fix broken doc links (#170)

* fix broken doc links

* typo

* docstrings

Author: CarloLucibello

docs/src/index.md

@@ -2,7 +2,7 @@
 
 ## An optimisation rule
 
-A new optimiser must overload two functions, [`apply!`](@ref) and [`init`](@ref).
+A new optimiser must overload two functions, [`apply!`](@ref Optimisers.apply!) and [`init`](@ref Optimisers.init).
 These act on one array of parameters:
 
 ```julia
@@ -33,8 +33,8 @@ It of course also makes it easier to store the state.
 
 ## Usage with [Flux.jl](https://github.com/FluxML/Flux.jl)
 
-To apply such an optimiser to a whole model, [`setup`](@ref) builds a tree containing any initial
-state for every trainable array. Then at each step, [`update`](@ref) uses this and the gradient
+To apply such an optimiser to a whole model, [`setup`](@ref Optimisers.setup) builds a tree containing any initial
+state for every trainable array. Then at each step, [`update`](@ref Optimisers.update) uses this and the gradient
 to adjust the model:
 
 ```julia
@@ -142,10 +142,10 @@ end;
 
 Optimisers.jl uses [Functors.jl](https://fluxml.ai/Functors.jl) to walk the `struct`s
 making up the model, for which they must be annotated `@functor Type`. 
-By default optimisation will alter all [`isnumeric`](@ref) arrays. 
+By default optimisation will alter all [`isnumeric`](@ref Optimisers.isnumeric) arrays. 
 
 If some arrays of a particular layer should not be treated this way,
-you can define a method for [`trainable`](@ref)
+you can define a method for [`trainable`](@ref Optimisers.trainable)
 
 ```julia
 struct Layer{T}
@@ -239,7 +239,7 @@ from StaticArrays.jl.
 ## Obtaining a flat parameter vector
 
 Instead of a nested tree-like structure, sometimes is is convenient to have all the
-parameters as one simple vector. Optimisers.jl contains a function [`destructure`](@ref)
+parameters as one simple vector. Optimisers.jl contains a function [`destructure`](@ref Optimisers.destructure)
 which creates this vector, and also creates way to re-build the original structure
 with new parameters. Both flattening and re-building may be used within `gradient` calls.
 
@@ -270,7 +270,7 @@ st, flat = Optimisers.update(st, flat, ∇flat)
 
 Here `flat` contains only the 283 trainable parameters, while the non-trainable
 ones are preserved inside `re`, an object of type `Restructure`.
-When defining new layers, these can be specified if necessary by overloading [`trainable`](@ref).
+When defining new layers, these can be specified if necessary by overloading [`trainable`](@ref Optimisers.trainable).
 By default, all numeric arrays visible to [Functors.jl](https://github.com/FluxML/Functors.jl)
 are assumed to contain trainable parameters.
 Tied parameters (arrays appearing in different layers) are included only once in `flat`.

src/Optimisers.jl

@@ -48,7 +48,7 @@ apply!
     Optimisers.init(rule::RuleType, parameters) -> state
 
 Sets up the initial state for a given optimisation rule, and an array of parameters.
-This and [`apply!`](@ref) are the two functions which any new optimisation rule must define.
+This and [`apply!`](@ref Optimisers.apply!) are the two functions which any new optimisation rule must define.
 
 # Examples
 ```jldoctest
@@ -70,8 +70,8 @@ init
     Optimisers.setup(rule, model) -> state_tree
 
 Initialises the given optimiser for every trainable parameter within the model.
-Returns a tree of the relevant states, which must be passed to [`update`](@ref)
-or [`update!`](@ref).
+Returns a tree of the relevant states, which must be passed to [`update`](@ref Optimisers.update)
+or [`update!`](@ref Optimisers.update!).
 
 # Example
 ```jldoctest
@@ -112,9 +112,9 @@ setup
 
 Uses the optimiser and the gradient to change the trainable parameters in the model.
 Returns the improved model, and the optimiser states needed for the next update.
-The initial tree of states comes from [`setup`](@ref).
+The initial tree of states comes from [`setup`](@ref Optimisers.setup).
 
-See also [`update!`](@ref), which will be faster for models of ordinary `Array`s or `CuArray`s.
+See also [`update!`](@ref Optimisers.update!), which will be faster for models of ordinary `Array`s or `CuArray`s.
 
 # Example
 ```jldoctest
@@ -136,9 +136,9 @@ update
 
 Uses the optimiser and the gradient to change the trainable parameters in the model.
 Returns the improved model, and the optimiser states needed for the next update.
-The initial tree of states comes from [`setup`](@ref).
+The initial tree of states comes from [`setup`](@ref Optimisers.setup).
 
-This is used in exactly the same manner as [`update`](@ref), but because it may mutate
+This is used in exactly the same manner as [`update`](@ref Optimisers.update), but because it may mutate
 arrays within the old model (and the old state), it will be faster for models of ordinary
 `Array`s or `CuArray`s. However, you should not rely on the old model being fully updated
 but rather use the returned model.

src/destructure.jl

@@ -5,7 +5,7 @@ const NoT = NoTangent()
 """
     destructure(model) -> vector, reconstructor
 
-Copies all [`trainable`](@ref), [`isnumeric`](@ref) parameters in the model
+Copies all [`trainable`](@ref Optimisers.trainable), [`isnumeric`](@ref Optimisers.isnumeric) parameters in the model
 to a vector, and returns also a function which reverses this transformation.
 Differentiable.
 
@@ -34,7 +34,7 @@ end
 """
     Restructure(Model, ..., length)
 
-This is what [`destructure`](@ref) returns, and `re(p)` will re-build the model with
+This is what [`destructure`](@ref Optimisers.destructure) returns, and `re(p)` will re-build the model with
 new parameters from vector `p`. If the model is callable, then `re(x, p) == re(p)(x)`.
 
 # Example