Prompt tuning for multi-label classification with watsonx

This notebook is part of a series of notebooks, designed to guide you on how to utilize watsonx.ai Large Language Models (LLMs) for text classification.

The scenario presented in this notebook assumes you have a few hundred or thousand labeled text elements in a multi-label setup, that is, each text element is labeled to zero or more classes from a given taxonomy. Compared to the Prompt tuning for multi-class classification with watsonx notebook that demonstrated the multi-class scenario where each text element is classfied to exactly one class, in this notebook we handle the case of generating a list of predicted classes, where the list can be either empty, contain one class, or contain multiple classes.

Some familiarity with Python is helpful. This notebook runs with Python 3.10.