ai2dキャプション
コレクションでコンテンツを整理
必要に応じて、コンテンツの保存と分類を行います。
このデータセットは主に AI2D データセットに基づいています (ここを参照)。
AI2D-Caption データセットのアノテーション プロセスについては、論文のセクション 4.1を参照してください。
スプリット | 例 |
---|
'auditor_llm_training_examples' | 30 |
'gpt4v' | 4,903 |
'llava_15' | 4,902 |
'planner_llm_training_examples' | 30 |
'test' | 75 |
FeaturesDict({
'caption': Text(shape=(), dtype=string),
'entities': Sequence({
'bounds': BBoxFeature(shape=(4,), dtype=float32),
'cat': ClassLabel(shape=(), dtype=int64, num_classes=10),
'from': Text(shape=(), dtype=string),
'id': Text(shape=(), dtype=string),
'label': Text(shape=(), dtype=string),
'to': Text(shape=(), dtype=string),
'type': ClassLabel(shape=(), dtype=int64, num_classes=5),
}),
'image': Image(shape=(None, None, 3), dtype=uint8, description=The image of the diagram.),
'image_filename': Text(shape=(), dtype=string),
'layout': ClassLabel(shape=(), dtype=int64, num_classes=7),
'relationships': Sequence(Text(shape=(), dtype=string)),
'topic': ClassLabel(shape=(), dtype=int64, num_classes=4),
})
特徴 | クラス | 形 | Dタイプ | 説明 |
---|
| 特徴辞書 | | | |
キャプション | 文章 | | 弦 | |
エンティティ | 順序 | | | |
エンティティ/境界 | BBox機能 | (4,) | float32 | |
エンティティ/猫 | クラスラベル | | int64 | |
エンティティ/から | 文章 | | 弦 | |
エンティティ/ID | 文章 | | 弦 | |
エンティティ/ラベル | 文章 | | 弦 | |
エンティティ/宛先 | 文章 | | 弦 | |
エンティティ/タイプ | クラスラベル | | int64 | |
画像 | 画像 | (なし、なし、3) | uint8 | ダイアグラムのイメージ。 |
画像ファイル名 | 文章 | | 弦 | 画像のファイル名。例: "1337.png" |
レイアウト | クラスラベル | | int64 | |
人間関係 | シーケンス(テキスト) | (なし、) | 弦 | |
トピック | クラスラベル | | int64 | |

@inproceedings{Zala2024DiagrammerGPT,
author = {Abhay Zala and Han Lin and Jaemin Cho and Mohit Bansal},
title = {DiagrammerGPT: Generating Open-Domain, Open-Platform Diagrams via LLM Planning},
year = {2024},
booktitle = {COLM},
}
特に記載のない限り、このページのコンテンツはクリエイティブ・コモンズの表示 4.0 ライセンスにより使用許諾されます。コードサンプルは Apache 2.0 ライセンスにより使用許諾されます。詳しくは、Google Developers サイトのポリシーをご覧ください。Java は Oracle および関連会社の登録商標です。
最終更新日 2024-11-18 UTC。
[null,null,["最終更新日 2024-11-18 UTC。"],[],[],null,["# ai2dcaption\n\n\u003cbr /\u003e\n\n- **Description**:\n\nThis dataset is primarily based off the AI2D Dataset (see\n[here](https://prior.allenai.org/projects/diagram-understanding)).\n\nSee [Section 4.1](https://arxiv.org/pdf/2310.12128) of our paper for the\nAI2D-Caption dataset annotation process.\n\n- **Homepage** :\n \u003chttps://huggingface.co/datasets/abhayzala/AI2D-Caption\u003e\n\n- **Source code** :\n [`tfds.datasets.ai2dcaption.Builder`](https://github.com/tensorflow/datasets/tree/master/tensorflow_datasets/datasets/ai2dcaption/ai2dcaption_dataset_builder.py)\n\n- **Versions**:\n\n - **`1.0.0`** (default): Initial release.\n- **Download size** : `Unknown size`\n\n- **Dataset size** : `2.01 GiB`\n\n- **Auto-cached**\n ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):\n No\n\n- **Splits**:\n\n| Split | Examples |\n|-----------------------------------|----------|\n| `'auditor_llm_training_examples'` | 30 |\n| `'gpt4v'` | 4,903 |\n| `'llava_15'` | 4,902 |\n| `'planner_llm_training_examples'` | 30 |\n| `'test'` | 75 |\n\n- **Feature structure**:\n\n FeaturesDict({\n 'caption': Text(shape=(), dtype=string),\n 'entities': Sequence({\n 'bounds': BBoxFeature(shape=(4,), dtype=float32),\n 'cat': ClassLabel(shape=(), dtype=int64, num_classes=10),\n 'from': Text(shape=(), dtype=string),\n 'id': Text(shape=(), dtype=string),\n 'label': Text(shape=(), dtype=string),\n 'to': Text(shape=(), dtype=string),\n 'type': ClassLabel(shape=(), dtype=int64, num_classes=5),\n }),\n 'image': Image(shape=(None, None, 3), dtype=uint8, description=The image of the diagram.),\n 'image_filename': Text(shape=(), dtype=string),\n 'layout': ClassLabel(shape=(), dtype=int64, num_classes=7),\n 'relationships': Sequence(Text(shape=(), dtype=string)),\n 'topic': ClassLabel(shape=(), dtype=int64, num_classes=4),\n })\n\n- **Feature documentation**:\n\n| Feature | Class | Shape | Dtype | Description |\n|-----------------|----------------|-----------------|---------|---------------------------------|\n| | FeaturesDict | | | |\n| caption | Text | | string | |\n| entities | Sequence | | | |\n| entities/bounds | BBoxFeature | (4,) | float32 | |\n| entities/cat | ClassLabel | | int64 | |\n| entities/from | Text | | string | |\n| entities/id | Text | | string | |\n| entities/label | Text | | string | |\n| entities/to | Text | | string | |\n| entities/type | ClassLabel | | int64 | |\n| image | Image | (None, None, 3) | uint8 | The image of the diagram. |\n| image_filename | Text | | string | Image filename. e.g. \"1337.png\" |\n| layout | ClassLabel | | int64 | |\n| relationships | Sequence(Text) | (None,) | string | |\n| topic | ClassLabel | | int64 | |\n\n- **Supervised keys** (See\n [`as_supervised` doc](https://www.tensorflow.org/datasets/api_docs/python/tfds/load#args)):\n `None`\n\n- **Figure**\n ([tfds.show_examples](https://www.tensorflow.org/datasets/api_docs/python/tfds/visualization/show_examples)):\n\n- **Examples** ([tfds.as_dataframe](https://www.tensorflow.org/datasets/api_docs/python/tfds/as_dataframe)):\n\nDisplay examples... \n\n- **Citation**:\n\n @inproceedings{Zala2024DiagrammerGPT,\n author = {Abhay Zala and Han Lin and Jaemin Cho and Mohit Bansal},\n title = {DiagrammerGPT: Generating Open-Domain, Open-Platform Diagrams via LLM Planning},\n year = {2024},\n booktitle = {COLM},\n }"]]