Machine Learning in Elixir: Model output "iris_features" => nil, not tuple

Pragmatic Bookshelf PragProg Customers
,
2

Thanks for pointing this out! To get the {nil, 4} you can specify a shape in the input:

Axon.input("iris_features", shape: {nil, 4})
|> Axon.dense(3, activation: :softmax)
1 Like