Carlos Mocholí
Carlos Mocholí
Can you share your LightningModule code?
You need to define the metrics with a `ModuleDict`: https://torchmetrics.readthedocs.io/en/stable/pages/overview.html#metrics-and-devices
Can you share the colab link now? Does it not happen on CPU or GPU?
Any ideas @kaushikb11? A bug impacting optimization would be of high priority.
Probably a duplicate of https://github.com/Lightning-AI/lightning/issues/13108 but for Lite
Related internal discussion: https://linear.app/lightning-ai/issue/LAI2-10220/lightning-quickstart-app-experience-feedback
Looks to me like this is a duplicate of the proposal in https://github.com/Lightning-AI/lightning/issues/13577
@awaelchli In this PR, `tpu_available` is deleted in favor of `TPUAccelerator.is_available()`. In most places, we want to check `XLA_AVAILABLE` as that's what needs to be `True` to successfully import `torch_xla`....
@awaelchli I'll need you to run MPS tests locally with this branch
If this is going to be sequenced so the properties are introduced first, and then used. I would like that you open the follow-up PR already so that we can...