Eval this module with a numpy interface Same as a call to __call__ except all Variable input/outputs are replaced with numpy equivalents

:param value: some value, x :param pre_tanh_value: arctanh(x) :return:

:param inputs: :param obs_distribution_params: :return: log probability of input under decoder

Undo internal replay buffer representation changes: save images as bytes

Ask a yes/no question via raw_input() and return their answer. "question" is a string that is presented to the user. "default" is the presume

Returns relative probability from the log probabilities. They're not exactly equal to the probability, but relative scalings between them are

The following value for the following keys will be a 2D array, with the first dimension corresponding to the time dimension. - observati

Sampling in the reparameterization case.

Format a fixed width table for pretty printing. >>> print(tabulate([[1, 2.34], [-56, "8.999"], ["2", "10001"]])) --- --------- 1

This is useful to override in case the wrapped env has some funky __getstate__ that doesn't play well with overriding __getattr__.

Initializes WrappedLoggedPrioritizedReplayBuffer.

Initializes WrappedLoggedReplayBuffer.

Initialize the FixedReplayBuffer class. Args: data_dir: str, log Directory from which to load the replay buffer. replay_suffix: int,

Symbols after a decimal point, -1 if the string lacks the decimal point. >>> _afterpoint("123.45") 2 >>> _afterpoint("1001") -1 >

[string] -> [padded_string] >>> list(map(str,_align_column(["12.345", "-1234.5", "1.23", "1234.5", "1e+234", "1.0e234"], "decimal"))) [' 12

Build an op used as a target for the Q-value. Returns: target_q_op: An op calculating the Q-value.

Build an op used as a target for the Q-value. Returns: target_q_op: An op calculating the Q-value.

Format a value accoding to its type. Unicode is supported: >>> hrow = ['\u0431\u0443\u043a\u0432\u0430', '\u0446\u0438\u0444\u0440\u0430'] ;

Produce a plain-text representation of the table.

Loads a OutOfGraphReplayBuffer replay buffer.

Loads multiple checkpoints into a list of replay buffers.

Transform a supported data type to a list of lists, and a list of headers. Supported tabular data types: * list-of-lists or another iterable

Center string. >>> _padboth(6, '\u044f\u0439\u0446\u0430') == ' \u044f\u0439\u0446\u0430 ' True

Flush right. >>> _padleft(6, '\u044f\u0439\u0446\u0430') == ' \u044f\u0439\u0446\u0430' True

Flush left. >>> _padright(6, '\u044f\u0439\u0446\u0430') == '\u044f\u0439\u0446\u0430 ' True

Return a segment of a horizontal line with optional colons which indicate column's alignment (as in `pipe` output format).

Run training phase.

Save statistics as tensorboard summaries. Args: iteration: int, The current iteration number. num_episodes_eval: int, number of evalu

Train model.

The least generic type (type(None), int, float, str, unicode). >>> _type(None) is type(None) True >>> _type("foo") is type("") True

Add a path to the replay buffer. This default implementation naively goes through every step, but you may want to optimize t

Add a transition tuple.