CUDA 10.2 and TF-gpu 2.1 support

churchlab / UniRep

UniRep model, usage, and examples.

338 stars 96 forks source link

CUDA 10.2 and TF-gpu 2.1 support #19

Closed ptynecki closed 3 years ago

ptynecki commented 4 years ago

Good morning,

I'm wondering if the UniRep dev core team has a plan to upgrade the code base and support the newest CUDA, cuDNN and Tensorflow-GPU environment?

Thank you for your reply.

Umbriona commented 4 years ago

You could try the conversion tool that tensorflow provides to upgrade projects from TF1 to TF2.

surgebiswas commented 3 years ago

Sorry for the slow response, but we are not developing further on this repository. I'd refer you to a JAX re-implementation that others in the community are maintaining: https://github.com/ElArkk/jax-unirep. Of course this isn't in Tensorflow, but it's more up to date and likely compatible with newest CUDA/cuDNN.