


























I just updated this package to the latest version on PyPI (1.8.0.post3) and was able to import pytorch_lightning into a REPL after installation.
I had to add a new dependency, python-lightning-utils, but otherwise did not see that any other dependencies were unavailable as previous comments indicated. I did run namcap, but I have only installed this on my desktop PC and one server so far. Please alert me to any issues you encounter.
I get a ModuleNotFoundError: No module named 'deprecate' error
Orphaned AUR Packages: python-torchmetrics
Dependencies orphaned?
Seem to lose some dependency?
ModuleNotFoundError: `jsonargparse` is not installed but it is required for the CLI. Install it with `pip install -U jsonargparse[signatures]`.
Hi @hottea, could you list python-torchvision as optdepends? It's not mentioned in their requirements.txt, only in requirements/examples.txt, and it's quite the large package that has to be compiled at install. Thank you for maintaining the package :)
hi @hottea :). Pytorch_lightning makes its metrics module into a new package torchmetrics. According to requirements.txt of pytorch_lightning, it should be added into depends of PKGBUILD. If it is missing, the code will raise ModuleNotFoundError: No module named 'torchmetrics' when you used metrics from pytorch_lightning.
@hottea tensorboard should not be an optdepend, as it is imported by many pl modules even if not used. If tensorboard is missing, most code will not run.
@tobol Would you like to report this to upstream?
The folder /usr/lib/python3.9/site-packages/tests should be deleted in the package() step
此内容由惯性聚合(RSS阅读器)自动聚合整理,仅供阅读参考。 原文来自 — 版权归原作者所有。
Missing a dependency?
Also, the official package works with
but the arch version you have to do