3. LatentMAS LatentMAS introduces a framework enabling language model agents to collaborate directly within a continuous latent space rather than relying on text-based communication.
RESEARCH
-
INTELLECT-3: 106B MoE Model Achieves State-of-the-Art Performance
By
–
1. INTELLECT-3
— DAIR.AI (@dair_ai) 30 novembre 2025
A 106B-parameter Mixture-of-Experts model (12B active) trained with large-scale reinforcement learning, achieving state-of-the-art performance for its size across math, code, science, and reasoning benchmarks.https://t.co/8tphkxnVzy1. INTELLECT-3 A 106B-parameter Mixture-of-Experts model (12B active) trained with large-scale reinforcement learning, achieving state-of-the-art performance for its size across math, code, science, and reasoning benchmarks.
-
Top AI Papers of the Week: Latest Research Highlights
By
–
Top AI Papers of the Week (Nov 24 – 30): – LatentMAS
– INTELLECT-3
– OmniScientist
– Lightweight End-to-End OCR
– Evolution Strategies at Hyperscale
– Training LLMs with Reasoning Traces
– Cognitive Foundations for Reasoning in LLMs Read on for more: -
Building Strong AI Communities Over Wasteful Alternatives
By
–
I encourage strong community The rest are a waste of time
-

Swarm Intelligence: The Collective Learning Secret Revealed
By
–
Swarm intelligence has a secret. The standard approach treats swarms as collections of independent learners. Each agent makes decisions, learns from outcomes, and coordination somehow emerges. More agents, more complexity, more mystery. But what if the swarm itself is the
-
Memory in AI Agents: The Evolution Beyond RAG
By
–
RAG was never the end goal.
— Akshay 🚀 (@akshay_pachaar) 30 novembre 2025
Memory in AI agents is where everything is heading. Let me break down this evolution in the simplest way possible.
RAG (2020-2023):
– Retrieve info once, generate response
– No decision-making, just fetch and answer
– Problem: Often retrieves… pic.twitter.com/EN8cfP6qflRAG was never the end goal. Memory in AI agents is where everything is heading. Let me break down this evolution in the simplest way possible. RAG (2020-2023): – Retrieve info once, generate response
– No decision-making, just fetch and answer
– Problem: Often retrieves -
LeCun: LLMs Not a Bubble but Limited Alone
By
–
Yann LeCun affirme que les LLM ne sont pas une bulle, ni en termes de valeur ni d’investissement : ils vont faire tourner énormément d’applications utiles et justifier les infrastructures massives qu’on construit aujourd’hui.
— VISION IA (@vision_ia) 30 novembre 2025
Selon lui, la vraie bulle, c’est de croire que les… pic.twitter.com/zNTE3ONXisYann LeCun asserts that LLMs are not a bubble, neither in terms of value nor investment: they will power a vast array of useful applications and justify the massive infrastructures we're building today. According to him, the real bubble is the belief that LLMs, on their own,
-

NVIDIA NVQLink: Connecting Quantum Computing and GPU
By
–
NVIDIA Introduces NVQLink — Connecting #Quantum and #GPU Computing for 17 Quantum Builders and Nine Scientific Labs https://nvidianews.nvidia.com/news/nvidia-nvqlink-quantum-gpu-computing
… @nvidia
#Quantum #CES2026 #AI #IoT #GTC -

VQ-Insight: Progressive Learning for AI Video Quality Evaluation
By
–
How do we fix the messy state of AI video quality evaluation? Enter VQ-Insight: it uses progressive learning (starting with image quality, then adding temporal skills, and linking directly to generation models) plus multi-layered rewards to judge everything from frame
-

Top Machine Learning and Deep Learning Researchers to Follow
By
–
Top #MachineLearning, #DeepLearning Researchers to Follow OutSide and On Twitter! by @MirrorReview1
! #BigData #Analytics #DataScience #AI #MachineLearning #IoT #IIoT #PyTorch #Python #RStats #TensorFlow #Java #JavaScript #ReactJS #GoLang #CloudComputing #Serverless
