Self-Attentive

Implementation of "A Structured Self-Attentive Sentence Embedding"

Result

Accuracy: 0.85

Visualization

Multi annotation vector - multi r hops

Single annotation vector - single r hop

Reference

https://arxiv.org/abs/1703.03130
https://github.com/flrngel/Self-Attentive-tensorflow