PeopleofSuperintelligence

Researcher

Hugo Touvron

Hugo Touvron is a researcher who is first author of the LLaMA paper and worked on Llama 2 and Code Llama. His doctoral work from 2019 to 2022 centered on architectures and training for visual understanding. A paper posted in December 2020, which he leads as first author, trains a convolution-free transformer on ImageNet alone. The LLaMA paper, from February 2023, introduces foundation language models from 7B to 65B parameters. In May 2025 he was still a research scientist at Meta, one of three original Llama authors who remained there.

Now at
Meta
Works
39

Publication numbers from OpenAlex, as of 15 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2019–2022

    A PhD on visual understanding

    He studies for a PhD on architectures and training for visual understanding from 2019 to 2022.

  2. December 2020

    Data-efficient image transformers

    The paper, posted in December 2020, trains a convolution-free transformer on ImageNet alone, and he is its first author.

  3. February 2023

    The LLaMA foundation models

    In February 2023 he is first author of LLaMA, a paper that introduces foundation language models from 7B to 65B parameters.

  4. July 2023

    Llama 2 open models

    A July 2023 paper on Llama 2 covers pretrained and fine-tuned models from 7 to 70 billion parameters, with him as first author.

  5. August 2023

    Code Llama models

    Code Llama, posted in August 2023, is a family of large language models for code based on Llama 2, and he is one of its authors.

  6. May 2025

    A research scientist at Meta

    In May 2025 he is one of three original Llama authors still at Meta, where he works as a research scientist.

Sources and credits

Photo: We have no published photo of Hugo Touvron yet. Send us a free-licensed photo.

  1. Matthieu Cord - My Chordettes Research band cord.isir.upmc.fr. Checked 15 October 2026.
  2. [2012.12877] Training data-efficient image transformers & distillation through attention arxiv.org. Checked 15 October 2026.
  3. [2302.13971] LLaMA: Open and Efficient Foundation Language Models arxiv.org. Checked 15 October 2026.
  4. [2307.09288] Llama 2: Open Foundation and Fine-Tuned Chat Models arxiv.org. Checked 15 October 2026.
  5. [2308.12950] Code Llama: Open Foundation Models for Code arxiv.org. Checked 15 October 2026.
  6. Meta's Llama AI Team Has Been Bleeding Talent. Many Joined Mistral. - Business Insider businessinsider.com. Checked 15 October 2026.