- 説明:
D4RL は、オフライン強化学習用のオープンソース ベンチマークです。トレーニングおよびベンチマーク アルゴリズム用の標準化された環境とデータセットを提供します。
データセットはRLDS 形式に従ってステップとエピソードを表します。
追加ドキュメント:コード付きの論文について調べる
構成の説明: タスクとそのバージョンの詳細については、 https://github.com/rail-berkeley/d4rl/wiki/Tasks#gymを参照してください。
バージョン:
-
1.0.0: 初期リリース。 -
1.1.0: is_last を追加しました。 -
1.2.0(デフォルト): 次の観察を考慮するために更新されました。
-
監視キー(
as_superviseddocを参照):None図( tfds.show_examples ): サポートされていません。
引用:
@misc{fu2020d4rl,
title={D4RL: Datasets for Deep Data-Driven Reinforcement Learning},
author={Justin Fu and Aviral Kumar and Ofir Nachum and George Tucker and Sergey Levine},
year={2020},
eprint={2004.07219},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
d4rl_mujoco_walker2d/v0-expert (デフォルト設定)
ダウンロードサイズ:
78.41 MiBデータセットのサイズ:
98.64 MiB自動キャッシュ(ドキュメント): はい
分割:
| スプリット | 例 |
|---|---|
'train' | 1,628 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v0-medium
ダウンロードサイズ:
80.83 MiBデータセットのサイズ:
99.72 MiB自動キャッシュ(ドキュメント): はい
分割:
| スプリット | 例 |
|---|---|
'train' | 5,315 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v0-medium-expert
ダウンロードサイズ:
159.24 MiBデータセットのサイズ:
198.36 MiB自動キャッシュ(ドキュメント):
shuffle_files=False(トレイン) の場合のみ分割:
| スプリット | 例 |
|---|---|
'train' | 6,943 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v0-mixed
ダウンロードサイズ:
8.42 MiBデータセットのサイズ:
10.06 MiB自動キャッシュ(ドキュメント): はい
分割:
| スプリット | 例 |
|---|---|
'train' | 501 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v0-ランダム
ダウンロードサイズ:
78.41 MiBデータセットのサイズ:
112.04 MiB自動キャッシュ(ドキュメント): はい
分割:
| スプリット | 例 |
|---|---|
'train' | 50,988 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v1-expert
ダウンロードサイズ:
143.06 MiBデータセットのサイズ:
452.72 MiB自動キャッシュ(ドキュメント): いいえ
分割:
| スプリット | 例 |
|---|---|
'train' | 1,003 |
- 機能の構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'policy': FeaturesDict({
'fc0': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 17), dtype=float32),
}),
'fc1': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 256), dtype=float32),
}),
'last_fc': FeaturesDict({
'bias': Tensor(shape=(6,), dtype=float32),
'weight': Tensor(shape=(6, 256), dtype=float32),
}),
'last_fc_log_std': FeaturesDict({
'bias': Tensor(shape=(6,), dtype=float32),
'weight': Tensor(shape=(6, 256), dtype=float32),
}),
'nonlinearity': string,
'output_distribution': string,
}),
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float32,
'qpos': Tensor(shape=(9,), dtype=float32),
'qvel': Tensor(shape=(9,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| アルゴリズム | テンソル | 弦 | ||
| 反復 | テンソル | int32 | ||
| ポリシー | 特徴辞書 | |||
| ポリシー/fc0 | 特徴辞書 | |||
| ポリシー/fc0/バイアス | テンソル | (256,) | float32 | |
| ポリシー/fc0/重み | テンソル | (256, 17) | float32 | |
| ポリシー/fc1 | 特徴辞書 | |||
| ポリシー/FC1/バイアス | テンソル | (256,) | float32 | |
| ポリシー/fc1/重み | テンソル | (256、256) | float32 | |
| ポリシー/last_fc | 特徴辞書 | |||
| ポリシー/last_fc/バイアス | テンソル | (6,) | float32 | |
| ポリシー/last_fc/weight | テンソル | (6,256) | float32 | |
| ポリシー/last_fc_log_std | 特徴辞書 | |||
| ポリシー/last_fc_log_std/bias | テンソル | (6,) | float32 | |
| ポリシー/last_fc_log_std/weight | テンソル | (6,256) | float32 | |
| ポリシー/非線形性 | テンソル | 弦 | ||
| ポリシー/出力_配布 | テンソル | 弦 | ||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float32 | ||
| ステップ/情報/qpos | テンソル | (9,) | float32 | |
| ステップ/情報/qvel | テンソル | (9,) | float32 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v1-medium
ダウンロードサイズ:
144.23 MiBデータセットのサイズ:
510.08 MiB自動キャッシュ(ドキュメント): いいえ
分割:
| スプリット | 例 |
|---|---|
'train' | 1,207 |
- 機能の構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'policy': FeaturesDict({
'fc0': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 17), dtype=float32),
}),
'fc1': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 256), dtype=float32),
}),
'last_fc': FeaturesDict({
'bias': Tensor(shape=(6,), dtype=float32),
'weight': Tensor(shape=(6, 256), dtype=float32),
}),
'last_fc_log_std': FeaturesDict({
'bias': Tensor(shape=(6,), dtype=float32),
'weight': Tensor(shape=(6, 256), dtype=float32),
}),
'nonlinearity': string,
'output_distribution': string,
}),
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float32,
'qpos': Tensor(shape=(9,), dtype=float32),
'qvel': Tensor(shape=(9,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| アルゴリズム | テンソル | 弦 | ||
| 反復 | テンソル | int32 | ||
| ポリシー | 特徴辞書 | |||
| ポリシー/fc0 | 特徴辞書 | |||
| ポリシー/fc0/バイアス | テンソル | (256,) | float32 | |
| ポリシー/fc0/重み | テンソル | (256, 17) | float32 | |
| ポリシー/fc1 | 特徴辞書 | |||
| ポリシー/FC1/バイアス | テンソル | (256,) | float32 | |
| ポリシー/fc1/重み | テンソル | (256、256) | float32 | |
| ポリシー/last_fc | 特徴辞書 | |||
| ポリシー/last_fc/バイアス | テンソル | (6,) | float32 | |
| ポリシー/last_fc/weight | テンソル | (6,256) | float32 | |
| ポリシー/last_fc_log_std | 特徴辞書 | |||
| ポリシー/last_fc_log_std/bias | テンソル | (6,) | float32 | |
| ポリシー/last_fc_log_std/weight | テンソル | (6,256) | float32 | |
| ポリシー/非線形性 | テンソル | 弦 | ||
| ポリシー/出力_配布 | テンソル | 弦 | ||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float32 | ||
| ステップ/情報/qpos | テンソル | (9,) | float32 | |
| ステップ/情報/qvel | テンソル | (9,) | float32 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v1-medium-expert
ダウンロードサイズ:
286.69 MiBデータセットのサイズ:
342.46 MiB自動キャッシュ(ドキュメント): いいえ
分割:
| スプリット | 例 |
|---|---|
'train' | 2,209 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float32,
'qpos': Tensor(shape=(9,), dtype=float32),
'qvel': Tensor(shape=(9,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float32 | ||
| ステップ/情報/qpos | テンソル | (9,) | float32 | |
| ステップ/情報/qvel | テンソル | (9,) | float32 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v1-medium-replay
ダウンロードサイズ:
84.37 MiBデータセットのサイズ:
52.10 MiB自動キャッシュ(ドキュメント): はい
分割:
| スプリット | 例 |
|---|---|
'train' | 1,093 |
- 機能の構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float64),
'discount': float64,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(9,), dtype=float64),
'qvel': Tensor(shape=(9,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float64),
'reward': float64,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| アルゴリズム | テンソル | 弦 | ||
| 反復 | テンソル | int32 | ||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float64 | |
| 歩数/割引 | テンソル | float64 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float64 | ||
| ステップ/情報/qpos | テンソル | (9,) | float64 | |
| ステップ/情報/qvel | テンソル | (9,) | float64 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float64 | |
| 歩数/報酬 | テンソル | float64 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v1-フル-リプレイ
ダウンロードサイズ:
278.95 MiBデータセットのサイズ:
171.66 MiB自動キャッシュ(ドキュメント):
shuffle_files=False(トレイン) の場合のみ分割:
| スプリット | 例 |
|---|---|
'train' | 1,888 |
- 機能の構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float64),
'discount': float64,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(9,), dtype=float64),
'qvel': Tensor(shape=(9,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float64),
'reward': float64,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| アルゴリズム | テンソル | 弦 | ||
| 反復 | テンソル | int32 | ||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float64 | |
| 歩数/割引 | テンソル | float64 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float64 | ||
| ステップ/情報/qpos | テンソル | (9,) | float64 | |
| ステップ/情報/qvel | テンソル | (9,) | float64 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float64 | |
| 歩数/報酬 | テンソル | float64 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v1-ランダム
ダウンロードサイズ:
132.36 MiBデータセットのサイズ:
192.06 MiB自動キャッシュ(ドキュメント):
shuffle_files=False(トレイン) の場合のみ分割:
| スプリット | 例 |
|---|---|
'train' | 48,790 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float32,
'qpos': Tensor(shape=(9,), dtype=float32),
'qvel': Tensor(shape=(9,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float32 | ||
| ステップ/情報/qpos | テンソル | (9,) | float32 | |
| ステップ/情報/qvel | テンソル | (9,) | float32 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v2-expert
ダウンロードサイズ:
219.89 MiBデータセットのサイズ:
452.16 MiB自動キャッシュ(ドキュメント): いいえ
分割:
| スプリット | 例 |
|---|---|
'train' | 1,001 |
- 機能の構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'policy': FeaturesDict({
'fc0': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 17), dtype=float32),
}),
'fc1': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 256), dtype=float32),
}),
'last_fc': FeaturesDict({
'bias': Tensor(shape=(6,), dtype=float32),
'weight': Tensor(shape=(6, 256), dtype=float32),
}),
'last_fc_log_std': FeaturesDict({
'bias': Tensor(shape=(6,), dtype=float32),
'weight': Tensor(shape=(6, 256), dtype=float32),
}),
'nonlinearity': string,
'output_distribution': string,
}),
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(9,), dtype=float64),
'qvel': Tensor(shape=(9,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| アルゴリズム | テンソル | 弦 | ||
| 反復 | テンソル | int32 | ||
| ポリシー | 特徴辞書 | |||
| ポリシー/fc0 | 特徴辞書 | |||
| ポリシー/fc0/バイアス | テンソル | (256,) | float32 | |
| ポリシー/fc0/重み | テンソル | (256, 17) | float32 | |
| ポリシー/fc1 | 特徴辞書 | |||
| ポリシー/FC1/バイアス | テンソル | (256,) | float32 | |
| ポリシー/fc1/重み | テンソル | (256、256) | float32 | |
| ポリシー/last_fc | 特徴辞書 | |||
| ポリシー/last_fc/バイアス | テンソル | (6,) | float32 | |
| ポリシー/last_fc/weight | テンソル | (6,256) | float32 | |
| ポリシー/last_fc_log_std | 特徴辞書 | |||
| ポリシー/last_fc_log_std/bias | テンソル | (6,) | float32 | |
| ポリシー/last_fc_log_std/weight | テンソル | (6,256) | float32 | |
| ポリシー/非線形性 | テンソル | 弦 | ||
| ポリシー/出力_配布 | テンソル | 弦 | ||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float64 | ||
| ステップ/情報/qpos | テンソル | (9,) | float64 | |
| ステップ/情報/qvel | テンソル | (9,) | float64 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v2-フル-リプレイ
ダウンロードサイズ:
271.91 MiBデータセットのサイズ:
171.66 MiB自動キャッシュ(ドキュメント):
shuffle_files=False(トレイン) の場合のみ分割:
| スプリット | 例 |
|---|---|
'train' | 1,888 |
- 機能の構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(9,), dtype=float64),
'qvel': Tensor(shape=(9,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| アルゴリズム | テンソル | 弦 | ||
| 反復 | テンソル | int32 | ||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float64 | ||
| ステップ/情報/qpos | テンソル | (9,) | float64 | |
| ステップ/情報/qvel | テンソル | (9,) | float64 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v2-medium
ダウンロードサイズ:
221.50 MiBデータセットのサイズ:
505.58 MiB自動キャッシュ(ドキュメント): いいえ
分割:
| スプリット | 例 |
|---|---|
'train' | 1,191 |
- 機能の構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'policy': FeaturesDict({
'fc0': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 17), dtype=float32),
}),
'fc1': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 256), dtype=float32),
}),
'last_fc': FeaturesDict({
'bias': Tensor(shape=(6,), dtype=float32),
'weight': Tensor(shape=(6, 256), dtype=float32),
}),
'last_fc_log_std': FeaturesDict({
'bias': Tensor(shape=(6,), dtype=float32),
'weight': Tensor(shape=(6, 256), dtype=float32),
}),
'nonlinearity': string,
'output_distribution': string,
}),
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(9,), dtype=float64),
'qvel': Tensor(shape=(9,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| アルゴリズム | テンソル | 弦 | ||
| 反復 | テンソル | int32 | ||
| ポリシー | 特徴辞書 | |||
| ポリシー/fc0 | 特徴辞書 | |||
| ポリシー/fc0/バイアス | テンソル | (256,) | float32 | |
| ポリシー/fc0/重み | テンソル | (256, 17) | float32 | |
| ポリシー/fc1 | 特徴辞書 | |||
| ポリシー/FC1/バイアス | テンソル | (256,) | float32 | |
| ポリシー/fc1/重み | テンソル | (256、256) | float32 | |
| ポリシー/last_fc | 特徴辞書 | |||
| ポリシー/last_fc/バイアス | テンソル | (6,) | float32 | |
| ポリシー/last_fc/weight | テンソル | (6,256) | float32 | |
| ポリシー/last_fc_log_std | 特徴辞書 | |||
| ポリシー/last_fc_log_std/bias | テンソル | (6,) | float32 | |
| ポリシー/last_fc_log_std/weight | テンソル | (6,256) | float32 | |
| ポリシー/非線形性 | テンソル | 弦 | ||
| ポリシー/出力_配布 | テンソル | 弦 | ||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float64 | ||
| ステップ/情報/qpos | テンソル | (9,) | float64 | |
| ステップ/情報/qvel | テンソル | (9,) | float64 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v2-medium-expert
ダウンロードサイズ:
440.79 MiBデータセットのサイズ:
342.45 MiB自動キャッシュ(ドキュメント): いいえ
分割:
| スプリット | 例 |
|---|---|
'train' | 2,191 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(9,), dtype=float64),
'qvel': Tensor(shape=(9,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float64 | ||
| ステップ/情報/qpos | テンソル | (9,) | float64 | |
| ステップ/情報/qvel | テンソル | (9,) | float64 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v2-medium-replay
ダウンロードサイズ:
82.32 MiBデータセットのサイズ:
52.10 MiB自動キャッシュ(ドキュメント): はい
分割:
| スプリット | 例 |
|---|---|
'train' | 1,093 |
- 機能の構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(9,), dtype=float64),
'qvel': Tensor(shape=(9,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| アルゴリズム | テンソル | 弦 | ||
| 反復 | テンソル | int32 | ||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float64 | ||
| ステップ/情報/qpos | テンソル | (9,) | float64 | |
| ステップ/情報/qvel | テンソル | (9,) | float64 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_walker2d/v2-ランダム
ダウンロードサイズ:
206.10 MiBデータセットのサイズ:
192.11 MiB自動キャッシュ(ドキュメント):
shuffle_files=False(トレイン) の場合のみ分割:
| スプリット | 例 |
|---|---|
'train' | 48,908 |
- 機能の構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(6,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(9,), dtype=float64),
'qvel': Tensor(shape=(9,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(17,), dtype=float32),
'reward': float32,
}),
})
- 機能ドキュメント:
| 特徴 | クラス | 形 | Dタイプ | 説明 |
|---|---|---|---|---|
| 特徴辞書 | ||||
| ステップ | データセット | |||
| ステップ/アクション | テンソル | (6,) | float32 | |
| 歩数/割引 | テンソル | float32 | ||
| 手順/情報 | 特徴辞書 | |||
| ステップ/情報/action_log_probs | テンソル | float64 | ||
| ステップ/情報/qpos | テンソル | (9,) | float64 | |
| ステップ/情報/qvel | テンソル | (9,) | float64 | |
| ステップ/is_first | テンソル | ブール | ||
| ステップ/is_last | テンソル | ブール | ||
| ステップ/is_terminal | テンソル | ブール | ||
| ステップ/観察 | テンソル | (17,) | float32 | |
| 歩数/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):