Julia-XAI · adrhill · Oct 11, 2024 · Oct 11, 2024
diff --git a/Project.toml b/Project.toml
@@ -11,7 +11,6 @@ Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 XAIBase = "9b48221d-a747-4c1b-9860-46a1d8ba24a7"
-Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [compat]
 ADTypes = "1"
@@ -21,5 +20,4 @@ Random = "<0.0.1, 1"
 Reexport = "1"
 Statistics = "<0.0.1, 1"
 XAIBase = "4"
-Zygote = "0.6"
 julia = "1.10"
diff --git a/benchmark/Project.toml b/benchmark/Project.toml
@@ -4,6 +4,7 @@ ExplainableAI = "4f1bc3e1-d60d-4ed0-9367-9bdff9846d3b"
 Flux = "587475ba-b771-5e3f-ad9e-33799f191a9c"
 PkgBenchmark = "32113eaa-f34f-5b0d-bd6c-c81e245fc73d"
 PkgJogger = "10150987-6cc1-4b76-abee-b1c1cbd91c01"
+Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [compat]
 BenchmarkTools = "1"

diff --git a/benchmark/bench_jogger.jl b/benchmark/bench_jogger.jl
@@ -1,4 +1,5 @@
 using BenchmarkTools
+using Zygote
 using Flux
 using ExplainableAI
 

diff --git a/docs/Project.toml b/docs/Project.toml
@@ -11,6 +11,7 @@ ImageShow = "4e3cecfd-b093-5904-9786-8bbb286a6a31"
 Literate = "98b081ad-f1c9-55d3-8b20-4c87d4299306"
 MLDatasets = "eb30cadb-4394-5ae3-aed4-317e484a6458"
 VisionHeatmaps = "27106da1-f8bc-4ca8-8c66-9b8289f1e035"
+Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [compat]
-VisionHeatmaps = "1.4"
+VisionHeatmaps = "1.4"
diff --git a/docs/src/literate/augmentations.jl b/docs/src/literate/augmentations.jl
@@ -8,6 +8,7 @@
 # and start out by loading the same pre-trained LeNet5 model and MNIST input data:
 using ExplainableAI
 using VisionHeatmaps
+using Zygote
 using Flux
 
 using BSON # hide

diff --git a/docs/src/literate/example.jl b/docs/src/literate/example.jl
@@ -7,7 +7,6 @@
 
 # For this first example, we already have loaded a pre-trained LeNet5 model
 # to look at explanations on the MNIST dataset.
-using ExplainableAI
 using Flux
 
 using BSON # hide
@@ -41,8 +40,11 @@ input = reshape(x, 28, 28, 1, :);
 #md #     (width, height, channels, batch), which is Flux.jl's convention.
 
 # ## Explanations
-# We can now select an analyzer of our choice and call [`analyze`](@ref)
-# to get an [`Explanation`](@ref):
+# We can now select an analyzer of our choice and call [`analyze`](@ref) to get an [`Explanation`](@ref).
+# Note that for gradient-based optimizers, a backend for automatic differentiation must be loaded, by default [Zygote.jl](https://github.com/FluxML/Zygote.jl):
+using ExplainableAI
+using Zygote
+
 analyzer = InputTimesGradient(model)
 expl = analyze(input, analyzer);
 

diff --git a/docs/src/literate/heatmapping.jl b/docs/src/literate/heatmapping.jl
@@ -10,6 +10,7 @@
 # We start out by loading the same pre-trained LeNet5 model and MNIST input data:
 using ExplainableAI
 using VisionHeatmaps
+using Zygote
 using Flux
 
 using BSON # hide

diff --git a/src/ExplainableAI.jl b/src/ExplainableAI.jl
@@ -11,7 +11,6 @@ using Random: AbstractRNG, GLOBAL_RNG
 # Automatic differentiation
 using ADTypes: AbstractADType, AutoZygote
 using DifferentiationInterface: value_and_pullback
-using Zygote
 const DEFAULT_AD_BACKEND = AutoZygote()
 
 include("bibliography.jl")

diff --git a/test/Project.toml b/test/Project.toml
@@ -13,3 +13,4 @@ ReferenceTests = "324d217c-45ce-50fc-942e-d289b448e8cf"
 StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 XAIBase = "9b48221d-a747-4c1b-9860-46a1d8ba24a7"
+Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
diff --git a/test/test_batches.jl b/test/test_batches.jl
@@ -1,4 +1,5 @@
 using ExplainableAI
+using Zygote
 using Test
 
 using Flux

diff --git a/test/test_cnn.jl b/test/test_cnn.jl
@@ -1,4 +1,5 @@
 using ExplainableAI
+using Zygote
 using Test
 using ReferenceTests