Yohann trained @AIatMeta’s TRIBE v2, a trimodal model (video, audio and language) to predict human brain activity based on 1,000 hours of fMRI data across 720 subjects.
Read his paper here: https://t.co/L7JRlyeAdJ
It's coming home... well, I am at least. Back in Paris next week for @RaiseSummit + @MACHINASUMMIT.
Coming with @joinaicircle and hosting a few dinners.
Also recording Field Notes while I'm there - researchers in world models, computer vision, robotics, open source + small models, reach out!
Welcome Dinner with Centific: https://t.co/JkvL3Vg8sZ
Roundtable dinner on AI Benchmarking Across Languages with Lilt: https://t.co/0YkyL4BDUe
Happy Hour for members and friends: https://t.co/hzUV4h5jo1
Best comic app: https://t.co/Pu8YQ75XSX
AI Circle is officially #londonmaxxing 🇬🇧
Riding the #LondonTechWeek hype with the G2i x @nvidia Boat Party - the AI frontier, on the Thames.
RSVP below before we're full - limited spots!
In academia we learn math and algorithms that teach us how to best model the world, but we rarely learn how to build a production grade harness that can handle 200 QPS without race conditions lol.
Building agentic systems is a software engineering problem as much as a Research one. We need to start treating our training environments with the same rigor we treat production -- seeing enough RL envs get thrown out on this is why I wrote this series :)
London views, a boat party, and a discussion on post-training. The dream?
Closing out @LDNTechWeek on a boat with @joinaicircle, @g2i_co and @nvidia next week.
We will be "navigating the frontier of AI" (yes, we like a good play on words) with
@talfanevans (@cursive_ai)
Eujeong Choi (@MistralAI)
@matteosarcinell (@nvidia)
and more, then sunset drinks.
Researchers, this one is for you. Sign up -> https://t.co/ExaFWVzN8F
Researchers from @Meta, @cohere, @MIT, founder of @world_lica, and VP of Product @contralabs_ai shared how model divergence provides meaningful signal during post-training in frontier creative ai.