fixes

src/layers/losses.jl

@@ -27,7 +27,7 @@ The loss corresponding to mean squared logarithmic errors, calculated as
 The `ϵ` term provides numerical stability.
 Penalizes an under-predicted estimate more than an over-predicted estimate.
 """
-msle(ŷ, y; agg=mean, ϵ=eps(eltype(ŷ))) = agg((log.(ŷ .+ ϵ) .- log.(y .+ ϵ)).^2)
+msle(ŷ, y; agg=mean, ϵ=epseltype(ŷ)) = agg((log.(ŷ .+ ϵ) .- log.(y .+ ϵ)).^2)
 
 
 """
@@ -64,7 +64,7 @@ calculated as
 
 See also: [`Flux.logitcrossentropy`](@ref), [`Flux.binarycrossentropy`](@ref), [`Flux.logitbinarycrossentropy`](@ref)
 """
-function crossentropy(ŷ, y; dims=1, agg=mean, ϵ=eps(eltype(ŷ)), weight=nothing)
+function crossentropy(ŷ, y; dims=1, agg=mean, ϵ=epseltype(ŷ), weight=nothing)
     agg(.-wsum(weight, y .* log.(ŷ .+ ϵ); dims=dims))
 end
 
@@ -94,7 +94,7 @@ Typically, the prediction `ŷ` is given by the output of a [`sigmoid`](@ref) ac
 
 See also: [`Flux.crossentropy`](@ref), [`Flux.logitcrossentropy`](@ref), [`Flux.logitbinarycrossentropy`](@ref)
 """
-function binarycrossentropy(ŷ, y; agg=mean, ϵ=eps(eltype(ŷ)))
+function binarycrossentropy(ŷ, y; agg=mean, ϵ=epseltype(ŷ))
     agg(@.(-y*log(ŷ+ϵ) - (1-y)*log(1-ŷ+ϵ)))
 end
 
@@ -128,21 +128,21 @@ from the other.
 It is always non-negative and zero only when both the distributions are equal
 everywhere.
 """
-function kldivergence(ŷ, y; dims=1, agg=mean, ϵ=eps(eltype(ŷ)))
+function kldivergence(ŷ, y; dims=1, agg=mean, ϵ=epseltype(ŷ))
   entropy = agg(sum(y .* log.(y .+ ϵ), dims=dims))
   cross_entropy = crossentropy(ŷ, y; dims=dims, agg=agg, ϵ=ϵ)
   return entropy + cross_entropy
 end
 
 """
-    poisson_loss(ŷ, y; agg=mean)
+    poisson_loss(ŷ, y; agg=mean, ϵ=eps(eltype(ŷ))))
 
 # Return how much the predicted distribution `ŷ` diverges from the expected Poisson
 # distribution `y`; calculated as `sum(ŷ .- y .* log.(ŷ)) / size(y, 2)`.
 REDO
 [More information.](https://peltarion.com/knowledge-center/documentation/modeling-view/build-an-ai-model/loss-functions/poisson).
 """
-poisson_loss(ŷ, y; agg=mean) = agg(ŷ .- y .* log.(ŷ))
+poisson_loss(ŷ, y; agg=mean, ϵ=epseltype(ŷ)) = agg(ŷ .- y .* log.(ŷ .+ ϵ))
 
 @deprecate poisson poisson_loss
 

src/utils.jl

@@ -5,6 +5,8 @@ nfan(n_out, n_in) = n_in, n_out # In case of Dense kernels: arranged as matrices
 nfan(dims...) = prod(dims[1:end-2]) .* (dims[end-1], dims[end]) # In case of convolution kernels
 
 ofeltype(x, y) = convert(float(eltype(x)), y)
+epseltype(x) = eps(float(eltype(x)))
+
 
 """
     glorot_uniform(dims...)

test/layers/stateless.jl

@@ -107,7 +107,7 @@ const ϵ = 1e-7
     for T in (Float32, Float64)
       y = rand(T, 2)
       ŷ = rand(T, 2)
-      for f in (mse, crossentropy, logitcrossentropy, Flux.kldivergence, Flux.hinge, Flux.poisson,
+      for f in (mse, crossentropy, logitcrossentropy, Flux.kldivergence, Flux.hinge, Flux.poisson_loss,
               Flux.mae, Flux.huber_loss, Flux.msle, Flux.squared_hinge, Flux.dice_coeff_loss, Flux.tversky_loss)
         fwd, back = Flux.pullback(f, ŷ, y)
         @test fwd isa T