2000 character limit reached
Think Again Networks and the Delta Loss
Published 26 Apr 2019 in cs.CL and cs.LG | (1904.11816v2)
Abstract: This short paper introduces an abstraction called Think Again Networks (ThinkNet) which can be applied to any state-dependent function (such as a recurrent neural network).
Paper Prompts
Sign up for free to create and run prompts on this paper.