Researcher

Lucile Saulnier

Lucile Saulnier is a machine-learning engineer whose work links open-source NLP with collaborative multilingual training and Mistral's language and vision models. At Hugging Face, she presented BigScience's collaborative training work and developed and supported open-source tools while contributing to NLP research. Her papers include the ROOTS corpus, Mistral 7B, Mixtral, Pixtral, Magistral, and Ministral 3. As of October 2026, she is an AI Specialist at Mistral AI.

Timeline

  1. 2022

    BigScience at Hugging Face

    At the Teratec Forum in 2022, she presents BigScience's collaborative work to train a large multilingual language model as a machine-learning engineer at Hugging Face. She develops and supports open-source tools and contributes to NLP research projects, including collaborative training and BigScience.

  2. 2023

    The ROOTS corpus

    In 2023, she coauthors The BigScience ROOTS Corpus paper, which describes a 1.6TB dataset spanning 59 languages used to train the 176-billion-parameter BLOOM model.

  3. 2023

    Mistral 7B

    She coauthors Mistral 7B in 2023, a 7-billion-parameter language model that uses grouped-query attention and sliding-window attention.

  4. 2024

    Mixtral of Experts

    In 2024, she coauthors Mixtral of Experts, which introduces Mixtral 8x7B, a sparse mixture-of-experts language model with eight feed-forward experts per layer and a router that selects two experts for each token.

  5. 2024

    Pixtral 12B

    She coauthors Pixtral 12B in 2024, a 12-billion-parameter multimodal language model trained to understand natural images and documents, with a vision encoder that handles images at their natural resolution and aspect ratio.

  6. 2025

    Magistral reasoning model

    In 2025, she coauthors Magistral, Mistral AI's first reasoning model, with a scalable reinforcement-learning pipeline that relies on Mistral's own models and infrastructure.

  7. January 2026

    Ministral 3

    In January 2026, she coauthors Ministral 3, a parameter-efficient family of dense language models for compute- and memory-constrained applications, in 3B, 8B, and 14B sizes.

  8. As of October 2026

    AI specialist at Mistral AI

    As of October 2026, she is an AI Specialist at Mistral AI.

Career

  1. NowMistral AIAI Specialist
  2. From 2022Hugging FaceMachine-learning engineer

Sources and credits

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

  1. Teratec 2022 Forum, Workshop 08 teratec.fr. Checked 21 October 2026.
  2. Introduction · Hugging Face huggingface.co. Checked 21 October 2026.
  3. The BigScience ROOTS Corpus: A 1.6TB Composite Multilingual Dataset arxiv.org. Checked 21 October 2026.
  4. Mistral 7B arxiv.org. Checked 21 October 2026.
  5. Mixtral of Experts arxiv.org. Checked 21 October 2026.
  6. Pixtral 12B arxiv.org. Checked 21 October 2026.
  7. Magistral arxiv.org. Checked 21 October 2026.
  8. Ministral 3 arxiv.org. Checked 21 October 2026.
  9. SaulLu (Lucile Saulnier) · GitHub github.com. Checked 21 October 2026.