Hugging Face's logo

Datasets:
TIGER-Lab
/
packages_python_filtered

Modalities:
Tabular
Text
Formats:
csv
Languages:
English
Size:
1K - 10K
ArXiv:
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
packages_python_filtered / README.md
lllqaq's picture
lllqaq
Update dataset card for packages_python_filtered
d15b762 verified
preview code
|
raw
history blame contribute delete
4.65 kB
metadata 
license: mit
pretty_name: SWE-Next Seed Repository List
language:
  - en
size_categories:
  - 1K<n<10K
configs:
  - config_name: default
    data_files:
      - split: train
        path: packages_python_filtered.csv

SWE-Next: Scalable Real-World Software Engineering Tasks for Agents

Paper Project Page Code Dataset SFT Trajs Model 7B Model 14B

packages_python_filtered

This repository contains packages_python_filtered.csv, the seed repository list used by SWE-Next. The file contains 3,971 Python package / repository entries that serve as the starting point for large-scale repository mining and execution-grounded task synthesis.

Each row links a package-oriented seed entry to a GitHub repository and includes lightweight metadata used by the collection pipeline, such as stars, language, whether the repository was downloaded, and merged-PR counts.

Overview

SWE-Next begins from this seed list of 3,971 repositories, then mines merged pull requests, executes candidate base/merged commit pairs, and filters them into the final execution-grounded SWE dataset. This CSV is therefore the upstream repository inventory that defines the initial search space for the pipeline.

Format

The CSV has the following columns:

Column Description
pypi_name PyPI package or seed package name associated with the repository
repo_name GitHub repository in owner/repo format
local_path Original local clone path used during collection
stars Repository stars at collection time
downloaded Whether the repository was successfully downloaded
primary_language Primary repository language
pr_count Number of pull requests observed for the repository

Example rows:

pypi_name,repo_name,local_path,stars,downloaded,primary_language,pr_count
vdtool,yt-dlp/yt-dlp,...,120643,True,Python,4349
django-squad,django/django,...,84411,True,Python,20318

Files

  • packages_python_filtered.csv: seed repository list for SWE-Next collection

Usage

This artifact is mainly useful for:

  • reproducing the initial repository search space of SWE-Next,
  • analyzing repository-scale coverage before task synthesis,
  • selecting subsets of repositories for custom collection runs.

Load it with pandas:

import pandas as pd

df = pd.read_csv("hf://datasets/TIGER-Lab/packages_python_filtered/packages_python_filtered.csv")
print(df.head())

Relationship to the SWE-Next Release

This repo contains the seed repository list used by SWE-Next. Related artifacts are available separately:

  • Repository summary with NEW_COMMIT_BETTER counts: TIGER-Lab/new_commit_better_repos
  • Final task dataset: TIGER-Lab/SWE-Next
  • SFT trajectories: TIGER-Lab/SWE-Next-SFT-Trajectories
  • Project code: github.com/TIGER-AI-Lab/SWE-Next

Citation 

@misc{liang2026swenextscalablerealworldsoftware,
      title={SWE-Next: Scalable Real-World Software Engineering Tasks for Agents},
      author={Jiarong Liang and Zhiheng Lyu and Zijie Liu and Xiangchao Chen and Ping Nie and Kai Zou and Wenhu Chen},
      year={2026},
      eprint={2603.20691},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      url={https://arxiv.org/abs/2603.20691},
}