View Source Axon.Quantization.Layers (Axon v0.7.0)

Quantized Layer Implementations.

Summary

Functions

weight_only_quantized_dense(input, kernel, bias \\ 0, opts \\ [])

Weight-only quantized version of a dense layer.

Functions

Link to this function

weight_only_quantized_dense(input, kernel, bias \\ 0, opts \\ [])

View Source

Weight-only quantized version of a dense layer.

It expects the input kernel to be an Axon.Quantization.QTensor.