打印或保存 tf.keras 模型的输出，与输入配对

Question

def locations_model(...):
    input_shape = image_shape + (3,)
    base_model = tf.keras.applications.MobileNetV2(...)                                                                base_model.trainable = False 
    inputs = tf.keras.Input(...)  
... ...     
    outputs = tfl.Dense(5, activation = "softmax")(x)
    model = tf.keras.Model(inputs, outputs)
 
    return model

The code above is just to show inputs and outputs in a tf.keras model that classifies input images into 5 categories. How can I save the output category ("y_pred") for every input image?

The simple statement ypreds = model(inputs) or ypreds = model.predict(inputs) produces a set of 5-element arrays that add to 1, ie, probabilities.
The question therefore is how to output the predicted categories, which in this case are integers: 0-4, instead of the probabilities. Update: this was answer by Apostolova for the question "Get class labels from Keras functional model" by Lodzz, as test_probas = model.predict(test_data) test_classes = probas.argmax(axis = -1)

Answer 1

Thanks for the confirmation @EduardoriosChicago . I am mentioning your answer here for the benefit of the community.

The code is

probas = model(x_in); x_classes = probas.argmax( axis = - 1)