Minor documentation build errors

Author: darsnack

src/convnets/convnext.jl

@@ -8,7 +8,7 @@ Creates a single block of ConvNeXt.
 
   - `planes`: number of input channels.
   - `drop_path_rate`: Stochastic depth rate.
-  - `λ`: Init value for LayerScale
+  - `λ`: Initial value for [`LayerScale`](#)
 """
 function convnextblock(planes, drop_path_rate = 0.0, λ = 1.0f-6)
     layers = SkipConnection(Chain(DepthwiseConv((7, 7), planes => planes; pad = 3),
@@ -33,7 +33,8 @@ Creates the layers for a ConvNeXt model.
   - `depths`: list with configuration for depth of each block
   - `planes`: list with configuration for number of output channels in each block
   - `drop_path_rate`: Stochastic depth rate.
-  - `λ`: Init value for [LayerScale](https://arxiv.org/abs/2103.17239)
+  - `λ`: Initial value for [`LayerScale`](#) 
+         ([reference](https://arxiv.org/abs/2103.17239))
   - `nclasses`: number of output classes
 """
 function convnext(depths, planes; inchannels = 3, drop_path_rate = 0.0, λ = 1.0f-6,

src/convnets/inception.jl

@@ -332,7 +332,7 @@ Creates an Inceptionv4 model.
 
 !!! warning
     
-    `Inceptionv4`` does not currently support pretrained weights.
+    `Inceptionv4` does not currently support pretrained weights.
 """
 struct Inceptionv4
     layers::Any
@@ -464,8 +464,8 @@ Creates an InceptionResNetv2 model.
   - `nclasses`: the number of output classes.
 
 !!! warning
-    
-        `InceptionResNetv2` does not currently support pretrained weights.
+
+    `InceptionResNetv2` does not currently support pretrained weights.
 """
 struct InceptionResNetv2
     layers::Any

src/convnets/resnet.jl

@@ -226,10 +226,6 @@ See also [`Metalhead.resnet`](#).
   - `depth`: depth of the ResNet model. Options include (18, 34, 50, 101, 152).
   - `nclasses`: the number of output classes
 
-!!! warning
-    
-    Only `ResNet(50)` currently supports pretrained weights.
-
 For `ResNet(18)` and `ResNet(34)`, the parameter-free shortcut style (type `:A`)
 is used in the first block and the three other blocks use type `:B` connection
 (following the implementation in PyTorch). The published version of

src/convnets/resnext.jl

@@ -112,9 +112,8 @@ Create a ResNeXt model with specified configuration. Currently supported values
 Set `pretrain = true` to load the model with pre-trained weights for ImageNet.
 
 !!! warning
-    
 
-`ResNeXt` does not currently support pretrained weights.
+    `ResNeXt` does not currently support pretrained weights.
 
 See also [`Metalhead.resnext`](#).
 """

src/convnets/vgg.jl

@@ -154,10 +154,6 @@ Create a VGG style model with specified `depth`. Available values include (11, 1
 ([reference](https://arxiv.org/abs/1409.1556v6)).
 See also [`VGG`](#).
 
-!!! warning
-    
-    `VGG` does not currently support pretrained weights.
-
 # Arguments
 
   - `pretrain`: set to `true` to load pre-trained model weights for ImageNet

src/layers/conv.jl

@@ -154,7 +154,7 @@ Squeeze and excitation layer used by MobileNet variants
 
   - `channels`: the number of input/output feature maps
   - `reduction = 4`: the reduction factor for the number of hidden feature maps
-    (must be >= 1)
+    (must be ≥ 1)
 """
 function squeeze_excite(channels, reduction = 4)
     @assert (reduction>=1) "`reduction` must be >= 1"
@@ -182,7 +182,7 @@ Create a basic inverted residual block for MobileNet variants
   - `stride`: The stride of the convolutional kernel, has to be either 1 or 2
   - `reduction`: The reduction factor for the number of hidden feature maps
     in a squeeze and excite layer (see [`squeeze_excite`](#)).
-    Must be >= 1 or `nothing` for no squeeze and excite layer.
+    Must be ≥ 1 or `nothing` for no squeeze and excite layer.
 """
 function invertedresidual(kernel_size, inplanes, hidden_planes, outplanes,
                           activation = relu; stride, reduction = nothing)

src/utilities.jl

@@ -39,7 +39,7 @@ cat_channels(xy...) = cat(xy...; dims = Val(3))
 """
     inputscale(λ; activation = identity)
 
-Scale the input by a scalar λ and applies an activation function to it.
+Scale the input by a scalar `λ` and applies an activation function to it.
 Equivalent to `activation.(λ .* x)`.
 """
 inputscale(λ; activation = identity) = x -> _input_scale(x, λ, activation)

src/vit-based/vit.jl

@@ -79,7 +79,8 @@ Creates a Vision Transformer (ViT) model.
 
 # Arguments
 
-  - `mode`: the model configuration, one of [:tiny, :small, :base, :large, :huge, :giant, :gigantic]
+  - `mode`: the model configuration, one of
+            `[:tiny, :small, :base, :large, :huge, :giant, :gigantic]`
   - `imsize`: image size
   - `inchannels`: number of input channels
   - `patch_size`: size of the patches