{"slug": "agbcc-slade-agbcc-neural-matching-decompilation", "title": "agbcc: SLaDe-AGBCC: Neural Matching Decompilation", "summary": "A research project called SLaDe-AGBCC is testing whether large language models can perform matching decompilation of Game Boy Advance agbcc assembly, optimizing for C code that recompiles to byte-identical output rather than readable structure. The project, which credits Claude Fable 5 for co-authorship on implementation and diagnostic tasks, treats C recovery as a folding problem inspired by AlphaFold 2 and uses a compiler-in-the-loop recycling signal, with recent v4 recycling experiments showing progress in loop-based matching. The author states the core hypothesis — that transformers can defeat the agbcc optimizer to achieve byte-exactness — is being tested rather than assumed, drawing on a curated corpus of GBA decompilation projects and decomp.me data.", "body_md": "## Evidence at a glance\n\nLoading source evidence…\n\n## What was done\n\nSLaDe-AGBCC is a research project using LLMs to perform matching decompilation on GBA agbcc assembly. It treats C recovery as a folding problem, aiming for byte-identical output via compiler-in-the-loop recycling.\n\nSLaDe-AGBCC is a research project investigating neural matching decompilation for the Game Boy Advance. Unlike traditional decompilers, it does not prioritize readable structure or name recovery; instead, it optimizes for producing C code that recompiles to byte-identical assembly when processed by the `agbcc` compiler. The project draws inspiration from AlphaFold 2, treating C recovery as a folding problem and utilizing a compiler-in-the-loop recycling signal to refine predictions. The author reports that the core hypothesis—that transformers can defeat the `agbcc` optimizer to achieve byte-exactness—is currently being tested rather than assumed. Development involves an automated pipeline where disassembled objects are converted to C and iteratively refined. The project relies on a curated corpus of GBA decompilation projects and decomp.me data. Recent experiments include v4 recycling, which demonstrated progress in loop-based matching, though the author notes that recycling must be explicitly trained. The project documentation credits Claude Fable 5 for co-authorship on various implementation and diagnostic tasks.\n\nSources: [Source 1](https://raw.githubusercontent.com/jrickey/slade-agbcc/HEAD/README.md)\n\n## Reproduce it\n\n```\nuv venv && uv pip install -e . && cd vendor/agbcc && ./build.sh && cd - && python scripts/measure_corpus.py\n```\n\n## Community\n\nAsk a question, share what worked, or follow the next milestone.\n\n[Your library](https://reversedwithai.com/library)", "url": "https://wpnews.pro/news/agbcc-slade-agbcc-neural-matching-decompilation", "canonical_source": "https://reversedwithai.com/p/agbcc-f0856640", "published_at": "2026-10-28 23:01:02+00:00", "updated_at": "2026-10-28 23:41:01.791328+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "machine-learning", "ai-research", "developer-tools"], "entities": ["SLaDe-AGBCC", "agbcc", "Game Boy Advance", "AlphaFold 2", "Claude Fable 5", "decomp.me"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/agbcc-slade-agbcc-neural-matching-decompilation", "markdown": "https://wpnews.pro/news/agbcc-slade-agbcc-neural-matching-decompilation.md", "text": "https://wpnews.pro/news/agbcc-slade-agbcc-neural-matching-decompilation.txt", "jsonld": "https://wpnews.pro/news/agbcc-slade-agbcc-neural-matching-decompilation.jsonld"}}