Service training in Caila

Caila allows for the creation of services based on user data. Examples of fittable services:

• Intent classifier * Aimyvoice voice customisation
• LLM fine-tuning

For a user, the service training process in Caila consists of the following steps:

• Find a necessary base service, for example intent classifier.
• Find out in what form the dataset needs to be prepared. You can do this via the documentation on the service page and the general dataset types register.
• Upload dataset. You can do it separately or directly on the fitting launch form.
• Start training.
• Check that the resulting derived service works.