iiinova — where your digital self keeps evolving

Benoît Sagot

@bensagot.bsky.social

82 followers 70 following 8 posts

Directeur de recherche at Inria, former invited professor at Collège de France, co-founder of opensquare

Posts Replies Media

Benoît Sagot @bensagot.bsky.social · 12/11/2025

Codebase (Gapetron, Apache-2 licence): github.com/NathanGodey/...

github.com

GitHub - NathanGodey/gapetron

Contribute to NathanGodey/gapetron development by creating an account on GitHub.

020

Benoît Sagot @bensagot.bsky.social · 12/11/2025

Models (OpenRAIL-M licence): huggingface.co/collections/...

huggingface.co

Gaperon - a almanach Collection

Our French-English LLM suite (SFT models are coming soon)

111

Benoît Sagot @bensagot.bsky.social · 12/11/2025

Thanks also to GENCI @gencifrance.bsky.social and CINES for compute support.

110

Benoît Sagot @bensagot.bsky.social · 12/11/2025

Congratulations to Nathan Godey @nthngdy.bsky.social, Wissam Antoun @wissamantoun.bsky.social and Rian Touchent, who did most of the work, supervised by Djamé Seddah @zehavoc.bsky.social, myself, Éric de La Clergerie and Rachel Bawden @rachelbawden.bsky.social (in order of decreasing implication).

110

Benoît Sagot @bensagot.bsky.social · 12/11/2025

Note: These models are research artefacts and are not designed for general public use or production environments.

100

Benoît Sagot @bensagot.bsky.social · 12/11/2025

If you’d like to find out what we discovered, I encourage you to read Nathan's thread (reposted in the first post of this thread) as well as the paper, where we describe our experiments and findings in detail: arxiv.org/pdf/2510.25771

arxiv.org

100

Benoît Sagot @bensagot.bsky.social · 12/11/2025

Our main goal with this project was to deepen our understanding of language models' training dynamics and of the impact of the properties of their pretraining data. The results we obtained led us to take a closer look at a phenomenon whose impact is often underestimated: data contamination.

100

Benoît Sagot @bensagot.bsky.social · 12/11/2025

I'm proud to share that at @inriaparisnlp.bsky.social we have released Gaperon — a suite of generative language models trained on French, English and code data, the largest of which has 24 billion parameters. Both the models and the code are being published under open licences. Short thread🧵

185

Reposted by Benoît Sagot

Inria Paris NLP (ALMAnaCH team) @inriaparisnlp.bsky.social · 18/06/2025

We are delighted to announce our next seminar by Hal Daumé III @haldaume3.bsky.social ‬(@univofmaryland.bsky.social, currently on sabbatical at @sorbonne-universite.fr) entitled "Fairness and Trustworthiness in Generative Al" on Friday 20th June at 11am CEST.

293

Reposted by Benoît Sagot

Pierre Chambon @pierrechambon.bsky.social · 20/03/2025

Does your LLM truly comprehend the complexity of the code it generates? 🥰 Introducing our new non-saturated (for at least the coming week? 😉) benchmark: ✨BigO(Bench)✨ - Can LLMs Generate Code with Controlled Time and Space Complexity? Check out the details below !👇

143

Reposted by Benoît Sagot

Nathan Godey @nthngdy.bsky.social · 06/03/2025

🚀 New Paper Alert! 🚀 We introduce Q-Filters, a training-free method for efficient KV Cache compression! It is compatible with FlashAttention and can compress along generation which is particularly useful for reasoning models ⚡ TLDR: we make Streaming-LLM smarter using the geometry of attention

1207