Datasets:
Convert data to HF datasets format
#9
by
thompsonmj
- opened
Currently, the data in this repository is provided as split ZIP files which must be concatenated and unzipped before use.
This format works, but is non-standard and possibly inconvenient.
The idea is to convert this into an HF datasets
library native format, providing access directly through the datasets
library. E.g.:
from datasets import load_dataset
ds = load_dataset("imageomics/kabr")
We will also provide an optional script to 'export' the data into its original hierarchical directory structure for a potentially less abstract, more accessible format, which will also be compatible with the image2*.py
utilities.