AI People Wiki

Builders & open source

Marie-Anne Lachaux

Co-author of Meta's LLaMA and Llama 2 papers and of Mistral 7B; first author on unsupervised code translation

Marie-Anne Lachaux is a co-author of LLaMA: Open and Efficient Foundation Language Models (2023), written at Meta AI, and of Llama 2: Open Foundation and Fine-Tuned Chat Models (2023), from GenAI at Meta. Earlier, listed with Facebook AI Research, Lachaux was first author of Unsupervised Translation of Programming Languages (2020), which trains a neural transcompiler on monolingual open-source code to translate functions between C++, Java and Python and releases a test set with unit tests. Lachaux also 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

At a glance

Current role
None confirmed in the sources
Work listed
3
Sources cited
6
Profile reviewed

Research

Lachaux's early research at Facebook AI Research focused on machine learning for source code. As first author of "Unsupervised Translation of Programming Languages" in 2020, Lachaux and colleagues trained a neural transcompiler on monolingual open-source code to translate functions between C++, Java and Python. In 2021 Lachaux co-authored DOBF, a deobfuscation pre-training objective for programming languages, with Baptiste Roziere, Marc Szafraniec and Guillaume Lample.36

Lachaux then worked on large language models, appearing among the authors of Meta's LLaMA paper in February 2023 and the Llama 2 paper in July 2023. Lachaux is also a co-author of the Mistral 7B paper in October 2023 and the Mixtral of Experts paper in January 2024, both describing models released under the Apache 2.0 license.1245

Work and contributions

  • LLaMA and Llama 2 Open weights

    Co-author of LLaMA (2023), whose abstract says all models were released to the research community, and of Llama 2 (2023), which develops and releases pretrained and chat models from 7 billion to 70 billion parameters.12

  • Unsupervised Translation of Programming Languages Published research

    First author of the 2020 paper (with Baptiste Roziere, Lowik Chanussot and Guillaume Lample), written at Facebook AI Research, on a neural transcompiler trained without parallel data.3

  • Mistral 7B and Mixtral 8x7B 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

Timeline

  1. Released "Unsupervised Translation of Programming Languages" as first author.3
  2. Co-authored DOBF, a deobfuscation pre-training objective for programming languages.6
  3. Co-authored the LLaMA paper on open and efficient foundation language models.1
  4. Co-authored the Llama 2 paper on open foundation and fine-tuned chat models.2
  5. Co-authored the Mistral 7B paper.4
  6. Co-authored the Mixtral of Experts paper.5

Documented roles

The sources do not confirm a current role; these are the roles they document.

Companies in their history

  • Meta 106 people in the wiki

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

6 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. LLaMA: Open and Efficient Foundation Language Models (arXiv 2302.13971), author list and Meta AI affiliationarXiv · Paper · published Feb 27, 2023 · accessed Oct 21, 2026
  2. Llama 2: Open Foundation and Fine-Tuned Chat Models (arXiv 2307.09288), author list and GenAI, Meta affiliationarXiv · Paper · published Jul 18, 2023 · accessed Oct 21, 2026
  3. Unsupervised Translation of Programming Languages (arXiv 2006.03511), Facebook AI Research affiliationarXiv · Paper · published Jun 5, 2020 · accessed Oct 21, 2026
  4. Mistral 7B (arXiv 2310.06825)arXiv · Paper · published Oct 10, 2023 · accessed Oct 21, 2026
  5. Mixtral of Experts (arXiv 2401.04088)arXiv · Paper · published Jan 8, 2024 · accessed Oct 21, 2026
  6. DOBF: A Deobfuscation Pre-Training Objective for Programming LanguagesarXiv · Paper · published Feb 15, 2021 · accessed Oct 21, 2026