Is there an equivalent of LazyLinear in Flux.jl?

Specific Domains Machine Learning
1

LazyLinear

It seems that a similar option in Flux.jl is @autosize. But we still need to manually pass in the first input layer dimension.