Question on the RL performance on wan2.2.

[chenkang455]

Hi Hunyuan Team,

Thanks for open-sourcing this great work! I was wondering where I could find the evaluation results for the RL-finetuned models, such as Wan2.2. Is there a table comparing the performance before and after RL fine-tuning?

Thanks again!

[Jayce-Ping]

Since video models are heavy and the rollout is also expensive, it takes a while to see the fully converage with the latest codebase. We will update the results later in this channel.