Clean some doctests

Author: Saransh-cpp

docs/src/gpu.md

@@ -60,22 +60,13 @@ Dense(10 => 5)      # 55 parameters
 julia> x = rand(10) |> gpu
 10-element CuArray{Float32, 1, CUDA.Mem.DeviceBuffer}:
  0.066846445
- 0.15663664
- 0.60529673
- 0.13574456
- 0.8381178
- 0.914712
- 0.30007496
- 0.7228498
- 0.11965257
+ ⋮
  0.76706964
 
 julia> m(x)
 5-element CuArray{Float32, 1, CUDA.Mem.DeviceBuffer}:
  -0.99992573
-  0.08930303
-  0.4309393
- -0.5205649
+ ⋮
  -0.547261
 ```
 
@@ -85,27 +76,13 @@ The analogue `cpu` is also available for moving models and data back off of the
 julia> x = rand(10) |> gpu
 10-element CuArray{Float32, 1, CUDA.Mem.DeviceBuffer}:
  0.8019236
- 0.03534455
- 0.48466054
- 0.8991991
- 0.9516907
- 0.8011185
- 0.12432273
- 0.114268765
- 0.07955447
+ ⋮
  0.7766742
 
 julia> x |> cpu
 10-element Vector{Float32}:
  0.8019236
- 0.03534455
- 0.48466054
- 0.8991991
- 0.9516907
- 0.8011185
- 0.12432273
- 0.114268765
- 0.07955447
+ ⋮
  0.7766742
 ```
 

docs/src/models/overview.md

@@ -75,8 +75,7 @@ julia> predict(x_train)
 In order to make better predictions, you'll need to provide a *loss function* to tell Flux how to objectively *evaluate* the quality of a prediction. Loss functions compute the cumulative distance between actual values and predictions. 
 
 ```jldoctest overview
-julia> loss(x, y) = Flux.Losses.mse(predict(x), y)
-loss (generic function with 1 method)
+julia> loss(x, y) = Flux.Losses.mse(predict(x), y);
 
 julia> loss(x_train, y_train)
 122.64734f0

docs/src/models/regularisation.md

@@ -14,18 +14,15 @@ julia> using Flux.Losses: logitcrossentropy
 julia> m = Dense(10 => 5)
 Dense(10 => 5)      # 55 parameters
 
-julia> loss(x, y) = logitcrossentropy(m(x), y)
-loss (generic function with 1 method)
+julia> loss(x, y) = logitcrossentropy(m(x), y);
 ```
 
 We can apply L2 regularisation by taking the squared norm of the parameters , `m.weight` and `m.bias`.
 
 ```jldoctest regularisation
-julia> penalty() = sum(abs2, m.weight) + sum(abs2, m.bias)
-penalty (generic function with 1 method)
+julia> penalty() = sum(abs2, m.weight) + sum(abs2, m.bias);
 
-julia> loss(x, y) = logitcrossentropy(m(x), y) + penalty()
-loss (generic function with 1 method)
+julia> loss(x, y) = logitcrossentropy(m(x), y) + penalty();
 ```
 
 When working with layers, Flux provides the `params` function to grab all
@@ -35,8 +32,7 @@ parameters at once. We can easily penalise everything with `sum`:
 julia> Flux.params(m)
 Params([Float32[0.34704182 -0.48532376 … -0.06914271 -0.38398427; 0.5201164 -0.033709668 … -0.36169025 -0.5552353; … ; 0.46534058 0.17114447 … -0.4809643 0.04993277; -0.47049698 -0.6206029 … -0.3092334 -0.47857067], Float32[0.0, 0.0, 0.0, 0.0, 0.0]])
 
-julia> sqnorm(x) = sum(abs2, x)
-sqnorm (generic function with 1 method)
+julia> sqnorm(x) = sum(abs2, x);
 
 julia> sum(sqnorm, Flux.params(m))
 8.34994f0
@@ -52,11 +48,9 @@ Chain(
   Dense(32 => 10),                      # 330 parameters
 )                   # Total: 6 arrays, 104_938 parameters, 410.289 KiB.
 
-julia> sqnorm(x) = sum(abs2, x)
-sqnorm (generic function with 1 method)
+julia> sqnorm(x) = sum(abs2, x);
 
-julia> loss(x, y) = logitcrossentropy(m(x), y) + sum(sqnorm, Flux.params(m))
-loss (generic function with 1 method)
+julia> loss(x, y) = logitcrossentropy(m(x), y) + sum(sqnorm, Flux.params(m));
 
 julia> loss(rand(28^2), rand(10))
 300.76693683244997

docs/src/saving.md

@@ -107,8 +107,7 @@ Chain(
 julia> evalcb = throttle(30) do
          # Show loss
          @save "model-checkpoint.bson" model
-       end
-(::Flux.var"#throttled#109"{Flux.var"#throttled#105#110"{Bool, Bool, var"#1#2", Int64}}) (generic function with 1 method)
+       end;
 ```
 
 This will update the `"model-checkpoint.bson"` file every thirty seconds.