LATIDIA · Agentes

Citando a Ted Sanders

En general, el rendimiento de LLM es increíblemente multidimensional. No hay garantía de un orden fijo en el que el modelo que es mejor en matemáticas sea también el modelo que es mejor en chistes. A pesar de esa complejidad, gran parte de la varia

La noticia

In general, LLM performance is incredibly multidimensional. There's no guarantee of a fixed ordering where the model that is best at math is also the model that's best at jokes. Despite that complexity, much of the variation in performance is explained by just a single variable: model size. In general, models that are bigger will (a) be more performant / reliable, (b) cost more per token, and (c) have better recall of esoteric facts. As you note, our models are ordered: Luna < Terra < Sol < Astra [ ... ] These days, version number signifies a few things at once: the order of release, the relationship to other releases,