Apply for community grant: Academic project (gpu)

#1
by errai34 - opened
UniverseTBD org

Dear Hugging Face Team,

We've just fine-tuned the first multi-modal large model for Astronomy, the AstroLlava, and we are now writing the paper to submit it to the ML in the Wild ICML 2024 Workshop. To evaluate and release the model for the community, we would greatly benefit from a community grant for an NVidia A10G large, which is where the model is currently running from. We are aware that the resources are limited, and we don't want to overstep our bounds, but we think this model will truly inspire the community to think about the great beyond and it would be great to be able to host it freely here.

Our project UniverseTBD is currently being supported by the commitment of a few junior researchers with limited funding. This would really make a difference for our mission to democratise science for everyone. Thank you so much.

Kindest regards,
the UTBD Team

Hi @errai34 , we've assigned a10g to this Space as a grant for now. I think this Space is based on the LLaVA 1.6 Space, but would it be possible to migrate to code like this Space? The reason I'm asking this is because we recently started using ZeroGPU as the default hardware for grants, but the original LLaVA 1.6 Space is launching a separate process for model serving, which is not compatible with ZeroGPU, while the LLaVA-NeXT Space is compatible with Zero. Also, now gr.ChatInterface supports multimodal inputs, so it would be nice to use it.
I'll add you to the ZeroGPU explorers org so you can assign ZeroGPU yourself, so it would be nice if you could see if you can migrate your Space to use ZeroGPU.

UniverseTBD org

You are a star saver thank you so so much. I will talk to our team and see how we can move there! Thank you soooo much!!

Sign up or log in to comment