- 説明:
論文で説明されているデータセット: Unnatural Instructions: Tuning Language Models with (Almost) No Human Labor (2022)。オプションの制約 / LLM によって生成された再定式化を含む、自然言語命令のセットが含まれています。
ホームページ: https://github.com/orhonovich/unnatural-instructions
ソースコード:
tfds.text.unnatural_instructions.UnnaturalInstructionsバージョン:
-
0.0.1(デフォルト): 初期リリース。追加の処理を使用する必要があるため、命令/入力を省略します。 Instruction_with_inputs と reformulations には、命令とコンテキストが含まれます。
-
ダウンロードサイズ:
17.48 MiBデータセットのサイズ:
154.71 MiB自動キャッシュ(ドキュメント):
shuffle_files=Falseの場合のみ (トレーニング)スプリット:
| スプリット | 例 |
|---|---|
'train' | 66,010 |
- 機能構造:
FeaturesDict({
'id': Text(shape=(), dtype=string),
'instances': Sequence({
'constraints': Text(shape=(), dtype=string),
'input': Text(shape=(), dtype=string),
'instruction_with_input': Text(shape=(), dtype=string),
'output': Text(shape=(), dtype=string),
}),
'instruction': Text(shape=(), dtype=string),
'reformulations': Sequence({
'input': Text(shape=(), dtype=string),
'instruction': Text(shape=(), dtype=string),
'instruction_with_input': Text(shape=(), dtype=string),
'output': Text(shape=(), dtype=string),
}),
})
- 機能のドキュメント:
| 特徴 | クラス | 形 | Dtype | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ID | 文章 | ストリング | たとえば、一意の識別子。 | |
| インスタンス | 順序 | |||
| インスタンス/制約 | 文章 | ストリング | タスク固有の制約。 | |
| インスタンス/入力 | 文章 | ストリング | 指定された命令のプレースホルダーに入力される入力。 | |
| インスタンス/instruction_with_input | 文章 | ストリング | プレースホルダーに提供される入力を含む命令。 | |
| インスタンス/出力 | 文章 | ストリング | 特定のタスクのターゲット出力。 | |
| 命令 | 文章 | ストリング | 入力用のプレースホルダー付きの命令。 | |
| 再定式化 | 順序 | |||
| 再定式化/入力 | 文章 | ストリング | 指定された命令のプレースホルダーに入力される入力。 | |
| 再定式化/指示 | 文章 | ストリング | 入力用のプレースホルダー付きの命令。 | |
| 再定式化/instruction_with_input | 文章 | ストリング | プレースホルダーに提供される入力を含む命令。 | |
| 再定式化/出力 | 文章 | ストリング | 特定のタスクのターゲット出力。 |
監視されたキー(
as_superviseddocを参照):None図( tfds.show_examples ): サポートされていません。
例( tfds.as_dataframe ):
- 引用:
@misc{honovich2022unnatural,
title = {Unnatural Instructions: Tuning Language Models with (Almost) No Human Labor},
author = {Honovich, Or and Scialom, Thomas and Levy, Omer and Schick, Timo},
url = {https://arxiv.org/abs/2212.09689},
publisher = {arXiv},
year={2022}
}