A category of language models that leverage deep - learning algorithms and are trained on vast textual datasets, which can reach sizes of multiple terabytes. LLMs can be broadly classified into two types: generative and discriminative.
Generative LLMs are those models capable of generating text. For example, they can provide answers to questions or even compose essays on specific topics. These are generally unsupervised or semi-supervised learning models that forecast the appropriate response for a given task. On the other hand, discriminative LLMs are supervised learning models mainly used for text classification. For instance, they can determine whether a piece of text is written by a human or AI.