Spaces:
Sleeping
Sleeping
Apply for community grant: Academic project (gpu and storage)
#1
by
dinhanhx
- opened
I'm building a Gradio app to demonstrate my prompting vision language model for English and Vietnamese. It was for my master thesis. The code, the model, the data and everything else is publicly opened at https://github.com/dinhanhx/velvet/ .
The model weight is around 6 GiBs. And the GPU ram needed for inference is less than 12 GiBs. It has BLOOMZ 1.7B inside. Therefore, I would like to ask for NVIDIA T4 and a bit of persistent storage.
I just push gradio app with model. The inference time with CPU was very long (more than few minutes). Therefore, may I have access to NVIDIA T4?
It took around 1254 seconds ~ 20 minutes on average .
dinhanhx
changed discussion status to
closed