---
dataset_info:
  features:
  - name: text
    dtype: string
  - name: labels
    sequence: int64
  splits:
  - name: train
    num_bytes: 5011317.077050539
    num_examples: 22208
  - name: test
    num_bytes: 1253054.9229494615
    num_examples: 5553
  download_size: 2405205
  dataset_size: 6264372
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
license: mit
task_categories:
- text-classification
task_ids:
- multi-label-classification
language:
- en
tags:
- mtg
- multilabel
- magic
pretty_name: Magic the Gathering Color Identity Multilabel Classification
size_categories:
- 10K<n<100K
---
This dataset was made specifically for multilabel classification using the following process:

1. Downloading https://mtgjson.com/api/v5/AtomicCards.json.bz2 on January 10, 2024
2. Encoding color identity of each card into the `labels` feature
```python
colors = ['B', 'G', 'R', 'U', 'W']
b = [1, 0, 0, 0, 0]
bw = [1, 0, 0, 0, 1]
gru = [0, 1, 1, 1, 0]
# and so on
```
3. Concatenating card name and card text into the `text` feature
4. `split = ds['train'].train_test_split(test_size=0.2)`
5. `split.push_to_hub("mtg-coloridentity-multilabel-classification")`