AV-GRPO

AV-GRPO is a modality-anchored online diffusion reinforcement learning framework for joint audio-video generation. It enables full-parameter or LoRA training of the 22B LTX-2.3 model on 8 A800 GPUs.

This repository contains the model introduced in AV-GRPO: Modality-Anchored Decoupling Diffusion Reinforcement Learning for Joint Audio-Video Generation.

For training, inference, and usage details, please refer to the GitHub repository.