Cohere For AI
Cohere For AI is Cohere's 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.
Cohere For AI
Cohere For AI is Cohere's 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.
Fundamental research lab
We work at the frontier of AI progress with the goal of solving cutting edge scientific problems. We see contributions to traditional conferences and publications in journals as an important part of our work, but also support efforts that go “beyond the research paper” and encourage scientific communication through different mediums. We drive the creation of new research spaces and breakthroughs that changes where, how and by whom research is done. We believe that technology is powerful, and empowering different perspectives ensures responsible innovation.
Open Science Initiative
We’re not just another research group. We are a hybrid lab with both a dedicated research staff and support for open science initiatives. We collaborate openly with independent researchers all over the world to conduct top-tier ML research.
Our open science research community is a space where researchers, engineers, linguists, social scientists, and lifelong learners connect and collaborate with each other. We come together from over 100 countries around the world and support large and small scale research collaborations.
Our models
State of the Art, Accessible Research LLM
Aya Expanse - 8B
State of the Art Research LLM
Aya Expanse - 32B
Massively Multilingual Research LLM
Aya
MODEL WEIGHTS FOR DEMOCRATIZING RESEARCH ACCESS
C4AI Command R - 104B
MODEL WEIGHTS FOR DEMOCRATIZING RESEARCH ACCESS
C4AI Command R - 35B
Our papers
Mix Data or Merge Models? Optimizing for Diverse Multi-Task Learning
In this work, we explore model merging in a diverse multi-task setting, combining safety and general-purpose tasks within a multilingual context. Overall, our comprehensive study of merging approaches provides a useful framework for building strong and safe multilingual models.
Nexus: Specialization meets Adaptability for Efficiently Training Mixture of Experts
In our latest work, we ask “How to make Mixture-of-Expert models more efficient, specialized, and adaptable at once?” We introduce Nexus, which enables specialization and adaptability within the efficient upcycling framework.
Multilingual Arbitrage: Optimizing Data Pools to Accelerate Multilingual Progress
Can you surpass individual model performance by sampling parts of the distribution strategically from a pool of models?
We introduce “multilingual arbitrage” to describe capitalizing on performance variations to produce large gains in performance.
To Code, or Not To Code? Exploring Impact of Code in Pre-training
Including code in the pre-training data mixture, even for models not specifically designed for code, has become a common practice in LLMs pre-training. e ask "what is the impact of code data used in pre-training on a large variety of downstream tasks beyond code generation".
Consent in Crisis: The Rapid Decline of the AI Data Commons
General-purpose AI systems are built on massive swathes of public web data, assembled into corpora such as C4, RefinedWeb, and Dolma. To our knowledge, we conduct the first, large-scale, longitudinal audit of the consent protocols for the web domains underlying AI training corpora.
How Does Quantization Affect Multilingual LLMs?
Quantization techniques are widely used to improve inference speed and deployment of large language models. While a wide body of work examines the impact of quantized LLMs on English tasks, none have examined the effect
of quantization across languages.
RLHF Can Speak Many Languages: Unlocking Multilingual Preference Optimization for LLMs
We introduce a novel, scalable method for generating high-quality multilingual feedback data to balance data coverage. We establish the benefits of cross-lingual transfer and increased dataset size in preference training.
LLM See, LLM Do: Guiding Data Generation to Target Non-Differentiable Objectives
Our work exhaustively characterizes the impact of passive inheritance of model properties by systematically studying the consequences of synthetic data integration.
The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harm
We explore the viability of different alignment approaches when balancing dual objectives: addressing and optimizing for a non-homogeneous set of languages and cultural preferences while minimizing both global and local harms.
Our programs
Advancing the NLP space through our programs.
ACCELERATING MULTILINGUAL AI THROUGH OPEN SCIENCE
Introducing Aya
About
Aya is a global initiative led by Cohere For AI to advance the state-of-art in multilingual AI and bridge gaps between people and cultures across the world. An open science project to create new models and datasets that expand the number of languages covered by AI, Aya involves over 3,000 independent researchers across 119 countries.
Exploring the unknown together
Scholars program
About
Our Scholars Program provides the opportunity to work alongside some of the best research and engineering experts in the world. We have created an open and supportive environment that provides an alternative point of entry into machine learning research.
academic support
Research grant
Benefits
Cohere For AI research grants are designed to support academic partners who are conducting research with the goal of releasing a peer-reviewed scientific artifact. Our program provides academic partners, developers, researchers, and other members of our community with subsidized access to the Cohere API.
ACCELERATING MULTILINGUAL AI THROUGH OPEN SCIENCE
Introducing Aya
About
Aya is a global initiative led by Cohere For AI to advance the state-of-art in multilingual AI and bridge gaps between people and cultures across the world. An open science project to create new models and datasets that expand the number of languages covered by AI, Aya involves over 3,000 independent researchers across 119 countries.
Exploring the unknown together
Scholars program
About
Our Scholars Program provides the opportunity to work alongside some of the best research and engineering experts in the world. We have created an open and supportive environment that provides an alternative point of entry into machine learning research.
academic support
Research grant
Benefits
Cohere For AI research grants are designed to support academic partners who are conducting research with the goal of releasing a peer-reviewed scientific artifact. Our program provides academic partners, developers, researchers, and other members of our community with subsidized access to the Cohere API.
Past events and videos
Research is inherently a human endeavor, and our event series provide insights from beginning to breakthrough.
Video
Cong Lu: The AI Scientist
Video
Fireside Chat: Max Welling
Video
C4AI Expedition Aya - Closing Ceremony
Video
AI & Technical Governance: Saffron Huang and Tina M. Park, PhD
Video
Panayiotis Panayiotou: Curricula for Learning Robust Policies...
Video
Arthur Conmy: Mechanistic Interpretability Research Frontiers
Meet our research team
Our staff brings together machine learning experts to contribute to progress in machine learning through fundamental research. We are committed to open collaboration, and empowering more points of entry into machine learning research through our scholars program.
Sara hooker
head, Cohere for ai
Marzieh Fadaee
Senior Research Scientist
Julia Kreutzer
SENIOR RESEARCH SCIENTIST
Ahmet Üstün
Senior Research Scientist
Beyza Ermis
Senior Research Scientist
Madeline Smith
Operations and Community Lead
Aidan Peppin
Policy & Responsible AI Lead
Brittawnya Prince
Operations Associate
Arielle Salman Bailey
Operations Associate
Saurabh Dash
Research Engineer
Daniel D'souza
Research Engineer
Alejandro Salamanca
Open Science Research Engineer
Shivalika Singh
Open Science Research Engineer
Aakanksha
Research Scholar
Viraat Aryabumi
Research Scholar
John Dang
Research Scholar
Oliver Nan
Research Scholar
Luísa Shimabucoro
Research Scholar
Arash Ahmadian Dehkordi
Research Scholar
Frequently Asked Questions
What’s C4AI’s origin story?
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 original 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. Watch the C4AI history video here.
Do you charge for your educational programs or community membership?
We do not charge for participating in any of our programs, and are committed to supporting educational outreach programs, which include compute resources and infrastructure needed to participate in machine learning research.
are you hiring for research positions or interns?
Our full list of positions are listed here.
How can I stay in touch?
To stay up to date on upcoming talks, sign up for our mailing list.
You can also apply to join our open science community or follow us on LinkedIn and Twitter.
What is Aya?
Aya is a state-of-the-art, open source, massively multilingual research LLM covering 101 languages – including more than 50 previously underserved languages. Learn more here.
Join our open science community