Open pmeier opened 2 years ago
Hi @pmeier,
I'd like to work on this.
Hey @federicopozzi33 and thanks for the interest. Unfortunately, we are currently not looking for contributions on porting more datasets. We need to make sure that the new architecture works well in an end-to-end setting. For that we are still running benchmarks.
Furthermore, torchdata
is in the process of introducing a new data loader: https://github.com/pytorch/data/tree/main/torchdata/dataloader2. Before that is not stable-ish, we cannot run our own reference scripts and thus have little idea if our current dataset design is sufficient or not.
If all these blockers are resolved, we will happily accept a PR porting this dataset to the new API. I'll ping you here when this is the case and if you are still interested then, you can pick it up.
cc @pmeier @bjuncek