---
dataset_info:
  features:
  - name: image_id
    dtype: int64
  - name: caption_id
    dtype: int64
  - name: caption
    dtype: string
  - name: height
    dtype: int64
  - name: width
    dtype: int64
  - name: file_name
    dtype: string
  - name: coco_url
    dtype: string
  - name: image
    dtype: image
  splits:
  - name: train
    num_bytes: 95110372499.827
    num_examples: 591753
  - name: validation
    num_bytes: 3666613561.754
    num_examples: 25014
  - name: test
    num_bytes: 6468072328.66
    num_examples: 40670
  download_size: 74303820184
  dataset_size: 105245058390.241
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
---

The [COCO](https://huggingface.co/datasets/ydshieh/coco_dataset_script) dataset without needing to run remote code, so it is compatible with datasets >= 4.0.0.