Mark APIs in app framework as experimental

Lightning-AI / pytorch-lightning

Pretrain, finetune ANY AI model of ANY size on multiple GPUs, TPUs with zero code changes.

https://lightning.ai

Apache License 2.0

28.54k stars 3.39k forks source link

Open awaelchli opened 2 years ago

awaelchli commented 2 years ago

Decide which parts of the app framework are experimental and change rapidly. Mark them as experimental in the docs and or source code.

cc @borda @rohitgr7 @Borda @tchaton @carmocca

Borda commented 2 years ago

@lantiga do you have a suggestion where and eventually what all classes all be marked so?

awaelchli commented 2 years ago

Example of such changes that happen quickly after the release: #15314 Marking the class there experimental would be good to justify the removals.

Borda commented 2 years ago

shall be done for #15547

lantiga commented 2 years ago

All of app should be marked as experimental, we don't make any commitment to not ship changes in patch releases for this part.

lantiga commented 2 years ago

How would we go about marking?

I'm good with standardized a string / tag in docstring? Or should we reach out to something more involved like a decorator?