Come for the trade, stay for the dazzling 60-minute education on the state of AI. ( h/t @snowmaker )
RESEARCH
-
Visual Guide to Reinforcement Learning Concepts and Multi-Agent Training
By
–
The best visual explanation of Reinforcement Learning (RL) I've ever seen! 😂
— Pascal Bornet (@pascal_bornet) 27 janvier 2025
It illustrates many key RL concepts like:
– Exploration and exploitation
– Imitation learning
– Reward mechanisms
– Distributed training
– Multi-agent cooperation
It's like watching multiple trials… pic.twitter.com/wNp4aGNdxuThe best visual explanation of Reinforcement Learning (RL) I've ever seen! It illustrates many key RL concepts like:
– Exploration and exploitation
– Imitation learning
– Reward mechanisms
– Distributed training
– Multi-agent cooperation It's like watching multiple trials -

Deepseek Vindicated: China’s AI Engineering Supremacy Emerges
By
–
In my book AI Superpowers, I predicted that US will lead breakthroughs, but China will be better and faster in engineering. Many people simplified that to be “China will beat US”. And many claimed I was wrong with GenAI. With the recent Deepseek releases, I feel vindicated.
-
Supervised Fine-Tuning vs Reinforcement Learning: A Learning Analogy
By
–
“Imagine teaching a child to ride a bike. You could give them a detailed manual (Supervised Fine Tuning), but they'll likely learn better by trying it themselves (Reinforcement Learning), falling, getting up, & gradually improving.” – @McDonaghMatthew ELI5 on DeepSeek, link
-
Data Mix Specification Impact on Model Performance
By
–
unclear tbh, data mix is not specified. probably is the big reason for performance
-

CNRS criticized for poor internet foresight in 1998
By
–
Le scandale LUCIE n’est pas isolé @H_liberty2 nous rappelle à quel point le @CNRS a été rempli de bras cassés en informatique En 1998 le CNRS prévoyait qu’Internet était sans avenir Et que le Fax resterait le cœur des échanges écrits Le CNRS devrait s’excuser
-

Baichuan-M1-14B: Medical AI Model for Healthcare Evidence
By
–
It has unlocked a medical evidence-based model, enabling a complete end-to-end service from medical evidence retrieval to deep reasoning. To better promote AI healthcare ecosystem, we also launched the industry's first open-source medical enhancement large model, Baichuan-M1-14B.
-

China Launches Baichuan-M1 Deep Thinking Model Surpassing o1
By
–
We released China's first full-scenario deep thinking model, Baichuan-M1-preview. It’s currently the only one in China that simultaneously possesses reasoning capabilities in 3 areas: language, vision, and search. It has surpassed o1-preview in multiple authoritative evaluations.
-
France faces technological setback in artificial intelligence race
By
–
Intelligence artificielle : « La France est à deux doigts de la déroute » Merci à @Challenges pour ce long entretien où j’évoque le décrochage technologique de la France et de l’Europe, et les (dernières) chances de sursaut.
