Using classes

The model uses the classes you provide for labeling the texts present in your dataset file.

If you don't have the classes in your dataset file, add them. You must specify at least two classes.

Adding classes and their descriptions

  1. Click the + Add classes button in the Settings area.

  2. The Describe at least 2 classes for your task popup will open. Enter the following data:

    • Name is the class name that will be used for prompt generation and for labeling. Names should be unique and it also helps if they are short (ideally — one token).

    • Description is the class description that will be used for prompt generation. The description is an optional field, but it greatly helps the model better understand the context and the task and leads to better labeling results. We recommend that you to provide it for each class.

  3. Click + Add to add more classes or Delete class to remove the class from the list and discard it.

  4. Click Save classes to add the classes to your project.

You will see the classes in the Settings area.

Editing existing classes

To edit the classes present in the dataset file or added manually, click the Edit classes icon next to the Classes field in the Settings area. Then follow the steps in the Adding classes and their descriptions section to edit or delete classes and their descriptions.

Next steps

Last updated: September 5, 2023

Toloka LLMRegistering and signing inAdding OpenAI API keyTerminology
Working with Toloka LLM
Workflow overview
Working with datasets
Problem definition
Iterating on quality
Creating endpoint