Xlwings UDF don't creat tensorflow model

General Discussion
,
#1

Dear all,

I want to create an User Defined Function (UDF) in Excel (using xlwings) to handle TensorFlow models in Python. I installed Python, tensorflow and xlwings and created the following Python script:

# Importe as bibliotecas necessárias
import numpy as np
import xlwings as xw
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

#Função para criar e treinar uma rede neural
@xw.func
def treinar_rede():

    # Defina os dados de treinamento (exemplo simplificado)
    x_train = np.array([1, 2, 3, 4])
    y_train = np.array([3, 5, 7, 9])

    # Expanda as dimensões dos dados de entrada
    x_train = np.expand_dims(x_train, axis=-1)
    y_train = np.expand_dims(y_train, axis=-1)

    # Crie o modelo da rede neural
    model = keras.Sequential([
        layers.Dense(10, activation='relu'),  # Camada oculta com 10 neurônios e função de ativação ReLU
        layers.Dense(1),  # Camada de saída com 1 neurônio
    ])

    # Compile o modelo
    model.compile(optimizer='adam', loss='mean_squared_error')

    # Treine o modelo
    model.fit(x_train, y_train, epochs=10, verbose=0)  

    # Salvando o modelo
    model.save('meu_modelo.keras')
    
    return "modelo treinado!"

Then, I created the following VBA code in Excel with the same name of Python script:

Sub RunNeuralNetwork()

Range("A1").Value = ""

Range("A1").Value = treinar_rede()

End Sub

So, when I execute the Macro, I receive the text “modelo treinado!” in cell A1 as expected but, when I look in the directory of spreadsheet (which is the same of the Python script), the model “meu_modelo.keras” doesn’t appear. What is wrong?