If I pass validation data to the fit() function, do we need model.evaluate()?-CodePudding

 history = model.fit(
        train_x, train_y,
        validation_data=(validate_x, validate_y),
        batch_size=BATCH_SIZE,
        epochs=EPOCH_SIZE,
        callbacks=[callbacks_vector],
        shuffle=True,
        verbose=2
    )

If I pass validation data to the fit() function as above, do I actually need to call model.evaluate() later?

If I passed validation data to fit, my data is already validated. Right?

CodePudding user response：

Because validation data and test data are quite different for machine learning, check this out : https://stats.stackexchange.com/questions/19048/what-is-the-difference-between-test-set-and-validation-set .

And model.evaluate() is used on test data.