Onze wereld verbinden 🌍 We zijn trots om ✨Aya Expanse✨ te presenteren Probeer het uit via SMS in Nederland: (+31) 97006520757 of wereldwijd op WhatsApp : +1 431-302-8498 Onze WhatsApp-integratie omvat nu de beeldinterpretatiemogelijkheden van Aya Vision! Probeer het nu door een afbeelding op te nemen in je volgende tekst naar Aya: https://lnkd.in/gWBGtgwn
Cohere For AI
Technology, Information and Internet
We’re Cohere's research lab and community exploring the unknown, together.
About us
Who we are Cohere For AI is Cohere's non-profit research lab that seeks to solve complex machine learning problems. We support fundamental research that explores the unknown, and are focused on creating more points of entry into machine learning research. Our community is a space where researchers, engineers, linguists, social scientists and lifelong learners connect and collaborate with each other. We come together from all over the world and welcome you whether you are a mentor, dropout, just getting started, PhD, masters, undergraduate, unaffiliated, industry, academic or not really sure. We are excited to support community-driven research and to be shaped by our members' interests. Where we’ve come from In 2017, a team of friends, classmates, and engineers started a distributed research collaboration, with a focus on creating a medium for early-career AI enthusiasts to engage with experienced researchers – they called it “for.ai.” Two of those co-founding members, Aidan Gomez and Ivan Zhang, later went on to co-found Cohere, and many of the founding members went on to do exciting things (pursuing PhDs, working at industry and academic labs). At the time, For AI was one of the first community-driven research groups to support independent researchers around the world. Today, Cohere is proud to reintroduce For AI as Cohere For AI, a dedicated research lab and community for exploring the unknown, together.
- Website
-
https://cohere.com/research
External link for Cohere For AI
- Industry
- Technology, Information and Internet
- Company size
- 11-50 employees
- Founded
- 2022
- Specialties
- research, machine learning, and open science
Updates
-
The C4AI Research Grant program supports a variety of research and ML for good projects worldwide. Today we’re highlighting Understanding stereotypes in LLMs in Argentina, led by Luciana Benotti and Guido Ivetta ✨ This team developed a tool and methodology to help social science experts in Latin America analyze biases in AI language models, overcoming technical barriers. They are now working with Argentinian secondary school teachers and students to build community-driven datasets that capture local stereotypes, aiding in the detection of discriminatory behaviors and hate speech. With support from the Cohere For AI research grant, they have integrated Cohere’s models into their tool and benchmark them against other open-source models to assess stereotype biases in language models. Read more about this project: https://lnkd.in/epfFtPiM Apply for a research API grant: https://lnkd.in/gawdasUq
-
-
We're excited to welcome Nataniel Ruiz, Research Scientist at Google, for a session on "Virtual Generative Worlds: From Camera Control to Unbounded Character Simulation" - be sure to join us on March 19th! Thanks to Ahmad Anis, our Regional Asia community lead for organizing this event ✨ Learn more: https://lnkd.in/gBse-_NJ
-
-
We’re excited to bring back Expedition Aya 🌍. A 6-week open build challenge to facilitate new collaborations and research projects worldwide. Over the 6 weeks, our goal is to inspire and support each other in launching innovative projects in global teams. We are excited for people to connect, exchange ideas, and push the boundaries of AI research. Lead your own research idea, or join a team of researchers worldwide. Sign-ups are now open! By participating in Expedition Aya, you will: 🌍 connect ML researchers worldwide, 💻 gain access to API credits, 🤝 get support to launch your research ideas, ✨ be eligible for limited edition Expedition Aya swag Interested in joining the expedition, but don’t have a team? We’re hosting three social connections events to help you team-up with like-minded researchers. Our Discord server is another great place to share ideas and connect with others. Join us in expanding the world that AI sees. Learn more here: https://lnkd.in/gnNkEDgS Join us on Discord: https://lnkd.in/gxaj5GBb
-
Join our Reinforcement Learning group next week on Tuesday, March 18th for a session with James Staley on "Gains on all Trajectories: Demonstration-agnostic learning with model-based RL." Thanks to Rahul Narava and Gusti Triandi Winata for organizing this event 👏 Learn more: https://lnkd.in/ghZKHyae
-
-
Our grants program enables researchers worldwide to access our models through the Cohere API. 🌎 Among these, is our latest state-of-art multimodal, multilingual model - Aya Vision. ✨ Apply now: https://lnkd.in/gW_bUHjx
Today, we release Aya Vision. Aya Vision 🌿 sets a new frontier for multilingual understanding – consistently outperforms other leading open-weights multimodal models in image understanding tasks while preserving text multilingual performance 🌍. **Aya 🐭 outperforms much larger models. 🐘** Aya 8B outperforms Llama 90B, a model **11x** the size. Aya 32B outperforms Qwen 72B, Llama 90B and Molmo 72B, models which are **2x** the size. As part of our commitment to access, we release both Aya 8B and 32B open-weights. Access Aya Vision open-weights on Hugging Face and Kaggle. We are also making it available for free on WhatsApp: https://lnkd.in/gUppEEtf Find more information in our blog: https://lnkd.in/gpgCZzqU Connect with our wider research efforts: https://lnkd.in/ebfdwzcn
-
-
Conectando nuestro mundo 🌎 Presentamos orgullosamente ✨Aya Vision ✨ Pruébalo en WhatsApp +14313028498 ¡Nuestra integración de WhatsApp ahora incluye las capacidades de interpretación de imágenes de Aya Vision! Pruébalo ahora incluyendo una imagen en tu próximo mensaje de texto para Aya:https://lnkd.in/gWBGtgwn ¡Gracias a Alejandro Salamanca por compartir!
-
-
What is the real-world impact of personalized preference learning methods? We introduce a multi-faceted evaluation framework for personalization algorithms that goes beyond accuracy. We examine different personalized preference learning methods on ✅ Performance, ⚖️ Fairness, ⚠️ Unintended Effects, and 🔄 Adaptability. We find that fine-tuning individual reward models is a strong baseline. Collaborative learning methods, such as Personalized Reward Modeling, achieve up to 6% gain over the baseline, with an additional advantage of better adaptability to new users. However, we find that personalization can lead to misalignment and up to a 20% decline in chat, reasoning, and safety benchmarks measured by Reward Bench. These findings highlight the critical need for holistic evaluation approaches to advance the development of more effective and inclusive preference learning systems. 📜 Paper link: https://lnkd.in/gRgHNP2y Great collaboration with researchers from University of Cambridge Language Technology Lab and Cohere For AI, led by: Yijiang (River) Dong, Tiancheng Hu, Yinhong Liu, Ahmet Üstün, Nigel Collier.
-
-
How did we add multimodal capabilities to Aya without sacrificing multilingual performance? 👁️💬 The Aya Vision team, including Ahmet Üstün, Sara Hooker, Saurabh Dash, Beyza Ermis and Yiyang Nan, discuss the challenges & breakthroughs behind this best-in-class research model. https://lnkd.in/gnbvifn7
Aya Vision - The Challenges & Breakthroughs
https://www.youtube.com/
-
Our Regional Africa group is looking forward to hosting Ahmad Rufai Yusuf for a session on "Leveraging LLMs for Advanced Drug Form Classification in MARMAR" on March 18th. Thanks to Kato Steven and Bronson Bakunga for organizing this event 🤗 Learn more: https://lnkd.in/gzPHtwpX
-