search

MingSun-Tse / Collaborative-Distillation

[CVPR'20] Collaborative Distillation for Ultra-Resolution Universal Style Transfer (PyTorch)
MIT License
185 stars 23 forks source link

CUDA + cuDNN #7

Closed nickyvdz-art closed 4 years ago

nickyvdz-art commented 4 years ago

Hi,

Thank you so much for this! Not used it yet, looking forward to it. Unfortunately I am not a coder and quite clueless and new to all of this (python etc.), however I did get cysmith/neural-style-tf to work and can do about 1920px max.

So, apologies for any 'silly' questions - what versions of CUDA + cuDNN should be installed for this?

Many thanks again

MingSun-Tse commented 4 years ago

Hi,

On my desktop, it is CUDA10.2 without cuDNN (it's fine if you don't install it). CUDA mainly depends on the OS and driver version. So it is really case by case. A good try is to use anaconda to install the old-version pytorch (https://pytorch.org/get-started/previous-versions/). It will take care of the most version issues.

Best,

nickyvdz-art commented 4 years ago

Thank you sir for the quick reply; I will try to use it this evening; kind regards