Research Scientist @Meta Superintelligence Labs. ex-@GoogleDeepMind - Gemma. PhD @IP_Paris_ | @Mila_Quebec | MSc MVA | @ENS_ULM
Paris
Joined October 2015
- Tweets172
- Following314
- Followers543
- Likes1.4K
Thomas Mesnard retweeted
Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple.
Thomas Mesnard retweeted
MUSE WONDER
this is the culmination of over a year of research and infrastructure work in msl
since we originally shipped muse spark in april, we have continued to scale our models across pre-training and RL, now reaching the frontier 🏆
see our model's improvements on the artificial analysis index:
A pretty dramatic shift in how we interact with AI: a personal assistant that actually does things for you 24/7.
Muse has been part of my daily life for months now and it’s hard to imagine going back.
1/ today we're rolling out Muse, our new personal ai assistant. Muse is always-on, wicked fast, can use a browser, connect to your apps, and is designed to be secure.
try it now: muse.ai.
Thomas Mesnard retweeted
Excited to share what we have been cooking in the past few months: Muse Glimmer: research.meta.ai/blog/introd…
Thomas Mesnard retweeted
Introducing Muse Glimmer, an open-weight 30B-parameter model optimized for local, always-on agent workflows.
Muse Glimmer delivers strong performance on key agentic use cases and benchmarks compared with leading models in its size category, and is designed to run entirely on consumer hardware like a Mac or PCs with performant GPUs.
In keeping with our long tradition of sharing fundamental AI research, we’re releasing model weights under a permissive Apache 2.0 license.
🧵👇
Thomas Mesnard retweeted
1/ today we're releasing muse spark, the first model from MSL. nine months ago we rebuilt our ai stack from scratch. new infrastructure, new architecture, new data pipelines. muse spark is the result of that work, and now it powers meta ai. 🧵
Thomas Mesnard retweeted
An exciting milestone for AI in science: Our C2S-Scale 27B foundation model, built with @Yale and based on Gemma, generated a novel hypothesis about cancer cellular behavior, which scientists experimentally validated in living cells.
With more preclinical and clinical tests, this discovery may reveal a promising new pathway for developing therapies to fight cancer.
Thrilled to announce the world’s most capable differentially private LLM!
Huge congratulations to the entire team — and special kudos to Amer Sinha for his outstanding contributions.
Take a look!
services.google.com/fh/files…
VaultGemma is a release of an open model trained from scratch with differential privacy.
The blog post below and the full tech report linked from the tech report have some nice analyses to present a scaling law for differentially private language models:
Blog: research.google/blog/vaultge…
Paper: arxiv.org/abs/2501.18914
Thomas Mesnard retweeted
Introducing EmbeddingGemma, our newest open model that can run completely on-device. It's the top model under 500M parameters on the MTEB benchmark and comparable to models nearly 2x its size – enabling state-of-the-art embeddings for search, retrieval + more.
Thomas Mesnard retweeted
Introducing EmbeddingGemma🎉
🔥With only 308M params, this is the top open model under 500M
🌏Trained on 100+ languages
🪆Flexible embeddings (768 to 128 dims) with Matryoshka
🤗Works with your favorite open tools
🤏Runs with as little as 200MB
developers.googleblog.com/en…
Thomas Mesnard retweeted
We've had a busy August!
August at Google DeepMind was like 🧞♂️ 🖼️ 🍌 🚀 🔍 🤏🏻
- Nano Banana (Gemini 2.5 Flash Image)
- Gemini Embedding
- Veo 3 Fast
- Genie 3
- Imagen 4 Fast
- Gemma 3 270M
- Perch 2
- Kaggle Game Arena
- Gemini API Url Context
- AI Studio Builder (UI Rework, Prompt Suggestions, GitHub integration …)
- AI studio UI (Model Picker, Chat, Scrolling…)
and much more!
🚀 After Gemma 3 1B, we’re going tiny: Gemma 270M — fast, on-device, low-power & privacy-first.
A new vision: smaller models with strong instruction-following & finetuning, ideal for low-latency edge apps & automation.
Congrats to everyone involved!
huggingface.co/google/gemma-…
Introducing Gemma 3 270M 🔥
🤏A tiny model! Just 270 million parameters
🧠 Very strong instruction following
🤖 Fine-tune in just a few minutes, with a large vocabulary to serve as a high-quality foundation
developers.googleblog.com/en…
Thomas Mesnard retweeted
Looking for a small or medium sized VLM? PaliGemma 2 spans more than 150x of compute!
Not sure yet if you want to invest the time 🪄finetuning🪄 on your data? Give it a try with our ready-to-use "mix" checkpoints:
🤗 huggingface.co/blog/paligemm…
🎤 developers.googleblog.com/en…
Thomas Mesnard retweeted
Thrilled to launch the PaliGemma 2 Mix models today! Try out class-leading vision-language models from 3B to 28B =). #gemmaverse
PaliGemma 2 mix is an upgraded vision-language model that supports image captioning, OCR, image Q&A, object detection, and segmentation. With sizes from 3B-28B parameters, there's a model for everyone. Get started. → goo.gle/430HnDe
ALT The text "PaliGemma 2 mix" is centered on a dark background with smaller images and text snippets surrounding it, showcasing examples of image captioning and location identification.
🥳🥳🥳
PaliGemma 2 mix is an upgraded vision-language model that supports image captioning, OCR, image Q&A, object detection, and segmentation. With sizes from 3B-28B parameters, there's a model for everyone. Get started. → goo.gle/430HnDe
ALT The text "PaliGemma 2 mix" is centered on a dark background with smaller images and text snippets surrounding it, showcasing examples of image captioning and location identification.
Thomas Mesnard retweeted
🚀🚀PaliGemma 2 is our updated and improved PaliGemma release using the Gemma 2 models and providing new pre-trained checkpoints for the full cross product of {224px,448px,896px} resolutions and {3B,10B,28B} model sizes.
1/7
Thomas Mesnard retweeted
Thrilled to announce the launch of PaliGemma 2, a state-of-the-art update to our PaliGemma vision-language models! These models are based on Gemma 2, and are available today in 2B, 9B, and 27B sizes.
developers.googleblog.com/en…
huggingface.co/blog/paligemm…
