diff --git a/src/layers/stateless.jl b/src/layers/stateless.jl
index 5298b634bb..06c8b6a4a9 100644
--- a/src/layers/stateless.jl
+++ b/src/layers/stateless.jl
@@ -35,27 +35,21 @@ Per default, `dims` is the last dimension.
 
 # Examples
 ```jldoctest
-julia> x = [9, 10, 20, 60];
+julia> using Statistics
 
-julia> Flux.std(x)
-24.01908963026423
+julia> x = [9, 10, 20, 60];
 
 julia> y = Flux.normalise(x);
 
-julia> Flux.std(y)
-1.1546999832655012
+julia> isapprox(std(y), 1, atol=0.2) && std(y) != std(x)
+true
 
 julia> x = rand(1:100, 10, 2);
 
-julia> Flux.std(x, dims=1)
-1×2 Matrix{Float64}:
- 28.5324  34.6425
-
 julia> y = Flux.normalise(x, dims=1);
 
-julia> Flux.std(y, dims=1)
-1×2 Matrix{Float64}:
- 1.05409  1.05409
+julia> isapprox(std(y, dims=1), ones(1, 2), atol=0.2) && std(y, dims=1) != std(x, dims=1)
+true
 ```
 """
 @inline function normalise(x::AbstractArray; dims=ndims(x), ϵ=ofeltype(x, 1e-5))
diff --git a/src/losses/functions.jl b/src/losses/functions.jl
index c0278bd395..ebf2de2301 100644
--- a/src/losses/functions.jl
+++ b/src/losses/functions.jl
@@ -401,7 +401,7 @@ distribution `y`; calculated as -
 ```jldoctest
 julia> y_model = [1, 3, 3];  # data should only take integral values
 
-julia> poisson_loss(y_model, 1:3)
+julia> Flux.poisson_loss(y_model, 1:3)
 0.5023128522198171
 ```
 """
@@ -430,14 +430,14 @@ julia> y_pred = [0.1, 0.3, 1, 1.5];
 julia> Flux.hinge_loss(y_pred, y_true)
 0.55
 
-julia> Flux.hinge_loss(y_pred[1], y_true[1])  # same sign but |ŷ| < 1
-0.9
+julia> Flux.hinge_loss(y_pred[1], y_true[1]) != 0  # same sign but |ŷ| < 1
+true
 
-julia> Flux.hinge_loss(y_pred[end], y_true[end])  # same sign but |ŷ| >= 1 -> loss = 0
-0.0
+julia> Flux.hinge_loss(y_pred[end], y_true[end]) == 0  # same sign but |ŷ| >= 1
+true
 
-julia> Flux.hinge_loss(y_pred[2], y_true[2])  # opposite signs -> loss != 0
-1.3
+julia> Flux.hinge_loss(y_pred[2], y_true[2]) != 0 # opposite signs
+true
 ```
 """
 function hinge_loss(ŷ, y; agg = mean)
@@ -465,14 +465,14 @@ julia> y_pred = [0.1, 0.3, 1, 1.5];
 julia> Flux.squared_hinge_loss(y_pred, y_true)
 0.625
 
-julia> Flux.squared_hinge_loss(y_pred[1], y_true[1])  # same sign but |ŷ| < 1
-0.81
+julia> Flux.squared_hinge_loss(y_pred[1], y_true[1]) != 0
+true
 
-julia> Flux.squared_hinge_loss(y_pred[end], y_true[end])  # same sign and |ŷ| >= 1 -> loss = 0
-0.0
+julia> Flux.squared_hinge_loss(y_pred[end], y_true[end]) == 0
+true
 
-julia> Flux.squared_hinge_loss(y_pred[2], y_true[2])  # opposite signs -> loss != 0
-1.6900000000000002
+julia> Flux.squared_hinge_loss(y_pred[2], y_true[2]) != 0
+true
 ```
 """
 function squared_hinge_loss(ŷ, y; agg = mean)
@@ -527,8 +527,8 @@ julia> ŷ_fnp = [1, 1, 0, 1, 1, 0];  # 1 false negative, 1 false positive -> 2
 julia> Flux.tversky_loss(ŷ_fnp, y)
 0.19999999999999996
 
-julia> Flux.tversky_loss(ŷ_fp, y)  # should be smaller than tversky_loss(ŷ_fnp, y), as FN is given more weight
-0.1071428571428571
+julia> Flux.tversky_loss(ŷ_fp, y) < Flux.tversky_loss(ŷ_fnp, y)  # FN is given more weight
+true
 ```
 """
 function tversky_loss(ŷ, y; β = ofeltype(ŷ, 0.7))
diff --git a/src/utils.jl b/src/utils.jl
index 15525c6357..c67d1a6270 100644
--- a/src/utils.jl
+++ b/src/utils.jl
@@ -535,7 +535,7 @@ Flux
 
 julia> for i = 1:4  # sleeps for 1 second -> the function can be called in alternate iterations
            a()
-           sleep(1)
+           sleep(1.5)
        end
 Flux
 Flux
diff --git a/test/losses.jl b/test/losses.jl
index 2ca697a657..c1061bb388 100644
--- a/test/losses.jl
+++ b/test/losses.jl
@@ -163,8 +163,8 @@ y = [1.0 0.5 0.3 2.4]
 end
 
 @testset "tversky_loss" begin
-  @test Flux.tversky_loss(ŷ, y) ≈ -0.06772009029345383
-  @test Flux.tversky_loss(ŷ, y, β=0.8) ≈ -0.09490740740740744
+  @test Flux.tversky_loss(ŷ, y) ≈ 0.028747433264887046
+  @test Flux.tversky_loss(ŷ, y, β=0.8) ≈ 0.050200803212851364
   @test Flux.tversky_loss(y, y) ≈ -0.5576923076923075
 end