A Dynamic Execution Neural Network Processor for Fine-Grained Mixed-Precision Model Training Based on Online Quantization Sensitivity Analysis
https://sscs.ieee.org/wp-content/themes/movedo/images/empty/thumbnail.jpg
150
150
https://secure.gravatar.com/avatar/8935a7dcd6741d8e23d45bb15c1470a8?s=96&d=mm&r=g
Abstract:
BSTcontrol As neural network (NN) training cost red has been growing exponentially over the past decade, developing high-speed and energy-efficient training methods has become an urgent task. Fine-grained mixed-precision low-bit training is the most promising way for high-efficiency training, but it needs dedicated processor designs to overcome the overhead in …