{
  "added_at": "2026-10-20",
  "arxiv_id": "2503.07215",
  "arxiv_version_reviewed": "v3",
  "authors": [
    "Peipei Liu",
    "Jian Sun",
    "Rongkang Sun",
    "Li Chen",
    "Zhaoteng Yan",
    "Xiaoling Zhang",
    "Dawei Wang",
    "Dapeng Sun",
    "Peizheng Zhang",
    "Dan Li"
  ],
  "bibtex_url": "https://awesomerea.org/papers/citations/paper-codeinverter-suite.bib",
  "citation": {
    "container_title": "Proceedings of the 41st IEEE/ACM International Conference on Automated Software Engineering",
    "pages": "3278-3289",
    "publisher": "ACM",
    "sources": [
      "https://api.crossref.org/works/10.1145/3832783.3837554"
    ]
  },
  "code_url": "https://github.com/LiuPeiP-CS/CodeInverter",
  "date_basis": "Online and print publication date in ACM’s Crossref deposit; arXiv first submission is recorded separately.",
  "date_basis_zh": "采用 ACM 向 Crossref 提交的在线及印刷出版日期，另行保留 arXiv 首次提交日期。",
  "date_precision": "day",
  "description_en": "CodeInverter studies function-level LLM decompilation using control-flow graphs and memory-to-data mappings extracted with IDA/objdump. It presents a prompt workflow, training corpus and model family, evaluated through compilation, benchmark input/output tests and edit similarity.",
  "description_zh": "CodeInverter 研究函数级 LLM 反编译，使用 IDA／objdump 提取的控制流图和内存到数据映射补充输入。论文提出提示工作流、训练语料与模型系列，并以编译、基准输入输出测试和编辑相似性评估结果。",
  "doi": "10.1145/3832783.3837554",
  "entry_url": "https://awesomerea.org/papers/paper-codeinverter-suite/",
  "entry_url_zh": "https://awesomerea.org/zh/papers/paper-codeinverter-suite/",
  "evidence_urls": [
    "https://api.crossref.org/works/10.1145/3832783.3837554",
    "https://arxiv.org/abs/2503.07215v3",
    "https://arxiv.org/pdf/2503.07215v3",
    "https://arxiv.org/html/2503.07215v3",
    "https://arxiv.org/abs/2503.07215v1",
    "https://arxiv.org/abs/2503.07215v2",
    "https://github.com/LiuPeiP-CS/CodeInverter/blob/548e59c5db07bd108659bf2bfb069f1cfd0c9298/README.md",
    "https://zenodo.org/api/records/19351453",
    "https://huggingface.co/CodeInverter/CIM-1.3b/raw/1f7745e19c5420a8c5896f5c31d6938f83454be4/README.md",
    "https://github.com/LiuPeiP-CS/CodeInverter/blob/548e59c5db07bd108659bf2bfb069f1cfd0c9298/code4train/train/run_ms2decllm_train.sh",
    "https://github.com/LiuPeiP-CS/CodeInverter/blob/548e59c5db07bd108659bf2bfb069f1cfd0c9298/code4train/LICENSE",
    "https://github.com/LiuPeiP-CS/CodeInverter/blob/548e59c5db07bd108659bf2bfb069f1cfd0c9298/code4train/LICENSE-MODEL"
  ],
  "first_arxiv_date": "2025-03-10",
  "id": "paper-codeinverter-suite",
  "json_url": "https://awesomerea.org/papers/paper-codeinverter-suite/record.json",
  "kind": "research",
  "markdown_url": "https://awesomerea.org/papers/paper-codeinverter-suite/index.md",
  "markdown_url_zh": "https://awesomerea.org/zh/papers/paper-codeinverter-suite/index.md",
  "metadata_notes": "The v3 title and ten-author order match ACM’s Crossref deposit. The repository README at 548e59c still cites v2/2025. Direct ACM access returned HTTP 403. Artifact snapshots are separately versioned. The training directory contains MIT and DeepSeek notices; the CIM-1.3B card declares CC BY 4.0.",
  "metadata_notes_zh": "v3 的标题和十位作者顺序与 ACM 的 Crossref 元数据一致，仓库 548e59c 版本的 README 仍引用 v2／2025。直接访问 ACM 返回 HTTP 403。各产物使用独立版本。训练目录包含 MIT 与 DeepSeek 许可说明，CIM-1.3B 模型卡声明 CC BY 4.0。",
  "pdf_url": "https://arxiv.org/pdf/2503.07215v3",
  "project_url": "https://huggingface.co/CodeInverter",
  "publication_date": "2026-10-07",
  "publication_status": "conference-paper",
  "record_type": "paper",
  "runtime_tested": false,
  "scope_limitations_en": "Evaluation covers individual C functions under 32/64-bit GCC settings. Whole-program recovery, aggressive optimization and protected binaries remain limitations. The text reports one epoch (15 steps), while the pinned launch example sets one epoch and 100 warmup steps; the exact training schedule remains unresolved.",
  "scope_limitations_zh": "评估针对 32／64 位 GCC 设置下的单个 C 函数。完整程序恢复、激进优化和受保护二进制仍有局限。正文写有 1 个 epoch（15 步），固定版本启动示例设置 1 个 epoch 及 100 个预热步，尚无法据此核实精确训练日程。",
  "targets": [
    "binary"
  ],
  "title": "The CodeInverter Suite: Structure- and Data-Aware Binary Decompilation with Efficient LLMs",
  "title_zh": "CodeInverter 套件：利用高效大语言模型进行结构与数据感知的二进制反编译",
  "url": "https://dl.acm.org/doi/10.1145/3832783.3837554",
  "venue": "ASE 2026",
  "verification": "source-reviewed",
  "verified_at": "2026-10-19",
  "why_fits_en": "Adds structure- and data-informed LLM decompilation with separate compilation, execution-test and similarity evidence.",
  "why_fits_zh": "补充结合程序结构与数据上下文的 LLM 反编译研究，并区分编译、执行测试与相似性证据。",
  "year": 2026
}