Fasttext automated prameter tuning training set

Question

I am using te automated tuning code for classification models in the fasttext library, and I cannot find whether the final model it gives you is trained only on the training set or on both the training and validation set.

For example, when, if you run this command:

>>./fasttext supervised -input cooking.train -output model_cooking -autotune-validation cooking.valid

Does it give you a model trained on cooking.train only or on both cooking.train and cooking.valid .

Thanks in advance

Answer 1

The model is only trained on the training set. The validation set is only for you to measure the models ability to generalize, ie to check whether the model has actually learned the concept you want it to learn rather than just memorizing the training data.