Zhemal Khamidun
Hamidun News editorial desk led by Editor-in-Chief Zhemal Khamidun (CPO of AlpinaGPT). Daily curation across 50+ global AI labs, research feeds, and industry reports in 6 languages.
LumiBot Raises Tens of Millions of Yuan to Build Dexterous Robot Hands With Visuo-Tactile Sensors
Chinese startup LumiBot, founded by a former Wall Street investor, has closed an angel round of tens of millions of yuan to build dexterous robot hands with its own…
Ant Group Takes Stake in AI Robotics Startup Lexiang Intelligent
An Ant Group investment arm has joined the shareholders of Lexiang Intelligent, a young Chinese startup building intelligent robots and AI applications.
Yuewen Unveils Buddy AI Agents for Writers, Comic Dramas and Copyright Work
China Literature (Yuewen) has released three vertical Buddy AI agents for writing books, adapting them into comic dramas and handling copyright.
AI Decodes the Meaning of Speech From Brain Activity and Transfers to New People Without Retraining
Prefix-GRPO: Training Small Language Model Agents on Trajectory Prefixes
OOD Benchmark Leak: How a Contaminated Class Dragged a Detector's AUROC Down to 0.326
Graph Neural Networks vs. Gradient Boosting: Predicting Arsenic in US Water
mLSTM and Orthogonalized Reads: The Recall Gain Is Trainability, Not Memory
An arXiv preprint finds that orthogonalizing the mLSTM memory matrix (five Newton–Schulz iterations) improves trainability, not data storage, and the effect can be…
GLEAM: Spectral Losses for Turbulence Forecasting on Unstructured Meshes
Researchers extended spectral loss functions to unstructured meshes by replacing Fourier bands with graph Laplacian bands. Their GLEAM method forecasts turbulent…
SAE Interpretability Evaluation: The Measurement Pipeline Matters More Than the Model Architecture
FNO and UNO Neural Operators for Calculating Two-Dimensional Neutron Flux
SUM: A Server-Side Fix for Forgetting in Federated Continual Learning, Up to 22% Gain
STN-TGAT: A Graph Neural Network for Ranking S&P 500 Stocks and Building Portfolios
ECoM Reasoning: Compressed Reasoning Speeds Up Voice Models on Spoken Math
Researchers unveil ECoM Reasoning, a compressed-reasoning method that raises spoken-math accuracy of speech language models by 21% while using only 40% of the text…
Can Transformers Select a Model? An arXiv Bayesian Test at 2.8M Parameters
Researchers tested whether transformers can do more than filter data within a single hypothesis class and actually select the class itself. They found a sharp…