Cycle option for StreamingDataLoader

[Aceticia]

🚀 Feature

A function or an argument in StreamingDataLoader to cycle the passed in StreamingDataset.

Motivation

Many training scenarios in CV involve training models with multiple epochs, while wanting to control the exact number of steps being trained, independent of the underlying dataset size. E.g., given a CombinedStreamingDataset of some length, restart its iterations when it is exhausted.

Pitch

I'm not quite sure how this should be done - maybe in iter method of StreamingDataLoader, we can catch the final iteration and restart it?

[tchaton]

You could check PyTorch Lightning Cycle Loaders: https://lightning.ai/docs/pytorch/stable/api/lightning.pytorch.utilities.combined_loader.html

Or create your own wrapper that iterates for a given number of steps.