Tweeted By @hardmaru
MetaInit: Initializing learning by learning to initialize
— hardmaru (@hardmaru) December 18, 2019
They propose a strategy to automatically identify good initial parameters, and show that deep architectures *without* batch norm or residual connections can be trained to get near SOTA results. 🔥https://t.co/Ei94S99ZbZ pic.twitter.com/RnYc7YghDR