Datasets:
Tasks:
Image Classification
Formats:
parquet
Sub-tasks:
multi-class-image-classification
Size:
< 1K
License:
Convert to parquet (loading scripts are no longer supported)
#5
by davanstrien HF Staff - opened
datasets no longer supports dataset loading scripts, so load_dataset('biglam/illustrated_ads') fails with RuntimeError: Dataset scripts are no longer supported.
This PR converts the dataset to parquet and removes illustrated_ads.py. The schema, the ads.csv/sample.csv join, and the box/ocr parsing are all carried over from the loading script unchanged. Images are embedded in the parquet instead of being referenced by path, so the dataset no longer depends on Zenodo being reachable at load time.
Source data unchanged: https://doi.org/10.5281/zenodo.5838410
davanstrien changed pull request status to merged