I read the MLP and LSTM tutorials of Lux.jl, and I noticed that the loss functions of these two tutorials need to be defined manually, which I think is a bit troublesome.
Flux.jl has some built-in loss functions. What is the motivation for Lux.jl not to include built-in loss functions? Is it because Flux.jl does not define the loss function as a separate Julia package, making it difficult to reuse?
I read the MLP and LSTM tutorials of Lux.jl, and I noticed that the loss functions of these two tutorials need to be defined manually, which I think is a bit troublesome.
Flux.jl has some built-in loss functions. What is the motivation for Lux.jl not to include built-in loss functions? Is it because Flux.jl does not define the loss function as a separate Julia package, making it difficult to reuse?