2026 · Preprint
ASAP: Assembly-Source Aligned Pseudocode Refinement For Binary Decompilation
ASAP aligns assembly and source representations during training, compresses assembly features into tokens, and supplies them alongside decompiler pseudocode to an LLM refiner. Masking and an assembly-advantage objective encourage the refiner to use low-level evidence.
Scope and limitations
Evaluation uses C adaptations of HumanEval and MBPP, x86-64 GCC 11.4 at O0–O3, and Ghidra 11.0.3 pseudocode. Re-execution means passing supplied tests, not proving equivalence. Identifier/comment recovery is not a dedicated component, and frontier-scale model training is not exhaustively evaluated.
Why it fits this collection
Adds an assembly-informed refinement approach that explicitly documents the limits of its execution-test evidence.
Publication and review details
- Venue
- arXiv
- Status
- Preprint
- Publication / posting date
- September 30, 2026
- Date source
- First arXiv submission date from the reviewed submission history; no formal publication was verified.
- First arXiv submission
- September 30, 2026
- arXiv
- 2610.06900
- Reviewed arXiv version
- v1
Both the abstract history and v1 HTML give 2026-09-30, despite the 2610 identifier prefix; the date was not inferred from that prefix. The article is CC BY 4.0. No author artifact URL or code/model/data license was verified from the reviewed sources and focused search; referenced training datasets retain their own terms.
Sources reviewed
Source review only; experiments were not rerun for this collection. Cite the original authors using the linked paper or BibTeX.