Diffusion Models & GRPO Reinforcement Learning
11 repos
Generative model optimization using reinforcement learning techniques, primarily focused on diffusion-based image generation and text-to-image synthesis. The cluster centers on GRPO (Group Relative Policy Optimization) and related RL methods applied to diffusion models for improving generation quality, with implementations spanning PyTorch, Paddle, and various visual generation tasks including dance motion synthesis. Repositories here explore how to leverage reward signals and policy optimization to enhance both text-to-image and visual generative models.
Python · 9
reinforcement-learning ·1,245
diffusion ·1,245
grpo ·1,245
text2image ·66
diffusers ·12
safetensors ·12
text-to-image ·4
