Asynchronously call a method of the external environment. Args: name: Name of the method to call. *args: Positional arguments to forw

Returns weighted target value from sampled actions. This calculation is proposed in BCQ paper for the first time. `x` should be shaped with `(

Returns the resulting tensor when all operations of this layer are applied to input_tensor. If input_tensor is 2D, this method returns a 3D t

Constructs the variables of this fully-connected layer. Returns: None

contains observation

Returns registered action scaler object. Args: name: regsitered scaler type name. kwargs: scaler arguments. Returns:

Returns registered encoder factory object. Args: name: regsitered encoder factory type name. kwargs: encoder arguments. Retu

Returns registered Q function factory object. Args: name: registered Q function factory type name. kwargs: Q function arguments.

Returns registered scaler object. Args: name: regsitered scaler type name. kwargs: scaler arguments. Returns: scaler

Extract the latent distribution from a prior or posterior state.

Extract the latent distribution from a prior or posterior state.

Gets scaling parameters from environment. Args: env: gym environment.

Combine all leaves of a nested structure into a tuple. The nested structure can consist of any combination of tuples, lists, and dicts. Dictionar

Returns the current activation function for this layer. Arguments: as_func: (bool) Determines whether the returned value is the s

Returns the current activation function for this layer. Arguments: as_func: (bool) Determines whether the returned value is the s

Computes the observations for the agents. Each agent receives information about cars in it's vision range (a surrounding 3 × 3 neighb

Returns cartpole dataset and environment. The dataset is automatically downloaded to ``d3rlpy_data/cartpole.h5`` if it does not exist. A

Returns d4rl dataset and envrironment. The dataset is provided through d4rl. .. code-block:: python from d3rlpy.datasets import get_d4

Returns the list of losses corresponding to the weight decay imposed on each weight of the network. Returns: the list of weight decay

Return diagnostic information as ordered dictionary. Records mean and standard deviation of Q-function and state value function, and

Return diagnostic information as ordered dictionary. Records mean and standard deviation of Q-function and state value function, and

Returns pendulum dataset and environment. The dataset is automatically downloaded to ``d3rlpy_data/pendulum.h5`` if it does not exist. A

Returns the variables of this layer.

Reference : Appendix F => Network Architecture & Appendix A : Proposition A.2 in https://arxiv.org/pdf/1805.11593.pdf (Page-11)

Run the schedule for a specified number of steps and yield scores. Call the operation of the current phase until the global step reaches the

Launches the example on autoscaled ray cluster through ray exec_cmd. This handles basic validation and sanity checks for the experiment, and

Should return the legal actions at each turn, if it is not available, it can return the whole action space. At each turn, the game ha