# LeDex: Training LLMs to Better Self-Debug and Explain Code

arXiv: https://arxiv.org/abs/2405.18649 · HTML: https://mlsanity.com/paper/2405.18649/

## Key facts

- Cited by 40 papers in the map (data through 14 Oct 2026).
- Builds on 28 of its 42 references that are in the map.
- First posted on arXiv 28 May 2024.

## Builds on (most cited first, 28 of 28)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [Competition-Level Code Generation with AlphaCode](https://mlsanity.com/paper/2203.07814/) (2022)
- [Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation](https://mlsanity.com/paper/2305.01210/) (2023)
- [Teaching Large Language Models to Self-Debug](https://mlsanity.com/paper/2304.05128/) (2023)
- [CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation](https://mlsanity.com/paper/2109.00859/) (2021)
- [Measuring Coding Challenge Competence With APPS](https://mlsanity.com/paper/2105.09938/) (2021)
- [Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation](https://mlsanity.com/paper/2004.09813/) (2020)
- [CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning](https://mlsanity.com/paper/2207.01780/) (2022)
- [CodeT5+: Open Code Large Language Models for Code Understanding and Generation](https://mlsanity.com/paper/2305.07922/) (2023)
- [OpenCodeInterpreter: Integrating Code Generation with Execution and Refinement](https://mlsanity.com/paper/2402.14658/) (2024)
- [Practical Program Repair in the Era of Large Pre-trained Language Models](https://mlsanity.com/paper/2210.14179/) (2022)
- [CodeGen2: Lessons for Training LLMs on Programming and Natural Languages](https://mlsanity.com/paper/2305.02309/) (2023)
- [Large Language Models are Few-shot Testers: Exploring LLM-based General Bug Reproduction](https://mlsanity.com/paper/2209.11515/) (2022)
- [Self-Edit: Fault-Aware Code Editor for Code Generation](https://mlsanity.com/paper/2305.04087/) (2023)
- [Large Language Models are Zero-Shot Fuzzers: Fuzzing Deep-Learning Libraries via Large Language Models](https://mlsanity.com/paper/2212.14834/) (2022)
- [Execution-based Code Generation using Deep Reinforcement Learning](https://mlsanity.com/paper/2301.13816/) (2023)
- [Impact of Code Language Models on Automated Program Repair](https://mlsanity.com/paper/2302.05020/) (2023)
- [Improving Code Generation by Training with Natural Language Feedback](https://mlsanity.com/paper/2303.16749/) (2023)
- [How Effective Are Neural Networks for Fixing Security Vulnerabilities](https://mlsanity.com/paper/2305.18607/) (2023)
- [Training Language Models with Language Feedback](https://mlsanity.com/paper/2204.14146/) (2022)
- [CYCLE: Learning to Self-Refine the Code Generation](https://mlsanity.com/paper/2403.18746/) (2024)
- [A Deep Dive into Large Language Models for Automated Bug Localization and Repair](https://mlsanity.com/paper/2404.11595/) (2024)
- [TOGLL: Correct and Strong Test Oracle Generation with LLMs](https://mlsanity.com/paper/2405.03786/) (2024)
- [Leveraging Print Debugging to Improve Code Generation in Large Language Models](https://mlsanity.com/paper/2401.05319/) (2024)

## Built on by (most cited first, 30 of 40)

- [Enhancing Code LLMs with Reinforcement Learning in Code Generation: A Survey](https://mlsanity.com/paper/2412.20367/) (2024)
- [Hierarchy-of-Groups Policy Optimization for Long-Horizon Agentic Tasks](https://mlsanity.com/paper/2602.22817/) (2026)
- [From Code to Correctness: Closing the Last Mile of Code Generation with Hierarchical Debugging](https://mlsanity.com/paper/2410.01215/) (2024)
- [Vibe Coding vs. Agentic Coding: Fundamentals and Practical Implications of Agentic AI](https://mlsanity.com/paper/2505.19443/) (2025)
- [Internet of Agents: Fundamentals, Applications, and Challenges](https://mlsanity.com/paper/2505.07176/) (2025)
- [AlgoTune: Can Language Models Speed Up General-Purpose Numerical Programs?](https://mlsanity.com/paper/2507.15887/) (2025)
- [Advancing Tool-Augmented Large Language Models via Meta-Verification and Reflection Learning](https://mlsanity.com/paper/2506.04625/) (2025)
- [Revisit Self-Debugging with Self-Generated Tests for Code Generation](https://mlsanity.com/paper/2501.12793/) (2025)
- [An LLM Agent for Automatic Geospatial Data Analysis](https://mlsanity.com/paper/2410.18792/) (2024)
- [VisCoder: Fine-Tuning LLMs for Executable Python Visualization Code Generation](https://mlsanity.com/paper/2506.03930/) (2025)
- [R4ec: A Reasoning, Reflection, and Refinement Framework for Recommendation Systems](https://mlsanity.com/paper/2507.17249/) (2025)
- [RefineBench: Evaluating Refinement Capability of Language Models via Checklists](https://mlsanity.com/paper/2511.22173/) (2025)
- [RepoST: Scalable Repository-Level Coding Environment Construction with Sandbox Testing](https://mlsanity.com/paper/2503.07358/) (2025)
- [Multi-Programming Language Sandbox for LLMs](https://mlsanity.com/paper/2410.23074/) (2024)
- [LATTE: Improving Latex Recognition for Tables and Formulae with Iterative Refinement](https://mlsanity.com/paper/2409.14201/) (2024)
- [ReflexiCoder: Teaching Large Language Models to Self-Reflect on Generated Code and Self-Correct It via Reinforcement Learning](https://mlsanity.com/paper/2603.05863/) (2026)
- [PatchWorld: Gradient-Free Optimization of Executable World Models for Agent Environments](https://mlsanity.com/paper/2605.30880/) (2026)
- [CodeEvo: Interaction-Driven Synthesis of Code-centric Data through Hybrid and Iterative Feedback](https://mlsanity.com/paper/2507.22080/) (2025)
- [How Many Tries Does It Take? Iterative Self-Repair in LLM Code Generation Across Model Scales and Benchmarks](https://mlsanity.com/paper/2604.10508/) (2026)
- [NL-Debugging: Exploiting Natural Language as an Intermediate Representation for Code Debugging](https://mlsanity.com/paper/2505.15356/) (2025)
- [Code Reasoning for Software Engineering Tasks: A Survey and A Call to Action](https://mlsanity.com/paper/2506.13932/) (2025)
- [InCoder-32B-Thinking: Industrial Code World Model for Thinking](https://mlsanity.com/paper/2604.03144/) (2026)
- [Falsification, Not Exposure: An Internally Preregistered Placebo-Controlled Decomposition of Self-Repair Feedback in Frozen Small Code Models](https://mlsanity.com/paper/2606.31511/) (2026)
- [Agentic-R1: Distilled Dual-Strategy Reasoning](https://mlsanity.com/paper/2507.05707/) (2025)
- [MapCoder-Lite: Distilling Multi-Agent Coding into a Single Small LLM](https://mlsanity.com/paper/2509.17489/) (2025)
- [VDebugger: Harnessing Execution Feedback for Debugging Visual Programs](https://mlsanity.com/paper/2406.13444/) (2024)
- [To Trade or Not to Trade: An Agentic Approach to Estimating Market Risk Improves Trading Decisions](https://mlsanity.com/paper/2507.08584/) (2025)
- [Self-Correction Distillation for Structured Data Question Answering](https://mlsanity.com/paper/2511.07998/) (2025)
- [Credit-Budgeted ICPC-Style Coding: When Agents Must Pay for Every Decision](https://mlsanity.com/paper/2604.10182/) (2026)
- [REFLEX: Reflective Evolution from LLM Experience](https://mlsanity.com/paper/2606.16496/) (2026)

## Abstract (arXiv)

In the domain of code generation, self-debugging is crucial. It allows LLMs to refine their generated code based on execution feedback. This is particularly important because generating correct solutions in one attempt proves challenging for complex tasks. Prior works on self-debugging mostly focus on prompting methods by providing LLMs with few-shot examples, which work poorly on small open-sourced LLMs. In this work, we propose LeDex, a training framework that significantly improves the self-debugging capability of LLMs. Intuitively, we observe that a chain of explanations on the wrong code followed by code refinement helps LLMs better analyze the wrong code and do refinement. We thus propose an automated pipeline to collect a high-quality dataset for code explanation and refinement by generating a number of explanations and refinement trajectories from the LLM itself or a larger teacher model and filtering via execution verification. We perform supervised fine-tuning (SFT) and further reinforcement learning (RL) on both success and failure trajectories with a novel reward design considering code explanation and refinement quality. SFT improves the pass@1 by up to 15.92% and pass@10 by 9.30% over four benchmarks. RL training brings additional up to 3.54% improvement on pass@1 and 2.55% improvement on pass@10. The trained LLMs show iterative refinement ability and can keep refining code continuously. Lastly, our human evaluation shows that the LLMs trained with our framework generate more useful code explanations and help developers better understand bugs in source code.