model.py

import tensorflow as tf

BATCH_SIZE = 10

def create_model_for_training():
    model = tf.keras.models.Sequential([
        # The input layer expects an array of shape (batch_size, 10).
        tf.keras.layers.Dense(units=64, input_shape=(None, 10), name='input_layer'),
        tf.keras.layers.Activation('relu'),

        # The first hidden layer has 64 units.
        tf.keras.layers.Dense(units=64, name='hidden_layer1'),
        tf.keras.layers.Activation('relu'),

        # The second hidden layer has 64 units.
        tf.keras.layers.Dense(units=64, name='hidden_layer2'),
        tf.keras.layers.Activation('relu'),

        # The output layer has 35 units.
        tf.keras.layers.Dense(units=36, name='output_layer'),
    ])
    model.compile(optimizer='adam', loss='mse', metrics=['accuracy'])
    return model

def load_model(model_fp):
    model = tf.keras.models.load_model(model_fp)
    return model

def get_batch_size():
    return BATCH_SIZE