🚀 The feature

Currently HuggingFaceHubReaderIterDataPipe is a wrapper for load_dataset() (https://huggingface.co/docs/datasets/v2.6.1/en/package_reference/loading_methods#datasets.load_dataset). load_dataset() however supports many more optional arguments.
If we want to fully support downloading from HuggingFace we need to add these options to HuggingFaceHubReaderIterDataPipe as well.

Motivation, pitch

E.g. the oscar-dataset requires you to also provide a name, which is not possible when using HuggingFaceHubReaderIterDataPipe.

I could provide a fix for this.

Alternatives

No response

Additional context

No response