WinBuzzer

AI Coding: Microsoft’s 7B X-Coder Outperforms 14B Rivals on Synthetic Data

Microsoft and Tsinghua University have developed a 7B-parameter AI coding model that outperforms 14B rivals using only ...

MemRL outperforms RAG on complex agent benchmarks without fine-tuning

MemRL separates stable reasoning from dynamic memory, giving AI agents continual learning abilities without model fine-tuning ...
EurekAlert!

RDHNet: Addressing rotational and permutational symmetries in continuous multi-agent systems

"Welcome to the world of RDHNet, a groundbreaking approach to multi-agent reinforcement learning (MARL) introduced by Dongzi Wang and colleagues from the College of Computer Science at the National ...
IEEE

Survey on Large Language Model-Enhanced Reinforcement Learning: Concept, Taxonomy, and Methods

Abstract: With extensive pretrained knowledge and high-level general capabilities, large language models (LLMs) emerge as a promising avenue to augment reinforcement learning (RL) in aspects, such as ...
Interesting Engineering

US researchers build fall-safe biped robots to advance real-world reinforcement learning

HybridLeg robots Olaf and Snogie use impact-safe design and self-recovery to enable scalable, real-world hardware ...

Why reinforcement learning plateaus without representation depth (and other key takeaways from NeurIPS 2025)

Why reinforcement learning plateaus without representation depth (and other key takeaways from NeurIPS 2025) ...

Dopamine under control: Precision regulation of inhibition shapes learning, memory and mental health

For decades, dopamine has been celebrated in neuroscience as the quintessential "reward molecule"—a chemical herald of ...
EurekAlert!

Multi-constraint reinforcement learning in complex robot environments

FPMCO decomposes multi-constraint RL into KL-projection sub-problems, achieving higher reward with lower computing than second-order rivals on the new SCIG robotics benchmark.
ZDNet

True agentic AI is years away - here's why and how we get there

Today's AI agents don't meet the definition of true agents. Key missing elements are reinforcement learning and complex memory. It will take at least five years to get AI agents where they need to be.
Microsoft

Agent Lightning: Adding reinforcement learning to AI agents without code rewrites

AI agents are reshaping software development, from writing code to carrying out complex instructions. Yet LLM-based agents are prone to errors and often perform poorly on complicated, multi-step tasks ...
Scientific American

Learning Another Language May Slow Brain Aging, Huge New Study Finds

Speaking multiple languages could slow down brain ageing and help to prevent cognitive decline, a study of more than 80,000 people has found. The work, published in Nature Aging on 10 November, ...