linear#

classtorch.ao.nn.quantized.functional.linear(input,weight,bias=None,scale=None,zero_point=None)[source]#

Applies a linear transformation to the incoming quantized data:$y=x{A}^T+b$y=xAT+b.SeeLinear

Note

Current implementation packs weights on every call, which has penalty on performance.If you want to avoid the overhead, useLinear.

Parameters

• input (Tensor) – Quantized input of typetorch.quint8

• weight (Tensor) – Quantized weight of typetorch.qint8

• bias (Tensor) – None or fp32 bias of typetorch.float

• scale (double) – output scale. If None, derived from the input scale

• zero_point (python:long) – output zero point. If None, derived from the input zero_point

Return type

Tensor

Shape:

• Input:$(N,\ast ,in\mathrm{\_}features)$(N,∗,in_features) where* means any number ofadditional dimensions

• Weight:$(out\mathrm{\_}features,in\mathrm{\_}features)$(out_features,in_features)

• Bias:$(out\mathrm{\_}features)$(out_features)

• Output:$(N,\ast ,out\mathrm{\_}features)$(N,∗,out_features)