Awesome_MOPD

KO | EN

gitlite — search

#ios #awesome-list #devops #http #csharp #rust #reactjs #framework #pdf #cpp

(★ 9)

A curated list of papers and technical reports related to Multi-Teacher On-Policy Distillation (MOPD) and On-Policy Distillation (OPD).

File Explorer

ZIP Download

README.md

# Use via CDN

jsDelivr

jsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.

Version

File

Link

Example

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

A curated list of papers and technical reports related to **Multi-Teacher On-Policy Distillation (MOPD)** and On-Policy Distillation (OPD).

## 1. Blogs

- Multi-Teacher On-Policy Distillation [![Blog](https://img.shields.io/badge/Notion-Blog-000000.svg)](https://yumoxu.notion.site/multi-teacher-on-policy-distillation)

## 2. Papers

### Foundations & Capability Integration

- Consolidating RLVR Capabilities Across Domains: A Deep Dive into Fusion Paradigms [![arXiv](https://img.shields.io/badge/arXiv-2608.27409-b31b1b.svg)](http://arxiv.org/abs/2608.27409)
- Open-MOPD: Diagnosing and Fixing Capability Imbalance in Multi-Teacher On-Policy Distillation [![arXiv](https://img.shields.io/badge/arXiv-2608.19098-b31b1b.svg)](http://arxiv.org/abs/2608.19098)
- Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent [![arXiv](https://img.shields.io/badge/arXiv-2606.30616-b31b1b.svg)](http://arxiv.org/abs/2606.30616)
- MOPD: Multi-Teacher On-Policy Distillation for Capability Integration in LLM Post-Training [![arXiv](https://img.shields.io/badge/arXiv-2606.30406-b31b1b.svg)](http://arxiv.org/abs/2606.30406)
- Co-Evolving Policy Distillation [![arXiv](https://img.shields.io/badge/arXiv-2604.27083-b31b1b.svg)](http://arxiv.org/abs/2604.27083)
- ORBIT: On-policy Exploration-Exploitation for Controllable Multi-Budget Reasoning [![arXiv](https://img.shields.io/badge/arXiv-2601.08310-b31b1b.svg)](http://arxiv.org/abs/2601.08310)

### Methods & Optimization

- Eliciting Weak-to-Strong Generalization with On-Policy Reverse Distillation [![arXiv](https://img.shields.io/badge/arXiv-2609.08798-b31b1b.svg)](http://arxiv.org/abs/2609.08798)
- Learn from Whoever Is Right: Answer-Verified Multi-Teacher Distillation for Multi-Domain LLMs [![arXiv](https://img.shields.io/badge/arXiv-2609.02548-b31b1b.svg)](http://arxiv.org/abs/2609.02548)
- Preserving General Capabilities during Domain Specialization with Uncertainty-Calibrated MOPD [![arXiv](https://img.shields.io/badge/arXiv-2608.26735-b31b1b.svg)](http://arxiv.org/abs/2608.26735)
- D3-MOPD: Adaptive Dynamic Domain ScheDuling for Efficient Multi-Teacher Distillation [![arXiv](https://img.shields.io/badge/arXiv-2608.24987-b31b1b.svg)](http://arxiv.org/abs/2608.24987)
- Beyond the Best Teacher: Expanding and Compressing the Reasoning Solution Manifold [![arXiv](https://img.shields.io/badge/arXiv-2607.27770-b31b1b.svg)](http://arxiv.org/abs/2607.27770)
- Weak-to-Strong On-Policy Distillation [![arXiv](https://img.shields.io/badge/arXiv-2607.26246-b31b1b.svg)](http://arxiv.org/abs/2607.26246)
- H-OPD: Confidence Aware Heterogeneous Multi-Teacher Multimodal On-policy Distillation [![arXiv](https://img.shields.io/badge/arXiv-2607.02592-b31b1b.svg)](http://arxiv.org/abs/2607.02592)
- DOPD: Dual On-policy Distillation [![arXiv](https://img.shields.io/badge/arXiv-2606.30626-b31b1b.svg)](http://arxiv.org/abs/2606.30626)
- Counteraction-Aware Multi-Teacher On-Policy Distillation for General Capability Recovery with Domain Preservation [![arXiv](https://img.shields.io/badge/arXiv-2605.27115-b31b1b.svg)](http://arxiv.org/abs/2605.27115)
- Uni-OPD: Unifying On-Policy Distillation with a Dual-Perspective Recipe [![arXiv](https://img.shields.io/badge/arXiv-2605.03677-b31b1b.svg)](http://arxiv.org/abs/2605.03677)

### Analysis & Failure Modes

- Rethinking On-Policy Distillation of Large Language Models II: One Training Example [![arXiv](https://img.shields.io/badge/arXiv-2609.04172-b31b1b.svg)](http://arxiv.org/abs/2609.04172)
- Every Coin Has Two Sides: On the Dual Nature of Generalization in On-Policy Distillation of Large Language Models [![arXiv](https://img.shields.io/badge/arXiv-2608.16647-b31b1b.svg)](http://arxiv.org/abs/2608.16647)
- The Physics of Multi-Turn Long-Horizon Planning: From Pre-training to Post-training via Single- and Multi-Teacher On-Policy Agentic Distillation [![arXiv](https://img.shields.io/badge/arXiv-2607.24720-b31b1b.svg)](http://arxiv.org/abs/2607.24720)
- When Top-K Misses the Decision: Tool-Call Drift in Multi-Teacher On-Policy Distillation [![arXiv](https://img.shields.io/badge/arXiv-2607.07050-b31b1b.svg)](http://arxiv.org/abs/2607.07050)

### Applications In Different Domains

- Poly-OPD: Heterogeneous Multi-Teacher On-Policy Distillation for Capability-Selectable Flow Models [![arXiv](https://img.shields.io/badge/arXiv-2608.04349-b31b1b.svg)](http://arxiv.org/abs/2608.04349)
- Cross-Domain Hybrid OPD for Generalizable Search Agents [![arXiv](https://img.shields.io/badge/arXiv-2608.02101-b31b1b.svg)](http://arxiv.org/abs/2608.02101)
- UI-MOPD: Multi-Platform On-Policy Distillation for Continual GUI Agent Learning [![arXiv](https://img.shields.io/badge/arXiv-2607.04425-b31b1b.svg)](http://arxiv.org/abs/2607.04425)
- DanceOPD: On-Policy Generative Field Distillation [![arXiv](https://img.shields.io/badge/arXiv-2606.27377-b31b1b.svg)](http://arxiv.org/abs/2606.27377)
- CollectionLoRA: Collecting 50 Effects in 1 LoRA via Multi-Teacher On-Policy Distillation [![arXiv](https://img.shields.io/badge/arXiv-2605.25378-b31b1b.svg)](http://arxiv.org/abs/2605.25378)

## 3. Technical Reports

- Kimi K3: Open Frontier Intelligence [![arXiv](https://img.shields.io/badge/arXiv-2607.24653-b31b1b.svg)](http://arxiv.org/abs/2607.24653)
- Solar Open 2 Technical Report [![arXiv](https://img.shields.io/badge/arXiv-2607.20062-b31b1b.svg)](http://arxiv.org/abs/2607.20062)
- OvisOCR2 Technical Report [![arXiv](https://img.shields.io/badge/arXiv-2607.13639-b31b1b.svg)](http://arxiv.org/abs/2607.13639)
- Mach-Mind-4-Flash Technical Report [![arXiv](https://img.shields.io/badge/arXiv-2607.09375-b31b1b.svg)](http://arxiv.org/abs/2607.09375)
- KAT-Coder-V2.5 Technical Report [![arXiv](https://img.shields.io/badge/arXiv-2607.05471-b31b1b.svg)](http://arxiv.org/abs/2607.05471)
- Unified Audio Intelligence Without Regressing on Text Intelligence [![arXiv](https://img.shields.io/badge/arXiv-2607.05196-b31b1b.svg)](http://arxiv.org/abs/2607.05196)
- Qwen-Image-2.0-RL Technical Report [![arXiv](https://img.shields.io/badge/arXiv-2606.27608-b31b1b.svg)](http://arxiv.org/abs/2606.27608)
- NebulaExp-8B: An Empirical Post-Training Pipeline via Full-Scale Ablation Research [![arXiv](https://img.shields.io/badge/arXiv-2606.26671-b31b1b.svg)](http://arxiv.org/abs/2606.26671)
- DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence [![arXiv](https://img.shields.io/badge/arXiv-2606.19348-b31b1b.svg)](http://arxiv.org/abs/2606.19348)
- Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning [![arXiv](https://img.shields.io/badge/arXiv-2606.15007-b31b1b.svg)](http://arxiv.org/abs/2606.15007)
- Kwai Keye-VL-2.0 Technical Report [![arXiv](https://img.shields.io/badge/arXiv-2606.10651-b31b1b.svg)](http://arxiv.org/abs/2606.10651)
- Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation [![arXiv](https://img.shields.io/badge/arXiv-2603.19220-b31b1b.svg)](http://arxiv.org/abs/2603.19220)
- GLM-5: from Vibe Coding to Agentic Engineering [![arXiv](https://img.shields.io/badge/arXiv-2602.15763-b31b1b.svg)](http://arxiv.org/abs/2602.15763)
- Baichuan-M3: Modeling Clinical Inquiry for Reliable Medical Decision-Making [![arXiv](https://img.shields.io/badge/arXiv-2602.06570-b31b1b.svg)](http://arxiv.org/abs/2602.06570)
- MiMo-V2-Flash Technical Report [![arXiv](https://img.shields.io/badge/arXiv-2601.02780-b31b1b.svg)](http://arxiv.org/abs/2601.02780)

## 4. Contributing

Feel free to open a pull request to add related works.

Awesome_MOPD

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

Download README