| sources |
|
|
|---|---|---|
| type | Organization | |
| description | Google Research is the research division of Google involved in the Transformer paper. |
[[entities/google-research|Google Research]] is a research organization at Google that co-authored [[summaries/attention-is-all-you-need]] and contributed to the development of the [[concepts/transformer-models|Transformer]] architecture.
In this paper, Google Research authors include Niki Parmar and Jakob Uszkoreit. The document presents the Transformer, an attention-only sequence transduction model that removes recurrence and convolutions in favor of [[concepts/attention-mechanisms|attention mechanisms]].
- Google Research authors helped design and evaluate the Transformer.
- The paper reports state-of-the-art translation results on [[entities/wmt-2014|WMT 2014]] tasks, including [[entities/wmt-2014-english-german|English-German]] and [[entities/wmt-2014-english-french|English-French]].
- The work was presented at [[entities/nips-2017|NIPS 2017]].
- The paper also references the Tensor2Tensor codebase, associated with [[entities/tensor2tensor|tensor2tensor]].
- [[summaries/attention-is-all-you-need]]
- [[concepts/transformer-models]]
- [[concepts/attention-mechanisms]]
- [[entities/google]]
- [[entities/google-brain]]
- [[entities/tensor2tensor]]