FS#65604 - [python-pytorch-cuda] has too many dependencies
Attached to Project:
Community Packages
Opened by Jacob Gorm Hansen (jacobgorm) - Sunday, 23 February 2020, 19:51 GMT
Last edited by Sven-Hendrik Haase (Svenstaro) - Monday, 16 March 2020, 14:15 GMT
Opened by Jacob Gorm Hansen (jacobgorm) - Sunday, 23 February 2020, 19:51 GMT
Last edited by Sven-Hendrik Haase (Svenstaro) - Monday, 16 March 2020, 14:15 GMT
|
Details
Description:
Compared to the official releases of pytorch from pytorch.org, this package depends on QT5, FFMPEG, and OpenCV, all of which have large dependency trees by themselves. This precludes the use of arch's pytorch in kind of docker or AWS scenario. My suggestion would be to mirror the config settings that come default with pytorch.org's PIP packages, instead of enabling every possible option. |
This task depends upon
Closed by Sven-Hendrik Haase (Svenstaro)
Monday, 16 March 2020, 14:15 GMT
Reason for closing: Won't fix
Monday, 16 March 2020, 14:15 GMT
Reason for closing: Won't fix
Let's discuss.
Wrt using in Docker, e.g., for pushing images to the cloud, you will want as few dependencies as possible, to shorten your docker image build times and the amount of data to down- and upload.