LazyLinear#

classtorch.nn.LazyLinear(out_features,bias=True,device=None,dtype=None)[source]#

Atorch.nn.Linear module wherein_features is inferred.

In this module, theweight andbias are oftorch.nn.UninitializedParameterclass. They will be initialized after the first call toforward is done and themodule will become a regulartorch.nn.Linear module. Thein_features argumentof theLinear is inferred from theinput.shape[-1].

Check thetorch.nn.modules.lazy.LazyModuleMixin for further documentationon lazy modules and their limitations.

Parameters

• out_features (int) – size of each output sample

• bias (UninitializedParameter) – If set toFalse, the layer will not learn an additive bias.Default:True

Variables

• weight (torch.nn.parameter.UninitializedParameter) – the learnable weights of the module of shape$(\text{out\_features},\text{in\_features})$(out_features,in_features). The values areinitialized from$\mathcal{U}(-\sqrt{k},\sqrt{k})$U(−k​,k​), where$k=\frac{1}{\text{in\_features}}$k=in_features1​

• bias (torch.nn.parameter.UninitializedParameter) – the learnable bias of the module of shape$(\text{out\_features})$(out_features).Ifbias isTrue, the values are initialized from$\mathcal{U}(-\sqrt{k},\sqrt{k})$U(−k​,k​) where$k=\frac{1}{\text{in\_features}}$k=in_features1​

cls_to_become[source]#

alias ofLinear

initialize_parameters(input)[source]#

Infersin_features based oninput and initializes parameters.

reset_parameters()[source]#

Resets parameters based on their initialization used in__init__.