Hi all! I am trying to construct a loss function in Flux (v0.11.2) that regularizes the parameters relative to some number (not 0). Below is a minimal working example, does anybody know why the gradients are `nothing`

?

```
using Flux
m = Chain(Dense(2, 10, relu), Dense(10, 10, relu), Dense(10, 1))
ps = Flux.params(m)
function project_norm() # basically trying to do sum(||x - p|| for all p in params) given a specific x
t = 0.0f0
for k=1:length(ps)
for j=1:length(ps[k])
t += (1.0f0 - ps[k][j])^2
end
end
return t
end
gs = gradient(ps) do
project_norm()
end
gs[ps[1]] # returns nothing?
```