| sources |
|
|
|---|---|---|
| type | Organization | |
| description | Google Brain is a Google research team involved in the Transformer paper. |
Google Brain is a research organization within Google that appears in the author affiliations of [[summaries/attention-is-all-you-need]]. In that paper, multiple coauthors list Google Brain as their home institution, and the team is credited with major contributions to the development, implementation, and evaluation of the [[concepts/transformer-models|Transformer]] architecture.
The paper credits Google Brain personnel for several core aspects of the work, including:
- proposing and evaluating self-attention as a replacement for recurrent layers
- designing and implementing the first Transformer models
- proposing scaled dot-product attention and multi-head attention
- building and improving the codebase used for experiments
- supporting efficient inference and visualization work
Several authors are listed with Google Brain affiliations, including:
- Ashish Vaswani
- Noam Shazeer
- Łukasz Kaiser
The paper also notes collaborative work across [[entities/google-research|Google Research]] and other institutions.
Google Brain's work in this paper is closely tied to:
- [[concepts/attention-mechanisms|attention mechanisms]]
- [[concepts/transformer-models|Transformer models]]
- [[concepts/positional-encoding|positional encoding]]
- [[summaries/attention-is-all-you-need]]
In this document, Google Brain is presented as the institutional setting for a major breakthrough in neural machine translation and sequence modeling, and as a central contributor to the Transformer architecture that became foundational in later NLP systems.