---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: label
    dtype:
      class_label:
        names:
          '0': '0'
          '1': '1'
          '2': '2'
          '3': '3'
          '4': '4'
          '5': '5'
          '6': '6'
          '7': '7'
          '8': '8'
          '9': '9'
  - name: source_index
    dtype: int32
  splits:
  - name: train
    num_bytes: 17155689
    num_examples: 60000
  - name: test
    num_bytes: 2555832
    num_examples: 8920
  download_size: 19348835
  dataset_size: 19711521
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
---

# tsilva/mnist

## Dataset Summary

This dataset is a Hugging Face packaged version of the classic MNIST handwritten digits benchmark. It contains grayscale 28x28 images of digits `0` through `9`, split into the standard `train` and `test` partitions.

Each row includes:

- `image`: a 28x28 grayscale digit image
- `label`: the digit class from `0` to `9`
- `source_index`: the original position of the example inside the source MNIST split

## Splits

- `train`: 60,000 examples
- `test`: 8,920 examples, balanced to `892` examples per class

## Source

The underlying data comes from the original MNIST release maintained by Yann LeCun and collaborators and downloaded here through `torchvision.datasets.MNIST`.

- MNIST homepage: http://yann.lecun.com/exdb/mnist/
- TorchVision dataset docs: https://pytorch.org/vision/stable/generated/torchvision.datasets.MNIST.html

## Intended Use

This dataset is suitable for standard handwritten digit classification baselines, representation learning experiments, and as a clean reference set for synthetic corruption studies.

## Load Example

```python
from datasets import load_dataset

ds = load_dataset("tsilva/mnist")
print(ds["train"][0])
```