[ PAPER ] · 2019
Poly-encoders: Transformer Architectures and Pre-training Strategies for Fast and Accurate Multi-sentence Scoring
Samuel Humeau, Kurt Shuster, M. Lachaux, J. Weston
248 citationsno open PDF · spoken briefing available
No open-access PDF, so we'll narrate a briefing from the abstract instead.
[ TLDR ]
This work develops a new transformer architecture, the Poly-encoder, that learns global rather than token level self-attention features and achieves state-of-the-art results on three existing tasks.