d4rl_mujoco_hopper

  • Descrição :

D4RL é um benchmark de código aberto para aprendizagem por reforço offline. Ele fornece ambientes e conjuntos de dados padronizados para algoritmos de treinamento e benchmarking.

Os conjuntos de dados seguem o formato RLDS para representar etapas e episódios.

@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_hopper/v0-expert (configuração padrão)

  • Tamanho do download : 51.56 MiB

  • Tamanho do conjunto de dados : 64.10 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 1.029
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v0-medium

  • Tamanho do download : 51.74 MiB

  • Tamanho do conjunto de dados : 64.68 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 3.064
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v0-medium-expert

  • Tamanho do download : 62.01 MiB

  • Tamanho do conjunto de dados : 77.25 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 2.277
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v0-mixed

  • Tamanho do download : 10.48 MiB

  • Tamanho do conjunto de dados : 13.15 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 1.250
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v0-random

  • Tamanho do download : 51.83 MiB

  • Tamanho do conjunto de dados : 66.06 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 8.793
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v1-expert

  • Tamanho do download : 93.19 MiB

  • Tamanho do conjunto de dados : 608.03 MiB

  • Armazenado em cache automaticamente ( documentação ): Não

  • Divisões :

Dividir Exemplos
'train' 1.836
  • Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'iteration': int32,
    'policy': FeaturesDict({
        'fc0': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 11), dtype=float32),
        }),
        'fc1': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 256), dtype=float32),
        }),
        'last_fc': FeaturesDict({
            'bias': Tensor(shape=(3,), dtype=float32),
            'weight': Tensor(shape=(3, 256), dtype=float32),
        }),
        'last_fc_log_std': FeaturesDict({
            'bias': Tensor(shape=(3,), dtype=float32),
            'weight': Tensor(shape=(3, 256), dtype=float32),
        }),
        'nonlinearity': string,
        'output_distribution': string,
    }),
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float32,
            'qpos': Tensor(shape=(6,), dtype=float32),
            'qvel': Tensor(shape=(6,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
algoritmo Tensor corda
iteração Tensor int32
política RecursosDict
política/fc0 RecursosDict
política/fc0/bias Tensor (256,) float32
política/fc0/peso Tensor (256, 11) float32
política/fc1 RecursosDict
política/fc1/preconceito Tensor (256,) float32
política/fc1/peso Tensor (256, 256) float32
política/last_fc RecursosDict
política/last_fc/bias Tensor (3,) float32
política/last_fc/peso Tensor (3, 256) float32
política/last_fc_log_std RecursosDict
política/last_fc_log_std/bias Tensor (3,) float32
política/last_fc_log_std/peso Tensor (3, 256) float32
política/não linearidade Tensor corda
política/output_distribution Tensor corda
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float32
etapas/informações/qpos Tensor (6,) float32
passos/informações/qvel Tensor (6,) float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v1-medium

  • Tamanho do download : 92.03 MiB

  • Tamanho do conjunto de dados : 1.78 GiB

  • Armazenado em cache automaticamente ( documentação ): Não

  • Divisões :

Dividir Exemplos
'train' 6.328
  • Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'iteration': int32,
    'policy': FeaturesDict({
        'fc0': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 11), dtype=float32),
        }),
        'fc1': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 256), dtype=float32),
        }),
        'last_fc': FeaturesDict({
            'bias': Tensor(shape=(3,), dtype=float32),
            'weight': Tensor(shape=(3, 256), dtype=float32),
        }),
        'last_fc_log_std': FeaturesDict({
            'bias': Tensor(shape=(3,), dtype=float32),
            'weight': Tensor(shape=(3, 256), dtype=float32),
        }),
        'nonlinearity': string,
        'output_distribution': string,
    }),
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float32,
            'qpos': Tensor(shape=(6,), dtype=float32),
            'qvel': Tensor(shape=(6,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
algoritmo Tensor corda
iteração Tensor int32
política RecursosDict
política/fc0 RecursosDict
política/fc0/bias Tensor (256,) float32
política/fc0/peso Tensor (256, 11) float32
política/fc1 RecursosDict
política/fc1/preconceito Tensor (256,) float32
política/fc1/peso Tensor (256, 256) float32
política/last_fc RecursosDict
política/last_fc/bias Tensor (3,) float32
política/last_fc/peso Tensor (3, 256) float32
política/last_fc_log_std RecursosDict
política/last_fc_log_std/bias Tensor (3,) float32
política/last_fc_log_std/peso Tensor (3, 256) float32
política/não linearidade Tensor corda
política/output_distribution Tensor corda
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float32
etapas/informações/qpos Tensor (6,) float32
passos/informações/qvel Tensor (6,) float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v1-medium-expert

  • Tamanho do download : 184.59 MiB

  • Tamanho do conjunto de dados : 230.24 MiB

  • Armazenado em cache automaticamente ( documentação ): somente quando shuffle_files=False (train)

  • Divisões :

Dividir Exemplos
'train' 8.163
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float32,
            'qpos': Tensor(shape=(6,), dtype=float32),
            'qvel': Tensor(shape=(6,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float32
etapas/informações/qpos Tensor (6,) float32
passos/informações/qvel Tensor (6,) float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v1-medium-replay

  • Tamanho do download : 55.65 MiB

  • Tamanho do conjunto de dados : 34.78 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 1.151
  • Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'iteration': int32,
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float64),
        'discount': float64,
        'infos': FeaturesDict({
            'action_log_probs': float64,
            'qpos': Tensor(shape=(6,), dtype=float64),
            'qvel': Tensor(shape=(6,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float64),
        'reward': float64,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
algoritmo Tensor corda
iteração Tensor int32
passos Conjunto de dados
etapas/ação Tensor (3,) float64
passos/desconto Tensor float64
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float64
etapas/informações/qpos Tensor (6,) float64
passos/informações/qvel Tensor (6,) float64
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float64
passos/recompensa Tensor float64

d4rl_mujoco_hopper/v1-full-replay

  • Tamanho do download : 183.32 MiB

  • Tamanho do conjunto de dados : 114.78 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 2.907
  • Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'iteration': int32,
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float64),
        'discount': float64,
        'infos': FeaturesDict({
            'action_log_probs': float64,
            'qpos': Tensor(shape=(6,), dtype=float64),
            'qvel': Tensor(shape=(6,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float64),
        'reward': float64,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
algoritmo Tensor corda
iteração Tensor int32
passos Conjunto de dados
etapas/ação Tensor (3,) float64
passos/desconto Tensor float64
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float64
etapas/informações/qpos Tensor (6,) float64
passos/informações/qvel Tensor (6,) float64
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float64
passos/recompensa Tensor float64

d4rl_mujoco_hopper/v1-random

  • Tamanho do download : 91.11 MiB

  • Tamanho do conjunto de dados : 130.73 MiB

  • Armazenado em cache automaticamente ( documentação ): somente quando shuffle_files=False (train)

  • Divisões :

Dividir Exemplos
'train' 45.265
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float32,
            'qpos': Tensor(shape=(6,), dtype=float32),
            'qvel': Tensor(shape=(6,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float32
etapas/informações/qpos Tensor (6,) float32
passos/informações/qvel Tensor (6,) float32
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v2-expert

  • Tamanho do download : 145.37 MiB

  • Tamanho do conjunto de dados : 390.40 MiB

  • Armazenado em cache automaticamente ( documentação ): Não

  • Divisões :

Dividir Exemplos
'train' 1.028
  • Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'iteration': int32,
    'policy': FeaturesDict({
        'fc0': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 11), dtype=float32),
        }),
        'fc1': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 256), dtype=float32),
        }),
        'last_fc': FeaturesDict({
            'bias': Tensor(shape=(3,), dtype=float32),
            'weight': Tensor(shape=(3, 256), dtype=float32),
        }),
        'last_fc_log_std': FeaturesDict({
            'bias': Tensor(shape=(3,), dtype=float32),
            'weight': Tensor(shape=(3, 256), dtype=float32),
        }),
        'nonlinearity': string,
        'output_distribution': string,
    }),
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float64,
            'qpos': Tensor(shape=(6,), dtype=float64),
            'qvel': Tensor(shape=(6,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
algoritmo Tensor corda
iteração Tensor int32
política RecursosDict
política/fc0 RecursosDict
política/fc0/bias Tensor (256,) float32
política/fc0/peso Tensor (256, 11) float32
política/fc1 RecursosDict
política/fc1/preconceito Tensor (256,) float32
política/fc1/peso Tensor (256, 256) float32
política/last_fc RecursosDict
política/last_fc/bias Tensor (3,) float32
política/last_fc/peso Tensor (3, 256) float32
política/last_fc_log_std RecursosDict
política/last_fc_log_std/bias Tensor (3,) float32
política/last_fc_log_std/peso Tensor (3, 256) float32
política/não linearidade Tensor corda
política/output_distribution Tensor corda
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float64
etapas/informações/qpos Tensor (6,) float64
passos/informações/qvel Tensor (6,) float64
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v2-full-replay

  • Tamanho do download : 179.29 MiB

  • Tamanho do conjunto de dados : 115.04 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 3.515
  • Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'iteration': int32,
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float64,
            'qpos': Tensor(shape=(6,), dtype=float64),
            'qvel': Tensor(shape=(6,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
algoritmo Tensor corda
iteração Tensor int32
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float64
etapas/informações/qpos Tensor (6,) float64
passos/informações/qvel Tensor (6,) float64
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v2-medium

  • Tamanho do download : 145.68 MiB

  • Tamanho do conjunto de dados : 702.57 MiB

  • Armazenado em cache automaticamente ( documentação ): Não

  • Divisões :

Dividir Exemplos
'train' 2.187
  • Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'iteration': int32,
    'policy': FeaturesDict({
        'fc0': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 11), dtype=float32),
        }),
        'fc1': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 256), dtype=float32),
        }),
        'last_fc': FeaturesDict({
            'bias': Tensor(shape=(3,), dtype=float32),
            'weight': Tensor(shape=(3, 256), dtype=float32),
        }),
        'last_fc_log_std': FeaturesDict({
            'bias': Tensor(shape=(3,), dtype=float32),
            'weight': Tensor(shape=(3, 256), dtype=float32),
        }),
        'nonlinearity': string,
        'output_distribution': string,
    }),
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float64,
            'qpos': Tensor(shape=(6,), dtype=float64),
            'qvel': Tensor(shape=(6,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
algoritmo Tensor corda
iteração Tensor int32
política RecursosDict
política/fc0 RecursosDict
política/fc0/bias Tensor (256,) float32
política/fc0/peso Tensor (256, 11) float32
política/fc1 RecursosDict
política/fc1/preconceito Tensor (256,) float32
política/fc1/peso Tensor (256, 256) float32
política/last_fc RecursosDict
política/last_fc/bias Tensor (3,) float32
política/last_fc/peso Tensor (3, 256) float32
política/last_fc_log_std RecursosDict
política/last_fc_log_std/bias Tensor (3,) float32
política/last_fc_log_std/peso Tensor (3, 256) float32
política/não linearidade Tensor corda
política/output_distribution Tensor corda
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float64
etapas/informações/qpos Tensor (6,) float64
passos/informações/qvel Tensor (6,) float64
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v2-medium-expert

  • Tamanho do download : 290.43 MiB

  • Tamanho do conjunto de dados : 228.28 MiB

  • Armazenado em cache automaticamente ( documentação ): somente quando shuffle_files=False (train)

  • Divisões :

Dividir Exemplos
'train' 3.214
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float64,
            'qpos': Tensor(shape=(6,), dtype=float64),
            'qvel': Tensor(shape=(6,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float64
etapas/informações/qpos Tensor (6,) float64
passos/informações/qvel Tensor (6,) float64
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v2-medium-replay

  • Tamanho do download : 72.34 MiB

  • Tamanho do conjunto de dados : 46.51 MiB

  • Armazenado em cache automaticamente ( documentação ): Sim

  • Divisões :

Dividir Exemplos
'train' 2.041
  • Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'iteration': int32,
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float64,
            'qpos': Tensor(shape=(6,), dtype=float64),
            'qvel': Tensor(shape=(6,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
algoritmo Tensor corda
iteração Tensor int32
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float64
etapas/informações/qpos Tensor (6,) float64
passos/informações/qvel Tensor (6,) float64
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32

d4rl_mujoco_hopper/v2-random

  • Tamanho do download : 145.46 MiB

  • Tamanho do conjunto de dados : 130.72 MiB

  • Armazenado em cache automaticamente ( documentação ): somente quando shuffle_files=False (train)

  • Divisões :

Dividir Exemplos
'train' 45.240
  • Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_probs': float64,
            'qpos': Tensor(shape=(6,), dtype=float64),
            'qvel': Tensor(shape=(6,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(11,), dtype=float32),
        'reward': float32,
    }),
})
  • Documentação de recursos :
Recurso Aula Forma Tipo D Descrição
RecursosDict
passos Conjunto de dados
etapas/ação Tensor (3,) float32
passos/desconto Tensor float32
etapas/informações RecursosDict
etapas/informações/action_log_probs Tensor float64
etapas/informações/qpos Tensor (6,) float64
passos/informações/qvel Tensor (6,) float64
passos/é_primeiro Tensor bool
passos/é_último Tensor bool
etapas/is_terminal Tensor bool
etapas/observação Tensor (11,) float32
passos/recompensa Tensor float32