2 0 Advanced

.. list-table:: adv. user 2.0 :widths: 40 40 20 :header-rows: 1

• • If
  • Then
  • Ref
• • used the torchdistx package and integration in Trainer
  • materialize the model weights manually, or follow our :doc:guide for initializing large models <../../advanced/model_init>
  • PR17995_
• • defined def training_step(self, dataloader_iter, batch_idx) in LightningModule
  • remove batch_idx from the signature and expect dataloader_iter to return a triplet (batch, batch_idx, dataloader_idx)
  • PR18390_
• • defined def validation_step(self, dataloader_iter, batch_idx) in LightningModule
  • remove batch_idx from the signature and expect dataloader_iter to return a triplet (batch, batch_idx, dataloader_idx)
  • PR18390_
• • defined def test_step(self, dataloader_iter, batch_idx) in LightningModule
  • remove batch_idx from the signature and expect dataloader_iter to return a triplet (batch, batch_idx, dataloader_idx)
  • PR18390_
• • defined def predict_step(self, dataloader_iter, batch_idx) in LightningModule
  • remove batch_idx from the signature and expect dataloader_iter to return a triplet (batch, batch_idx, dataloader_idx)
  • PR18390_
• • used batch = next(dataloader_iter) in LightningModule *_step hooks
  • use batch, batch_idx, dataloader_idx = next(dataloader_iter)
  • PR18390_
• • relied on automatic detection of Kubeflow environment
  • use Trainer(plugins=KubeflowEnvironment()) to explicitly set it on a Kubeflow cluster
  • PR18137_

.. _pr17995: https://github.com/Lightning-AI/pytorch-lightning/pull/17995 .. _pr18390: https://github.com/Lightning-AI/pytorch-lightning/pull/18390 .. _pr18137: https://github.com/Lightning-AI/pytorch-lightning/pull/18390