スキャン

  • 説明:

さまざまな分割を伴う SCAN タスク。

SCAN は、構成学習とゼロショット一般化を研究するための単純な言語駆動型ナビゲーション タスクのセットです。

ほとんどの分割については、 https://github.com/brendenlake/SCANで説明されています。MCD 分割については、 https: //arxiv.org/abs/1912.09713.pdf を参照してください。

基本的な使い方:

data = tfds.load('scan/length')

より高度な例:

import tensorflow_datasets as tfds
from tensorflow_datasets.datasets.scan import scan_dataset_builder

data = tfds.load(
    'scan',
    builder_kwargs=dict(
        config=scan_dataset_builder.ScanConfig(
            name='simple_p8', directory='simple_split/size_variations')))
FeaturesDict({
    'actions': Text(shape=(), dtype=string),
    'commands': 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
行動文章ストリング
コマンド文章ストリング
@inproceedings{Lake2018GeneralizationWS,
  title={Generalization without Systematicity: On the Compositional Skills of
         Sequence-to-Sequence Recurrent Networks},
  author={Brenden M. Lake and Marco Baroni},
  booktitle={ICML},
  year={2018},
  url={https://arxiv.org/pdf/1711.00350.pdf},
}
@inproceedings{Keysers2020,
  title={Measuring Compositional Generalization: A Comprehensive Method on
         Realistic Data},
  author={Daniel Keysers and Nathanael Sch\"{a}rli and Nathan Scales and
          Hylke Buisman and Daniel Furrer and Sergii Kashubin and
          Nikola Momchev and Danila Sinopalnikov and Lukasz Stafiniak and
          Tibor Tihon and Dmitry Tsarkov and Xiao Wang and Marc van Zee and
          Olivier Bousquet},
  note={Additional citation for MCD splits},
  booktitle={ICLR},
  year={2020},
  url={https://arxiv.org/abs/1912.09713.pdf},
}

スキャン/シンプル (デフォルト設定)

  • ダウンロードサイズ: 17.82 MiB

  • データセットのサイズ: 4.47 MiB

  • スプリット:

スプリット例
'test' 4,182
'train' 16,728

スキャン/addprim_jump

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 4.53 MiB

  • スプリット:

スプリット例
'test' 7,706
'train' 14,670

scan/addprim_turn_left

  • ダウンロードサイズ: 17.82 MiB

  • データセットのサイズ: 4.58 MiB

  • スプリット:

スプリット例
'test' 1,208
'train' 21,890

スキャン/filler_num0

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 3.20 MiB

  • スプリット:

スプリット例
'test' 1,173
'train' 15,225

スキャン/filler_num1

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 3.51 MiB

  • スプリット:

スプリット例
'test' 1,173
'train' 16,290

スキャン/filler_num2

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 3.84 MiB

  • スプリット:

スプリット例
'test' 1,173
'train' 17,391

スキャン/filler_num3

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 4.17 MiB

  • スプリット:

スプリット例
'test' 1,173
'train' 18,528

スキャン/長さ

  • ダウンロードサイズ: 17.82 MiB

  • データセットのサイズ: 4.47 MiB

  • スプリット:

スプリット例
'test' 3,920
'train' 16,990

スキャン/template_around_right

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 4.17 MiB

  • スプリット:

スプリット例
'test' 4,476
'train' 15,225

scan/template_jump_around_right

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 4.17 MiB

  • スプリット:

スプリット例
'test' 1,173
'train' 18,528

スキャン/template_opposite_right

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 4.22 MiB

  • スプリット:

スプリット例
'test' 4,476
'train' 15,225

スキャン/template_right

  • ダウンロードサイズ: 17.82 MiB

  • データセットサイズ: 4.26 MiB

  • スプリット:

スプリット例
'test' 4,476
'train' 15,225

スキャン/mcd1

  • ダウンロードサイズ: 17.89 MiB

  • データセットサイズ: 1.89 MiB

  • スプリット:

スプリット例
'test' 1,045
'train' 8,365

スキャン/mcd2

  • ダウンロードサイズ: 17.89 MiB

  • データセットサイズ: 1.84 MiB

  • スプリット:

スプリット例
'test' 1,045
'train' 8,365

スキャン/mcd3

  • ダウンロードサイズ: 17.89 MiB

  • データセットサイズ: 1.87 MiB

  • スプリット:

スプリット例
'test' 1,045
'train' 8,365