expert specialization

mixture of expert (640) parameter efficiency (417) knowledge distillation (3725) model compression (3302) large language model (13587) neural network (6616) expert routing (44) instruction tuning (838) model architecture (324) load balancing (55)