BERT for unsupervised text tasks

This post discusses how we use BERT and similar self-attention architectures to address various text crunching tasks at Ether Labs. Self-attention architectures have caught the attention of NLP practi

>>阅读原文<<