System.Collections.Generic.KeyNotFoundException: The given key '1' was not present in the dictionary.

[wasd52030]

Description

Hi, I'm currently trying to run a f# program to train a simple CNN module

and here comes a problem :

Unhandled exception. System.Collections.Generic.KeyNotFoundException: The given key '1' was not present in the dictionary.
at System.Collections.Generic.Dictionary2.get_Item(TKey key) at Tensorflow.Keras.Engine.Node.MapArguments(Dictionary2 tensor_dict)
at Tensorflow.Keras.Engine.Functional.Call(Tensors inputs, Tensor state, Nullable1 training) at Tensorflow.Keras.Engine.Layer.Apply(Tensors inputs, Tensor state, Boolean training) at Tensorflow.Keras.Engine.Model.train_step(DataHandler data_handler, Tensors x, Tensors y) at Tensorflow.Keras.Engine.Model.train_step_function(DataHandler data_handler, OwnedIterator iterator) at Tensorflow.Keras.Engine.Model.FitInternal(DataHandler data_handler, Int32 epochs, Int32 verbose, List1 callbackList, Nullable1 validation_data, Func3 train_step_func)
at Tensorflow.Keras.Engine.Model.fit(NDArray x, NDArray y, Int32 batch_size, Int32 epochs, Int32 verbose, List1 callbacks, Single validation_split, Nullable1 validation_data, Boolean shuffle, Int32 initial_epoch, Int32 max_queue_size, Int32 workers, Boolean use_multiprocessing)
at Program.train(IModel model, NDArray x_train, NDArray y_train) in C:\Users\user\Downloads\fsCNN\Program.fs:line 40
at <StartupCode$fsCNN>.$Program.main@() in C:\Users\user\Downloads\fsCNN\Program.fs:line 47

and here is my code and environment

open type Tensorflow.KerasApi
open Tensorflow
open Tensorflow.NumPy
open type Tensorflow.Keras.Utils.np_utils

let layers = keras.layers

let input =
    keras.Input(Shape(28, 28, 1))
    |> layers.Conv2D(16, Shape(5, 5), activation = "relu").Apply
    |> layers.MaxPooling2D(Shape(2, 2)).Apply
    |> layers.Conv2D(36, Shape(5, 5), activation = "relu").Apply
    |> layers.MaxPooling2D(Shape(2, 2)).Apply
    |> layers.Flatten().Apply
    |> layers.Dense(128, activation = "relu").Apply

let output = input |> layers.Dense(10, activation = "relu").Apply
let model = keras.Model(input, output, "CNN")

let prepareData () =
    let (x_train, y_train, x_test, y_test) =
        keras.datasets.mnist.load_data().Deconstruct()

    let x_train = (x_train.reshape (Shape(-1, 28, 28, 1))).astype (TF_DataType.TF_FLOAT)
    let x_test = (x_test.reshape (Shape(-1, 28, 28, 1))).astype (TF_DataType.TF_FLOAT)

    let x_train = (x_train / 255).numpy ()
    let x_test = (x_test / 255).numpy ()


    let y_train = to_categorical (y_train, 10)
    let y_test = to_categorical (y_test, 10)

    (x_train, y_train, x_test, y_test)


let train (model: Tensorflow.Keras.Engine.IModel) (x_train: NDArray, y_train) =
    model.compile (keras.optimizers.Adam(), keras.losses.CategoricalCrossentropy(), metrics = [| "acc" |])

    model.fit (x_train, y_train, validation_split = 0.2f, batch_size = 200, epochs = 20, verbose = 1)
    |> ignore

    model.save ("fsCNN")


let (x_train, y_train, x_test, y_test) = prepareData ()
train model (x_train, y_train)

Windows 10 22H2(19045.4921)
dotnet sdk 8.0.204
NumSharp 0.30.0
SciSharp.TensorFlow.Redist-Windows-GPU 2.10.3
TensorFlow.Keras 0.10.5
TensorFlow.NET 0.100.5

and similar logic works in C#...

I don't know how to fix this problem.

I'd appreciate it if you could give me any suggestions or comments.

Thank you
Have a nice day!

[wasd52030]

ok, I solve the error

when build model, it should be write below

let cnn =
    let input = keras.Input(Shape(28, 28, 1))

    let modelFlow =
        input
        |> layers.Conv2D(16, Shape(5, 5), activation = "relu").Apply
        |> layers.MaxPooling2D(Shape(2,2)).Apply
        |> layers.Conv2D(36, Shape(5, 5), activation = "relu").Apply
        |> layers.MaxPooling2D(Shape(2,2)).Apply
        |> layers.Flatten().Apply
        |> layers.Dense(128, activation = "relu").Apply

    let output = modelFlow |> layers.Dense(10, activation = "softmax").Apply

    keras.Model(input, output, "CNN")