Closed shcheklein closed 9 months ago
Actually, I think it's better to be fixed right in the #749 since we are changing the lighting module there and here CI fails with mypy on it.
@shcheklein What issue did you hit using pytorch-lightning
? The codebase tries to be compatible with both lightning
and pytorch-lightning
, so wondering if there is something we need to address here.
I think the dvclive[all]
just doesn't install it. Since there is not pytorch-lightning
anymore AFAIR.
Okay, I see. It looks like they both still exist but lightning
is a superset that includes both pytorch-lightning
and fabric
. See https://github.com/Lightning-AI/pytorch-lightning/discussions/16688#discussioncomment-6750981.
To clarify, I think it us:
tf = ["tensorflow"]
xgb = ["xgboost"]
lgbm = ["lightgbm"]
huggingface = ["transformers", "datasets"]
catalyst = ["catalyst>22"]
fastai = ["fastai"]
lightning = ["lightning>=2.0", "torch"]
optuna = ["optuna"]
all = [
"dvclive[image,mmcv,tf,xgb,lgbm,huggingface,catalyst,fastai,lightning,optuna,plots,markdown]"
]
mind the lightning
and it's definition. all
was referring to non-existent extra in our config ASFAIU.
🤦 Got it, thanks for the clarification
Found a small issue while reviewing https://github.com/iterative/dvclive/pull/749
[X] ❗ I have followed the Contributing to DVCLive guide.
[X] 📖 If this PR requires documentation updates, I have created a separate PR (or issue, at least) in dvc.org and linked it here.
Thank you for the contribution - we'll try to review it as soon as possible. 🙏