Independent project. Not a U.S. government website.

Support our project (opens Buy Me a Coffee)
USASI

Victor Sanh

First author of DistilBERT and T0; co-author of the Transformers library paper and the Idefics2 model

  • Listed as the current company on Sanh's GitHub profile (no title stated), Rilix1

Victor Sanh's GitHub profile describes Sanh as an AI builder previously at Hugging Face and lists Rilix as the current company. At Hugging Face, Sanh was first author of DistilBERT, a smaller distilled version of BERT, and a co-author of the paper describing the open-source Transformers library. Sanh was also first author of the paper on multitask prompted training that introduced the T0 models, a senior contributor to the OBELICS dataset and IDEFICS models, and a co-author of the 2024 paper that released the Idefics2 vision-language model.124579

Work

  • TransformersOpen-source code

    Third-listed author of "Transformers: State-of-the-Art Natural Language Processing" (2019), the paper describing the open-source library, written with a Hugging Face affiliation.4

    Project page (external site: github.com)

  • DistilBERT (external site: huggingface.co)Open weights

    First author of the DistilBERT paper (2019), which pre-trains a smaller general-purpose model through knowledge distillation and tests it on device; the distilbert-base-uncased model card lists the Apache 2.0 license.23

  • T0 (external site: huggingface.co)Open weights

    First author of "Multitask Prompted Training Enables Zero-Shot Task Generalization" (2021), which fine-tunes an encoder-decoder model on a prompted multitask mixture and releases the trained models; the T0pp model card lists the Apache 2.0 license.56

  • OBELICS, IDEFICS and Idefics2 (external site: huggingface.co)Open weights

    Marked as an equal and senior contributor on the OBELICS paper (2023), which releases an open interleaved image-text dataset and the IDEFICS models, and last-listed author of the 2024 study that develops and releases the Idefics2 8B vision-language model, whose model card lists the Apache 2.0 license.78910

Other documented roles

Roles the sources document but that are not confirmed as current.

  • Listed with a Hugging Face affiliation on the DistilBERT (2019) and Idefics2 (2024) papers, Hugging Face1249

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.

Sources · reviewed Oct 13, 2026

  1. 1.
    VictorSanh (Victor Sanh) on GitHub (external site: github.com)

    GitHub · Official page · accessed Oct 13, 2026

  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
    bigscience/T0pp model card (license apache-2.0) (external site: huggingface.co)

    BigScience (Hugging Face) · Model card · accessed Oct 13, 2026

  7. 7.
  8. 8.
    HuggingFaceM4/OBELICS dataset card (license cc-by-4.0) (external site: huggingface.co)

    HuggingFaceM4 (Hugging Face) · Dataset card · accessed Oct 13, 2026

  9. 9.
    What matters when building vision-language models? (arXiv 2405.02246) (external site: arxiv.org)

    arXiv · Paper · published May 3, 2024 · accessed Oct 13, 2026

  10. 10.
    HuggingFaceM4/idefics2-8b model card (license apache-2.0) (external site: huggingface.co)

    HuggingFaceM4 (Hugging Face) · Model card · published Apr 9, 2024 · accessed Oct 13, 2026

Support Us

Help keep USASI useful.

Optional. No USASI account required. Payment takes place on the linked provider’s website (Buy Me a Coffee).

About supporting this project