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.
Agent Reinforcement Trainer: train multi-step agents for real-world tasks using GRPO. Give your agents on-the-job training. Reinforcement learning for Qwen3.6, GPT-OSS, Llama, and more!
| Date | Stars |
|---|---|
| 2026-07-24 | 10529 |
| 2026-07-25 | 10530 |
| 2026-07-28 | 10530 |
| 2026-07-30 | 10530 |
| 2026-07-31 | 10557 |
| 2026-08-01 | 10559 |
| 2026-08-02 | 10558 |
| 2026-08-03 | 10560 |
| 2026-08-04 | 10559 |
| 2026-08-05 | 10561 |
| 2026-08-06 | 10562 |
| 2026-08-08 | 10563 |
| 2026-08-09 | 10568 |
| 2026-08-10 | 10576 |
| 2026-08-11 | 10578 |
| 2026-08-12 | 10577 |
| 2026-08-13 | 10580 |
| 2026-08-14 | 10580 |
| 2026-08-15 | 10583 |
| 2026-08-16 | 10582 |
| 2026-08-17 | 10594 |
| 2026-08-18 | 10600 |
| 2026-08-19 | 10603 |
| 2026-08-20 | 10606 |
| 2026-08-21 | 10610 |
| 2026-08-22 | 10628 |
| 2026-08-23 | 10648 |
| 2026-08-24 | 10656 |
| 2026-08-25 | 10661 |
| 2026-08-26 | 10662 |
| 2026-08-27 | 10668 |
| 2026-08-28 | 10674 |
| 2026-08-29 | 10679 |
| 2026-08-30 | 10682 |
| 2026-08-31 | 10681 |
| 2026-09-01 | 10684 |
| 2026-09-02 | 10688 |
| 2026-09-03 | 10693 |
| 2026-09-04 | 10695 |
| 2026-09-05 | 10699 |
| 2026-09-06 | 10700 |
| 2026-09-07 | 10705 |
| 2026-09-09 | 10707 |
| 2026-09-10 | 10712 |
| 2026-09-11 | 10711 |
| 2026-09-12 | 10712 |
| 2026-09-13 | 10713 |
| 2026-09-14 | 10712 |
| 2026-09-15 | 10722 |
| 2026-09-16 | 10722 |
| 2026-09-17 | 10736 |
| 2026-09-18 | 10749 |
| 2026-09-19 | 10753 |
| 2026-09-20 | 10760 |
Today
+7 stars today
This week
+47 stars this week
This month
+150 stars this month
Momentum
90.0
growth rate 0.44%/day
<div align="center">
<a href="https://art.openpipe.ai"><picture>
<img alt="ART logo" src="https://github.com/openpipe/art/raw/main/assets/ART_logo.png" width="160px">
</picture></a>
<p align="center">
<h1>Agent Reinforcement Trainer</h1>
</p>
<p>
Train multi-step agents for real-world tasks using GRPO.
</p>
[![PRs-Welcome][contribute-image]][contribute-url]
[][pypi-url]
[](https://colab.research.google.com/github/openpipe/art-notebooks/blob/main/examples/2048/2048.ipynb)
[](https://discord.gg/EceeVdhpxD)
[](https://art.openpipe.ai)
</div>
## 🚀 W&B Training: Serverless RL
**W&B Training (Serverless RL)** is the first publicly available service for flexibly training models with reinforcement learning. It manages your training and inference infrastructure automatically, letting you focus on defining your data, environment and reward function—leading to faster feedback cycles, lower costs, and far less DevOps.
✨ **Key Benefits:**
- **40% lower cost** - Multiplexing on shared production-grade inference cluster
- **28% faster training** - Scale to 2000+ concurrent requests across many GPUs
- **Zero infra headaches** - Fully managed infrastructure that stays healthy
- **Instant deployment** - Every checkpoint instantly available via W&B Inference
```python
# Before: Hours of GPU setup and infra management
# RuntimeError: CUDA error: out of memory 😢
# After: Serverless RL with instant feedback
from art.serverless.backend import ServerlessBackend
model = art.TrainableModel(
project="voice-agent",
name="agent-001",
run_name="agent-001",
base_model="Qwen/Qwen3.6-27B"
)
backend = ServerlessBackend(
api_key="your_wandb_api_key"
)
model.register(backend)
# Edit and iterate in minutes, not hours!
```
[📖 Learn more about W&B Training →](https://docs.wandb.ai/guides/training)
## ART Overview
ART is an open-source RL framework that improves agent reliability by allowing LLMs to **learn from experience**. ART provides an ergonomic harness for integrating GRPO into any python application. For a quick hands-on introduction, run one of the notebooks below. When you're ready to learn more, check out the [docs](https://art.openpipe.ai).
## 📒 Notebooks
| Agent Task | Example Notebook | Description | Comparative Performance |
| ------------------- | -------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **ART•E [Serverless]** | [🏋️ Train agent](https://colab.research.google.com/github/openpipe/art-notebooks/blob/main/examples/art-e.ipynb) | Qwen 3.6 27B learns to search emails using RULER | <img src="https://github.com/openpipe/art/raw/main/assets/benchmarks/email_agent/accuracy-training-progress.svg" height="72"> [benchmarks](/dev/art-e/art_e/evaluate/display_benchmarks.ipynb) |
| **2048 [Serverless]** | [🏋️ Train agent](https://colab.research.google.com/github/openpipe/art-notebooks/blob/main/Excerpt of 14,157 characters
Read on GitHub
Brad Hilton · Ender Research Corp
480
312
Kyle Corbitt · OpenPipe · United States
181
Bohdan Kovalevskyi
76
72
Saumya Gandhi · CMU · United States
42
35
23
18
18
Cursor Agent
6
Angky William
5
2
Alaa Kaddour · @ubiai-incorporated · Tunisia
1
Benedict Aaron Tjandra · United Kingdom
1
1
Fiona Waters · Red Hat · Ireland
1
1
Josh Meyer
1
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:553f3700ebf4880a, topic:reinforcement-learning, topic:rl, desc:reinforcement learning
matched fp:553f3700ebf4880a, topic:lora, desc:grpo, readme:grpo