Skip to content

Dataset · Voxel51

Extended MNIST Dataset with Embeddings and Predictions

Voxel51/curated-mnist

Classic 70,000 MNIST digit images augmented with model embeddings and predictions.

What's inside

The dataset is based on the classic MNIST dataset (60,000 training and 10,000 test images) with English labels and an additional training set split. The data is enriched with model predictions, embeddings, and analytical fields generated in the FiftyOne library for a PyTorch data curation course. File formats are not specified in the description.

What you can do with it

  • Training and debugging baseline image classification models
  • Practicing label error detection and embedding analysis in FiftyOne
  • Testing PyTorch data curation pipelines
Tasks
Image classification
Languages
English
Size
10K–100K rows
Downloads this month
54.4K
Likes
0

How to load it

from datasets import load_dataset

ds = load_dataset("Voxel51/curated-mnist")