tf_agents.utils.tensor_normalizer.EMATensorNormalizer

View source on GitHub

TensorNormalizer with exponential moving avg. mean and var estimates.

Inherits From: TensorNormalizer

tf_agents.utils.tensor_normalizer.EMATensorNormalizer(
    tensor_spec, scope='normalize_tensor', norm_update_rate=0.001
)

Args
tensor_spec	The specs of the tensors to normalize.
scope	Scope for the tf.Module.

Methods

map_dtype

View source

map_dtype(
    dtype
)

normalize

View source

normalize(
    tensor, clip_value=5.0, center_mean=True, variance_epsilon=0.001
)

Applies normalization to tensor.

Args
tensor	Tensor to normalize.
clip_value	Clips normalized observations between +/- this value if clip_value > 0, otherwise does not apply clipping.
center_mean	If true, subtracts off mean from normalized tensor.
variance_epsilon	Epsilon to avoid division by zero in normalization.

Returns
normalized_tensor	Tensor after applying normalization.

update

View source

update(
    tensor, outer_dims=(0,)
)

Updates tensor normalizer variables.