jieliu/SD3.5-M-Flow-GRPO · Apply for community grant: Academic project (gpu)

Owner May 14

We propose Flow-GRPO, the first method integrating online reinforcement learning (RL) into flow matching models. Empirically, Flow-GRPO is effective across multiple text-to-image tasks. For complex compositions, RL-tuned SD3.5 generates nearly perfect object counts, spatial relations, and fine-grained attributes, boosting GenEval accuracy from 63% to 95%. In visual text rendering, its accuracy improves from 59% to 92%, significantly enhancing text generation. Flow-GRPO also achieves substantial gains in human preference alignment.

Our repository (https://github.com/yifan123/flow_grpo) has received 459 stars in just 5 days.

hysts

May 14

Hi @jieliu , looks like the Space is already using ZeroGPU. Didn't ZeroGPU work?

jieliu

Owner May 15

Yes, it seems no space GPU detected when I restart my space

hysts

May 15

@jieliu Decorating your function that uses GPU with @spaces.GPU would probably fix the issue. You might want to check out this page.

jieliu

Owner May 15

Thanks for your replay. I have fix this bug and zeroGPU works.