Add pretrained weights on ImageNet for some models

Artifacts.toml

@@ -1,3 +1,75 @@
+[resnet101]
+git-tree-sha1 = "68d563526ab34d3e5aa66b7d96278d2acde212f9"
+lazy = true
+
+    [[resnet101.download]]
+    sha256 = "0725f05db5772cfab1024b8d0d6c85ac1fc5a83eb6f0fe02b67b1e689d5a28db"
+    url = "https://huggingface.co/FluxML/resnet101/resolve/980158099e6917b74ade2b0a9599359f06057d21/resnet101.tar.gz"
+
+[resnet152]
+git-tree-sha1 = "85a97464b6cef66e1217ae829d3620651cffab47"
+lazy = true
+
+    [[resnet152.download]]
+    sha256 = "a8d30a735ef5649ec40a74a0515ee3d6774499267be06f5f2b372259c5ced8d6"
+    url = "https://huggingface.co/FluxML/resnet152/resolve/a66a3e1f5056179d167cb2165401950e3890b34d/resnet152.tar.gz"
+
+[resnet18]
+git-tree-sha1 = "4ced5a0338c0f0293940f1deb63e1c463125a6ff"
+lazy = true
+
+    [[resnet18.download]]
+    sha256 = "9444ef2285f507bd890d2ca852d663749f079110ed19b544e8d91f67f3cc6b83"
+    url = "https://huggingface.co/FluxML/resnet18/resolve/9b1c6c4f7c5dbe734d80d7d4b5f132ef58bf2467/resnet18.tar.gz"
+
+[resnet34]
+git-tree-sha1 = "485519977f375ca1770b3ff3971f61e438823f5a"
+lazy = true
+
+    [[resnet34.download]]
+    sha256 = "71ed75be6db0160af7f30be33e2f4a44836310949d1374267510e5803b1fb313"
+    url = "https://huggingface.co/FluxML/resnet34/resolve/0988ae2d4a86da06eefa6b61edf3e728861e286c/resnet34.tar.gz"
+
+[resnet50]
+git-tree-sha1 = "2973be0da60544080105756ecb3951cca2e007da"
+lazy = true
+
+    [[resnet50.download]]
+    sha256 = "60ad32eaf160444f3bfdb6f6d81ec1e5c36a3769be7df22aaa75127b16bb1501"
+    url = "https://huggingface.co/FluxML/resnet50/resolve/1529d6ddca42e3e705cb708c9de6f79188ce8ad5/resnet50.tar.gz"
+
+[resnext101_32x8d]
+git-tree-sha1 = "d13a85131b2c0c62ef2af79a09137d4e0760a685"
+lazy = true
+
+    [[resnext101_32x8d.download]]
+    sha256 = "aeb48f86f50ee8b0ca7dc01ca0ff5a2d2b2163e43c524203c4a8bd589db9bcc6"
+    url = "https://huggingface.co/FluxML/resnext101_32x8d/resolve/e060f030c445f644112efa2a00e3c544944046e1/resnext101_32x8d.tar.gz"
+
+[resnext101_64x4d]
+git-tree-sha1 = "db50f48614e673a40f98fb80a17688b34f42067a"
+lazy = true
+
+    [[resnext101_64x4d.download]]
+    sha256 = "89764dd7dc3b3432f0424cb592cec5d9db2fb802ab1646f0e3c2cca2b2e5386b"
+    url = "https://huggingface.co/FluxML/resnext101_64x4d/resolve/0d0485da04efe5a53289a560d105c42d3ca5435c/resnext101_64x4d.tar.gz"
+
+[resnext50_32x4d]
+git-tree-sha1 = "1e7a08a4acae690b635e8d1caa06e75eeb2dd2fe"
+lazy = true
+
+    [[resnext50_32x4d.download]]
+    sha256 = "084ccbc40fde07496c401ee2bc389b9cd1d60b1ac3b7ccbfde05479ea91ca707"
+    url = "https://huggingface.co/FluxML/resnext50_32x4d/resolve/150c52c9646fe697030d38ab2be767564fb4f28c/resnext50_32x4d.tar.gz"
+
+[squeezenet]
+git-tree-sha1 = "e2eeee109fda46470d657b13669cca09d5ef2f8c"
+lazy = true
+
+    [[squeezenet.download]]
+    sha256 = "aebfa06f44767e5ff728b7b67b2d01352b4618bd5d305c26e603aabcd5ba593d"
+    url = "https://huggingface.co/FluxML/squeezenet/resolve/01ef4221df5260bd992c669ab587eed74df0c39f/squeezenet.tar.gz"
+
 [vgg11]
 git-tree-sha1 = "78ffe7d74c475cc28175f9e23a545ce2f17b1520"
 lazy = true
@@ -30,42 +102,18 @@ lazy = true
     sha256 = "5fe26391572b9f6ac84eaa0541d27e959f673f82e6515026cdcd3262cbd93ceb"
     url = "https://huggingface.co/FluxML/vgg19/resolve/88e9056f60b054eccdc190a2eeb23731d5c693b6/vgg19.tar.gz"
 
-[resnet18]
-git-tree-sha1 = "7b555ed2708e551bfdbcb7e71b25001f4b3731c6"
-lazy = true
-
-    [[resnet18.download]]
-    sha256 = "d5782fd873a3072df251c7a4b3cf16efca8ee1da1180ff815bc107833f84bb26"
-    url = "https://huggingface.co/FluxML/resnet18/resolve/ef9c74047fda4a4a503b1f72553ec05acc90929f/resnet18.tar.gz"
-
-[resnet34]
-git-tree-sha1 = "e6e79666cd0fc81cd828508314e6c7f66df8d43d"
-lazy = true
-
-    [[resnet34.download]]
-    sha256 = "a8dec13609a86f7a2adac6a44b3af912a863bc2d7319120066c5fdaa04c3f395"
-    url = "https://huggingface.co/FluxML/resnet34/resolve/42061ddb463902885eea4fcc85275462a5445987/resnet34.tar.gz"
-
-[resnet50]
-git-tree-sha1 = "5c442ffd6c51a70c3bc36d849fca86beced446d4"
-lazy = true
-
-    [[resnet50.download]]
-    sha256 = "5325920ec91c2a4499ad7e659961f9eaac2b1a3a2905ca6410eaa593ecd35503"
-    url = "https://huggingface.co/FluxML/resnet50/resolve/10e601719e1cd5b0cab87ce7fd1e8f69a07ce042/resnet50.tar.gz"
-
-[resnet101]
-git-tree-sha1 = "694a8563ec20fb826334dd663d532b10bb2b3c97"
+[wideresnet101]
+git-tree-sha1 = "b881a9469fb230faff414ce9f983bc113061ab1c"
 lazy = true
 
-    [[resnet101.download]]
-    sha256 = "f4d737ce640957c30f76bfa642fc9da23e6852d81474d58a2338c1148e55bff0"
-    url = "https://huggingface.co/FluxML/resnet101/resolve/ea37819163cc3f4a41989a6239ce505e483b112d/resnet101.tar.gz"
+    [[wideresnet101.download]]
+    sha256 = "defa61fd80a988bb07bb9db00c692d8d0a30d95e6276add1413fb7f1f3aa2607"
+    url = "https://huggingface.co/FluxML/wideresnet101/resolve/ad4df1016bb5eba4c10e2d37b049ca5d2a455670/wideresnet101.tar.gz"
 
-[resnet152]
-git-tree-sha1 = "55eb883248a276d710d75ecaecfbd2427e50cc0a"
+[wideresnet50]
+git-tree-sha1 = "bbc6bc632e743c992784b5121dcb0f6082c66b1f"
 lazy = true
 
-    [[resnet152.download]]
-    sha256 = "57be335e6828d1965c9d11f933d2d41f51e5e534f9bfdbde01c6144fa8862a4d"
-    url = "https://huggingface.co/FluxML/resnet152/resolve/ba28814d5746643387b5c0e1d2269104e5e9bc8d/resnet152.tar.gz"
+    [[wideresnet50.download]]
+    sha256 = "7596a67b7aba762c2bfce8367055da79a8a3c117bd79ce11124c9f1f5a96c4e3"
+    url = "https://huggingface.co/FluxML/wideresnet50/resolve/5eca9979f5d9438a684b1e6a5b227f9d5611965a/wideresnet50.tar.gz"

src/convnets/densenet.jl

@@ -100,8 +100,7 @@ Create a DenseNet model
   - `nclasses`: the number of output classes
 """
 function densenet(nblocks::AbstractVector{<:Integer}; growth_rate::Integer = 32,
-                  reduction = 0.5,
-                  inchannels::Integer = 3, nclasses::Integer = 1000)
+                  reduction = 0.5, inchannels::Integer = 3, nclasses::Integer = 1000)
     return densenet(2 * growth_rate, [fill(growth_rate, n) for n in nblocks];
                     reduction, inchannels, nclasses)
 end
@@ -133,11 +132,11 @@ end
 function DenseNet(config::Integer; pretrain::Bool = false, growth_rate::Integer = 32,
                   reduction = 0.5, inchannels::Integer = 3, nclasses::Integer = 1000)
     _checkconfig(config, keys(DENSENET_CONFIGS))
-    model = densenet(DENSENET_CONFIGS[config]; growth_rate, reduction, inchannels, nclasses)
+    layers = densenet(DENSENET_CONFIGS[config]; growth_rate, reduction, inchannels, nclasses)
     if pretrain
-        loadpretrain!(model, string("DenseNet", config))
+        loadpretrain!(layers, string("densenet", config))
     end
-    return model
+    return DenseNet(layers)
 end
 
 (m::DenseNet)(x) = m.layers(x)

src/convnets/inception/inceptionv3.jl

@@ -1,5 +1,3 @@
-## Inceptionv3
-
 """
     inceptionv3_a(inplanes, pool_proj)
 
@@ -176,6 +174,7 @@ See also [`inceptionv3`](#).
 struct Inceptionv3
     layers::Any
 end
+@functor Inceptionv3
 
 function Inceptionv3(; pretrain::Bool = false, inchannels::Integer = 3,
                      nclasses::Integer = 1000)
@@ -186,8 +185,6 @@ function Inceptionv3(; pretrain::Bool = false, inchannels::Integer = 3,
     return Inceptionv3(layers)
 end
 
-@functor Inceptionv3
-
 (m::Inceptionv3)(x) = m.layers(x)
 
 backbone(m::Inceptionv3) = m.layers[1]

src/convnets/resnets/core.jl

@@ -280,15 +280,13 @@ function resnet_stages(get_layers, block_repeats::AbstractVector{<:Integer}, con
 end
 
 function resnet(img_dims, stem, get_layers, block_repeats::AbstractVector{<:Integer},
-                connection,
-                classifier_fn)
+                connection, classifier_fn)
     # Build stages of the ResNet
     stage_blocks = resnet_stages(get_layers, block_repeats, connection)
     backbone = Chain(stem, stage_blocks)
-    # Build the classifier head
+    # Add classifier to the backbone
     nfeaturemaps = Flux.outputsize(backbone, img_dims; padbatch = true)[3]
-    classifier = classifier_fn(nfeaturemaps)
-    return Chain(backbone, classifier)
+    return Chain(backbone, classifier_fn(nfeaturemaps))
 end
 
 function resnet(block_type::Symbol, block_repeats::AbstractVector{<:Integer};

src/convnets/resnets/resnet.jl

@@ -11,10 +11,6 @@ Creates a ResNet model with the specified depth.
   - `inchannels`: The number of input channels.
   - `nclasses`: the number of output classes
 
-!!! warning
-
-    `ResNet` does not currently support pretrained weights.
-
 Advanced users who want more configuration options will be better served by using [`resnet`](#).
 """
 struct ResNet
@@ -27,7 +23,7 @@ function ResNet(depth::Integer; pretrain::Bool = false, inchannels::Integer = 3,
     _checkconfig(depth, keys(RESNET_CONFIGS))
     layers = resnet(RESNET_CONFIGS[depth]...; inchannels, nclasses)
     if pretrain
-        loadpretrain!(layers, string("ResNet", depth))
+        loadpretrain!(layers, string("resnet", depth))
     end
     return ResNet(layers)
 end
@@ -52,10 +48,6 @@ The number of channels in outer 1x1 convolutions is the same.
   - `inchannels`: The number of input channels.
   - `nclasses`: the number of output classes
 
-!!! warning
-
-    `WideResNet` does not currently support pretrained weights.
-
 Advanced users who want more configuration options will be better served by using [`resnet`](#).
 """
 struct WideResNet

src/convnets/resnets/resnext.jl

@@ -1,23 +1,23 @@
 """
-    ResNeXt(depth::Integer; pretrain::Bool = false, cardinality = 32,
-            base_width = 4, inchannels::Integer = 3, nclasses::Integer = 1000)
+    ResNeXt(depth::Integer; pretrain::Bool = false, cardinality::Integer = 32,
+            base_width::Integer = 4, inchannels::Integer = 3, nclasses::Integer = 1000)
 
 Creates a ResNeXt model with the specified depth, cardinality, and base width.
 ((reference)[https://arxiv.org/abs/1611.05431])
 
 # Arguments
 
   - `depth`: one of `[18, 34, 50, 101, 152]`. The depth of the ResNet model.
-  - `pretrain`: set to `true` to load the model with pre-trained weights for ImageNet
+  - `pretrain`: set to `true` to load the model with pre-trained weights for ImageNet.
+    Supported configurations are:
+      - depth 50, cardinality of 32 and base width of 4.
+      - depth 101, cardinality of 32 and base width of 8.
+      - depth 101, cardinality of 64 and base width of 4.
   - `cardinality`: the number of groups to be used in the 3x3 convolution in each block.
   - `base_width`: the number of feature maps in each group.
   - `inchannels`: the number of input channels.
   - `nclasses`: the number of output classes
 
-!!! warning
-
-    `ResNeXt` does not currently support pretrained weights.
-
 Advanced users who want more configuration options will be better served by using [`resnet`](#).
 """
 struct ResNeXt
@@ -27,12 +27,12 @@ end
 
 (m::ResNeXt)(x) = m.layers(x)
 
-function ResNeXt(depth::Integer; pretrain::Bool = false, cardinality = 32,
-                 base_width = 4, inchannels::Integer = 3, nclasses::Integer = 1000)
+function ResNeXt(depth::Integer; pretrain::Bool = false, cardinality::Integer = 32,
+                 base_width::Integer = 4, inchannels::Integer = 3, nclasses::Integer = 1000)
     _checkconfig(depth, sort(collect(keys(RESNET_CONFIGS)))[3:end])
     layers = resnet(RESNET_CONFIGS[depth]...; inchannels, nclasses, cardinality, base_width)
     if pretrain
-        loadpretrain!(layers, string("ResNeXt", depth, "_", cardinality, "x", base_width))
+        loadpretrain!(layers, string("resnext", depth, "_", cardinality, "x", base_width, "d"))
     end
     return ResNeXt(layers)
 end

src/convnets/resnets/seresnet.jl

@@ -30,7 +30,7 @@ function SEResNet(depth::Integer; pretrain::Bool = false, inchannels::Integer =
     layers = resnet(RESNET_CONFIGS[depth]...; inchannels, nclasses,
                     attn_fn = squeeze_excite)
     if pretrain
-        loadpretrain!(layers, string("SEResNet", depth))
+        loadpretrain!(layers, string("seresnet", depth))
     end
     return SEResNet(layers)
 end
@@ -39,8 +39,8 @@ backbone(m::SEResNet) = m.layers[1]
 classifier(m::SEResNet) = m.layers[2]
 
 """
-    SEResNeXt(depth::Integer; pretrain::Bool = false, cardinality = 32, base_width = 4,
-              inchannels::Integer = 3, nclasses::Integer = 1000)
+    SEResNeXt(depth::Integer; pretrain::Bool = false, cardinality::Integer = 32,
+              base_width::Integer = 4, inchannels::Integer = 3, nclasses::Integer = 1000)
 
 Creates a SEResNeXt model with the specified depth, cardinality, and base width.
 ((reference)[https://arxiv.org/pdf/1709.01507.pdf])
@@ -67,13 +67,13 @@ end
 
 (m::SEResNeXt)(x) = m.layers(x)
 
-function SEResNeXt(depth::Integer; pretrain::Bool = false, cardinality = 32, base_width = 4,
-                   inchannels::Integer = 3, nclasses::Integer = 1000)
+function SEResNeXt(depth::Integer; pretrain::Bool = false, cardinality::Integer = 32,
+                   base_width::Integer = 4, inchannels::Integer = 3, nclasses::Integer = 1000)
     _checkconfig(depth, sort(collect(keys(RESNET_CONFIGS)))[3:end])
     layers = resnet(RESNET_CONFIGS[depth]...; inchannels, nclasses, cardinality, base_width,
                     attn_fn = squeeze_excite)
     if pretrain
-        loadpretrain!(layers, string("SEResNeXt", depth, "_", cardinality, "x", base_width))
+        loadpretrain!(layers, string("seresnext", depth, "_", cardinality, "x", base_width))
     end
     return SEResNeXt(layers)
 end

src/convnets/squeezenet.jl

@@ -62,10 +62,6 @@ Create a SqueezeNet
   - `inchannels`: number of input channels.
   - `nclasses`: the number of output classes.
 
-!!! warning
-
-    `SqueezeNet` does not currently support pretrained weights.
-
 See also [`squeezenet`](#).
 """
 struct SqueezeNet
@@ -77,7 +73,7 @@ function SqueezeNet(; pretrain::Bool = false, inchannels::Integer = 3,
                     nclasses::Integer = 1000)
     layers = squeezenet(; inchannels, nclasses)
     if pretrain
-        loadpretrain!(layers, "SqueezeNet")
+        loadpretrain!(layers, "squeezenet")
     end
     return SqueezeNet(layers)
 end

src/layers/mlp.jl

@@ -66,16 +66,19 @@ Creates a classifier head to be used for models.
 function create_classifier(inplanes::Integer, nclasses::Integer, activation = identity;
                            use_conv::Bool = false, pool_layer = AdaptiveMeanPool((1, 1)),
                            dropout_rate = nothing)
-    # Pooling
+    # Decide whether to flatten the input or not
     flatten_in_pool = !use_conv && pool_layer !== identity
     if use_conv
         @assert pool_layer === identity
         "`pool_layer` must be identity if `use_conv` is true"
     end
-    global_pool = flatten_in_pool ? [pool_layer, MLUtils.flatten] : [pool_layer]
+    classifier = []
+    flatten_in_pool ? push!(classifier, pool_layer, MLUtils.flatten) : 
+        push!(classifier, pool_layer)
+    # Dropout is applied after the pooling layer
+    isnothing(dropout_rate) ? nothing : push!(classifier, Dropout(dropout_rate))
     # Fully-connected layer
-    fc = use_conv ? Conv((1, 1), inplanes => nclasses, activation) :
-         Dense(inplanes => nclasses, activation)
-    drop = isnothing(dropout_rate) ? [] : [Dropout(dropout_rate)]
-    return Chain(global_pool..., drop..., fc)
+    use_conv ? push!(classifier, Conv((1, 1), inplanes => nclasses, activation)) :
+        push!(classifier, Dense(inplanes => nclasses, activation))
+    return Chain(classifier...)
 end

src/utilities.jl

@@ -73,6 +73,5 @@ end
 
 # Utility function for depth and configuration checks in models
 function _checkconfig(config, configs)
-    @assert config in configs
-    return "Invalid configuration. Must be one of $(sort(collect(configs)))."
+    @assert config in configs "Invalid configuration. Must be one of $(sort(collect(configs)))."
 end

test/convnets.jl

@@ -22,15 +22,14 @@ end
 
 @testset "ResNet" begin
     # Tests for pretrained ResNets
-    ## TODO: find a way to port pretrained models to the new ResNet API
     @testset "ResNet($sz)" for sz in [18, 34, 50, 101, 152]
         m = ResNet(sz)
         @test size(m(x_224)) == (1000, 1)
-        # if (ResNet, sz) in PRETRAINED_MODELS
-        #     @test acctest(ResNet(sz, pretrain = true))
-        # else
-        #     @test_throws ArgumentError ResNet(sz, pretrain = true)
-        # end
+        if (ResNet, sz) in PRETRAINED_MODELS
+            @test acctest(ResNet(sz, pretrain = true))
+        else
+            @test_throws ArgumentError ResNet(sz, pretrain = true)
+        end
     end
 
     @testset "resnet" begin
@@ -79,9 +78,9 @@ end
                 m = ResNeXt(depth; cardinality, base_width)
                 @test size(m(x_224)) == (1000, 1)
                 if (ResNeXt, depth, cardinality, base_width) in PRETRAINED_MODELS
-                    @test acctest(ResNeXt(depth, pretrain = true))
+                    @test acctest(ResNeXt(depth; cardinality, base_width, pretrain = true))
                 else
-                    @test_throws ArgumentError ResNeXt(depth, pretrain = true)
+                    @test_throws ArgumentError ResNeXt(depth; cardinality, base_width, pretrain = true)
                 end
                 @test gradtest(m, x_224)
                 _gc()