This programs aims to cultivate research interests and a good start for those who are starting their research journey 🐣

 #beginners-in-research

Monthly session. 3-5 guest speakers pitch research ideas in short talks and then spend an hour engaging in discussion with the community and/or exploring potential collaboration opportunities.

#mentorship

A program to bridge the gap in Frontier ML Research between the industry and academia!

#ml-industry

Leads: Alif, Sunitha

#safety-and-alignment

Group meets biweekly on Thursday

Leads: Mayank Bhaskar, Benedict Emoe-Kabu

#computer-vision on Discord

Group meets bi-weekly on Tuesdays at 11 AM EDT / 08:00 PST / 15:00 GMT / 17:00 CEST / 20:30 IST

#ml-efficiency on discord

We focus on systems and theory for efficient large scale deep learning. We invite guest speakers, organize reading groups and practical sessions.

Leads: Harsha, Bhavnick

Group meets biweekly on Wednesday at 

09:00 PST / 16:00 GMT / 18:00 CEST / 21:30 IST

#ml-math

Join us as we explore the world of mathematics together and dive into the fundamental math that drives the creation of machine learning algorithms. We welcome all skill levels to our page and cover concepts such as calculus, linear algebra, optimization and differential equations. See you there!

Leads: Katrina, Neel

Join us  bi- weekly on Monday at 12pm EST

#ml-theory

Leads: Andrej, Anier, Thang

We discuss theoretical research, with strong mathematical statements, making Machine Learning a more rigorous science.

Group meets biweekly on Thursday

at 18:00 CEST

#multimodal-ml

We explore the latest breakthroughs in models that understand text, images, audio, and video. From foundational concepts to advanced insights, join us through monthly newsletters, paper reading sessions, and hands-on code labs as we unpack the complexity behind multimodal intelligence.

Leads: Surya, Manasvi

Join Us Bi-Weekly on Saturday at 

10:00AM EST

We focus on the latest NLP papers and have many different assignments and discussions. We invite guest speakers, organize reading groups and practical sessions focusing on basics of nlp.

Leads: Siddhant, Vin

#nlp on Discord
Group meets bi-weekly on Saturday:

01:00 PM EDT / 10:30 AM PST / 03:00 PM GMT / 05:00 PM CEST / 10:30 PM IST

#reinforcement-learning
Weekly reading group on Saturdays

Our focus remains on engaging people in projects/papers and extensive collaboration in development of agent, benchmarks and many other papers.

We invite guest speakers, collaborate on projects and hold practical sessions.

Leads: Ram, Jebish, Siddhant

#ml-agents on Discord
Group meets bi-weekly on Saturday:

12:00 PM EDT / 09:00 AM PST / 02:00 PM GMT / 04:00 PM CEST / 09:30 PM IST

#ml-understanding
Dedicated to unraveling the inner workings of neural networks through the lens of mechanistic interpretability. We'll explore how features and circuits emerge, how techniques like probing, sparse autoencoders, and causal analysis help decode model behavior, and how these insights impact AI safety, model editing, and generation.

Weekly meetings, Friday 10 AM PT
Leads: Akanksha and Kabir

#geo-regional-africa 

Cohere Labs Geo Regional Africa is a vibrant community dedicated to advancing Artificial Intelligence across the African continent. 

#geo-regional-asia

Geo-Regional Asia hosts Guest Speakers from across Asia to share what they're working on and learn from each other.

 
Leads: Ahmad, Kanwal

 A chance to connect, share updates, celebrate successes, and welcome new members. Check out the slide decks from past socials here!

Last Friday of every month at 11:00 am ET
Zoom link

From Jan 2023 to Feb 2024 members across the C4AI Community and beyond collaborated to create Aya: a new, state-of-the-art, open-source multilingual dataset and model. Though this project has been de-prioritized since release, we are still accepting multilingual data submissions! Join the dedicated Aya Discord server and contribute your language!

Roads to Research is designed to demistify ML research, and showcase everything it takes to be a researcher, with a particular focus on elements of the process that are less often spotlighted.