Efficient scaling of large language models with mixture of experts and 3D analog in-memory computing
Julian Büchel, Athanasios Vasilopoulos, William Simon, Irem Boybat, Hsinyu Tsai, Geoffrey W. Burr, Hernan Castro, B. Filipiak, Manuel Le Gallo, Abbas Rahimi, Vijay Narayanan, Abu Sebastian
Topics & Concepts
Computer scienceBottleneckScalabilityInferenceSoftware deploymentMemory footprintLeverage (statistics)Computer architectureDistributed computingEmbedded systemArtificial intelligenceDatabaseOperating systemAdvanced Memory and Neural ComputingFerroelectric and Negative Capacitance DevicesNeural Networks and Reservoir Computing