施熠民 SHI YIMIN
施熠民 SHI YIMIN
> Isn't it equivalent on the computation graph when taking derivatives? FYI, see implementation (https://flower.dev/docs/apiref-flwr.html#server-strategy-fedprox) of [`flower` package](https://flower.dev). It seems same as the one in this repo. I think they...
Hi, @vaseline555, yes for the first part what I mean for W is the set of different layers' weights. In this repo's code it do calculates different layers‘ proximal term...
> independently Hi Adam, thanks for the reply and here's a quick response to the last point. Yes I think for the gradients generated in case A and B, the...