- 説明:
D4RL は、オフライン強化学習のオープンソース 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。
追加のドキュメント:コード を使用したペーパーの探索
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): 次の観測を考慮して更新されました。
-
監視されたキー(
as_supervised
doc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_ant/v0-expert (デフォルト設定)
ダウンロードサイズ:
131.34 MiB
データセットのサイズ:
464.94 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,288 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v0-中
ダウンロードサイズ:
131.39 MiB
データセットのサイズ:
464.78 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,122人 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v0-medium-expert
ダウンロードサイズ:
262.73 MiB
データセットサイズ:
929.71 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 2,410 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v0-mixed
ダウンロードサイズ:
104.63 MiB
データセットサイズ:
464.93 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,320 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v0-ランダム
ダウンロードサイズ:
139.50 MiB
データセットのサイズ:
464.97 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,377 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v1-expert
ダウンロードサイズ:
220.72 MiB
データセットサイズ:
968.63 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,033 |
- 機能構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'policy': FeaturesDict({
'fc0': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 111), dtype=float32),
}),
'fc1': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 256), dtype=float32),
}),
'last_fc': FeaturesDict({
'bias': Tensor(shape=(8,), dtype=float32),
'weight': Tensor(shape=(8, 256), dtype=float32),
}),
'last_fc_log_std': FeaturesDict({
'bias': Tensor(shape=(8,), dtype=float32),
'weight': Tensor(shape=(8, 256), dtype=float32),
}),
'nonlinearity': string,
'output_distribution': string,
}),
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float32,
'qpos': Tensor(shape=(15,), dtype=float32),
'qvel': Tensor(shape=(14,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 説明 |
---|---|---|---|---|
特徴辞書 | ||||
アルゴリズム | テンソル | ストリング | ||
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、111) | 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/重量 | 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 | (8) | float32 | |
ポリシー/last_fc/重量 | テンソル | (8,256) | float32 | |
ポリシー/last_fc_log_std | 特徴辞書 | |||
ポリシー/last_fc_log_std/バイアス | テンソル | (8) | float32 | |
ポリシー/last_fc_log_std/重量 | テンソル | (8,256) | float32 | |
ポリシー/非線形性 | テンソル | ストリング | ||
ポリシー/出力配布 | テンソル | ストリング | ||
ステップ | データセット | |||
ステップ/アクション | テンソル | (8) | float32 | |
ステップ/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 | (15) | float32 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v1-中
ダウンロードサイズ:
222.39 MiB
データセットサイズ:
1023.71 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,179 |
- 機能構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'policy': FeaturesDict({
'fc0': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 111), dtype=float32),
}),
'fc1': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 256), dtype=float32),
}),
'last_fc': FeaturesDict({
'bias': Tensor(shape=(8,), dtype=float32),
'weight': Tensor(shape=(8, 256), dtype=float32),
}),
'last_fc_log_std': FeaturesDict({
'bias': Tensor(shape=(8,), dtype=float32),
'weight': Tensor(shape=(8, 256), dtype=float32),
}),
'nonlinearity': string,
'output_distribution': string,
}),
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float32,
'qpos': Tensor(shape=(15,), dtype=float32),
'qvel': Tensor(shape=(14,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 説明 |
---|---|---|---|---|
特徴辞書 | ||||
アルゴリズム | テンソル | ストリング | ||
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、111) | 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/重量 | 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 | (8) | float32 | |
ポリシー/last_fc/重量 | テンソル | (8,256) | float32 | |
ポリシー/last_fc_log_std | 特徴辞書 | |||
ポリシー/last_fc_log_std/バイアス | テンソル | (8) | float32 | |
ポリシー/last_fc_log_std/重量 | テンソル | (8,256) | float32 | |
ポリシー/非線形性 | テンソル | ストリング | ||
ポリシー/出力配布 | テンソル | ストリング | ||
ステップ | データセット | |||
ステップ/アクション | テンソル | (8) | float32 | |
ステップ/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 | (15) | float32 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v1-medium-expert
ダウンロードサイズ:
442.25 MiB
データセットサイズ:
1.13 GiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 2,211 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float32,
'qpos': Tensor(shape=(15,), dtype=float32),
'qvel': Tensor(shape=(14,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (15) | float32 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v1-medium-replay
ダウンロードサイズ:
132.05 MiB
データセットのサイズ:
175.27 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' | 485 |
- 機能構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float64),
'discount': float64,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(15,), dtype=float64),
'qvel': Tensor(shape=(14,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float64),
'reward': float64,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 説明 |
---|---|---|---|---|
特徴辞書 | ||||
アルゴリズム | テンソル | ストリング | ||
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 | (8) | float64 | |
ステップ/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 | (15) | float64 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float64 | |
ステップ/報酬 | テンソル | float64 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v1-full-replay
ダウンロードサイズ:
437.57 MiB
データセットのサイズ:
580.09 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,319 |
- 機能構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float64),
'discount': float64,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(15,), dtype=float64),
'qvel': Tensor(shape=(14,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float64),
'reward': float64,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 説明 |
---|---|---|---|---|
特徴辞書 | ||||
アルゴリズム | テンソル | ストリング | ||
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 | (8) | float64 | |
ステップ/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 | (15) | float64 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float64 | |
ステップ/報酬 | テンソル | float64 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v1-ランダム
ダウンロードサイズ:
225.18 MiB
データセットのサイズ:
583.83 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 5,741 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float32,
'qpos': Tensor(shape=(15,), dtype=float32),
'qvel': Tensor(shape=(14,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (15) | float32 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v2-expert
ダウンロードサイズ:
355.94 MiB
データセットサイズ:
969.38 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,035 |
- 機能構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'policy': FeaturesDict({
'fc0': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 111), dtype=float32),
}),
'fc1': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 256), dtype=float32),
}),
'last_fc': FeaturesDict({
'bias': Tensor(shape=(8,), dtype=float32),
'weight': Tensor(shape=(8, 256), dtype=float32),
}),
'last_fc_log_std': FeaturesDict({
'bias': Tensor(shape=(8,), dtype=float32),
'weight': Tensor(shape=(8, 256), dtype=float32),
}),
'nonlinearity': string,
'output_distribution': string,
}),
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(15,), dtype=float64),
'qvel': Tensor(shape=(14,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 説明 |
---|---|---|---|---|
特徴辞書 | ||||
アルゴリズム | テンソル | ストリング | ||
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、111) | 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/重量 | 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 | (8) | float32 | |
ポリシー/last_fc/重量 | テンソル | (8,256) | float32 | |
ポリシー/last_fc_log_std | 特徴辞書 | |||
ポリシー/last_fc_log_std/バイアス | テンソル | (8) | float32 | |
ポリシー/last_fc_log_std/重量 | テンソル | (8,256) | float32 | |
ポリシー/非線形性 | テンソル | ストリング | ||
ポリシー/出力配布 | テンソル | ストリング | ||
ステップ | データセット | |||
ステップ/アクション | テンソル | (8) | float32 | |
ステップ/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 | (15) | float64 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v2-full-replay
ダウンロードサイズ:
428.57 MiB
データセットのサイズ:
580.09 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,319 |
- 機能構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(15,), dtype=float64),
'qvel': Tensor(shape=(14,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 説明 |
---|---|---|---|---|
特徴辞書 | ||||
アルゴリズム | テンソル | ストリング | ||
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 | (8) | float32 | |
ステップ/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 | (15) | float64 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v2-中
ダウンロードサイズ:
358.81 MiB
データセットサイズ:
1.01 GiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 1,203 |
- 機能構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'policy': FeaturesDict({
'fc0': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 111), dtype=float32),
}),
'fc1': FeaturesDict({
'bias': Tensor(shape=(256,), dtype=float32),
'weight': Tensor(shape=(256, 256), dtype=float32),
}),
'last_fc': FeaturesDict({
'bias': Tensor(shape=(8,), dtype=float32),
'weight': Tensor(shape=(8, 256), dtype=float32),
}),
'last_fc_log_std': FeaturesDict({
'bias': Tensor(shape=(8,), dtype=float32),
'weight': Tensor(shape=(8, 256), dtype=float32),
}),
'nonlinearity': string,
'output_distribution': string,
}),
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(15,), dtype=float64),
'qvel': Tensor(shape=(14,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 説明 |
---|---|---|---|---|
特徴辞書 | ||||
アルゴリズム | テンソル | ストリング | ||
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、111) | 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/重量 | 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 | (8) | float32 | |
ポリシー/last_fc/重量 | テンソル | (8,256) | float32 | |
ポリシー/last_fc_log_std | 特徴辞書 | |||
ポリシー/last_fc_log_std/バイアス | テンソル | (8) | float32 | |
ポリシー/last_fc_log_std/重量 | テンソル | (8,256) | float32 | |
ポリシー/非線形性 | テンソル | ストリング | ||
ポリシー/出力配布 | テンソル | ストリング | ||
ステップ | データセット | |||
ステップ/アクション | テンソル | (8) | float32 | |
ステップ/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 | (15) | float64 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v2-中-専門家
ダウンロードサイズ:
713.67 MiB
データセットサイズ:
1.13 GiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 2,237 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(15,), dtype=float64),
'qvel': Tensor(shape=(14,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (15) | float64 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v2-medium-replay
ダウンロードサイズ:
130.16 MiB
データセットのサイズ:
175.27 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' | 485 |
- 機能構造:
FeaturesDict({
'algorithm': string,
'iteration': int32,
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(15,), dtype=float64),
'qvel': Tensor(shape=(14,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 説明 |
---|---|---|---|---|
特徴辞書 | ||||
アルゴリズム | テンソル | ストリング | ||
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 | (8) | float32 | |
ステップ/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 | (15) | float64 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):
d4rl_mujoco_ant/v2-ランダム
ダウンロードサイズ:
366.66 MiB
データセットサイズ:
583.90 MiB
自動キャッシュ(ドキュメント): いいえ
スプリット:
スプリット | 例 |
---|---|
'train' | 5,822 |
- 機能構造:
FeaturesDict({
'steps': Dataset({
'action': Tensor(shape=(8,), dtype=float32),
'discount': float32,
'infos': FeaturesDict({
'action_log_probs': float64,
'qpos': Tensor(shape=(15,), dtype=float64),
'qvel': Tensor(shape=(14,), dtype=float64),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': Tensor(shape=(111,), dtype=float32),
'reward': float32,
}),
})
- 機能のドキュメント:
特徴 | クラス | 形 | Dtype | 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 | (8) | float32 | |
ステップ/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 | (15) | float64 | |
ステップ/情報/qvel | テンソル | (14) | 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 | (111) | float32 | |
ステップ/報酬 | テンソル | float32 |
- 例( tfds.as_dataframe ):