model.py

import os
import random
import numpy as np

# disable tensorflow warnings
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'

import tensorflow as tf
from tensorflow import keras
from keras.datasets import mnist

# Set the random seed for reproducibility, remember these lines :)
SEED = 42
random.seed(SEED)
np.random.seed(SEED)
tf.random.set_seed(SEED)

# Load the dataset from keras.datasets (so noone would need to download it manually from any sources)
(x_train, y_train), (x_test, y_test) = mnist.load_data()

# Preprocess the dataset
x_train = x_train.astype('float32') / 255.0
x_test = x_test.astype('float32') / 255.0

# Define the model architecture
model = keras.Sequential([
    keras.layers.Flatten(input_shape=(28, 28)),
    keras.layers.Dense(128, activation='relu'),
    keras.layers.Dense(10, activation='softmax')
])

# Compile and train the model
# target in one-hot categorical_crossentropy -> [0,0,1,0,0,0,0,0,0]
# target can be as integer sparse_categorical_crossentropy -> 3
model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])

# 4-epoch is overfitting, 3-rd is okay
model.fit(x_train, y_train, validation_data=(x_test, y_test), epochs=4, shuffle=True, batch_size=32)

# Evaluate the model
print('\n')
_, test_accuracy = model.evaluate(x_test, y_test)
print('Test accuracy:', test_accuracy)

# Save the model
model.save('artifacts/models/mnist_model.h5')