🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

I am excited to attend ACL2024 to present our proud paper "Pride and Prejudice: LLM Amplifies Self-Bias in Self-Refinement" (We received scores of 5, 4.5, 4 and a meta-review of 5🤲). I am looking for full-time industry positions. Here is a list of my first-author publications

Super proud of the work of José Maria Pombal in Tower! He just started his PhD on MT and MT evaluation along with me and Andre Martins! Give him a follow to get updates on what’s to come!

Really cool work from Antonio Farinhas Haau-Sing Li 李 效丞 and Andre Martins which set some theoretical ground to explain why MBR and spending more during inference actually improves quality.

The good news continue 🥳 We are delighted that Tower has been selected as a oral spotlight presentation at Conference on Language Modeling (top 2% of submissions)! Check out our paper here: openreview.net/pdf?id=EHPns3h…

We are releasing the first 🇪🇺LLM’s! New models are coming (9B and 22B) as well as strong instruct models! Stay tuned!

Researchers from industry and academia have released EuroLLM, the first #LLM that supports all #EU languages 🇪🇺 and several additional languages 🌍 and excels 🚀 at machine #translation. slator.com/first-large-la… Pedro Martins Manuel Faysse Andre Martins Ricardo Rei

Me and Nuno M. Guerreiro will be giving a keynote at WMT this Friday at 14:00 about why research on MT is still fascinating in the era of LLMs! Don’t miss it!

In the morning, we will also present our work on Tower v2 which ranked first in 8/11 language pairs on WMT 2024 General shared task! "Tower-v2: Unbabel-IST 2024 Submission for the General MT Shared Task", WMT, Nov 15, 11:00-12:00

Super excited to announce the launch of Widn.ai! 🎉 This results from several years of work in Translation Evaluation and the application of that knowledge to LLM research, leading to the creation of Tower. Now, Tower is accessible and ready for anyone!

💻 Thank you Slator for inviting Unbabel to speak on the current state of AI in translations. Key takeaways: 🔶 LLMs will continue to outperform NMT 🔶 LLM research will become evaluation centric 🔶 MT 2.0 will focus more on persoanlization and transcreation

Extremely happy with the release of EuroLLM 9B and with its results! This model is truly Multilingual, supporting 35 languages and covering all European Union official languages! Follow the 🧵 below for more details..👇👇

3) We have another poster in the main conference, “QUEST: Quality-Aware Metropolis-Hastings Sampling for Machine Translation” with Gonçalo Faria Sweta Agrawal Antonio Farinhas Ricardo Rei José G. C. de Souza (Thursday Dec 12 16:30-19:30).

I'm also happy to talk about widn.ai and #EuroLLM, feel free to DM me! 🚀

EuroLLM-9B is now ranking as the best LLM of its size (tied with Gemma) on the European LLM Leaderboard! Check it here: huggingface.co/spaces/openGPT…

Good to see European Commission promoting OS LLMs in Europe. However (1) "OpenEuroLLM" is appropriating a name (#EuroLLM) which already exists, (2) it is certainly *not* the "first family of open-source LLMs covering all EU languages" 🧵

🚨New machine translation dataset alert! 🚨We expanded the language coverage of WMT24 from 9 to 55 en->xx language pairs by collecting new reference translations for 46 languages in a dataset called WMT24++ Paper: arxiv.org/abs/2502.12404… Data: huggingface.co/datasets/googl…

🇪🇺 One month after the AI Action Summit 2025 in Paris, I am thrilled to announce EuroBERT, a family of multilingual encoder exhibiting the strongest multilingual performance for task such as retrieval, classification and regression over 15 languages, mathematics and code. ⬇️ 1/6

🧵 (3/7) 🌐 EuroBERT is open-source: 👉 Models (210M, 610M, 2.1B params) 👉 Training snapshots 👉 Full training framework Explore here: [huggingface.co/EuroBERT](huggingface.co/EuroBERT) Code coming soon! [github.com/Nicolas-BZRD/E…](github.com/Nicolas-BZRD/E…)

Happy to share our new work: ZSB is a framework to create benchmarks for any task and automatically evaluate any LLM. We show that it correlates highly with human evaluation on some of the most used benchmarks like chatbot arena

We just released M-Prometheus, a suite of strong open multilingual LLM judges at 3B, 7B, and 14B parameters! Check out the models and training data on Huggingface: huggingface.co/collections/Un… and our paper: arxiv.org/abs/2504.04953