tidy up, add NEWS

Author: mcabbott

NEWS.md

@@ -2,6 +2,8 @@
 
 ## v0.13.7
 * Added [`@autosize` macro](https://github.com/FluxML/Flux.jl/pull/2078)
+* New macro `Flux.@layer` which should be used in place of `@functor`.
+  This also adds `show` methods for pretty printing.
 
 ## v0.13.4
 * Added [`PairwiseFusion` layer](https://github.com/FluxML/Flux.jl/pull/1983)

src/layers/macro.jl

@@ -4,42 +4,43 @@
     @layer :expand Chain
     @layer BatchNorm trainable=(β,γ)
     @layer Struct children=(α,β) trainable=(β,)
-    
-This macro replaces most uses of `@functor` in Flux 0.14. Its basic purpose is the same:
+
+This macro replaces most uses of `@functor`. Its basic purpose is the same:
 When you define a new layer, this tells Flux to explore inside it
 to see the parameters it trains, and also to move them to the GPU, change precision, etc.
 Like `@functor`, this assumes your struct has the default constructor, to enable re-building.
 
-Some "keywords" allow control of the recursion:
+Some "keywords" allow control of the recursion. 
 * If some fields look like parameters but should not be trained,
   then `trainable` lets you specify fields to include, and ignore the rest.
 * You can likewise add restructions to Functors's `children` (although this is seldom a good idea).
 
+The defaults are `fieldnames(T)` for both. They must be subsets of this, and `trainable` must be a subset of `children`.
+
 It also handles overloads of `show` for pretty printing.
 * By default, it adds methods to 3-arg `Base.show` to treat your layer much like `Dense` or `Conv`.
 * If your layer is a container, more like `Chain` or `Parallel`, then `:expand` makes `show` unfold its contents.
 * To disable all `show` overloads, there is an `:ignore` option too.
 
+Note that re-running the macro with different options does not overwrite all methods, you will need to restart.
 (You probably still want to define 2-arg `show(io::IO, x::Layer)`, the macro does not touch this.)
 
-Note that re-running the macro with different options does not overwrite all methods, you will need to restart.
-  
 # Example
 ```jldoctest
 julia> struct Trio; a; b; c end
 
-julia> tri = Trio(Dense([1.1 2.2],), Dense([3.3;;], false), Dropout(0.4))
-Trio(Dense(1 => 1, tanh), Dense(1 => 1; bias=false), Dropout(0.4))
+julia> tri = Trio(Dense([1.1 2.2], [0.0], tanh), Dense([3.3;;], false), Dropout(0.4))
+Trio(Dense(2 => 1, tanh), Dense(1 => 1; bias=false), Dropout(0.4))
 
-julia> Flux.destructure(tri)  # parameters not visible to Flux
+julia> Flux.destructure(tri)  # parameters are not yet visible to Flux
 (Bool[], Restructure(Trio, ..., 0))
 
 julia> Flux.@layer :expand Trio
 
 julia> Flux.destructure(tri)  # now gpu, train!, etc will see inside too
 ([1.1, 2.2, 0.0, 3.3], Restructure(Trio, ..., 4))
 
-julia> tri
+julia> tri  # and layer is printed like Chain
 Trio(
   Dense(2 => 1),                        # 3 parameters
   Dense(1 => 1; bias=false),            # 1 parameters
@@ -58,7 +59,7 @@ macro layer(exs...)
   elseif exs[1] == QuoteNode(:ignore)
     exs[2:end]
   elseif exs[1] isa QuoteNode
-    error("before the type, only accepted options are `:expand` and `:ignore`")
+    error("`@layer` accepts only two options before the layer type, `:expand` and `:ignore` (to control `show`)")
   else
     push!(out.args, _macro_layer_show(esc(exs[1])))
     exs
@@ -76,12 +77,14 @@ macro layer(exs...)
   for j in 1:length(rest)
     j == i && continue
     ex = rest[j]
-    Meta.isexpr(ex, :(=)) || error("expected keyword = fields")
+    Meta.isexpr(ex, :(=)) || error("The macro `@layer` expects here `keyword = (fields...,)`, got $ex")
 
     name = if ex.args[1] == :trainable
       :(Optimisers.trainable)
+    elseif ex.args[1] == :functor
+      error("Can't use `functor=(...)` as a keyword to `@layer`. Use `childen=(...)` to define a method for `functor`.")
     else
-      @warn "trying to define a method for `$(ex.args[1])` in your scope... this is experimental" maxlog=1
+      @warn "Trying to define a method for `$(ex.args[1])` in your scope... this is experimental" maxlog=1
       esc(ex.args[1])
     end
     push!(out.args, _macro_trainable(esc(type), name, ex.args[2]))
@@ -94,7 +97,7 @@ end
 
 function _check_new_macro(x::T) where T
   Functors.isleaf(x) && return
-  @warn "This type should now use Flux.@layer instead of @functor" T maxlog=1 _id=hash(T)
+  Base.depwarn("This type should probably now use `Flux.@layer` instead of `@functor`: $T", Symbol("@functor"))
 end
 _check_new_macro(::Tuple) = nothing  # defined by Functors.jl, not by users
 _check_new_macro(::NamedTuple) = nothing
@@ -159,11 +162,10 @@ function _macro_trainable(type, fun, fields)
   gets = [:(getfield(x, $f)) for f in quoted]
   quote
     $fun(x::$type) = NamedTuple{$symbols}(($(gets...),))
-    # Flux.trainable(x::$type) = NamedTuple{$symbols}(($(gets...),))  # ?? scope is weird
   end
 end
 _macro_trainable(type, fun, field::Union{Symbol,QuoteNode}) = _macro_trainable(type, fun, :(($field,)))  # lets you forget a comma
 
 _noquotenode(s::Symbol) = s
 _noquotenode(q::QuoteNode) = q.value  # lets you write trainable=(:x,:y) instead of (x,y)
-_noquotenode(ex) = error("expected a symbol, got $ex")
+_noquotenode(ex) = error("expected a symbol here, as a field name, but got $ex")

src/layers/normalise.jl

@@ -104,7 +104,6 @@ function Dropout(p; dims=:, rng = default_rng_value())
 end
 
 @layer Dropout trainable=()
-# trainable(a::Dropout) = (;)
 
 function (a::Dropout)(x)
   _isactive(a) || return x
@@ -159,7 +158,6 @@ AlphaDropout(p, active) = AlphaDropout(p, active, default_rng_value())
 AlphaDropout(p; rng = default_rng_value()) = AlphaDropout(p, nothing, rng)
 
 @layer AlphaDropout trainable=()
-# trainable(a::AlphaDropout) = (;)
 
 function (a::AlphaDropout)(x::AbstractArray{T}) where T
   _isactive(a) || return x
@@ -355,7 +353,6 @@ function BatchNorm(chs::Int, λ=identity;
 end
 
 @layer BatchNorm trainable=(β,γ)
-# trainable(bn::BatchNorm) = hasaffine(bn) ? (β = bn.β, γ = bn.γ) : (;)
 
 function (BN::BatchNorm)(x)
   @assert size(x, ndims(x)-1) == BN.chs
@@ -449,7 +446,6 @@ function InstanceNorm(chs::Int, λ=identity;
 end
 
 @layer InstanceNorm trainable=(β,γ)
-# trainable(in::InstanceNorm) = hasaffine(in) ? (β = in.β, γ = in.γ) : (;)
 
 function (l::InstanceNorm)(x)
   @assert ndims(x) > 2
@@ -528,7 +524,6 @@ mutable struct GroupNorm{F,V,N,W}
 end
 
 @layer GroupNorm trainable=(β,γ)
-# trainable(gn::GroupNorm) = hasaffine(gn) ? (β = gn.β, γ = gn.γ) : (;)
 
 function GroupNorm(chs::Int, G::Int, λ=identity;
               initβ=zeros32, initγ=ones32,

src/layers/show.jl

@@ -55,7 +55,6 @@ end
 _show_leaflike(x) = isleaf(x)  # mostly follow Functors, except for:
 _show_leaflike(::Tuple{Vararg{<:Number}}) = true         # e.g. stride of Conv
 _show_leaflike(::Tuple{Vararg{<:AbstractArray}}) = true  # e.g. parameters of LSTMcell
-# _show_leaflike(::Scale) = true                           # appears inside LayerNorm
 _show_leaflike(::AbstractArray{<:Number}) = true         # e.g. transposed arrays
 
 _show_children(x) = trainable(x)

test/layers/macro.jl

@@ -10,8 +10,6 @@ module MacroTest
   @layer Trio trainable=(a,b) test=(c) # should be (c,) but it lets you forget
 
   struct TwoThirds; a; b; c; end
-  @layer :expand TwoThirds children=(a,c) trainable=(a)
-
 end
 
 @testset "@layer macro" begin
@@ -33,6 +31,9 @@ end
   @test MacroTest.test(m3) == (c = [3.0],)
 
   m23 = MacroTest.TwoThirds([1 2], [3 4], [5 6])
+  # Check that we can use the macro with a qualified type name, outside the defining module:
+  Flux.@layer :expand MacroTest.TwoThirds children=(:a,:c) trainable=(:a)  # documented as (a,c) but allow quotes
+
   @test Functors.children(m23) == (a = [1 2], c = [5 6])
   m23re = Functors.functor(m23)[2]((a = [10 20], c = [50 60]))
   @test m23re isa MacroTest.TwoThirds