search

zhaoyl18 / SEIKO

SEIKO is a novel reinforcement learning method to efficiently fine-tune diffusion models in an online setting. Our methods outperform all baselines (PPO, classifier-based guidance, direct reward backpropagation) for fine-tuning Stable Diffusion.
https://arxiv.org/abs/2402.16359
MIT License
14 stars 0 forks source link

Experiment in Molecules #1

Closed nmsl121381 closed 3 weeks ago

nmsl121381 commented 2 months ago

Hello, Zhao Yulai, Thank you very much for sharing your code for fine-tuning the diffusion model. But I noticed that this code base contains the results of fine-tuning the images and aesthetics. In your paper, Appendix D2 discusses the fine-tuning experiments on molecule generation, which does not seem to be in the existing code base. Can you continue to share the code related to the experiment of fine-tuning the molecular generation diffusion model? Thank you very much.

zhaoyl18 commented 3 weeks ago

Hi, Thanks for your interest in our work. We plan to release mol experiments after images and proteins.

Best,