Spaces:

huzey

/

ncut-pytorch

like 7

Running on Zero

Hi HF team,

Is that possible for us to use the Dynamic GPU?
We am building a spectral clustering toolkit, it can be applied to unsupervised segmentation.

We have setup the space, it's running on the free-tier CPU instance now.
The model is optimized for speed, it can run in under 60s in the free 2 vCPU machine. We also implemented caching to reduce the compute loads.

The project page is here: https://ncut-pytorch.readthedocs.io/en/latest/
Example usages:

Best,
Huzheng