The previous chapters have provided the components needed to construct complete, text-based LLM applications, from generation and RAG to autonomous agents. This final chapter offers a look at several specialized capabilities for extending your applications further.

You will be introduced to three distinct topics:

• Multimodal Inputs: Learn how to process images alongside text using vision-language models, moving from purely textual interactions.
• Application Evaluation: Address the challenge of measuring quality by using built-in utilities to benchmark the performance of your LLM outputs against reference standards.
• Fine-Tuning Preparation: See how to use the toolkit's helper functions to format custom datasets, which is the first step in training a specialized model.

These sections provide a practical starting point for incorporating more sophisticated features into your projects.