遇见数据集

demeleww/evm-decompiler-dataset

收藏
Hugging Face2026-04-27 更新2026-05-03 收录
官方服务:

资源简介:

EVM反编译器数据集,用于训练EVM智能合约反编译模型,将三地址码(TAC)转换为Solidity和Huff汇编代码。数据集包含100,000个示例,其中训练集95,000个,测试集5,000个。大部分示例(98,854个)将TAC转换为Solidity,573个转换为Huff汇编,另有573个同时转换为两者。数据格式支持ChatML消息(系统/用户/助手)用于SFT,以及prompt/completion用于向后兼容。

The EVM Decompiler Dataset is designed for training EVM smart contract decompilation models that convert Three-Address Code (TAC) into Solidity and Huff assembly code. The dataset contains a total of 100,000 examples, including 95,000 training samples and 5,000 test samples. Among these examples, 98,854 convert TAC to Solidity, 573 convert TAC to Huff assembly, and an additional 573 examples support conversion to both formats. The dataset supports ChatML message format with system, user, and assistant roles for Supervised Fine-Tuning (SFT), as well as the prompt/completion format for backward compatibility.

提供机构:
demeleww
二维码
社区交流群
二维码
科研交流群
商业服务