def decode(self,
latent_tensor,
observation_shape,
is_training,
reuse=False):
"""Decodes the latent_tensor to an observation."""
return architectures.make_decoder(latent_tensor,
observation_shape,
is_training=is_training,
reuse=reuse)
def forward_pass(self, latent_tensor, observation_shape, is_training):
"""Decodes the latent_tensor to an observation."""
return architectures.make_decoder(latent_tensor,
observation_shape,
is_training=is_training)
def decode(self, latent_tensor, observation_shape, is_training):
"""Decodes the latent_tensor to an observation without features."""
return get_return_v(
architectures.make_decoder(latent_tensor,
observation_shape,
is_training=is_training), 1)
def decode_with_gfeats(self, latent_tensor, observation_shape,
is_training):
"""Decodes the latent_tensor to an observation."""
return architectures.make_decoder(latent_tensor,
observation_shape,
is_training=is_training)