MNIST Classification with SimpleChains
SimpleChains.jl is an excellent framework for training small neural networks. In this tutorial we will demonstrate how to use the same API as Lux.jl to train a model using SimpleChains.jl. We will use the tutorial from SimpleChains.jl as a reference.
Package Imports
using Lux, MLUtils, Optimisers, Zygote, OneHotArrays, Random, Statistics, Printf, Reactant
using MLDatasets: MNIST
using SimpleChains: SimpleChains
Reactant.set_default_backend("cpu")
Precompiling Reactant...
19010.0 ms ✓ HTTP
90000.6 ms ✓ Reactant
2 dependencies successfully precompiled in 109 seconds. 78 already precompiled.
Precompiling OptimisersReactantExt...
17517.8 ms ✓ Reactant → ReactantStatisticsExt
19762.0 ms ✓ Optimisers → OptimisersReactantExt
2 dependencies successfully precompiled in 20 seconds. 88 already precompiled.
Precompiling LuxCoreReactantExt...
17161.3 ms ✓ LuxCore → LuxCoreReactantExt
1 dependency successfully precompiled in 17 seconds. 85 already precompiled.
Precompiling MLDataDevicesReactantExt...
17368.5 ms ✓ MLDataDevices → MLDataDevicesReactantExt
1 dependency successfully precompiled in 18 seconds. 82 already precompiled.
Precompiling LuxLibReactantExt...
16838.5 ms ✓ Reactant → ReactantSpecialFunctionsExt
17520.3 ms ✓ LuxLib → LuxLibReactantExt
17753.4 ms ✓ Reactant → ReactantKernelAbstractionsExt
16663.6 ms ✓ Reactant → ReactantArrayInterfaceExt
4 dependencies successfully precompiled in 34 seconds. 158 already precompiled.
Precompiling WeightInitializersReactantExt...
16900.9 ms ✓ WeightInitializers → WeightInitializersReactantExt
1 dependency successfully precompiled in 17 seconds. 96 already precompiled.
Precompiling ReactantAbstractFFTsExt...
17084.2 ms ✓ Reactant → ReactantAbstractFFTsExt
1 dependency successfully precompiled in 17 seconds. 82 already precompiled.
Precompiling ReactantOneHotArraysExt...
17660.2 ms ✓ Reactant → ReactantOneHotArraysExt
1 dependency successfully precompiled in 18 seconds. 104 already precompiled.
Precompiling ReactantNNlibExt...
19707.4 ms ✓ Reactant → ReactantNNlibExt
1 dependency successfully precompiled in 20 seconds. 103 already precompiled.
Precompiling LuxReactantExt...
11948.4 ms ✓ Lux → LuxReactantExt
1 dependency successfully precompiled in 12 seconds. 180 already precompiled.
Precompiling MLDatasets...
394.6 ms ✓ Glob
450.8 ms ✓ WorkerUtilities
505.4 ms ✓ BufferedStreams
338.1 ms ✓ PackageExtensionCompat
359.7 ms ✓ InternedStrings
1764.1 ms ✓ FileIO → HTTPExt
3016.6 ms ✓ DataDeps
451.9 ms ✓ StridedViews
371.0 ms ✓ StridedViews → StridedViewsPtrArraysExt
2283.3 ms ✓ MAT
2282.5 ms ✓ Pickle
18524.8 ms ✓ CSV
9274.5 ms ✓ MLDatasets
13 dependencies successfully precompiled in 29 seconds. 191 already precompiled.
Precompiling ReactantOffsetArraysExt...
16894.6 ms ✓ Reactant → ReactantOffsetArraysExt
1 dependency successfully precompiled in 17 seconds. 82 already precompiled.
2025-07-09 04:05:51.779296: I external/xla/xla/service/service.cc:153] XLA service 0x24933560 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2025-07-09 04:05:51.779412: I external/xla/xla/service/service.cc:161] StreamExecutor device (0): NVIDIA A100-PCIE-40GB MIG 1g.5gb, Compute Capability 8.0
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
I0000 00:00:1752033951.780209 1104355 se_gpu_pjrt_client.cc:1370] Using BFC allocator.
I0000 00:00:1752033951.780289 1104355 gpu_helpers.cc:136] XLA backend allocating 3825205248 bytes on device 0 for BFCAllocator.
I0000 00:00:1752033951.780335 1104355 gpu_helpers.cc:177] XLA backend will use up to 1275068416 bytes on device 0 for CollectiveBFCAllocator.
I0000 00:00:1752033951.793766 1104355 cuda_dnn.cc:471] Loaded cuDNN version 90800
Loading MNIST
function loadmnist(batchsize, train_split)
# Load MNIST
N = parse(Bool, get(ENV, "CI", "false")) ? 1500 : nothing
dataset = MNIST(; split=:train)
if N !== nothing
imgs = dataset.features[:, :, 1:N]
labels_raw = dataset.targets[1:N]
else
imgs = dataset.features
labels_raw = dataset.targets
end
# Process images into (H, W, C, BS) batches
x_data = Float32.(reshape(imgs, size(imgs, 1), size(imgs, 2), 1, size(imgs, 3)))
y_data = onehotbatch(labels_raw, 0:9)
(x_train, y_train), (x_test, y_test) = splitobs((x_data, y_data); at=train_split)
return (
# Use DataLoader to automatically minibatch and shuffle the data
DataLoader(collect.((x_train, y_train)); batchsize, shuffle=true, partial=false),
# Don't shuffle the test data
DataLoader(collect.((x_test, y_test)); batchsize, shuffle=false, partial=false),
)
end
loadmnist (generic function with 1 method)
Define the Model
lux_model = Chain(
Conv((5, 5), 1 => 6, relu),
MaxPool((2, 2)),
Conv((5, 5), 6 => 16, relu),
MaxPool((2, 2)),
FlattenLayer(3),
Chain(Dense(256 => 128, relu), Dense(128 => 84, relu), Dense(84 => 10)),
)
Chain(
layer_1 = Conv((5, 5), 1 => 6, relu), # 156 parameters
layer_2 = MaxPool((2, 2)),
layer_3 = Conv((5, 5), 6 => 16, relu), # 2_416 parameters
layer_4 = MaxPool((2, 2)),
layer_5 = Lux.FlattenLayer{Static.StaticInt{3}}(static(3)),
layer_6 = Chain(
layer_1 = Dense(256 => 128, relu), # 32_896 parameters
layer_2 = Dense(128 => 84, relu), # 10_836 parameters
layer_3 = Dense(84 => 10), # 850 parameters
),
) # Total: 47_154 parameters,
# plus 0 states.
We now need to convert the lux_model to SimpleChains.jl. We need to do this by defining the ToSimpleChainsAdaptor
and providing the input dimensions.
adaptor = ToSimpleChainsAdaptor((28, 28, 1))
simple_chains_model = adaptor(lux_model)
SimpleChainsLayer(
Chain(
layer_1 = Conv((5, 5), 1 => 6, relu), # 156 parameters
layer_2 = MaxPool((2, 2)),
layer_3 = Conv((5, 5), 6 => 16, relu), # 2_416 parameters
layer_4 = MaxPool((2, 2)),
layer_5 = Lux.FlattenLayer{Static.StaticInt{3}}(static(3)),
layer_6 = Chain(
layer_1 = Dense(256 => 128, relu), # 32_896 parameters
layer_2 = Dense(128 => 84, relu), # 10_836 parameters
layer_3 = Dense(84 => 10), # 850 parameters
),
),
) # Total: 47_154 parameters,
# plus 0 states.
Helper Functions
const lossfn = CrossEntropyLoss(; logits=Val(true))
function accuracy(model, ps, st, dataloader)
total_correct, total = 0, 0
st = Lux.testmode(st)
for (x, y) in dataloader
target_class = onecold(Array(y))
predicted_class = onecold(Array(first(model(x, ps, st))))
total_correct += sum(target_class .== predicted_class)
total += length(target_class)
end
return total_correct / total
end
accuracy (generic function with 1 method)
Define the Training Loop
function train(model, dev=cpu_device(); rng=Random.default_rng(), kwargs...)
train_dataloader, test_dataloader = dev(loadmnist(128, 0.9))
ps, st = dev(Lux.setup(rng, model))
vjp = dev isa ReactantDevice ? AutoEnzyme() : AutoZygote()
train_state = Training.TrainState(model, ps, st, Adam(3.0f-4))
if dev isa ReactantDevice
x_ra = first(test_dataloader)[1]
model_compiled = Reactant.with_config(;
dot_general_precision=PrecisionConfig.HIGH,
convolution_precision=PrecisionConfig.HIGH,
) do
@compile model(x_ra, ps, Lux.testmode(st))
end
else
model_compiled = model
end
### Lets train the model
nepochs = 10
tr_acc, te_acc = 0.0, 0.0
for epoch in 1:nepochs
stime = time()
for (x, y) in train_dataloader
_, _, _, train_state = Training.single_train_step!(
vjp, lossfn, (x, y), train_state
)
end
ttime = time() - stime
tr_acc =
accuracy(
model_compiled, train_state.parameters, train_state.states, train_dataloader
) * 100
te_acc =
accuracy(
model_compiled, train_state.parameters, train_state.states, test_dataloader
) * 100
@printf "[%2d/%2d] \t Time %.2fs \t Training Accuracy: %.2f%% \t Test Accuracy: \
%.2f%%\n" epoch nepochs ttime tr_acc te_acc
end
return tr_acc, te_acc
end
train (generic function with 2 methods)
Finally Training the Model
First we will train the Lux model
tr_acc, te_acc = train(lux_model, reactant_device())
[ 1/10] Time 389.09s Training Accuracy: 11.56% Test Accuracy: 14.84%
[ 2/10] Time 0.11s Training Accuracy: 30.23% Test Accuracy: 32.81%
[ 3/10] Time 0.10s Training Accuracy: 40.47% Test Accuracy: 45.31%
[ 4/10] Time 0.12s Training Accuracy: 52.11% Test Accuracy: 51.56%
[ 5/10] Time 0.10s Training Accuracy: 62.73% Test Accuracy: 65.62%
[ 6/10] Time 0.10s Training Accuracy: 69.14% Test Accuracy: 68.75%
[ 7/10] Time 0.10s Training Accuracy: 74.14% Test Accuracy: 71.09%
[ 8/10] Time 0.10s Training Accuracy: 77.27% Test Accuracy: 75.00%
[ 9/10] Time 0.10s Training Accuracy: 81.17% Test Accuracy: 77.34%
[10/10] Time 0.09s Training Accuracy: 83.44% Test Accuracy: 78.91%
Now we will train the SimpleChains model
tr_acc, te_acc = train(simple_chains_model)
[ 1/10] Time 896.41s Training Accuracy: 31.33% Test Accuracy: 31.25%
[ 2/10] Time 12.14s Training Accuracy: 52.50% Test Accuracy: 49.22%
[ 3/10] Time 12.13s Training Accuracy: 59.53% Test Accuracy: 51.56%
[ 4/10] Time 12.17s Training Accuracy: 70.31% Test Accuracy: 62.50%
[ 5/10] Time 12.21s Training Accuracy: 74.30% Test Accuracy: 71.09%
[ 6/10] Time 12.23s Training Accuracy: 80.86% Test Accuracy: 75.00%
[ 7/10] Time 12.24s Training Accuracy: 83.75% Test Accuracy: 78.12%
[ 8/10] Time 12.24s Training Accuracy: 85.08% Test Accuracy: 81.25%
[ 9/10] Time 12.25s Training Accuracy: 85.55% Test Accuracy: 81.25%
[10/10] Time 12.28s Training Accuracy: 87.34% Test Accuracy: 84.38%
On my local machine we see a 3-4x speedup when using SimpleChains.jl. The conditions of the server this documentation is being built on is not ideal for CPU benchmarking hence, the speedup may not be as significant and even there might be regressions.
Appendix
using InteractiveUtils
InteractiveUtils.versioninfo()
if @isdefined(MLDataDevices)
if @isdefined(CUDA) && MLDataDevices.functional(CUDADevice)
println()
CUDA.versioninfo()
end
if @isdefined(AMDGPU) && MLDataDevices.functional(AMDGPUDevice)
println()
AMDGPU.versioninfo()
end
end
Julia Version 1.11.5
Commit 760b2e5b739 (2025-04-14 06:53 UTC)
Build Info:
Official https://julialang.org/ release
Platform Info:
OS: Linux (x86_64-linux-gnu)
CPU: 48 × AMD EPYC 7402 24-Core Processor
WORD_SIZE: 64
LLVM: libLLVM-16.0.6 (ORCJIT, znver2)
Threads: 48 default, 0 interactive, 24 GC (on 2 virtual cores)
Environment:
JULIA_CPU_THREADS = 2
LD_LIBRARY_PATH = /usr/local/nvidia/lib:/usr/local/nvidia/lib64
JULIA_PKG_SERVER =
JULIA_NUM_THREADS = 48
JULIA_CUDA_HARD_MEMORY_LIMIT = 100%
JULIA_PKG_PRECOMPILE_AUTO = 0
JULIA_DEBUG = Literate
JULIA_DEPOT_PATH = /root/.cache/julia-buildkite-plugin/depots/01872db4-8c79-43af-ab7d-12abac4f24f6
This page was generated using Literate.jl.