---
dataset_info:
  features:
  - name: task_id
    dtype: string
  - name: train
    list:
    - name: input
      list:
        list: int64
    - name: output
      list:
        list: int64
  - name: test_input
    list:
      list: int64
  - name: test_output
    list:
      list: int64
  splits:
  - name: train
    num_bytes: 3631364
    num_examples: 386
  - name: evaluation
    num_bytes: 5994736
    num_examples: 381
  download_size: 340716
  dataset_size: 9626100
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: evaluation
    path: data/evaluation-*
---

# ARC AGI Dataset

![image/png](https://cdn-uploads.huggingface.co/production/uploads/63f2b7bcbe95ed4c9a9e7669/_xxU21Yh5_L1XRvtAgAIa.png)