- introduce that the goal is not to cover the models, but to use the two "types" of models as an example covering the two
example cases of a next token prediction and a classification task
- highlight the modular design of the framework and its adaptability to new and future models and data structures as input