https://github.com/Corleno/KEPO

KEPO: Knowledge-Enhanced Preference Optimization for Reinforcement Learning with Reasoning
https://github.com/Corleno/KEPO

Last synced: 15 days ago
JSON representation

KEPO: Knowledge-Enhanced Preference Optimization for Reinforcement Learning with Reasoning

• Host: GitHub
• URL: https://github.com/Corleno/KEPO
• Owner: Corleno
• Created: 2025-11-17T03:50:42.000Z (8 months ago)
• Default Branch: main
• Last Pushed: 2026-03-26T08:32:54.000Z (3 months ago)
• Last Synced: 2026-03-27T01:32:28.514Z (3 months ago)
• Language: Jupyter Notebook
• Homepage:
• Size: 33 MB
• Stars: 2
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesomeopd - KEPO - the-badge&logo=github&logoColor=white&labelColor=181717&color=ffd700" alt="Stars"> | 2026.01 | Industrial | [arXiv 2602.00400](https://arxiv.org/abs/2602.00400) | KEPO | (🤝 OPD-RL Hybrids — Inside-RL OPD / 🔁 Iterative Self-Bootstrapping)

README

          
KEPO: Knowledge-Enhanced Preference Optimization for Reinforcement Learning with Reasoning


This code is built upon the **Med-R1** repo, it supports GKD, GRPO and KEPO algorithns on sota models such as Qwen-3VL. The dataset mainly focus on the multimodal medical data.