Arthur Mensch
Co-founder and CEO of Mistral AI; equal-contribution author of DeepMind's Chinchilla scaling paper
- Co-founder and CEO, Mistral AI1
Arthur Mensch is a co-founder and the CEO of Mistral AI, according to the company's about page. Before Mistral, Mensch was one of four equal-contribution authors of Training Compute-Optimal Large Language Models (2022), the DeepMind paper that introduced Chinchilla and studied how model size and the number of training tokens should grow together under a fixed compute budget; the paper lists Mensch among its corresponding authors at DeepMind. Mensch was also an equal-contribution author of the retrieval-enhanced RETRO paper (2021), whose authors are all listed as from DeepMind. Mensch co-authored the Mistral 7B (2023) and Mixtral of Experts (2024) papers, whose abstracts say the models are released under the Apache 2.0 license.12345
Work
Chinchilla (Training Compute-Optimal Large Language Models) (external site: arxiv.org)Published research
One of four equal-contribution authors of the 2022 DeepMind paper, which trains over 400 language models to find that model size and training tokens should be scaled equally for compute-optimal training, and tests this with the 70B-parameter Chinchilla model.2
Mistral 7B and Mixtral 8x7B (external site: arxiv.org)Open weights
Co-author of the Mistral 7B (2023) and Mixtral of Experts (2024) papers; both abstracts say the models are released under the Apache 2.0 license.45
Other documented roles
Roles the sources document but that are not confirmed as current.
- Equal-contribution author of the RETRO (2021) and Chinchilla (2022) papers, written at DeepMind, DeepMind (now Google DeepMind)23
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.