struct AdaLNOut {
&FloatTensor out;
&FloatTensor gate;
}Fields
LN(x) * (1 + scale) + shift, [m, dim]
[m, dim]; all ones when not gated
struct AdaLNOut {
&FloatTensor out;
&FloatTensor gate;
}LN(x) * (1 + scale) + shift, [m, dim]
[m, dim]; all ones when not gated