improvements to stack

Project.toml

@@ -1,10 +1,11 @@
 name = "MLUtils"
 uuid = "f1d291b0-491e-4a28-83b9-f70985020b54"
 authors = ["Carlo Lucibello <carlo.lucibello@gmail.com> and contributors"]
-version = "0.3.1"
+version = "0.4.0"
 
 [deps]
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
+Compat = "34da2185-b29b-5c13-b0c7-acf172513d20"
 DataAPI = "9a962f9c-6df0-11e9-0e5d-c546b8b5ee8a"
 DelimitedFiles = "8bb1440f-4735-579b-a4ab-409b98df4dab"
 FLoops = "cc61a311-1640-44b5-9fba-1b764f453329"
@@ -20,6 +21,7 @@ Transducers = "28d57a85-8fef-5791-bfe6-a80928e7c999"
 
 [compat]
 ChainRulesCore = "1.0"
+Compat = "4.2"
 DataAPI = "1.0"
 DelimitedFiles = "1.0"
 FLoops = "0.2"

src/MLUtils.jl

@@ -21,7 +21,8 @@ import NNlib
 
 @traitdef IsTable{X}
 @traitimpl IsTable{X} <- Tables.istable(X)
-
+
+using Compat: stack
 
 include("observation.jl")
 export numobs,
@@ -75,7 +76,7 @@ export batch,
  rand_like,
  randn_like,
  rpad_constant,
- stack,
+ stack, # in Base since julia v1.9
  unbatch,
  unsqueeze,
  unstack,

src/deprecations.jl

@@ -1,5 +1,4 @@
 # Deprecated in v0.2
-@deprecate stack(x, dims) stack(x; dims=dims)
 @deprecate unstack(x, dims) unstack(x; dims=dims)
 @deprecate unsqueeze(x::AbstractArray, dims::Int) unsqueeze(x; dims=dims)
 @deprecate unsqueeze(dims::Int) unsqueeze(dims=dims)

src/utils.jl

@@ -5,6 +5,7 @@
 
 Return `x` reshaped into an array one dimensionality higher than `x`,
 where `dims` indicates in which dimension `x` is extended.
+`dims` can be an integer between 1 and `ndims(x)+1`.
 
 See also [`flatten`](@ref), [`stack`](@ref).
 
@@ -33,8 +34,9 @@ julia> unsqueeze(xs, dims=1)
  [1, 2] [3, 4] [5, 6]
 ```
 """
-function unsqueeze(x::AbstractArray; dims::Int)
- sz = ntuple(i -> i < dims ? size(x, i) : i == dims ? 1 : size(x, i - 1), ndims(x) + 1)
+function unsqueeze(x::AbstractArray{T,N}; dims::Int) where {T, N}
+ @assert 1 <= dims <= N + 1
+ sz = ntuple(i -> i < dims ? size(x, i) : i == dims ? 1 : size(x, i - 1), N + 1)
  return reshape(x, sz)
 end
 
@@ -55,51 +57,6 @@ _unsqueeze(x, dims) = unsqueeze(x; dims)
 
 Base.show_function(io::IO, u::Base.Fix2{typeof(_unsqueeze)}, ::Bool) = print(io, "unsqueeze(dims=", u.x, ")")
 
-"""
- stack(xs; dims)
-
-Concatenate the given array of arrays `xs` into a single array along the
-given dimension `dims`.
-
-See also [`stack`](@ref) and [`batch`](@ref).
-
-# Examples
-
-```jldoctest
-julia> xs = [[1, 2], [3, 4], [5, 6]]
-3-element Vector{Vector{Int64}}:
- [1, 2]
- [3, 4]
- [5, 6]
-
-julia> stack(xs, dims=1)
-3×2 Matrix{Int64}:
- 1 2
- 3 4
- 5 6
-
-julia> stack(xs, dims=2)
-2×3 Matrix{Int64}:
- 1 3 5
- 2 4 6
-
-julia> stack(xs, dims=3)
-2×1×3 Array{Int64, 3}:
-[:, :, 1] =
- 1
- 2
-
-[:, :, 2] =
- 3
- 4
-
-[:, :, 3] =
- 5
- 6
-```
-"""
-stack(xs; dims::Int) = cat(unsqueeze.(xs; dims)...; dims)
-
 """
  unstack(xs; dims)
 
@@ -329,17 +286,7 @@ end
 
 batchindex(xs, i) = (reverse(Base.tail(reverse(axes(xs))))..., i)
 
-function batch(xs::AbstractArray{<:AbstractArray})
- # Don't use stack(xs, dims=N+1), it is much slower.
- # Here we do reduce(vcat, xs) along with some reshapes.
- szxs = size(xs)
- @assert length(xs) > 0 "Minimum batch size is 1."
- szx = size(xs[1])
- @assert all(x -> size(x) == szx, xs) "All arrays must be of the same size."
- vxs = vec(vec.(xs))
- y = reduce(vcat, vxs)
- return reshape(y, szx..., szxs...)
-end
+batch(xs::AbstractArray{<:AbstractArray}) = stack(xs)
 
 function batch(xs::Vector{<:Tuple})
  @assert length(xs) > 0 "Input should be non-empty"

test/test_utils.jl

@@ -1,4 +1,17 @@
 
+"""
+Test gradients through zygote. 
+
+# Arguments
+
+- `f`: function to test
+- `xs`: inputs to `f`
+
+# Keyword Arguments
+Keyword arguments are passed to `rrule`.
+
+- `fkwargs`: keyword arguments to `f`
+"""
 function test_zygote(f, xs...; kws...)
  config = ZygoteRuleConfig()
  test_rrule(config, f, xs...; kws..., rrule_f = rrule_via_ad)

test/utils.jl

@@ -6,19 +6,30 @@
  @test @inferred(unsqueeze(x; dims=4)) == reshape(x, 2, 3, 2, 1)
 
  @test unsqueeze(dims=2)(x) == unsqueeze(x, dims=2)
+
+ @test_throws AssertionError unsqueeze(rand(2,2), dims=4)
 end
 
 @testset "stack and unstack" begin
  x = randn(3,3)
  stacked = stack([x, x], dims=2)
  @test size(stacked) == (3,2,3)
- @test_broken @inferred(stack([x, x], dims=2)) == stacked
+ @test @inferred(stack([x, x], dims=2)) == stacked
 
  stacked_array=[ 8 9 3 5; 9 6 6 9; 9 1 7 2; 7 4 10 6 ]
  unstacked_array=[[8, 9, 9, 7], [9, 6, 1, 4], [3, 6, 7, 10], [5, 9, 2, 6]]
  @test unstack(stacked_array, dims=2) == unstacked_array
  @test stack(unstacked_array, dims=2) == stacked_array
  @test stack(unstack(stacked_array, dims=1), dims=1) == stacked_array
+
+ for d in (1,2,3)
+ test_zygote(stack, [x,2x], fkwargs=(; dims=d), check_inferred=false)
+ end
+
+ # Issue #121
+ a = [[1] for i in 1:10000]
+ @test size(stack(a, dims=1)) == (10000, 1)
+ @test size(stack(a, dims=2)) == (1, 10000)
 end
 
 @testset "batch and unbatch" begin