ACL Findings2026
SynthFix: Adaptive Neuro-Symbolic Code Vulnerability Repair
Yifan Zhang, Jieyu Li, Kexin Pei, Yu Huang, Kevin Leach
Abstract
Large Language Models (LLMs) show promise for automated code repair but often struggle with the complex semantic and structural correctness required. We present SYNTHFIX, a hybrid neuro-symbolic framework that improves LLM-based vulnerability repair by unifying code synthesis with compiler-informed symbolic feedback. The core of our approach is an adaptive training strategy where a neural Router Model directs code samples to either Supervised Fine-Tuning (SFT) to learn common patterns or Reward Fine-Tuning (RFT) with symbolic rewards for complex, iterative refinement. On the FixJS (JavaScript) and Code-Flaws (C) benchmarks, SYNTHFIX achieves up to 18% relative improvement in Code-BLEU/CrystalBLEU and 32% in Exact Match over strong SFT and RFT baselines. Our results show that this adaptive combination of training strategies, which mirrors how developers alternate between pattern application and tool feedback, significantly improves the accuracy and efficiency of LLM-based vulnerability repair. Our code and data are available at https: //github.com/CoderDoge1108/SynthFix 1 .
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
- CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
- VulRepair: a T5-based automated software vulnerability repair cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
Builds on16
- CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and GenerationYue Wang, Weishi Wang, Shafiq R. Joty, Steven C. H. HoiEMNLP 2021 · 1,224 citations
- Scaling Laws for Reward Model OveroptimizationLeo Gao, John Schulman, Jacob HiltonICML 2023 · 963 citations
- CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement LearningHung Le, Yue Wang, Akhilesh Deepak Gotmare, Silvio Savarese et al.NeurIPS 2022 · 571 citations
- CodeGen: An Open Large Language Model for Code with Multi-Turn Program SynthesisErik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu et al.ICLR 2023 · 234 citations
- VulRepair: a T5-based automated software vulnerability repairMichael Fu, Chakkrit Tantithamthavorn, Trung Le, Van Nguyen et al.FSE 2022 · 206 citations
Related papers
- Synthetic Repo-level Bug Dataset for Training Automated Program Repair ModelsMinh V. T. Pham, Huy N. Phan, Nhat Hoang Phan, Cuong Chi Le et al.ICSE 2026
- INTENTFIX: Automated Logic Vulnerability Repair via LLM-Driven Intent ModelingJinseok Heo, Dongwook Choi, Jinyoung Kim, Misoo Kim et al.ICSE 2026
- Hit The Bullseye On The First Shot: Improving LLMs Using Multi-Sample Self-Reward Feedback for Vulnerability RepairRui Jiao, Yue Zhang, Jinku Li, Jianfeng MaASE 2025
- VerilogASTBench: Benchmark Construction of Verilog AST Dataset with Dual-Stage AST Semantic Enhancement FrameworkLuping Zhang, Chao Chen, Dapeng Yan, Hui Xu et al.FSE 2026
- Vul-R2: A Reasoning LLM for Automated Vulnerability RepairXin-Cheng Wen, Zirui Lin, Yijun Yang, Cuiyun Gao et al.ASE 2025 · 1 citation
