Model Candidates in Auototrain

in Autotrain, after you select a model to finetune, you are also asked to select the number of model candidates. I don’t understand what these candidates are and what the difference is between one candidate and another. How are these different candidates selected? maybe are they the same model I initially select trained with different splits of the training dataset?
Someone can explain, please!

Thank you in advance!

Split is the same for all model candidates within a project. The differences include the models themselves and hyperparameters used to train those models.

thank you for your answer! but I clearly selected a specific model before this step. Where can I see which model is actually selected and where can I see the hyperparameters chosen?

Thank you again very much

You will get all the different variations of the models generated using different hyperparameters but not the actual hyperparameters used to train the model. Important things like max_len will be available to you. :slight_smile: