bitorch.layersΒΆ

This submodule contains adapted pytorch layers that use quantization functions on their weights and activations before forwarding them. These layers use the quantization functions specified in the quantization submodule.

Functions

convert

Convert the given module to a new bitorch RuntimeMode.

Modules

bitorch.layers.bembedding

bitorch.layers.config

Config class for quantization layers.

bitorch.layers.debug_layers

bitorch.layers.extensions

This submodule contains objects needed to provide and manage custom layer implementations.

bitorch.layers.pact

bitorch.layers.pad

bitorch.layers.qactivation

bitorch.layers.qconv1d

Module containing the quantized 1d convolution layer

bitorch.layers.qconv2d

Module containing the quantized 2d convolution layer

bitorch.layers.qconv3d

Module containing the quantized 3d convolution layer

bitorch.layers.qconv_mixin

bitorch.layers.qembedding

bitorch.layers.qlinear

Module containing the quantized linear layer

bitorch.layers.register