- 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.
- Descrição da configuração : veja mais detalhes sobre a tarefa e suas versões em https://github.com/rail-berkeley/d4rl/wiki/Tasks#adroit 
- Página inicial : https://sites.google.com/view/d4rl-anonymous 
- Código fonte : - tfds.d4rl.d4rl_adroit_hammer.D4rlAdroitHammer
- Versões : -  1.0.0: Versão inicial.
-  1.1.0(padrão): Adicionado is_last.
 
-  
- Chaves supervisionadas (consulte o documento - as_supervised):- None
- Figura ( tfds.show_examples ): Não suportado. 
- Citação : 
@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_adroit_hammer/v0-human (configuração padrão)
- Tamanho do download : - 5.33 MiB
- Tamanho do conjunto de dados : - 6.10 MiB
- Armazenado em cache automaticamente ( documentação ): Sim 
- Divisões : 
| Dividir | Exemplos | 
|---|---|
| 'train' | 70 | 
- Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(26,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'qpos': Tensor(shape=(33,), dtype=float32),
            'qvel': Tensor(shape=(33,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(46,), dtype=float32),
        'reward': float32,
    }),
})
- Documentação de recursos :
| Recurso | Aula | Forma | Tipo D | Descrição | 
|---|---|---|---|---|
| RecursosDict | ||||
| passos | Conjunto de dados | |||
| etapas/ação | Tensor | (26,) | float32 | |
| passos/desconto | Tensor | float32 | ||
| etapas/informações | RecursosDict | |||
| etapas/informações/qpos | Tensor | (33,) | float32 | |
| passos/informações/qvel | Tensor | (33,) | float32 | |
| passos/é_primeiro | Tensor | bool | ||
| passos/é_último | Tensor | bool | ||
| etapas/is_terminal | Tensor | bool | ||
| etapas/observação | Tensor | (46,) | float32 | |
| passos/recompensa | Tensor | float32 | 
- Exemplos ( tfds.as_dataframe ):
d4rl_adroit_hammer/v0-clonado
- Tamanho do download : - 644.69 MiB
- Tamanho do conjunto de dados : - 538.97 MiB
- Armazenado em cache automaticamente ( documentação ): Não 
- Divisões : 
| Dividir | Exemplos | 
|---|---|
| 'train' | 5.594 | 
- Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(26,), dtype=float32),
        'discount': float64,
        'infos': FeaturesDict({
            'qpos': Tensor(shape=(33,), dtype=float64),
            'qvel': Tensor(shape=(33,), dtype=float64),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(46,), dtype=float64),
        'reward': float64,
    }),
})
- Documentação de recursos :
| Recurso | Aula | Forma | Tipo D | Descrição | 
|---|---|---|---|---|
| RecursosDict | ||||
| passos | Conjunto de dados | |||
| etapas/ação | Tensor | (26,) | float32 | |
| passos/desconto | Tensor | float64 | ||
| etapas/informações | RecursosDict | |||
| etapas/informações/qpos | Tensor | (33,) | float64 | |
| passos/informações/qvel | Tensor | (33,) | float64 | |
| passos/é_primeiro | Tensor | bool | ||
| passos/é_último | Tensor | bool | ||
| etapas/is_terminal | Tensor | bool | ||
| etapas/observação | Tensor | (46,) | float64 | |
| passos/recompensa | Tensor | float64 | 
- Exemplos ( tfds.as_dataframe ):
d4rl_adroit_hammer/v0-expert
- Tamanho do download : - 529.91 MiB
- Tamanho do conjunto de dados : - 737.00 MiB
- Armazenado em cache automaticamente ( documentação ): Não 
- Divisões : 
| Dividir | Exemplos | 
|---|---|
| 'train' | 5.000 | 
- Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(26,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_logstd': Tensor(shape=(26,), dtype=float32),
            'action_mean': Tensor(shape=(26,), dtype=float32),
            'qpos': Tensor(shape=(33,), dtype=float32),
            'qvel': Tensor(shape=(33,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(46,), dtype=float32),
        'reward': float32,
    }),
})
- Documentação de recursos :
| Recurso | Aula | Forma | Tipo D | Descrição | 
|---|---|---|---|---|
| RecursosDict | ||||
| passos | Conjunto de dados | |||
| etapas/ação | Tensor | (26,) | float32 | |
| passos/desconto | Tensor | float32 | ||
| etapas/informações | RecursosDict | |||
| etapas/informações/action_logstd | Tensor | (26,) | float32 | |
| etapas/informações/action_mean | Tensor | (26,) | float32 | |
| etapas/informações/qpos | Tensor | (33,) | float32 | |
| passos/informações/qvel | Tensor | (33,) | float32 | |
| passos/é_primeiro | Tensor | bool | ||
| passos/é_último | Tensor | bool | ||
| etapas/is_terminal | Tensor | bool | ||
| etapas/observação | Tensor | (46,) | float32 | |
| passos/recompensa | Tensor | float32 | 
- Exemplos ( tfds.as_dataframe ):
d4rl_adroit_hammer/v1-human
- Tamanho do download : - 5.35 MiB
- Tamanho do conjunto de dados : - 6.34 MiB
- Armazenado em cache automaticamente ( documentação ): Sim 
- Divisões : 
| Dividir | Exemplos | 
|---|---|
| 'train' | 25 | 
- Estrutura de recursos :
FeaturesDict({
    'steps': Dataset({
        'action': Tensor(shape=(26,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'board_pos': Tensor(shape=(3,), dtype=float32),
            'qpos': Tensor(shape=(33,), dtype=float32),
            'qvel': Tensor(shape=(33,), dtype=float32),
            'target_pos': Tensor(shape=(3,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(46,), dtype=float32),
        'reward': float32,
    }),
})
- Documentação de recursos :
| Recurso | Aula | Forma | Tipo D | Descrição | 
|---|---|---|---|---|
| RecursosDict | ||||
| passos | Conjunto de dados | |||
| etapas/ação | Tensor | (26,) | float32 | |
| passos/desconto | Tensor | float32 | ||
| etapas/informações | RecursosDict | |||
| etapas/informações/board_pos | Tensor | (3,) | float32 | |
| etapas/informações/qpos | Tensor | (33,) | float32 | |
| passos/informações/qvel | Tensor | (33,) | float32 | |
| etapas/informações/target_pos | Tensor | (3,) | float32 | |
| passos/é_primeiro | Tensor | bool | ||
| passos/é_último | Tensor | bool | ||
| etapas/is_terminal | Tensor | bool | ||
| etapas/observação | Tensor | (46,) | float32 | |
| passos/recompensa | Tensor | float32 | 
- Exemplos ( tfds.as_dataframe ):
d4rl_adroit_hammer/v1-clonado
- Tamanho do download : - 425.93 MiB
- Tamanho do conjunto de dados : - 1.68 GiB
- Armazenado em cache automaticamente ( documentação ): Não 
- Divisões : 
| Dividir | Exemplos | 
|---|---|
| 'train' | 3.606 | 
- Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'policy': FeaturesDict({
        'fc0': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(46, 256), dtype=float32),
        }),
        'fc1': FeaturesDict({
            'bias': Tensor(shape=(256,), dtype=float32),
            'weight': Tensor(shape=(256, 256), dtype=float32),
        }),
        'last_fc': FeaturesDict({
            'bias': Tensor(shape=(26,), dtype=float32),
            'weight': Tensor(shape=(256, 26), dtype=float32),
        }),
        'nonlinearity': string,
        'output_distribution': string,
    }),
    'steps': Dataset({
        'action': Tensor(shape=(26,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'board_pos': Tensor(shape=(3,), dtype=float32),
            'qpos': Tensor(shape=(33,), dtype=float32),
            'qvel': Tensor(shape=(33,), dtype=float32),
            'target_pos': Tensor(shape=(3,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(46,), dtype=float32),
        'reward': float32,
    }),
})
- Documentação de recursos :
| Recurso | Aula | Forma | Tipo D | Descrição | 
|---|---|---|---|---|
| RecursosDict | ||||
| algoritmo | Tensor | corda | ||
| política | RecursosDict | |||
| política/fc0 | RecursosDict | |||
| política/fc0/bias | Tensor | (256,) | float32 | |
| política/fc0/peso | Tensor | (46, 256) | 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 | (26,) | float32 | |
| política/last_fc/peso | Tensor | (256, 26) | float32 | |
| política/não linearidade | Tensor | corda | ||
| política/output_distribution | Tensor | corda | ||
| passos | Conjunto de dados | |||
| etapas/ação | Tensor | (26,) | float32 | |
| passos/desconto | Tensor | float32 | ||
| etapas/informações | RecursosDict | |||
| etapas/informações/board_pos | Tensor | (3,) | float32 | |
| etapas/informações/qpos | Tensor | (33,) | float32 | |
| passos/informações/qvel | Tensor | (33,) | float32 | |
| etapas/informações/target_pos | Tensor | (3,) | float32 | |
| passos/é_primeiro | Tensor | bool | ||
| passos/é_último | Tensor | bool | ||
| etapas/is_terminal | Tensor | bool | ||
| etapas/observação | Tensor | (46,) | float32 | |
| passos/recompensa | Tensor | float32 | 
- Exemplos ( tfds.as_dataframe ):
d4rl_adroit_hammer/v1-expert
- Tamanho do download : - 531.24 MiB
- Tamanho do conjunto de dados : - 843.54 MiB
- Armazenado em cache automaticamente ( documentação ): Não 
- Divisões : 
| Dividir | Exemplos | 
|---|---|
| 'train' | 5.000 | 
- Estrutura de recursos :
FeaturesDict({
    'algorithm': string,
    'policy': FeaturesDict({
        'fc0': FeaturesDict({
            'bias': Tensor(shape=(32,), dtype=float32),
            'weight': Tensor(shape=(32, 46), dtype=float32),
        }),
        'fc1': FeaturesDict({
            'bias': Tensor(shape=(32,), dtype=float32),
            'weight': Tensor(shape=(32, 32), dtype=float32),
        }),
        'last_fc': FeaturesDict({
            'bias': Tensor(shape=(26,), dtype=float32),
            'weight': Tensor(shape=(26, 32), dtype=float32),
        }),
        'last_fc_log_std': FeaturesDict({
            'bias': Tensor(shape=(26,), dtype=float32),
            'weight': Tensor(shape=(26, 32), dtype=float32),
        }),
        'nonlinearity': string,
        'output_distribution': string,
    }),
    'steps': Dataset({
        'action': Tensor(shape=(26,), dtype=float32),
        'discount': float32,
        'infos': FeaturesDict({
            'action_log_std': Tensor(shape=(26,), dtype=float32),
            'action_mean': Tensor(shape=(26,), dtype=float32),
            'board_pos': Tensor(shape=(3,), dtype=float32),
            'qpos': Tensor(shape=(33,), dtype=float32),
            'qvel': Tensor(shape=(33,), dtype=float32),
            'target_pos': Tensor(shape=(3,), dtype=float32),
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': Tensor(shape=(46,), dtype=float32),
        'reward': float32,
    }),
})
- Documentação de recursos :
| Recurso | Aula | Forma | Tipo D | Descrição | 
|---|---|---|---|---|
| RecursosDict | ||||
| algoritmo | Tensor | corda | ||
| política | RecursosDict | |||
| política/fc0 | RecursosDict | |||
| política/fc0/bias | Tensor | (32,) | float32 | |
| política/fc0/peso | Tensor | (32, 46) | float32 | |
| política/fc1 | RecursosDict | |||
| política/fc1/preconceito | Tensor | (32,) | float32 | |
| política/fc1/peso | Tensor | (32, 32) | float32 | |
| política/last_fc | RecursosDict | |||
| política/last_fc/bias | Tensor | (26,) | float32 | |
| política/last_fc/peso | Tensor | (26, 32) | float32 | |
| política/last_fc_log_std | RecursosDict | |||
| política/last_fc_log_std/bias | Tensor | (26,) | float32 | |
| política/last_fc_log_std/peso | Tensor | (26, 32) | float32 | |
| política/não linearidade | Tensor | corda | ||
| política/output_distribution | Tensor | corda | ||
| passos | Conjunto de dados | |||
| etapas/ação | Tensor | (26,) | float32 | |
| passos/desconto | Tensor | float32 | ||
| etapas/informações | RecursosDict | |||
| etapas/informações/action_log_std | Tensor | (26,) | float32 | |
| etapas/informações/action_mean | Tensor | (26,) | float32 | |
| etapas/informações/board_pos | Tensor | (3,) | float32 | |
| etapas/informações/qpos | Tensor | (33,) | float32 | |
| passos/informações/qvel | Tensor | (33,) | float32 | |
| etapas/informações/target_pos | Tensor | (3,) | float32 | |
| passos/é_primeiro | Tensor | bool | ||
| passos/é_último | Tensor | bool | ||
| etapas/is_terminal | Tensor | bool | ||
| etapas/observação | Tensor | (46,) | float32 | |
| passos/recompensa | Tensor | float32 | 
- Exemplos ( tfds.as_dataframe ):