00:04 Training Compact Models for Low Resource Entity Tagging using Pre-trained Language Models 05:41 Doubly Sparse: Sparse Mixture of Sparse Experts for Efficient Softmax Inference 14:24 Discovering Low-Precision Networks Close to Full-Precision Networks for Efficient Inference 21:50 Instant Quantization of Neural Networks using Monte Carlo Methods 27:21 Spoken Language Understanding on the Edge 32:39 Energy-Aware Neural Architecture Optimization With Splitting Steepest Descent 39:42 DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter 44:24 Algorithm-hardware Co-design for Deformable Convolution