Articoli correlati
Miscela di esperti
Architecture used by Mistral Large 4 in which a router selects a small group of experts per token; the model uses four of 128 experts plus one shared expert in most layers.
Articoli correlati
Architecture used by Mistral Large 4 in which a router selects a small group of experts per token; the model uses four of 128 experts plus one shared expert in most layers.