Implementation of the paper: "Mixture-of-Depths: Dynamically allocating compute in transformer-based language models"
#39 opened 2 months ago in kyegomez/Mixture-of-Depths
