REVIEW 9 cited by
LLM4Decompile: Decompiling Binary Code with Large Language Models
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Decompilation aims to convert binary code to high-level source code, but traditional tools like Ghidra often produce results that are difficult to read and execute. Motivated by the advancements in Large Language Models (LLMs), we propose LLM4Decompile, the first and largest open-source LLM series (1.3B to 33B) trained to decompile binary code. We optimize the LLM training process and introduce the LLM4Decompile-End models to decompile binary directly. The resulting models significantly outperform GPT-4o and Ghidra on the HumanEval and ExeBench benchmarks by over 100% in terms of re-executability rate. Additionally, we improve the standard refinement approach to fine-tune the LLM4Decompile-Ref models, enabling them to effectively refine the decompiled code from Ghidra and achieve a further 16.2% improvement over the LLM4Decompile-End. LLM4Decompile demonstrates the potential of LLMs to revolutionize binary code decompilation, delivering remarkable improvements in readability and executability while complementing conventional tools for optimal results. Our code, dataset, and models are released at https://github.com/albertan017/LLM4Decompile
Forward citations
Cited by 9 Pith papers
-
Agentic Vulnerability Reasoning on COTS Binaries
SLYP agentic pipeline discovers race condition vulnerabilities in Windows COM binaries and generates debugger-verified PoCs, scoring 0.973 F1 on a 40-case benchmark and finding 28 new confirmed vulnerabilities in prod...
-
Can Large Language Models Understand Intermediate Representations in Compilers?
Large language models can parse compiler intermediate representations but consistently fail at instruction-level reasoning such as control flow, loop handling, and exact execution simulation.
-
NotDec: WebAssembly Decompilation With Inter-Procedural Type Recovery
NotDec decompiles WebAssembly to readable C with 100% recompilability and 85.33% struct access recovery, far exceeding Ghidra, by combining Retypd type recovery with a new pointer/numeric differentiation graph.
-
CoDe-R: Refining Decompiler Output with LLMs via Rationale Guidance and Adaptive Inference
Rationale-guided fine-tuning plus dual-path adaptive inference lifts a 1.3B decompiler refiner to 50% average re-executability, a new lightweight SOTA on HumanEval-Decompile.
-
gigiProfiler: Diagnosing Performance Issues by Uncovering Application Resource Bottlenecks
gigiProfiler combines an LLM resource spotter with static validation to find application-level bottlenecks and root-cause variables by comparing buggy and normal executions.
-
When LLM Defenses Backfire: Characterizing Safety, Performance, and Cost Trade-offs
Across 11 defenses, 6 open LLMs, and multiple benchmarks, rule-based defenses best preserve task accuracy, conservative self-reflection drives over-refusal, and multi-round defenses dominate inference cost.
-
CodableLLM: Automating Decompiled and Source Code Mapping for LLM Dataset Generation
CodableLLM automatically aligns decompiled binary functions with source code functions to generate training datasets for code LLMs.
-
TraceRAG: A LLM-Based Framework for Explainable Android Malware Detection and Behavior Analysis
An LLM-based RAG framework that retrieves method-level Java code snippets to explain and detect malicious behavior in Android apps.
-
VulBinLLM: LLM-powered Vulnerability Detection for Stripped Binaries
An LLM pipeline that enriches decompiled code with vulnerability hints and then classifies CWEs reports high Juliet accuracy, but leaky hints and unmatched baselines weaken the claim.
Discussion (0). Continue with ORCID to comment.