Can diffusion models replace traditional LLMs to build lightning-fast AI agents? Researchers from Southeast University, Alibaba, and NTU Singapore just released a reality check on diffusion-based language models like LLaDA and Dream. They tested whether the speed of
@jiqizhixin
-
TextOp: Universal Cerebellum Transforms Streaming Text Into Robot Motion
By
–
Ever wonder why humanoid robots still feel like rigid puppets following a pre-recorded script? Professor Li Xuelong and the TeleAI team at China Telecom AI Research present TextOp to solve exactly that! They have built a universal cerebellum that turns streaming text into
-
TI-DPO: Gradient Attribution for Improved AI Model Training
By
–
What if your AI could ignore the noise and focus only on the words that actually matter for human preferences? Researchers from CASIA, ByteDance, and Microsoft Research introduce TI-DPO. This method moves beyond basic training by using gradient attribution to prioritize key
-
AI Agents Collaborating to Commit Social Media Fraud
By
–
What happens when AI agents start working together to scam people on social media? Researchers from Shanghai Jiao Tong University, Shanghai AI Lab, and Beihang University have introduced MultiAgentFraudBench to investigate this emerging threat. They simulated how groups of LLM
-
SiameseNorm: Training Efficiency Meets Model Performance
By
–
Why should AI developers have to choose between a model that trains easily and one that performs at the highest level? Researchers from Tsinghua University and Alibaba's Qwen team introduce SiameseNorm to solve this exact dilemma. The method uses a dual-path design that
-
GLM-5 Tech Report: Efficient Architecture Achieves State-of-the-Art Results
By
–
GLM-5 Tech Report is here! This new model from @Zai_org & @Tsinghua_Uni uses a more efficient architecture (DSA) to cut costs and a novel training system that lets the AI learn from complex, long-term tasks. It achieves state-of-the-art results on open benchmarks and
-
NVIDIA DreamZero Enables Robots Master Complex Tasks New Environments
By
–
What if robots could master complex tasks in totally new environments without needing thousands of repetitive demonstrations? NVIDIA presents DreamZero to do exactly that. The researchers built a World Action Model that learns physical dynamics by predicting future video frames
-
Chinese New Year: Wishing Researchers Breakthrough Success
By
–
Happy Year of the Horse! In Chinese, there is an idiom “馬到成功”, meaning success comes the moment the horse arrives. Wishing every researcher their own well-earned breakthroughs. We will also keep sharing more exciting research and AI stories from China.
-
How AI Agents Behave During Deep Search Analysis
By
–
Ever wonder how AI agents actually behave when they go on a deep-dive search for you? Researchers from Carnegie Mellon University, University of Lisbon, and NOVA University Lisbon analyzed 14 million real-world search requests to find out. They mapped the secret life of AI
-
LLMs Transform Data Cleaning From Manual to Automated Preparation
By
–
Could LLMs finally end the nightmare of manual data cleaning? Researchers from SJTU, Tsinghua, Microsoft Research, MIT, and Alibaba present a comprehensive survey on the future of application-ready data preparation. They detail a massive paradigm shift from rigid, rule-based