rclosson

Two import statements fill up my entire 50 GB GPU (independently of each other): `from alibi_detect.cd import SpotTheDiffDrift` and `from alibi_detect.utils.pytorch.data import TorchDataset` It seems odd to me that a...

alibi-detect: 0.9.0 tensorflow: 2.8.0 torch: 1.9.0+cu111  Likely relevant info: I'm running in a Jupyter notebook

I'm running on an NVIDIA Quadro RTX 8000, which is allocated to me through the determined.ai interface of sharing GPUs among AI researchers. (Sorry if I'm using the wrong terms...

Sorry, I'm not sure where I can test outside of a notebook given my specific environment (and the fact that I use shared/pooled resources with my colleagues, so it may...

SpotTheDiffDrift also not taking up all my GPU memory on the first import statement anymore. Thanks again.