Spaces:
Running
on
L40S
Apply for community grant: Academic project (gpu)
Dear Huggingface team,
We are the authors of "StdGEN: Semantic-Decomposed 3D Character Generation from Single Images" (CVPR 2025):
Project page: https://stdgen.github.io/
Huggingface page: https://huggingface.co/hyz317/StdGEN
We’ve recently released the interactive demo of our work on Huggingface. The demo has garnered community interest, including retweets/shoutouts from AK and Gradio’s official Twitter, highlighting its potential impact. Due to high demand, we’re now seeking community grant of L40S GPU resource to ensure stable, scalable performance for users experimenting with our model.
We’d be deeply grateful for your support in provisioning this hardware. Please let us know if any additional information would help evaluate our request.
Thank you for your time and for fostering such a vibrant ML community!
Hi @hysts ,
Thank you for your quick response! We truly appreciate the initial ZeroGPU grant—it’s been invaluable for early testing. we observed that ZeroGPU takes ~2x longer per case compared to L40S (4.2min vs. 2min). Given the ZeroGPU quota limits, this means users can currently generate only one model per day with ZeroGPU, which significantly impacts the interactivity of the demo.
We’d be extremely grateful if an L40S grant could be considered to improve performance. If that’s not possible right now, could you please help us switch back to ZeroGPU? (It seems this requires HF team assistance.)
Thanks again for your support!