98 - Analyzing Information Flow In Transformers, With Elena Voita by NLP Highlights published on 2019-12-09T17:51:08Z What function do the different attention heads serve in multi-headed attention models? In this episode, Lena describes how to use attribution methods to assess the importance and contribution of different heads in several tasks, and describes a gating mechanism to prune the number of effective heads used when combined with an auxiliary loss. Then, we discuss Lena’s work on studying the evolution of representations of individual tokens in transformers model. Lena’s homepage: https://lena-voita.github.io/ Blog posts: https://lena-voita.github.io/posts/acl19_heads.html https://lena-voita.github.io/posts/emnlp19_evolution.html Papers: https://arxiv.org/abs/1905.09418 https://arxiv.org/abs/1909.01380 Genre Science Comment by justHeuristic Probably the best piece of insight into deep NLP since Karpathy's RNN guide! Btw if you're looking for extended version, there are blog posts with cool visualizations for both papers in Lena's blog: lena-voita.github.io/posts.html 2019-12-09T19:47:48Z
