metarl.np.embeddings.encoder module¶

Base class for context encoder.

class Encoder[source]¶

Bases: abc.ABC

Base class of context encoders for training meta-RL algorithms.

get_latent(input_value)[source]¶

Encode an input value.

Parameters:	input_value (numpy.ndarray) – Input value of (input_dim, ) shape.
Returns:	Encoded embedding.
Return type:	numpy.ndarray

get_latents(input_values)[source]¶

Encode a batch of input values.

Parameters:	input_values (numpy.ndarray) – Input values of (N, input_dim) shape.
Returns:	Encoded embeddings.
Return type:	numpy.ndarray

input_dim¶

Dimension of the encoder input.

Type:	int

output_dim¶

Dimension of the encoder output (embedding).

Type:	int

reset(do_resets=None)[source]¶

Reset the encoder.

This is effective only to recurrent encoder. do_resets is effective only to vectoried encoder.

For a vectorized encoder, do_resets is an array of boolean indicating which internal states to be reset. The length of do_resets should be equal to the length of inputs.

Parameters:	do_resets (numpy.ndarray) – Bool array indicating which states to be reset.

spec¶

Input and output space.

Type:	metarl.InOutSpec

class StochasticEncoder[source]¶

Bases: metarl.np.embeddings.encoder.Encoder

An stochastic context encoders.

An stochastic encoder maps an input to a distribution, but not a deterministic vector.

distribution¶

Embedding distribution.

Type:	object