This is a very simple model, with a single layer that outputs to softmax.
Logistic regression can basically be thought of as a single layer neural network.
The data source is Fisher's classic dataset, retrieved from the UCI Machine Learning Repository.
cd into model-zoo/other/iris, start the Julia REPL and instantiate the environment:
julia> using Pkg; Pkg.activate("."); Pkg.instantiate()
Then train and evaluate the model:
julia> include("iris.jl")
Starting training.
Accuracy: 0.94
Confusion Matrix:
3×3 Array{Int64,2}:
16 0 0
0 16 1
0 2 15
julia>