Laurent Sartran
First author of Transformer Grammars at DeepMind and co-author of the SynJax library for JAX
Laurent Sartran is first author of Transformer Grammars (2022), a class of Transformer language models that compose representations of syntactic constituents recursively; the paper lists Sartran at DeepMind, and the code is published in the google-deepmind GitHub organization under the Apache-2.0 license. With Miloš Stanojević, Sartran co-authored SynJax, a JAX library of vectorized inference algorithms for structured distributions such as alignments, segmentations, and trees, also published by google-deepmind under Apache-2.0.12345
Work
Transformer Grammars (external site: github.com)Open-source code
First author of Transformer Grammars: Augmenting Transformer Language Models with Syntactic Inductive Biases at Scale (TACL 2022). The google-deepmind repository describes the approach, cites the paper as Sartran et al., and is published under the Apache-2.0 license.123
SynJax (external site: github.com)Open-source code
Co-author with Miloš Stanojević of SynJax: Structured Probability Distributions for JAX (2023); the google-deepmind repository cites the paper and is published under the Apache-2.0 license.45
Other documented roles
Roles the sources document but that are not confirmed as current.
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.