Skip to content
This repository was archived by the owner on May 1, 2026. It is now read-only.

Fix #21 and add export functionality.#22

Open
classner wants to merge 1 commit into
EmbarkStudios:mainfrom
classner:pr-fix-21
Open

Fix #21 and add export functionality.#22
classner wants to merge 1 commit into
EmbarkStudios:mainfrom
classner:pr-fix-21

Conversation

@classner
Copy link
Copy Markdown

Hi folks!

Thanks for the great work with the torch plugin! I had a stab at fixing the problem in #21 , and also added the functionality for pdm torch export [...] so that requirement files can be generated.

In the process I noticed that the +cpu build identifier is unfortunately not universal (it doesn't work for Mac builds) and think it's okay to omit it for CPU builds, just let me know if that's not the case.

@classner classner requested a review from tgolsson as a code owner February 26, 2024 04:00
@polarathene
Copy link
Copy Markdown

I noticed that the +cpu build identifier is unfortunately not universal (it doesn't work for Mac builds)

+cpu local identifier is only assigned to x86_64 (AMD64) packages, PyTorch doesn't publish the aarch64 (ARM64) packages with that as there is no other GPU variants supported on that platform that PyTorch publishes (if you need CUDA compatible torch packages for aarch64 hardware like the Jetson, you'd need to install torch another way such as the containers nvidia publishes).

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants