Spaces:
Sleeping
Apply for community grant: Academic project (gpu)
SEED is a multimodal llm model, which supports both text generation and image generation in a unified framework. https://github.com/AILab-CVC/SEED
Could you help to assign a grant GPU for this demo? It requires about 32G*2 GPU memory. Thanks
Hi
@AILab-CVCer
Unfortunately, we don't have multi-GPU environment on HF infra. Also, A100 is not available for a GPU grant. Is it possible for you to optimize memory usage and make it fit in a single A10G?
cc:
@akhaliq
@ysharma
Hi @AILab-CVCer Unfortunately, we don't have multi-GPU environment on HF infra. Also, A100 is not available for a GPU grant. Is it possible for you to optimize memory usage and make it fit in a single A10G?
Got it. What is the largest GPU memory we can use for granted GPU?
@AILab-CVCer We can assign an A10G as a grant. It has 24GB of VRAM.
@AILab-CVCer We can assign an A10G as a grant. It has 24GB of VRAM.
Thanks. Could you please help to assign a 24GB A10G as a grant?
@AILab-CVCer I've just assigned it.
@AILab-CVCer I've just assigned it.
Thanks！
When we try to deploy using GPU, it seems that no GPU is available. Could you please help to check it?
@AILab-CVCer
Not 100% sure, but I think that's because you are using python:3.11 as your base Docker image here. I guess it can be fixed by using a Docker image with the NVIDIA driver installed.
You might want to refer to other Spaces that are using Docker. For example, this one may be helpful.