mnist
The MNIST dataset consists of 70,000 28x28 black-and-white images in 10 classes (one for each digits), with 7,000 images per class. There are 60,000 training images and 10,000 test images.
Use this dataset
Load it over the Pelican Platform with Hugging Face
datasets.
# Consume with Hugging Face's `datasets` package
from datasets import load_dataset
dataset = load_dataset("parquet", data_files="pelican://uwdf-director.chtc.wisc.edu/wisc.edu/dsi/pytorch/mnist/train.parquet")
# Convert to the format of your choice, see
# https://huggingface.co/docs/datasets/use_with_pytorch
torch_dataset = dataset.with_format("torch")
torch_datasetplanned
A shorter form is planned once uploader identity and NetID integration land:
from pelican_data_loader import load_uw_data
dataset = load_uw_data("netid/dataset_key")
# Display the first record
list(dataset["train"].take(1))Not yet implemented — shown as a preview of the intended interface.
Danger zone
Deleting removes this dataset from the repository database and deletes its data and metadata files from S3.