested call budgets and one sampled-decoding setting.

Momentum

Related topics

Latest papers 105