mlsanityData to 7 Oct 2026

BoostAPR: Boosting Automated Program Repair via Execution-Grounded Reinforcement Learning with Dual Reward Models

arXiv 2605.09134 ↗First posted 9 May 2026cs.AI cs.SEcited by 1

On the map: LLM alignment / RLHF / preference optimisation → Process reward models for reasoning

Abstract (arXiv)

Reinforcement learning for program repair is hindered by sparse execution feedback and coarse sequence-level rewards that obscure which edits actually fix bugs. We present BoostAPR, a three-stage framework addressing these challenges: (1) supervised fine-tuning on execution-verified demonstrations with reasoning traces, (2) training dual reward models--a sequence-level assessor and a line-level credit allocator--from execution outcomes, and (3) PPO optimization where the line-level model redistributes rewards to critical edit regions. This line-level credit assignment operates at an intermediate granularity naturally suited to code changes. Trained on SWE-Gym and evaluated on four benchmarks, BoostAPR achieves 40.7% on SWE-bench Verified (+22.9pp over base model), 24.8% on Defects4J (Python-to-Java transfer), 84.5% on HumanEval-Java, and 95.0% on QuixBugs, achieving competitive results among open-source models with strong cross-language generalization.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in Process reward models for reasoning, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 35 of its 44 references that are in the map. First posted on arXiv 9 May 2026.

Builds on

35 of 44 references.

References · 35

  1. 2017Proximal Policy Optimization Algorithmscited by 14,441
  2. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  4. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
  5. 2023Let's Verify Step by Stepcited by 3,598
  6. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,193
  7. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 1,980
  8. 2015High-Dimensional Continuous Control Using Generalized Advantage Estimationcited by 1,916
  9. 2024SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineeringcited by 1,628
  10. 2024Qwen2.5-Coder Technical Reportcited by 1,151
  11. 2021Measuring Coding Challenge Competence With APPScited by 879
  12. 2024Agentless: Demystifying LLM-based Software Engineering Agentscited by 400
23 more
  1. 2022CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learningcited by 342
  2. 2024Training Software Engineering Agents and Verifiers with SWE-Gymcited by 300
  3. 2024AutoCodeRover: Autonomous Program Improvementcited by 272
  4. 2025SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolutioncited by 246
  5. 2024RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learningcited by 146
  6. 2022Practical Program Repair in the Era of Large Pre-trained Language Modelscited by 133
  7. 2023Keep the Conversation Going: Fixing 162 out of 337 bugs for 0.42 each using ChatGPTcited by 125
  8. 2023An Analysis of the Automatic Bug Fixing Performance of ChatGPTcited by 108
  9. 2025SWE-Fixer: Training Open-Source LLMs for Effective and Efficient GitHub Issue Resolutioncited by 87
  10. 2023Impact of Code Language Models on Automated Program Repaircited by 85
  11. 2024Dense Reward for Free in Reinforcement Learning from Human Feedbackcited by 77
  12. 2026MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Taskscited by 72
  13. 2024Lingma SWE-GPT: An Open Development-Process-Centric Language Model for Automated Software Improvementcited by 57
  14. 2023RepairLLaMA: Efficient Representations and Fine-Tuned Adapters for Program Repaircited by 50
  15. 2024TLCR: Token-Level Continuous Reward for Fine-grained Reinforcement Learning from Human Feedbackcited by 43
  16. 2023KNOD: Domain Knowledge Distilled Tree Decoder for Automated Program Repaircited by 13
  17. 2024MORepair: Teaching LLMs to Repair Code via Multi-Objective Fine-tuningcited by 13
  18. 2026Render-in-the-Loop: Vector Graphics Generation via Visual Self-Feedbackcited by 12
  19. 2025Eigenspectrum Analysis of Neural Networks without Aspect Ratio Biascited by 9
  20. 2025Reward Auditor: Inference on Reward Modeling Suitability in Real-World Perturbed Scenarioscited by 7
  21. 2026VAnim: Rendering-Aware Sparse State Modeling for Structure-Preserving Vector Animationcited by 6
  22. 2025Alleviating Attention Hacking in Discriminative Reward Modeling through Interaction Distillationcited by 5
  23. 2025DRAFT-RL: Multi-Agent Chain-of-Draft Reasoning for Reinforcement Learning-Enhanced LLMscited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Scope3Trace: Evidence-Based Identification and Extraction of Scope 3 GHG Emissions from Sustainability Reportscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024RePair: Automated Program Repair with Process-based Feedback
  2. 2026QiMeng-PRepair: Precise Code Repair via Edit-Aware Reward Optimization
  3. 2025Boosting Open-Source LLMs for Program Repair via Reasoning Transfer and LLM-Guided Reinforcement Learning
  4. 2026TaPR: Test-Aware Policy Refinement for Feedback-Conditioned Code Generation
  5. 2025The Art of Repair: Optimizing Iterative Program Repair with Instruction-Tuned Models
  6. 2026StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning
  7. 2026Anchored Self-Play for Code Repair
  8. 2025Accelerating Automatic Program Repair with Dual Retrieval-Augmented Fine-Tuning and Patch Generation on Large Language Models