Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
[NeurIPS 2025 D&B Spotlight] Scaling Data for SWE-agents
| Date | Stars |
|---|---|
| 2026-07-24 | 710 |
| 2026-07-25 | 711 |
| 2026-07-28 | 711 |
| 2026-07-30 | 711 |
| 2026-07-31 | 721 |
| 2026-08-06 | 721 |
Today
— stars today
This week
+10 stars this week
This month
— stars this month
Momentum
10.0
growth rate 1.41%/day
<p align="center">
<a href="https://swesmith.com/">
<img src="docs/assets/banner.png" style="height: 10em" alt="Kawhi the SWE-smith" />
</a>
</p>
<br>
<div align="center">
<strong>NeurIPS 2025 Datasets & Benchmarks Track - Spotlight 🔦</strong>
</div>
<br>
<div align="center">
<a href="https://www.python.org/">
<img alt="Build" src="https://img.shields.io/badge/Python-3.10+-1f425f.svg?color=purple">
</a>
<a href="https://copyright.princeton.edu/policy">
<img alt="License" src="https://img.shields.io/badge/License-MIT-blue">
</a>
<a href="https://badge.fury.io/py/swesmith">
<img src="https://badge.fury.io/py/swesmith.svg">
</a>
<a href="https://arxiv.org/abs/2504.21798">
<img src="https://img.shields.io/badge/arXiv-2504.21798-b31b1b.svg">
</a>
</div>
<hr />
SWE-smith is a toolkit for training [SWE-agents](https://github.com/SWE-agent/SWE-agent). You can:
* Turn any Github repository into a [SWE-gym](https://github.com/SWE-Gym/SWE-Gym).
* Create *unlimited* tasks (e.g., file localization, program repair, [SWE-bench](https://github.com/SWE-bench/SWE-bench)) for that repo.
* Train an LM to become a better SWE ([SWE-agent-LM-32B](https://huggingface.co/SWE-bench/SWE-agent-LM-32B)).
## ⚒️ Build Environments
If you're interested in turning a GitHub repository into a SWE-gym, install the package from [source](https://swesmith.com/getting_started/installation/).
> [!TIP]
> SWE-smith requires Docker to create execution environments. SWE-smith was developed and tested on Ubuntu 22.04.4 LTS.
> We do *not* plan on supporting Windows or MacOS.
You can then build a dataset for the repository by...
1. [Creating an environment](https://swesmith.com/guides/env_construction/#create-an-execution-environment)
2. [Synthesizing task instances](https://swesmith.com/guides/create_instances/)
3. [Keep tasks that break 1+ unit tests](https://swesmith.com/guides/harnesses/)
4. [Generating issue text for your tasks](https://swesmith.com/guides/issue_gen/)
## 🏋️ Train SWE-agent's
Training SWE-agent's using the [SWE-smith dataset](https://huggingface.co/datasets/SWE-bench/SWE-smith) is super simple.
```python
from swesmith.profiles import registry
from datasets import load_dataset
ds = load_dataset("SWE-bench/SWE-smith", split="train") # Loads all 52k task instances
for task in ds:
rp = registry.get_from_inst(task) # Get the RepoProfile for the task
container = rp.get_container(task) # Returns pointer to a Docker container with the task initialized
"""TODO: Train!"""
```
SWE-smith has been used to
* Fine-tune Qwen 2.5 Coder into SWE-agent-LM-32B (A +32% jump on SWE-bench Verified!) using [SWE-agent](https://github.com/SWE-agent/SWE-agent) [[Tutorial](https://swesmith.com/guides/train_swe_agent/)]
* Perform GRPO style reinforcement learning using [SkyRL](https://github.com/NovaSky-AI/SkyRL)
## 💿 Resources
* [52k Task Instances](https://huggingface.co/datasets/SWE-bench/SWE-smith)
* [SWE-agent-LM-32B](https://huggingface.co/SWE-bench/SWE-agent-LM-32B); **40.2%** pass@1 on [SWE-bench Verified](https://huggingface.co/datasets/SWE-bench/SWE-bench_Verified)!
* [26k SWE-agent Trajectories](https://huggingface.co/datasets/SWE-bench/SWE-smith-trajectories), including the 5k SWE-agent-LM-32B was trained on.
* [250+ Environments](https://github.com/SWE-bench/SWE-smith-envs), one Docker image per repo represented in SWE-smith.
And there's more coming!
## 💫 Contributions
We're actively working on several follow ups!
Check out the [Contributing Guide](CONTRIBUTING.md) for more.
Contact Person: [John Yang](https://john-b-yang.github.io/), [Kilian Lieret](https://lieret.net)
(Email: [[email protected]](mailto:[email protected]))
## 🪪 License
MIT. Check `LICENSE` for more information.
## ✍️ Citation
```bibtex
@inproceedings{yang2025swesmith,
title={SWE-smith: Scaling Data for Software Engineering Agents},
author={John Yang and Kilian Lieret and Carlos E. Jimenez and Alexander Wettig and Kabir Khandpur and Yanzhe Zhang and Excerpt of 5,195 characters
Read on GitHub
John Yang
94
27
21
Kilian Lieret · Meta · United States
15
Kevin Xiang Li · Stanford University
14
8
5
5
4
3
3
Marc-Alexandre Côté · Canada
3
3
1
1
Kabir
1
1
Ofir Press
1
1
Cesar Garcia · Argentina
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:cbf93a86923b9431, topic:training
matched fp:cbf93a86923b9431, topic:language-model
matched fp:cbf93a86923b9431, topic:agents