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.
A curated collection of papers and resources on On-Policy Distillation for Large Language Models.
| Date | Stars |
|---|---|
| 2026-07-31 | 481 |
| 2026-08-05 | 494 |
| 2026-08-06 | 494 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
<a name="readme-top"></a> <div align="center"> <a href="https://github.com/nick7nlp/Awesome-LLM-On-Policy-Distillation/stargazers"><img src="https://img.shields.io/github/stars/nick7nlp/Awesome-LLM-On-Policy-Distillation?style=for-the-badge" alt="Stars"></a> <a href="https://github.com/nick7nlp/Awesome-LLM-On-Policy-Distillation/network/members"><img src="https://img.shields.io/github/forks/nick7nlp/Awesome-LLM-On-Policy-Distillation?style=for-the-badge" alt="Forks"></a> <a href="https://github.com/nick7nlp/Awesome-LLM-On-Policy-Distillation/graphs/contributors"><img src="https://img.shields.io/github/contributors/nick7nlp/Awesome-LLM-On-Policy-Distillation?style=for-the-badge" alt="Contributors"></a> <a href="https://github.com/nick7nlp/Awesome-LLM-On-Policy-Distillation/blob/master/LICENSE"><img src="https://img.shields.io/github/license/nick7nlp/Awesome-LLM-On-Policy-Distillation?style=for-the-badge" alt="License"></a> </div> <br/> <h1 align="center">🔥 Awesome LLM On-Policy Distillation</h1> <p align="center"> <b>A curated collection of papers and resources on On-Policy Distillation for Large Language Models.</b> </p> <p align="center"> <a href="https://awesome.re"><img src="https://awesome.re/badge.svg" alt="Awesome"></a> <img src="https://img.shields.io/badge/Papers-262-blue" alt="Papers"> <img src="https://img.shields.io/github/last-commit/nick7nlp/Awesome-LLM-On-Policy-Distillation?label=Last%20Updated&color=green" alt="Last Updated"> <img src="https://img.shields.io/badge/Survey-V4-orange" alt="Survey V4"> </p> <p align="center"> <a href="https://nick7nlp.github.io/OPDHub/"><img src="https://img.shields.io/badge/🌐_Companion_Site-OPDHub-5698C3?style=for-the-badge" alt="OPDHub Companion Site"></a> <a href="https://arxiv.org/abs/2604.00626"><img src="https://img.shields.io/badge/📄_Survey-arXiv:2604.00626-b31b1b?style=for-the-badge" alt="arXiv"></a> </p> ## 🔥 News * **2026.07.13**: 📦 [**EasyOPD**](https://github.com/lds-ustc/EasyOPD) released — the first unified OPD framework covering 10+ methods (cross-tokenizer, self-distillation, step-wise) with one-line YAML switching, built on [verl](https://github.com/verl-project/verl). A companion toolkit of our survey. Paper: [arXiv:2607.11012](https://arxiv.org/abs/2607.11012) | [Demo Video](https://drive.google.com/file/d/1hgUeViTfSLEkAjsnj76PdLlgkBHRO-TS/view). * **2026.06.18**: 📊 Survey **V4** released with 72 new OPD papers, full-text review and AI-trace audit, updated taxonomy tree and method tables. Read on [arXiv](https://arxiv.org/abs/2604.00626). * **2026.06.02**: 🌐 [OPDHub](https://nick7nlp.github.io/OPDHub/) launched, a companion site with full-text search and multi-axis filters (section, loss, domain, signal source, rollout frequency, student size, year), plus a copy-ready BibTeX block. * **2026.05.18**: 🚀 Survey **V3** released, adding the §3 *Landscape and Method Selection* chapter and §7.4 *On-Policy vs Off-Policy Decision Framework*. Read on [arXiv](https://arxiv.org/abs/2604.00626). * **2026.05.12**: 🛠️ Survey **V2** released, adding the Hall of Fame, recommended reading orders by background, the Mermaid evolution timeline and taxonomy mindmap. Read on [arXiv](https://arxiv.org/abs/2604.00626). * **2026.04.01**: 📝 Initial release of the survey on [arXiv](https://arxiv.org/abs/2604.00626) with the first systematic taxonomy of OPD methods covering objective design, signal source, and training stabilization. ## 🤔 Why On-Policy? — The Core Problem <p align="center"> <img src="assets/opd-overview.png" alt="On-Policy Distillation: Teacher-Student Loop" width="640"> </p> Traditional off-policy distillation (e.g., SFT on teacher demonstrations) suffers from **exposure bias** and **train-test mismatch**: the student learns to predict the next token given perfect teacher prefixes, but during inference, it must condition on its own flawed generations. Errors compound rapidly. **On-policy distillation (OPD)** solves thi
Excerpt of 140,683 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:aa33f5bf80cd647e, topic:large-language-models, topic:llm
matched fp:aa33f5bf80cd647e, topic:knowledge-distillation
matched fp:aa33f5bf80cd647e, topic:rlhf
matched fp:aa33f5bf80cd647e, topic:awesome-list