James Laudon

mixture of expert (2) deep reinforcement learning (1) transformer architecture (1) catastrophic forgetting (1) few-shot learning (1) efficient computing (1) neural architecture (1) distribution shift (1) lifelong learning (1) language model (1) sparse activation (1) feed-forward network (1) expert routing (1) device placement (1) model efficiency (1) layer normalization (1) compiler optimization (1) computational graph (1) sparse gating (1) graph neural network (1)