Properly close the game.

Properly close the game.

Collects data via interaction with environment. Args: algo: algorithm object. env: gym-like environment. buffer : replay

Collects data via interaction with environment. If ``buffer`` is not given, ``ReplayBuffer`` will be internally created. Args:

Returns decayed :math:`\\epsilon`. Returns: :math:`\\epsilon`.

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

Computes sum of rewards. .. math:: R = \sum_{i=1} r_i Returns: flo

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

Returns PyTorch's Q function module. Args: encoder: an encoder module that processes the observation and action t

Returns PyTorch's Q function module. Args: encoder: an encoder module that processes the observation to obtain fe

Encodes numpy images into gif string. Args: images: A 5-D `uint8` `np.array` (or a list of 4-D images) of shape `[batch_size, time, heigh

Extract deterministic features from an observation.

Finalizes the network. Arguments: optimizer: (tf.train.Optimizer) An optimizer class from those available at tf.train.Optimizer.

Runs two ops, one for assigning the mean of the data to the internal mean, and another for assigning the standard deviation of the data to the

Runs two ops, one for assigning the mean of the data to the internal mean, and another for assigning the standard deviation of the data to the

Start training loop of online deep reinforcement learning. Args: env: gym-like environment. buffer : replay buffer.