LLMs as Idiomatic Decompilers: Recovering High-Level Code from x86-64 Assembly for Dart

TL;DR

This paper explores using small specialized LLMs to decompile x86-64 assembly into Dart, demonstrating competitive performance with larger models and analyzing the impact of synthetic and cross-language training data.

Contribution

It introduces a specialized 4B LLM for Dart decompilation, compares data augmentation strategies, and evaluates cross-lingual transfer effects.

Findings

The 4B model achieves 71.3 CODEBLEU on Dart decompilation.

Adding Swift data improves performance at 8B but not at 4B.

Small models can produce readable, idiomatic Dart code.

Abstract

Translating machine code into human-readable high-level languages is an open research problem in reverse engineering. Despite recent advancements in LLM-based decompilation to C, modern languages like Dart and Swift are unexplored. In this paper, we study the use of small specialized LLMs as an idiomatic decompiler for such languages. Additionally, we investigate the augmentation of training data using synthetic same-language examples, and compare it against adding human-written examples using related-language (Swift -> Dart). We apply CODEBLEU to evaluate the decompiled code readability and compile@k to measure the syntax correctness. Our experimental results show that on a 73-function Dart test dataset (representing diverse complexity levels), our 4B specialized model achieves 71.3 CODEBLEU (95% CI 65.5-77.1), approximately comparable to a ~480B code model (73.1; 67.4-78.8). On a subset of 34 natural Dart functions, it reaches compile@k5 = 79.4% (Wilson 95% CI 63.2-89.7), vs. 64.7% (47.9-78.5) for the base model; the difference is suggestive but not statistically significant at 0.05. Our results indicate that adding Swift training data helps at 8B but not at 4B, suggesting a capacity threshold for effective cross-lingual transfer. Our experimental results show that small specialized models can generate readable, idiomatic Dart with meaningful identifiers while using minimal compute.