You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for having a look. I believe I can already work around the first of the two linked issues, and I'm not sure the last one applies.
My core issue seems to be that the version of torch already installed has a pre-release identifier (I don't think the local version identifier is the issue here). I'm trying to use the system version by pinning torch to this version (2.0.0a0), but uv pip compile fails since it can't find it in any package index. I'm not sure of the best way to handle this.
Hi,
I'm currently working with the docker image
l4t-pytorch
, which already includes certain system libraries liketorch==2.0.0a0+ec3941ad.nv23.2
.My goal is to add extra dependencies on top of this and manage them with uv, while pinning the preinstalled version of torch.
I'm trying to this via a venv with system-site-packages and declaring a dependency on
torch==2.0.0a0
, which fails since it can't resolve dependencies.Is there any trick to do this or something I might have overlooked?
Thanks
The text was updated successfully, but these errors were encountered: