---
license: mit
dataset_info:
  features:
  - name: image
    dtype: image
  - name: text
    dtype: string
  splits:
  - name: train
    num_bytes: 2343966.0
    num_examples: 1000
  download_size: 2314338
  dataset_size: 2343966.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
- config_name: metadata
  data_files: metadata.jsonl
---
# Dataset Card for Cellular Automata

## Dataset Details

This dataset contains 1000 labeled images of cellular automata.

### Dataset Description

[Cellular Automaton](https://mathworld.wolfram.com/ElementaryCellularAutomaton.html) were described by
Stephen Wolphram in [A New King of Science](https://www.wolframscience.com/nks/).
Imagine you have a grid, like a checkerboard. Each square in the grid has a state - on or off, with the state of the square determining its color. There are rulesets (256 of them) that describe how the squares change their state depending on what's happening around them.

The python library [CellPyLib](https://github.com/lantunes/cellpylib) was used to generate the labeled images.

- **Curated by:** Kathy McGuiness

## Uses

One toy use-case is fine-tuning the [aMUSEd](https://huggingface.co/blog/amused) model.

## Dataset Creation

The dataset was created as a demo on how on to create a labeled image dataset.