Bilinear module — nn_bilinear • torch

Applies a bilinear transformation to the incoming data $y = x_{1}^{T} A x_{2} + b$

Usage

nn_bilinear(in1_features, in2_features, out_features, bias = TRUE)

Arguments

in1_features: size of each first input sample
in2_features: size of each second input sample
out_features: size of each output sample
bias: If set to FALSE, the layer will not learn an additive bias. Default: TRUE

Shape

Input1: $(N, *, H_{i n 1})$ $H_{i n 1} = in1\_features$ and $*$ means any number of additional dimensions. All but the last dimension of the inputs should be the same.
Input2: $(N, *, H_{i n 2})$ where $H_{i n 2} = in2\_features$ .
Output: $(N, *, H_{o u t})$ where $H_{o u t} = out\_features$ and all but the last dimension are the same shape as the input.

Attributes

weight: the learnable weights of the module of shape $(out\_features, in1\_features, in2\_features)$ . The values are initialized from $U (- \sqrt{k}, \sqrt{k})$ , where $k = \frac{1}{in1\_features}$
bias: the learnable bias of the module of shape $(out\_features)$ . If bias is TRUE, the values are initialized from $U (- \sqrt{k}, \sqrt{k})$ , where $k = \frac{1}{in1\_features}$

Examples

if (torch_is_installed()) {
m <- nn_bilinear(20, 30, 50)
input1 <- torch_randn(128, 20)
input2 <- torch_randn(128, 30)
output <- m(input1, input2)
print(output$size())
}
#> [1] 128  50