zea.models.layers

Layers used in zea.models

Functions

DownBlock(width, block_depth[, normalization])	Downsampling block with residual connections.
ResidualBlock(width[, normalization])	Residual block with swish activation.
UpBlock(width, block_depth[, normalization])	Upsampling block with residual connections.
sinusoidal_embedding(x, ...)	Sinusoidal embedding layer.

zea.models.layers.DownBlock(width, block_depth, normalization='batch_norm')

Downsampling block with residual connections.

zea.models.layers.ResidualBlock(width, normalization='batch_norm')

Residual block with swish activation.

Parameters:

• width – Number of filters.

• normalization – Normalization type. One of "batch_norm" or "group_norm". Defaults to "batch_norm".

zea.models.layers.UpBlock(width, block_depth, normalization='batch_norm')

Upsampling block with residual connections.

zea.models.layers.sinusoidal_embedding(x, embedding_min_frequency, embedding_max_frequency, embedding_dims)

Sinusoidal embedding layer.