CompRealVul_Bin Dataset

Dataset Summary

CompRealVul_Bin is a binary-level dataset derived from the CompRealVul_C dataset, designed for training and evaluating machine learning models on vulnerability detection in compiled binaries. This dataset contains cross-compiled .exe binaries corresponding to each C function in the source dataset, packaged as a single .zip archive.

The dataset bridges the gap between source-level vulnerability datasets and practical binary-level learning pipelines.

Key Features

• ✅ Matches samples from the original CompRealVul dataset
• ✅ Labeled with vulnerability information (binary class label)
• ✅ Suitable for binary-focused tasks such as vulnerability detection, similarity learning, and binary classification
• ✅ Provided as a ZIP archive for efficient download and storage

Dataset Structure

• CompRealVul_Bin.zip: Contains all compiled .exe binary files
• CompRealVul_metadata.csv: Metadata file with the following fields:

Column	Description
file_name	Name of the .exe file
function_name	The name of the target function in the .exe file
label	Vulnerability label (1.0 = vulnerable, 0.0 = non-vulnerable)

Example:

file_name,function_name,label
func_123.exe,func_123,1.0

Usage

You can load the dataset directly using the datasets library from Hugging Face:

import pandas as pd
import zipfile

# Load metadata
df = pd.read_csv("CompRealVul_metadata.csv")

# Access the ZIP archive
with zipfile.ZipFile("CompRealVul_Bin.zip", "r") as zip_ref:
    # Extract a specific binary
    binary_path = df.iloc[0]["binary_path"]
    zip_ref.extract(binary_path, "extracted_binaries/")

# Read binary data
with open(f"extracted_binaries/{binary_path}", "rb") as f:
    binary_data = f.read()

License

This dataset is released under the GPL-3.0.

Citation

@misc{comprealvul_bin,
  author       = {Compote},
  title        = {CompRealVul_Bin: A Binary Dataset for Vulnerability Detection},
  howpublished = {\url{https://huggingface.co/datasets/compAgent/CompRealVul_Bin}},
  year         = {2025}
}