Building Agent-Powered Applications

5

Generation, Question Answering, and Reasoning

In the previous chapter, we explored several core language tasks, such as summarization, text classification, information extraction, and multilingual processing. These tasks form the backbone of modern natural language processing (NLP) and have been studied for decades. Even before the emergence of transformer architectures and large language models (LLMs), researchers developed methods to solve many of these problems with varying degrees of success.

However, the arrival of LLMs has significantly expanded what machines can do with language. Tasks that once required specialized models, complex pipelines, or carefully engineered rules can now often be solved by a single model with the right prompt ...