cid 3667774f3ea7d2d91f7d8aaa7e22bf6887d710cc2a4442ff765ba3208d98eb27 · original · open access · via arxiv
Title: Attention Is All You Need Authors: Ashish Vaswani; Noam Shazeer; Niki Parmar; Jakob Uszkoreit; Llion Jones; Aidan N. Gomez; Lukasz Kaiser; Illia Polosukhin Venue: arXiv Year: 2017 Canonical: arxiv:1706.03762v7 URL: http://arxiv.org/abs/1706.03762v7 Collector: arxiv OA: true ABSTRACT The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.