LLM4Decompile studies language models trained to transform binary code or refine decompiler output. The paper is relevant when assessing whether an AI-generated program can be re-executed under benchmark tests. Read the task definitions, datasets and evaluation setup before interpreting its reported gains; a passing benchmark does not recover original source or prove equivalence for an unseen application.
What to inspect
Separate direct binary-to-code generation from refinement of an existing decompiler result. Check the benchmark input distribution, execution tests and exact model artifacts linked by the authors.
Practical limit
The arXiv record identifies the current version and EMNLP venue. Its performance figures are author-reported research results, not a guarantee for a new binary, compiler or architecture.
Requirements & evidence
- Resource type
- Paper
- Category
- Unknown
- Host platforms
- Unknown
- Targets
- Unknown
- Level
- Unknown
- Author(s)
- Hanzhuo Tan, Qi Luo, Jing Li, Yuqun Zhang
- Year
- 2024
- Venue
- EMNLP 2024
- Current version
- v3
- Version date
- 2024-10-22
- Topic
- LLM decompilation evaluated for re-executability
- Source checked
- 2026-11-04T10:35:59.748040+00:00
Author-linked code: https://github.com/albertan017/LLM4Decompile
Dataset: https://github.com/albertan017/LLM4Decompile
Models: https://github.com/albertan017/LLM4Decompile
Unknown fields have not been established by an author or publisher source.
Primary sources
- https://arxiv.org/abs/2403.05286
- https://raw.githubusercontent.com/albertan017/LLM4Decompile/main/README.md
Source descriptions establish documented scope. They do not establish that every workflow has been reproduced by AI Craft Pad.