Data Science Stack Exchange is a question and answer site for Data science professionals, Machine Learning specialists, and those interested in learning more about the field. It only takes a minute to sign up.
Sign up to join this community
Could someone explain why the loss returned is different than the loss printed during the evaluation?
They are the same in the Tensorflow documentation https://www.tensorflow.org/guide/keras/train_and_evaluate
Code:
results = model.evaluate(test_data, test_target, verbose=2)
print("test loss, test acc:", results)
Output:
45/1 - 0s - loss: 1.2592 - mae: 0.7602
test loss, test acc: [1.05335361427731, 0.76020277]
The model.evaluate function predicts the output for the given input and then computes the metrics function specified in the model.compile.Based on y_true and y_pred and returns the computed metric value as the output.
You can use model.metrics_names property of your model to find out what each of those values corresponds to.
I had the same problem and I couldn't understand why they were different. The problem is that the ProgbarLogger prints an average of the values (loss, regularization loss, other metrics), which are the values shown in the stdout like this:
45/1 - 0s - loss: 1.2592 - mae: 0.7602
While the values inside the History for the fit, or the scalar or list of scalars for evaluate, are the real values computed on your model. This can be changed with the stateful_metrics parameter of the ProgbarLogger, which will return the real values and not the averaged ones.
In your example could be done like this for loss:
results = model.evaluate(test_data,
test_target,
verbose=2,
callbacks=[
tf.keras.callbacks.ProgbarLogger(
count_mode="steps",
stateful_metrics=["loss"]
)
])
