@PierreColombo6

Omni-modal AI researcher. Creator of ColPali, BidirLM-omi, EuroBERT & EuroLLM. Ex-Prof@ Centrale (Paris-Saclay) · Ex-CSO@ https://xitter.cf/t.co/ncQ9gT2rpk (legaltech - SaulLM)

pierrecolombo.github.io/
Joined October 2020
  • Tweets649
  • Following1.2K
  • Followers550
6Media
🚀 Introducing SaulLM-141B and SaulLM-54B: The First Open Family of Legal Models. After #SaulLM-7B the family is growing! We are proud to unveil the latest innovations from our team: the SaulLM-141B and 54B generative AI models, specifically designed for the legal domain.
1
2
2
10
1,523
Pierre Colombo retweeted
Amélie Chatelain
@AmelieTabattaOct 12
Should you waste tokens having your coding agent use grep, or hand the search to a tiny local model? We release ColGrep-Agent-2B! A code-search agent that runs locally, searches with ColGrep, reads files, & returns locations + code to your main agent -> 34% fewer API tokens
31
47
12
262
20,145
BidirLM-Omni is live right now at COLM poster #28! Looking for an open source encoder training method that sits right on the Pareto frontier alongside @googlegemma recent omnimodal embedding? Come talk to us! 🚀
2
9
198
Pierre Colombo retweeted
Francis Bach@BachFrancisOct 8
The one and only Michael Jordan was interviewed last week by the French newspaper Libération. Here is an English translation. Enjoy! di.ens.fr/~fbach/MJordan_ITW…
6
40
7
235
42,493
Pierre Colombo retweeted
Théo Deschamps-Berger
@theodb_mlOct 7
Replying to @GoogleDeepMind @beirmug
Congrats on the release! 👀 Happy to see our BidirLM-Omni-2.5B sitting on the Pareto frontier of your MAEB chart 🎧 Presenting it at #COLM2026 tomorrow if anyone wants to discuss multimodal encoders
1
2
208
Pierre Colombo retweeted
Théo Deschamps-Berger
@theodb_mlOct 7
Replying to @osanseviero
Congrats on the release! 👀 Happy to see our BidirLM-Omni-2.5B sitting on the Pareto frontier of your MAEB chart 🎧 Presenting it at #COLM2026 tomorrow if anyone wants to discuss multimodal encoders
1
1
147
Pierre Colombo retweeted
Théo Deschamps-Berger
@theodb_mlOct 7
Replying to @tomaarsen @GoogleDeepMind
Happy to still see our BidirLM-Omni-2.5B sitting on the Pareto frontier of the MAEB chart 🎧 Presenting it at #COLM2026 tomorrow if anyone wants to discuss multimodal encoders
1
2
150
Huge congrats to @N1colAIs @theodb_ml and @gisship 🎉 @GoogleDeepMind just released EmbeddingGemma 2, and the @googlegemm launch blog benchmarks against BidirLM-Omni-2.5B on the Massive Audio Embedding Benchmark. Our model sits right on the Pareto frontier!
1
2
2
171
They're presenting BidirLM at @COLM_conf ! You should go find them and say hi. 📄 arxiv.org/abs/2604.02045 🤗 huggingface.co/BidirLM

BidirLM: From Text to Omnimodal Bidirectional Encoders by Adapting...

Transforming causal generative language models into bidirectional encoders offers a powerful alternative to BERT-style architectures. However, current approaches remain limited: they lack...

arxiv.org
1
41
Pierre Colombo retweeted
Théo Deschamps-Berger
@theodb_mlOct 7
I'm at #COLM2026 in San Francisco this week 🌉 I will be presenting BidirLM-Omni (/w @N1colAIs) open-source recipe that turns causal decoder LLMs (like Gemma 3 and Qwen3) into strong bidirectional encoders for text, image and audio
1
2
4
189
Pierre Colombo retweeted
Dario Shariatian
@dario_shaOct 1
🔥 Accelerate your diffusion LMs with Distribution Matching Distillation Introducing two methods for continuous diffusion LMs with gains at complementary sampling budgets: 🔺 Simplex-DMD at very few-step budgets 🎲 Reinforce-DMD at larger budgets 🔺 At 4 NFEs Simplex-DMD halves the strongest baseline’s generative perplexity at matched entropy (1/n)
2
25
4
101
6,909
Pierre Colombo retweeted
Manuel Faysse
@ManuelFaysseSep 17
I joined Google DeepMind last week to work on (hopefully) improving the Gemini models! The past few years have been incredibly fun and rewarding - working on great projects with amazing people whom I'll never thank enough. Excited to discover what lies ahead!
51
3
293
16,162
Pierre Colombo retweeted
Cohere Labs
@Cohere_LabsJul 6
We’re thrilled to share that research from Cohere Labs and @Cohere will be presented at ICML this week! 🥳 📄ModernVBERT: Towards Smaller Visual Document Retrievers. Congrats to authors Paul Teiletche Quentin Mace, Max Conti, Antonio Loison, Gautier Viaud, @PierreColombo6, Manuel Faysse (arxiv.org/pdf/2510.01149) 🔎Investigating continual pretraining in large language models: Insights and implications. Congrats to authors Çağatay Yıldız, Nishaanth Kanna Ravichandran, Nitin Sharma, Matthias Bethge, @beyzaermis (openreview.net/pdf?id=aKjJoE…) ⌚EAGER: Entropy-Aware GEneRation for Adaptive Inference-Time Scaling. Congrats to authors @daniel_sc4, Leonidas Zotos, Elisabetta Fersini, and Malvina Nissim, @ahmetustun89 (arxiv.org/abs/2510.11170) 📈When AI Benchmarks Plateau: A Systematic Study of Benchmark Saturation. Congrats to all 35+ authors including Cohere's @YiyangNan (arxiv.org/abs/2602.16763)
3
1
19
1,983
Awesome to see them build on the same tricks as BidirLM-Omni, like the decoder-to-encoder adaptation and cross-modality model merging. The synthetic data ablation is also a huge highlight with a +15 gain for code retrieval domains
1
4
1
8
769
Gemini Embedding 2 is out, and it's completely omnimodal 😎 Great to see the next chapter of encoders heading toward text, audio, and visual unlocking so many use cases arxiv.org/abs/2605.27295

Gemini Embedding 2: A Native Multimodal Embedding Model from Gemini

We introduce Gemini Embedding 2, a native multimodal embedding model that allows embedding video, audio, image, and text modalities in a unified representation space. We leverage the multimodal...

arxiv.org
2
18
99
5,825
👏 Congratulations to @cohere on Command A+ — a powerful new model optimized for NVIDIA Blackwell and trained using NVIDIA CUDA-X libraries. Proud to be a part of it! Learn more ⤵️
Cohere
@cohereMay 20
Introducing: Cohere Command A+ We’ve created our most powerful LLM yet, optimized it to run on as little hardware as possible, and released it open-source for all.

Meet Command A+

Available now on Hugging Face and vLLM

7
29
2
306
38,924
Pierre Colombo retweeted
Cohere
@cohereMay 20
Releasing open-source under the Apache 2.0 license. We want to give developers direct access to enterprise-grade agentic capabilities from experimentation to production. Sovereign AI. For all. Download Command A+: huggingface.co/CohereLabs/co… Or learn more: cohere.com/blog/command-a-pl…

CohereLabs/command-a-plus-05-2026-w4a4 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co
11
31
5
294
33,002
Pierre Colombo retweeted
Aidan Gomez
@aidangomezMay 20
Our first fully open source Apache 2 model :)
Cohere
@cohereMay 20
Introducing: Cohere Command A+ We’ve created our most powerful LLM yet, optimized it to run on as little hardware as possible, and released it open-source for all.

Meet Command A+

Available now on Hugging Face and vLLM

13
16
214
17,151
Cohere launches open weights model Command A+ that achieves 37 on the Artificial Analysis Intelligence Index The release of Command A+ places @Cohere in line with Claude 4.5 Haiku on the Intelligence Index, and just above NVIDIA Nemotron 3 Super and Gemini 3.1 Flash-Lite. Key Takeaways: ➤ Command A+ ranks first on AA-Omniscience Non-Hallucination at 86%, ~3 percentage points ahead of the next-best model. Its AA-Omniscience Accuracy is 9%, so the headline AA-Omniscience score lands at -4, demonstrating a similar archetype to Claude 4.5 Haiku, where the model knows its limits ➤ On Cohere’s API, Command A+ (~281 output tokens per second) is faster than several comparable open-weights and small to mid-sized proprietary models (e.g., GPT-5.4 nano, Claude 4.5 Haiku, and Grok 4.3), but still slower than Gemini 3.1 Flash-Lite Preview, which outputs 304 tokens per second ➤ Command A+ trails its peer set on scientific reasoning (HLE ~11%, GPQA Diamond ~76%) and on coding (Terminal-Bench Hard ~25%, SciCode ~38%), consistent with gaps on the hardest science and agentic coding benchmarks ➤ It supports visual reasoning and scores 63% on MMMU-Pro (between Claude 4.5 Haiku at 59% and GPT-5.4 nano (xhigh) at 65%)
13
23
5
253
35,244
Pierre Colombo retweeted
Cohere
@cohereMay 20
Introducing: Cohere Command A+ We’ve created our most powerful LLM yet, optimized it to run on as little hardware as possible, and released it open-source for all.

Meet Command A+

Available now on Hugging Face and vLLM

103
361
113
2,599
745,529
Pierre Colombo retweeted
Manuel Faysse
@ManuelFaysseMay 15
🚨 Do LLMs need to store everything they read in memory? To reduce KV cache size and improve decoding speeds, we propose Self-Pruned KV attention, a mechanism where the model learns to decide which KVs to write in the persistent KV cache, discarding all the rest! @AIatMeta🧵
9
45
8
200
23,498
Load more