Index

Layers API

The helia_edge.layers module provides classes to build neural network layers. For example, you can use the helia_edge.layers.preprocessing.AmplitudeWarp layer to apply amplitude warping to audio signals.

Available Layers

• Preprocessing Layers: Provides tf.data.Dataset preprocessing layers.
• Activations: Provides activation functions.
• Convolutional Layers: Provides convolutional layers.
• Gumbel Softmax Bottleneck: Provides Gumbel Softmax bottleneck layer.
• Normalization Layers: Provides normalization layers.
• Patching Layers: Provides patching layers.
• Residual Vector Quantizer: Provides residual vector quantizer layer.
• Squeeze-and-Excitation Layer: Provides squeeze-and-excitation layers.
• MBConv Layer: Provides popular mbconv block.
• Vector Quantizer: Provides vector quantizer layer.

Classes

Functions