Workaround Flux#1027

.gitignore

@@ -1 +1 @@
-/Manifest.toml
+Manifest.toml

Project.toml

@@ -1,22 +1,28 @@
 name = "LegolasFlux"
 uuid = "eb5f792d-d1b1-4535-bae3-d5649ec7daa4"
 authors = ["Beacon Biosignals, Inc."]
-version = "0.1.0"
+version = "0.1.1"
 
 [deps]
 Arrow = "69666777-d1a9-59fb-9406-91d4454c9d45"
+Functors = "d9f16b24-f501-4c13-a1f2-28368ffc5196"
 Legolas = "741b9549-f6ed-4911-9fbf-4a1c0c97f0cd"
 Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
 
 [compat]
 Arrow = "1"
+Flux = "0.12"
+Functors = "0.2.1"
 Legolas = "0.1, 0.2"
 Tables = "1"
 julia = "1.5"
 
 [extras]
 Flux = "587475ba-b771-5e3f-ad9e-33799f191a9c"
+Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
+StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
+Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["Test", "Flux"]
+test = ["Test", "Flux", "StableRNGs", "Statistics", "Random"]

README.md

@@ -4,10 +4,11 @@
 [![codecov](https://codecov.io/gh/beacon-biosignals/LegolasFlux.jl/branch/main/graph/badge.svg?token=NHYUL22HCC)](https://codecov.io/gh/beacon-biosignals/LegolasFlux.jl)
 
 LegolasFlux provides some simple functionality to use [Legolas.jl](https://github.com/beacon-biosignals/Legolas.jl/)'s
-extensible Arrow schemas as means to serialize Flux models using Flux's `params` and `loadparams!`.
+extensible Arrow schemas as means to serialize Flux models similarly to using Flux's `params` and `loadparams!`
+(instead, we export similar functions `fetch_weights` and `load_weights!` which handle layers like `BatchNorm` correctly for this purpose).
 
 The aim is to serialize only the numeric weights, *not* the code defining the model. This is a very different approach
-from e.g. BSON.jl, and hopefully much more robust.
+from e.g. BSON.jl, and hopefully much more robust. Note that in this package, we use `weights` to refer to the numeric arrays that are modified over the course of training a model; that includes biases as well as means and variances in e.g. BatchNorms (but not e.g. configuration settings).
 
 With this approach, however, if you change the code such that the weights are no longer valid (e.g. add a layer),
 you will not be able to load back the same model.
@@ -28,16 +29,16 @@ my_model = make_my_model()
 using LegolasFlux
 
 # We can save whatever other columns we'd like to as well as the `weights`.
-model_row = ModelRow(; weights = collect(params(cpu(my_model))), architecture_version = 1, loss = 0.5)
+model_row = ModelRow(; weights = fetch_weights(cpu(my_model)),
+                     architecture_version=1, loss=0.5)
 write_model_row("my_model.model.arrow", model_row)
 
 # Great! Later on, we want to re-load our model weights.
 fresh_model = make_my_model()
 
 model_row = read_model_row("my_model.model.arrow")
-Flux.loadparams!(fresh_model, collect(model_row.weights))
-# Now our params have been loaded back into `fresh_model`.
-# Note we needed to `collect` the weights before we use them.
+load_weights!(fresh_model, model_row.weights)
+# Now our weights have been loaded back into `fresh_model`.
 
 # We can also check out our other columns:
 model_row.loss # 0.5
@@ -47,6 +48,8 @@ model_row.loss # 0.5
 We can make use of the `architecture_version` column to specify a version number for the architectures, in order
 to keep track of for which architectures the weights are valid for.
 
+See [examples/digits.jl](examples/digits.jl) for a larger example.
+
 ## `LegolasFlux.ModelRow`
 
 A `LegolasFlux.ModelRow` is the central object of LegolasFlux. It acts as a Tables.jl-compatible row that can store the weights
@@ -78,4 +81,5 @@ one might name files produced by this row as e.g. `training_run.digits.model.arr
 Note in this example the schema is called `digits.model` instead of just say `digits`, since the package Digits might want to
 create other Legolas schemas as well at some point.
 
-Check out the [Legolas.jl](https://github.com/beacon-biosignals/Legolas.jl/) repo to see more about how its extensible schema system works.
+Check out the [Legolas.jl](https://github.com/beacon-biosignals/Legolas.jl/) repo to see more about how its extensible schema system works,
+and the example at [examples/digits.jl](examples/digits.jl).

examples/Project.toml

@@ -0,0 +1,7 @@
+[deps]
+Flux = "587475ba-b771-5e3f-ad9e-33799f191a9c"
+Legolas = "741b9549-f6ed-4911-9fbf-4a1c0c97f0cd"
+LegolasFlux = "eb5f792d-d1b1-4535-bae3-d5649ec7daa4"
+MLDatasets = "eb30cadb-4394-5ae3-aed4-317e484a6458"
+StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
+Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"

examples/digits.jl

@@ -0,0 +1,140 @@
+# Model modified from
+# https://discourse.julialang.org/t/how-to-drop-the-dropout-layers-in-flux-jl-when-assessing-model-performance/19924
+
+using Flux, Statistics, Random, Test
+# Uncomment to use MNIST data
+# using MLDatasets: MNIST
+using StableRNGs
+using Flux: onehotbatch, onecold, crossentropy, throttle
+using Base.Iterators: repeated, partition
+using Legolas, LegolasFlux
+
+# This should store all the information needed
+# to construct the model.
+Base.@kwdef struct DigitsConfig
+    seed::Int = 5
+    dropout_rate::Float32 = 0f1
+end
+
+# Here's our model object itself, just a `DigitsConfig` and
+# a `chain`. We keep the config around so it's easy to save out
+# later.
+struct DigitsModel
+    chain::Chain
+    config::DigitsConfig
+end
+
+# Ensure Flux can recurse into our model to find params etc
+Flux.@functor DigitsModel (chain,)
+
+# Construct the actual model from a config object. This is the only
+# constructor that should be used, to ensure the model is created just
+# from the config object alone.
+function DigitsModel(config::DigitsConfig=DigitsConfig())
+    dropout_rate = config.dropout_rate
+    Random.seed!(config.seed)
+    chain = Chain(Dropout(dropout_rate),
+                  Conv((3, 3), 1 => 32, relu),
+                  BatchNorm(32, relu),
+                  MaxPool((2, 2)),
+                  Dropout(dropout_rate),
+                  Conv((3, 3), 32 => 16, relu),
+                  Dropout(dropout_rate),
+                  MaxPool((2, 2)),
+                  Dropout(dropout_rate),
+                  Conv((3, 3), 16 => 10, relu),
+                  Dropout(dropout_rate),
+                  x -> reshape(x, :, size(x, 4)),
+                  Dropout(dropout_rate),
+                  Dense(90, 10),
+                  softmax)
+    return DigitsModel(chain, config)
+end
+
+# Our model acts on input just by applying the chain.
+(m::DigitsModel)(x) = m.chain(x)
+
+# Here, we define a schema extension of the `legolas-flux.model` schema.
+# We add our `DigitsConfig` object, as well as the epoch and accuracy.
+const DigitsRow = Legolas.@row("digits.model@1" > "legolas-flux.model@1",
+                               config::DigitsConfig,
+                               epoch::Union{Missing, Int},
+                               accuracy::Union{Missing, Float32})
+
+# Construct a `DigitsRow` from a model by collecting the weights.
+# This can then be saved with e.g. `LegolasFlux.write_model_row`.
+function DigitsRow(model::DigitsModel; epoch=missing, accuracy=missing)
+    return DigitsRow(; weights=fetch_weights(model), model.config, epoch, accuracy)
+end
+
+# Construct a `DigitsModel` from a row satisfying the `DigitsRow` schema,
+# i.e. one with a `weights` and `config::DigitsConfig`.
+# This could be the result of `LegolasFlux.read_model_row`.
+function DigitsModel(row)
+    m = DigitsModel(row.config)
+    load_weights!(m, row.weights)
+    return m
+end
+
+
+# Increase to get more training/test data
+N_train = 1_000
+N_test = 50
+
+##
+# to use MNIST data, uncomment these
+# train_x, train_y = MNIST.traindata(Float32, 1:N_train)
+# test_x,  test_y  = MNIST.testdata(Float32, 1:N_test)
+
+# Random data:
+rng = StableRNG(735)
+train_x = rand(rng, Float32, 28, 28, N_train)
+train_y = rand(rng, 0:9, N_train)
+test_x = rand(rng, Float32, 28, 28, N_test)
+test_y = rand(rng, 0:9, N_test)
+##
+
+# Partition into batches of size 32
+batch_size = 32
+train = [(reshape(train_x[:, :, I], 28, 28, 1, :), onehotbatch(train_y[I], 0:9))
+         for I in partition(1:N_train, batch_size)]
+
+tX = reshape(test_x, 28, 28, 1, :)
+tY = onehotbatch(test_y, 0:9)
+
+function accuracy(m, x, y)
+    testmode!(m)
+    val = mean(onecold(m(x)) .== onecold(y))
+    trainmode!(m)
+    return val
+end
+
+function train_model!(m; N = N_train)
+    loss = (x, y) -> crossentropy(m(x), y)
+    opt = ADAM()
+    evalcb = throttle(() -> @show(accuracy(m, tX, tY)), 5)
+    Flux.@epochs 1 Flux.train!(loss, params(m), Iterators.take(train, N), opt; cb=evalcb)
+    return accuracy(m, tX, tY)
+end
+
+m = DigitsModel()
+
+# increase N to actually train more than a tiny amount
+acc = train_model!(m; N=10)
+
+# Let's serialize out the weights into a `DigitsRow`.
+# We could save this here with `write_model_row`.
+row = DigitsRow(m; epoch=1, accuracy=acc)
+
+testmode!(m)
+input = tX[:, :, :, 1:1]
+output = m(input)
+label = tY[:, 1]
+
+# Let's now reconstruct the model from the `row` and check that we get
+# the same outputs.
+m2 = DigitsModel(row)
+testmode!(m2)
+output2 = m2(input)
+
+@test output ≈ output2

src/LegolasFlux.jl

@@ -1,11 +1,14 @@
 module LegolasFlux
 
 export write_model_row, read_model_row
+export fetch_weights, load_weights!
 
 using Legolas
 using Arrow
 using Arrow.ArrowTypes
 using Tables
+using Functors
+using Base: IdSet
 
 const LEGOLAS_SCHEMA = Legolas.Schema("legolas-flux.model@1")
 
@@ -110,5 +113,6 @@ function read_model_row(io_or_path)
     return only(rows)
 end
 
+include("functors.jl")
 
 end # module

src/functors.jl

@@ -0,0 +1,45 @@
+# Modified version of `fcollect` to use an `IdSet` cache so that
+# distinct arrays whose values happen to be duplicates are each kept.
+# <https://github.com/FluxML/Functors.jl/issues/16>
+function fcollect2(x; output=[], cache=IdSet(), exclude=_ -> false)
+    x in cache && return output
+    if !exclude(x)
+        push!(cache, x)
+        push!(output, x)
+        foreach(y -> fcollect2(y; cache=cache, output=output, exclude=exclude), Functors.children(x))
+    end
+    return output
+end
+
+"""
+    fetch_weights(m) -> Vector{Array}
+
+Returns the weights of a model by using `Functors.children` to recurse
+through the model, keeping any arrays found. The `@functor` macro defines
+`Functors.children` automatically so that should be sufficient to support
+custom types.
+
+Note that this function does not copy the results, so that e.g. mutating `fetch_weights(m)[1]` modifies the model.
+"""
+fetch_weights(m) = filter(x -> x isa Array, fcollect2(m))
+
+"""
+    load_weights!(m, xs)
+
+Load weights `xs` into the model `m`, using [`fetch_weights`](@ref).
+"""
+function load_weights!(m, xs)
+    model_weights = fetch_weights(m)
+    if length(model_weights) != length(xs)
+        throw(ArgumentError("Number of weights given ($(length(xs))) does not match number of weights model expects ($(length(model_weights)))"))
+    end
+    for (i, (p, x)) in enumerate(zip(model_weights, xs))
+        if size(p) != size(x)
+            throw(ArgumentError("For the $(i)th weight expected param size $(size(p)), got $(size(x))"))
+        end
+        copyto!(p, x)
+    end
+    return nothing
+end
+
+load_weights!(m, xs::Weights) = load_weights!(m, collect(xs))