rlu_rwrl

  • Description:

RL Unplugged is suite of benchmarks for offline reinforcement learning. The RL Unplugged is designed around the following considerations: to facilitate ease of use, we provide the datasets with a unified API which makes it easy for the practitioner to work with all data in the suite once a general pipeline has been established.

The datasets follow the RLDS format to represent steps and episodes.

Examples in the dataset represent SAR transitions stored when running a partially online trained agent as described in https://arxiv.org/abs/1904.12901 We follow the RLDS dataset format, as specified in https://github.com/google-research/rlds#dataset-format

We release 40 datasets on 8 tasks in total -- with no combined challenge and easy combined challenge on the cartpole, walker, quadruped, and humanoid tasks. Each task contains 5 different sizes of datasets, 1%, 5%, 20%, 40%, and 100%. Note that the smaller dataset is not guaranteed to be a subset of the larger ones. For details on how the dataset was generated, please refer to the paper.

@misc{gulcehre2020rl,
    title={RL Unplugged: Benchmarks for Offline Reinforcement Learning},
    author={Caglar Gulcehre and Ziyu Wang and Alexander Novikov and Tom Le Paine
        and  Sergio Gómez Colmenarejo and Konrad Zolna and Rishabh Agarwal and
        Josh Merel and Daniel Mankowitz and Cosmin Paduraru and Gabriel
        Dulac-Arnold and Jerry Li and Mohammad Norouzi and Matt Hoffman and
        Ofir Nachum and George Tucker and Nicolas Heess and Nando deFreitas},
    year={2020},
    eprint={2006.13888},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

rlu_rwrl/cartpole_swingup_combined_challenge_none_1_percent (default config)

  • Dataset size: 172.43 KiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 5
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_none_5_percent

  • Dataset size: 862.13 KiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 25
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_none_20_percent

  • Dataset size: 3.37 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 100
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_none_40_percent

  • Dataset size: 6.74 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 200
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_none_100_percent

  • Dataset size: 16.84 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 500
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_none_1_percent

  • Dataset size: 1.77 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 5
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_none_5_percent

  • Dataset size: 8.86 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 25
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_none_20_percent

  • Dataset size: 35.46 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 100
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_none_40_percent

  • Dataset size: 70.92 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 200
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_none_100_percent

  • Dataset size: 177.29 MiB

  • Auto-cached (documentation): Only when shuffle_files=False (train)

  • Splits:

Split Examples
'train' 500
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_none_1_percent

  • Dataset size: 6.27 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 50
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_none_5_percent

  • Dataset size: 31.34 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 250
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_none_20_percent

  • Dataset size: 125.37 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 1,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_none_40_percent

  • Dataset size: 250.75 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'train' 2,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_none_100_percent

  • Dataset size: 626.86 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'train' 5,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_none_1_percent

  • Dataset size: 69.40 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 200
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_none_5_percent

  • Dataset size: 346.98 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'train' 1,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_none_20_percent

Split Examples
'train' 4,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_none_40_percent

Split Examples
'train' 8,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_none_100_percent

Split Examples
'train' 20,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_easy_1_percent

  • Dataset size: 369.84 KiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 5
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_easy_5_percent

  • Dataset size: 1.81 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 25
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_easy_20_percent

  • Dataset size: 7.22 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 100
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_easy_40_percent

  • Dataset size: 14.45 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 200
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/cartpole_swingup_combined_challenge_easy_100_percent

  • Dataset size: 36.12 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 500
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(1,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'position': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(2,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (1,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/position Tensor (3,) tf.float32
steps/observation/velocity Tensor (2,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_easy_1_percent

  • Dataset size: 1.97 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 5
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_easy_5_percent

  • Dataset size: 9.83 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 25
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_easy_20_percent

  • Dataset size: 39.31 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 100
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_easy_40_percent

  • Dataset size: 78.63 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 200
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/quadruped_walk_combined_challenge_easy_100_percent

  • Dataset size: 196.57 MiB

  • Auto-cached (documentation): Only when shuffle_files=False (train)

  • Splits:

Split Examples
'train' 500
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(12,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'egocentric_state': Tensor(shape=(44,), dtype=tf.float32),
            'force_torque': Tensor(shape=(24,), dtype=tf.float32),
            'imu': Tensor(shape=(6,), dtype=tf.float32),
            'torso_upright': Tensor(shape=(1,), dtype=tf.float32),
            'torso_velocity': Tensor(shape=(3,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (12,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/egocentric_state Tensor (44,) tf.float32
steps/observation/force_torque Tensor (24,) tf.float32
steps/observation/imu Tensor (6,) tf.float32
steps/observation/torso_upright Tensor (1,) tf.float32
steps/observation/torso_velocity Tensor (3,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_easy_1_percent

  • Dataset size: 8.20 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 50
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_easy_5_percent

  • Dataset size: 40.98 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 250
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_easy_20_percent

  • Dataset size: 163.93 MiB

  • Auto-cached (documentation): Only when shuffle_files=False (train)

  • Splits:

Split Examples
'train' 1,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_easy_40_percent

  • Dataset size: 327.86 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'train' 2,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/walker_walk_combined_challenge_easy_100_percent

  • Dataset size: 819.65 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'train' 5,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(6,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'height': Tensor(shape=(1,), dtype=tf.float32),
            'orientations': Tensor(shape=(14,), dtype=tf.float32),
            'velocity': Tensor(shape=(9,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (6,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/height Tensor (1,) tf.float32
steps/observation/orientations Tensor (14,) tf.float32
steps/observation/velocity Tensor (9,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_easy_1_percent

  • Dataset size: 77.11 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'train' 200
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_easy_5_percent

  • Dataset size: 385.54 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'train' 1,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_easy_20_percent

Split Examples
'train' 4,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_easy_40_percent

Split Examples
'train' 8,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32

rlu_rwrl/humanoid_walk_combined_challenge_easy_100_percent

Split Examples
'train' 20,000
  • Feature structure:
FeaturesDict({
    'episode_return': tf.float32,
    'steps': Dataset({
        'action': Tensor(shape=(21,), dtype=tf.float32),
        'discount': Tensor(shape=(1,), dtype=tf.float32),
        'is_first': tf.bool,
        'is_last': tf.bool,
        'is_terminal': tf.bool,
        'observation': FeaturesDict({
            'com_velocity': Tensor(shape=(3,), dtype=tf.float32),
            'dummy-0': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-1': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-2': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-3': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-4': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-5': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-6': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-7': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-8': Tensor(shape=(1,), dtype=tf.float32),
            'dummy-9': Tensor(shape=(1,), dtype=tf.float32),
            'extremities': Tensor(shape=(12,), dtype=tf.float32),
            'head_height': Tensor(shape=(1,), dtype=tf.float32),
            'joint_angles': Tensor(shape=(21,), dtype=tf.float32),
            'torso_vertical': Tensor(shape=(3,), dtype=tf.float32),
            'velocity': Tensor(shape=(27,), dtype=tf.float32),
        }),
        'reward': Tensor(shape=(1,), dtype=tf.float32),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_return Tensor tf.float32
steps Dataset
steps/action Tensor (21,) tf.float32
steps/discount Tensor (1,) tf.float32
steps/is_first Tensor tf.bool
steps/is_last Tensor tf.bool
steps/is_terminal Tensor tf.bool
steps/observation FeaturesDict
steps/observation/com_velocity Tensor (3,) tf.float32
steps/observation/dummy-0 Tensor (1,) tf.float32
steps/observation/dummy-1 Tensor (1,) tf.float32
steps/observation/dummy-2 Tensor (1,) tf.float32
steps/observation/dummy-3 Tensor (1,) tf.float32
steps/observation/dummy-4 Tensor (1,) tf.float32
steps/observation/dummy-5 Tensor (1,) tf.float32
steps/observation/dummy-6 Tensor (1,) tf.float32
steps/observation/dummy-7 Tensor (1,) tf.float32
steps/observation/dummy-8 Tensor (1,) tf.float32
steps/observation/dummy-9 Tensor (1,) tf.float32
steps/observation/extremities Tensor (12,) tf.float32
steps/observation/head_height Tensor (1,) tf.float32
steps/observation/joint_angles Tensor (21,) tf.float32
steps/observation/torso_vertical Tensor (3,) tf.float32
steps/observation/velocity Tensor (27,) tf.float32
steps/reward Tensor (1,) tf.float32