New to Model Fine-Tuning: Post Training Epoch Selection

Have you ever trained a custom model?

Thus, you had to determine how many epochs (the number of times to cycle through the entire training set) were optimal for the training. Your options were either repeatedly train until you got the right number, or give up altogether.

Train first & select the optimal epoch later

Trying out different epochs and setting the right one as default, post-training, is now available via the Studio web app. Also available via API.

To set a default epoch, navigate to the "Custom models" page, click "view" on the custom model of your choice, then choose the epoch you like best and set it as default (in the item's menu).

Pro tip: Set the training with plenty of epochs, review each epoch's validation loss once the training is complete, and set the most promising one as default.