reveng-grp-2025/sorel20m-100k

SOREL-20M Subset Dataset

Dataset Description

This is a subset of the SOREL-20M dataset containing malware detection features and labels.

Dataset Summary

• Total samples: 196534
• Malicious samples: 99506
• Benign samples: 97028
• Feature dimensions: 2351 (EMBER v2 features)

Dataset Structure

The dataset contains the following columns:

• sha256: SHA256 hash of the original file
• label: Binary label (0=benign, 1=malicious)
• feature_0 to feature_2350: EMBER v2 feature vector
• Additional metadata columns as available

Features

The dataset contains 2351-dimensional feature vectors extracted using the EMBER v2 feature extraction process.

Usage

from datasets import load_dataset
import pandas as pd

# Load from parquet
df = pd.read_parquet("train.parquet")

# Or load as HuggingFace dataset
dataset = load_dataset("path/to/dataset")

Citation

If you use this dataset, please cite the original SOREL-20M paper:

@misc{harang2020sorel20m,
      title={SOREL-20M: A Large Scale Benchmark Dataset for Malicious PE Detection},
      author={Richard Harang and Ethan M. Rudd},
      year={2020},
      eprint={2012.07634},
      archivePrefix={arXiv},
      primaryClass={cs.CR}
}

License

Please refer to the original SOREL-20M dataset terms of use.