diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 4876530..a50bb30 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -22,7 +22,7 @@ jobs:
         version:
           - 'lts'
           - '1'
-          # - 'pre' # TODO: tests on pre-release version
+          - 'pre'
     steps:
       - uses: actions/checkout@v4
       - uses: julia-actions/setup-julia@v2
diff --git a/Project.toml b/Project.toml
index b65b8dc..ef0e644 100644
--- a/Project.toml
+++ b/Project.toml
@@ -22,4 +22,4 @@ Reexport = "1"
 Statistics = "<0.0.1, 1"
 XAIBase = "4"
 Zygote = "0.6"
-julia = "1.6"
+julia = "1.10"
diff --git a/README.md b/README.md
index e383bfd..e08b9cd 100644
--- a/README.md
+++ b/README.md
@@ -16,7 +16,7 @@ and [iNNvestigate][innvestigate-repo] for Keras models.
 [^1]: More specifically, models currently have to be differentiable with [Zygote.jl](https://github.com/FluxML/Zygote.jl).
 
 ## Installation 
-This package supports Julia ≥1.6. To install it, open the Julia REPL and run 
+This package supports Julia ≥1.10. To install it, open the Julia REPL and run 
 ```julia-repl
 julia> ]add ExplainableAI
 ```
diff --git a/src/ExplainableAI.jl b/src/ExplainableAI.jl
index f999b15..567971f 100644
--- a/src/ExplainableAI.jl
+++ b/src/ExplainableAI.jl
@@ -14,7 +14,6 @@ using DifferentiationInterface: value_and_pullback
 using Zygote
 const DEFAULT_AD_BACKEND = AutoZygote()
 
-include("compat.jl")
 include("bibliography.jl")
 include("input_augmentation.jl")
 include("gradient.jl")
diff --git a/src/compat.jl b/src/compat.jl
deleted file mode 100644
index aa98e07..0000000
--- a/src/compat.jl
+++ /dev/null
@@ -1,12 +0,0 @@
-# https://github.com/JuliaLang/julia/pull/39794
-if VERSION < v"1.7.0-DEV.793"
-    export Returns
-
-    struct Returns{V} <: Function
-        value::V
-        Returns{V}(value) where {V} = new{V}(value)
-        Returns(value) = new{Core.Typeof(value)}(value)
-    end
-
-    (obj::Returns)(args...; kw...) = obj.value
-end
diff --git a/test/Project.toml b/test/Project.toml
index 48d2ec4..08f0277 100644
--- a/test/Project.toml
+++ b/test/Project.toml
@@ -8,6 +8,6 @@ JuliaFormatter = "98e50ef6-434e-11e9-1051-2b60c6c9e899"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 ReferenceTests = "324d217c-45ce-50fc-942e-d289b448e8cf"
-Suppressor = "fd094767-a336-5f1f-9728-57cf17d0bbfb"
+StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 XAIBase = "9b48221d-a747-4c1b-9860-46a1d8ba24a7"
diff --git a/test/references/cnn/GradCAM_max.jld2 b/test/references/cnn/GradCAM_max.jld2
index 5031171..944423b 100644
Binary files a/test/references/cnn/GradCAM_max.jld2 and b/test/references/cnn/GradCAM_max.jld2 differ
diff --git a/test/references/cnn/Gradient_max.jld2 b/test/references/cnn/Gradient_max.jld2
index 8b774c4..8818624 100644
Binary files a/test/references/cnn/Gradient_max.jld2 and b/test/references/cnn/Gradient_max.jld2 differ
diff --git a/test/references/cnn/InputTimesGradient_max.jld2 b/test/references/cnn/InputTimesGradient_max.jld2
index dd54f3d..0951dea 100644
Binary files a/test/references/cnn/InputTimesGradient_max.jld2 and b/test/references/cnn/InputTimesGradient_max.jld2 differ
diff --git a/test/references/cnn/IntegratedGradients_max.jld2 b/test/references/cnn/IntegratedGradients_max.jld2
index 01b02f2..f78ec5a 100644
Binary files a/test/references/cnn/IntegratedGradients_max.jld2 and b/test/references/cnn/IntegratedGradients_max.jld2 differ
diff --git a/test/references/cnn/SmoothGrad_max.jld2 b/test/references/cnn/SmoothGrad_max.jld2
index 89c9f15..f0676f8 100644
Binary files a/test/references/cnn/SmoothGrad_max.jld2 and b/test/references/cnn/SmoothGrad_max.jld2 differ
diff --git a/test/runtests.jl b/test/runtests.jl
index e73db4d..f6b79ee 100644
--- a/test/runtests.jl
+++ b/test/runtests.jl
@@ -6,14 +6,13 @@ using Aqua
 using JET
 
 @testset "ExplainableAI.jl" begin
-    if VERSION >= v"1.10"
-        @info "Testing formalities..."
+    @testset verbose = true "Linting" begin
         @testset "Code formatting" begin
             @info "- running JuliaFormatter code formatting tests..."
             @test JuliaFormatter.format(ExplainableAI; verbose=false, overwrite=false)
         end
         @testset "Aqua.jl" begin
-            @info "- running Aqua.jl tests. These might print warnings from dependencies..."
+            @info "- running Aqua.jl tests..."
             Aqua.test_all(ExplainableAI; ambiguities=false)
         end
         @testset "JET tests" begin
diff --git a/test/test_batches.jl b/test/test_batches.jl
index 473073d..c7ba69a 100644
--- a/test/test_batches.jl
+++ b/test/test_batches.jl
@@ -5,7 +5,7 @@ using Flux
 using Random
 using Distributions: Laplace
 
-pseudorand(dims...) = rand(MersenneTwister(123), Float32, dims...)
+pseudorand(dims...) = rand(StableRNG(123), Float32, dims...)
 
 ## Test `fuse_batchnorm` on Dense and Conv layers
 ins = 20
@@ -15,16 +15,16 @@ batchsize = 15
 model = Chain(Dense(ins, 15, relu; init=pseudorand), Dense(15, outs, relu; init=pseudorand))
 
 # Input 1 with batch dimension
-input1 = rand(MersenneTwister(1), Float32, ins, 1)
+input1 = rand(StableRNG(1), Float32, ins, 1)
 # Input 2 with batch dimension
-input2 = rand(MersenneTwister(2), Float32, ins, 1)
+input2 = rand(StableRNG(2), Float32, ins, 1)
 # Batch containing inputs 1 & 2
 input_batch = cat(input1, input2; dims=2)
 
 ANALYZERS = Dict(
     "Gradient"            => Gradient,
     "InputTimesGradient"  => InputTimesGradient,
-    "SmoothGrad"          => m -> SmoothGrad(m, 5, 0.1, MersenneTwister(123)),
+    "SmoothGrad"          => m -> SmoothGrad(m, 5, 0.1, StableRNG(123)),
     "IntegratedGradients" => m -> IntegratedGradients(m, 5),
     "GradCAM"             => m -> GradCAM(m[1], m[2]),
 )
diff --git a/test/test_cnn.jl b/test/test_cnn.jl
index 6e7f6a8..b8ff8f3 100644
--- a/test/test_cnn.jl
+++ b/test/test_cnn.jl
@@ -4,21 +4,22 @@ using ReferenceTests
 
 using Flux
 using Random
+using StableRNGs: StableRNG
 using JLD2
 
 const GRADIENT_ANALYZERS = Dict(
     "InputTimesGradient"  => InputTimesGradient,
-    "SmoothGrad"          => m -> SmoothGrad(m, 5, 0.1, MersenneTwister(123)),
+    "SmoothGrad"          => m -> SmoothGrad(m, 5, 0.1, StableRNG(123)),
     "IntegratedGradients" => m -> IntegratedGradients(m, 5),
     "GradCAM"             => m -> GradCAM(m[1], m[2]),
 )
 
-pseudorand(dims...) = rand(MersenneTwister(123), Float32, dims...)
+pseudorand(dims...) = rand(StableRNG(123), Float32, dims...)
 
 input_size = (32, 32, 3, 1)
 input = pseudorand(input_size)
 
-init(dims...) = Flux.glorot_uniform(MersenneTwister(123), dims...)
+init(dims...) = Flux.glorot_uniform(StableRNG(123), dims...)
 
 model = Chain(
     Chain(