Merge branch 'master' into chainrules

.github/workflows/Downstream.yml

@@ -26,7 +26,7 @@ jobs:
           - {user: Chemellia, repo: AtomicGraphNets.jl, group: All}
           - {user: SciML, repo: DiffEqFlux.jl, group: Layers}
           - {user: SciML, repo: NeuralPDE.jl, group: NNPDE}
-
+          - {user: SciML, repo: OperatorLearning.jl, group: All}
     if: contains(github.event.pull_request.labels.*.name, 'run downstream test')
     steps:
       - uses: actions/checkout@v2

.github/workflows/ci.yml

@@ -22,10 +22,15 @@ jobs:
           - 'nightly'
         os:
           - ubuntu-latest
-          - macOS-latest
-          - windows-latest
         arch:
           - x64
+        include:
+          - os: windows-latest
+            version: '1'
+            arch: x64
+          - os: macOS-latest
+            version: '1'
+            arch: x64
     steps:
       - uses: actions/checkout@v2
       - uses: julia-actions/setup-julia@v1

NEWS.md

@@ -1,5 +1,13 @@
 # Flux Release Notes
 
+## v0.13
+* After a deprecations cycle, the datasets in `Flux.Data` have
+been removed in favour of MLDatasets.jl.
+* `params` is not exported anymore since it is a common name and is also exported by Distributions.jl
+* `flatten` is not exported anymore due to clash with Iterators.flatten.
+* Remove Juno.jl progress bar support as it is now obsolete.
+* `Dropout` gained improved compatibility with Int and Complex arrays and is now twice-differentiable.
+
 ## v0.12.10
 * `Dropout`/`AlphaDropout` now supports [user-specified RNGs](https://github.com/FluxML/Flux.jl/pull/1838)
 

Project.toml

@@ -1,50 +1,38 @@
 name = "Flux"
 uuid = "587475ba-b771-5e3f-ad9e-33799f191a9c"
-version = "0.12.9"
+version = "0.13.0-DEV"
 
 [deps]
-AbstractTrees = "1520ce14-60c1-5f80-bbc7-55ef81b5835c"
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
 ArrayInterface = "4fba245c-0d91-5ea0-9b3e-6abc04ee57a9"
 CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
-CodecZlib = "944b1d66-785c-5afd-91f1-9de20f533193"
-Colors = "5ae59095-9a9b-59fe-a467-6f913c188581"
-DelimitedFiles = "8bb1440f-4735-579b-a4ab-409b98df4dab"
 Functors = "d9f16b24-f501-4c13-a1f2-28368ffc5196"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 MacroTools = "1914dd2f-81c6-5fcd-8719-6d5c9610ff09"
 NNlib = "872c559c-99b0-510c-b3b7-b6c96a88d5cd"
 NNlibCUDA = "a00861dc-f156-4864-bf3c-e6376f28a68d"
-Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
-Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
 ProgressLogging = "33c8b6b6-d38a-422a-b730-caa89a2f386c"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
-SHA = "ea8e919c-243c-51af-8825-aaa63cd721ce"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
-ZipFile = "a5390f91-8eb1-5f08-bee0-b1d1ffed6cea"
 Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [compat]
-AbstractTrees = "0.3"
 Adapt = "3.0"
 ArrayInterface = "3.1, 4"
 CUDA = "3"
 ChainRulesCore = "1.12"
-CodecZlib = "0.7"
-Colors = "0.12"
 Functors = "0.2.1"
 MacroTools = "0.5"
-NNlib = "0.8"
+NNlib = "0.8.2"
 NNlibCUDA = "0.2"
 ProgressLogging = "0.1"
 Reexport = "0.2, 1.0"
 StatsBase = "0.33"
-ZipFile = "0.9"
 Zygote = "0.6.34"
 julia = "1.6"
 

docs/src/models/advanced.md

@@ -97,8 +97,8 @@ We can freeze a specific parameter of a specific layer which already entered a `
 by simply deleting it from `ps`:
 
 ```julia
-ps = params(m)
-delete!(ps, m[2].bias)
+ps = Flux.params(m)
+delete!(ps, m[2].bias) 
 ```
 
 ## Custom multiple input or output layer

docs/src/models/basics.md

@@ -39,7 +39,7 @@ julia> x = [2, 1];
 
 julia> y = [2, 0];
 
-julia> gs = gradient(params(x, y)) do
+julia> gs = gradient(Flux.params(x, y)) do
          f(x, y)
        end
 Grads(...)
@@ -83,7 +83,7 @@ To improve the prediction we can take the gradients of the loss with respect to
 ```julia
 using Flux
 
-gs = gradient(() -> loss(x, y), params(W, b))
+gs = gradient(() -> loss(x, y), Flux.params(W, b))
 ```
 
 Now that we have gradients, we can pull them out and update `W` to train the model.

docs/src/models/recurrence.md

@@ -160,7 +160,7 @@ data = zip(X,Y)
 Flux.reset!(m)
 [m(x) for x in seq_init]
 
-ps = params(m)
+ps = Flux.params(m)
 opt= ADAM(1e-3)
 Flux.train!(loss, ps, data, opt)
 ```

docs/src/saving.md

@@ -62,7 +62,7 @@ julia> using Flux
 julia> model = Chain(Dense(10,5,relu),Dense(5,2),softmax)
 Chain(Dense(10, 5, NNlib.relu), Dense(5, 2), NNlib.softmax)
 
-julia> weights = params(model);
+julia> weights = Flux.params(model);
 
 julia> using BSON: @save
 

docs/src/training/optimisers.md

@@ -14,7 +14,7 @@ loss(x, y) = sum((predict(x) .- y).^2)
 x, y = rand(5), rand(2) # Dummy data
 l = loss(x, y) # ~ 3
 
-θ = params(W, b)
+θ = Flux.params(W, b)
 grads = gradient(() -> loss(x, y), θ)
 ```
 

docs/src/training/training.md

@@ -64,7 +64,7 @@ At first glance it may seem strange that the model that we want to train is not
 
 ## Model parameters
 
-The model to be trained must have a set of tracked parameters that are used to calculate the gradients of the objective function. In the [basics](../models/basics.md) section it is explained how to create models with such parameters. The second argument of the function `Flux.train!` must be an object containing those parameters, which can be obtained from a model `m` as `params(m)`.
+The model to be trained must have a set of tracked parameters that are used to calculate the gradients of the objective function. In the [basics](../models/basics.md) section it is explained how to create models with such parameters. The second argument of the function `Flux.train!` must be an object containing those parameters, which can be obtained from a model `m` as `Flux.params(m)`.
 
 Such an object contains a reference to the model's parameters, not a copy, such that after their training, the model behaves according to their updated values.
 

src/Flux.jl

@@ -11,13 +11,13 @@ using Zygote: Params, @adjoint, gradient, pullback, @nograd
 export gradient
 using ChainRulesCore
 
-export Chain, Dense, Maxout, SkipConnection, Parallel, flatten,
+export Chain, Dense, Maxout, SkipConnection, Parallel,
        RNN, LSTM, GRU, GRUv3,
        SamePad, Conv, CrossCor, ConvTranspose, DepthwiseConv,
        AdaptiveMaxPool, AdaptiveMeanPool, GlobalMaxPool, GlobalMeanPool, MaxPool, MeanPool,
        Dropout, AlphaDropout, LayerNorm, BatchNorm, InstanceNorm, GroupNorm,
        Upsample, PixelShuffle,
-       params, fmap, cpu, gpu, f32, f64,
+       fmap, cpu, gpu, f32, f64,
        testmode!, trainmode!
 
 include("optimise/Optimise.jl")

src/data/Data.jl

@@ -6,62 +6,4 @@ using Base: @propagate_inbounds
 include("dataloader.jl")
 export DataLoader
 
-## TODO for v0.13: remove everything below ##############
-## Also remove the following deps:
-## AbstractTrees, ZipFiles, CodecZLib
-
-import ..Flux
-import SHA
-
-deprecation_message() = @warn("Flux's datasets are deprecated, please use the package MLDatasets.jl")
-
-function deps(path...)
-  if isnothing(@__DIR__) # sysimages
-    joinpath("deps", path...)
-  else
-    joinpath(@__DIR__, "..", "..", "deps", path...)
-  end
-end
-
-function download_and_verify(url, path, hash)
-    tmppath = tempname()
-    download(url, tmppath)
-    hash_download = open(tmppath) do f
-        bytes2hex(SHA.sha256(f))
-    end
-    if hash_download !== hash
-        msg  = "Hash Mismatch!\n"
-        msg *= "  Expected sha256:   $hash\n"
-        msg *= "  Calculated sha256: $hash_download"
-        error(msg)
-    end
-    mv(tmppath, path; force=true)
-end
-
-function __init__()
-  mkpath(deps())
-end
-
-include("mnist.jl")
-export MNIST
-
-include("fashion-mnist.jl")
-export FashionMNIST
-
-include("cmudict.jl")
-export CMUDict
-using .CMUDict; export cmudict
-
-include("tree.jl")
-include("sentiment.jl")
-export Sentiment
-
-include("iris.jl")
-export Iris
-
-include("housing.jl")
-export Housing
-
-#########################################
-
 end#module