Loading a h5ad file with HF datasets

🤗Datasets
1

Hi,
I’m working with files generated via the AnnData package, i.e my dataset is a .h5ad file containing dataframes and matrices.

I don’t see the .h5ad format in the list of HF Datasets supported extensions.

Can I still load a dataset of this format?

Best,
Nathan

2

Hi!

We don’t have a built-in H5 loader, so you can either use Dataset.from_generator to load the data or Dataset.from_dict/Dataset.from_pandas if the data is small enough to fit in memory