14 lines
347 B
Julia
14 lines
347 B
Julia
|
using Flux, MNIST
|
||
|
|
|
||
|
|
const data = collect(zip([trainfeatures(i) for i = 1:60_000],
|
||
|
|
[onehot(trainlabel(i), 1:10) for i = 1:60_000]))
|
||
|
|
const train = data[1:50_000]
|
||
|
|
const test = data[50_001:60_000]
|
||
|
|
|
||
|
|
const m = Sequence(
|
||
|
|
Input(784),
|
||
|
|
Dense(30), Sigmoid(),
|
||
|
|
Dense(10), Sigmoid())
|
||
|
|
|
||
|
|
@time Flux.train!(m, train, test, epoch = 30)
|