Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Researchers from Tencent, Tsinghua University, and HKUST developed FlashMemory-DeepSeek-V4, which uses Lookahead Sparse Attention to reduce memory consumption in large language models by predicting and storing only relev…
Researchers applied machine learning to whole-genome sequences of 299 Borrelia burgdorferi clinical isolates, identifying protein variants in five adhesins (BBK32, DbpA, OspC, P66, RevA) strongly associated with human di…
A developer post-trained a language model to roll a die, revealing that models default to outputting 4 due to training data bias. Standard reinforcement learning fails to encourage exploration, but adding an entropy bonu…
By 2026, AI workloads will consume roughly 40% of enterprise Kubernetes clusters, but the default scheduler is ill-suited for GPU-intensive tasks, leading to 30-45% GPU utilization rates and wasted costs. Platform teams …
A developer built a self-improvement loop for their OpenClaw agent that runs nightly at 2 AM, enabling the agent to learn from execution logs without human intervention. The system separates the executor from the critic,…
A BSIT student passionate about software engineering and machine learning began a consistent problem-solving journey on LeetCode, starting with the 'Merge Strings Alternately' problem. The student learned to traverse two…
Roboflow released a tutorial for building an automated surface defect inspection system for machined metal medical parts using an RF-DETR Small model trained on a metal surface defect dataset, achieving 66.5% mAP@50 and …
More than 200 civil-society organizations, led by Access Now and Amnesty International, are demanding an immediate ban on AI in military kill chains after reports revealed that Palantir's Maven Smart System, integrated w…
A developer built a chess bot using a transformer architecture with 11 million parameters, achieving a 1500 Elo rating from the model alone and 2100 Elo when combined with Monte Carlo Tree Search. The bot is available as…
Researchers used machine learning to re-examine old seismic data from Antarctica, discovering hundreds of previously unknown intermediate-depth earthquakes far from plate boundaries. The findings, published in Science, c…
Perion CEO Tal Jacobson argues that chief marketing officers lack credible answers to CFOs' demands for proof of advertising ROI due to fragmented marketing technology infrastructure that measures activity rather than bu…
The PyTorch Foundation has opened nominations for its 2026 Contributor Awards, recognizing individuals who strengthen projects like PyTorch, vLLM, DeepSpeed, Ray, Helion, and Safetensors through technical innovation, doc…
AI startup Aaru, cofounded by 21-year-old Ned Koh, accurately simulated nearly 2 million New York voters in the city's primary mayoral election, coming within 2,000 votes of the final count. The company uses AI agents bu…
A developer built a Naive Bayes spam filter from scratch using only word counts and Bayes' rule, without gradient descent or iterations. The filter visualizes which words push a message toward spam or ham, and an interac…
The Allen Institute for AI released MolmoMotion, a language-guided 3D motion forecasting model that predicts future 3D point trajectories of objects based on video frames and action descriptions. Alongside the model, the…
A new paper-thin, flexible AI sticker sensor node called Ghost Edge has been developed, built around the Nordic Semiconductor nRF54L15 wireless SoC. The device uses a flexible printed circuit board to conform to curved s…
A developer evaluated DeepSeek V4 Flash and Gemini 2.0 Pro for a high-volume ranking system processing 12 million inference calls daily, finding DeepSeek V4 Flash achieved a p99 latency of 1.18 seconds within a 2.4-secon…
Nvidia, Carnegie Mellon University, and UC Berkeley researchers developed ENPIRE, a system where AI coding agents autonomously train robots in dexterous grasping tasks. Using a fleet of eight robots that share results vi…
Maker Sritabh Priyadarshi built MiniSoul, a pocket-sized desktop companion robot that uses an ESP32-S3 SuperMini and a 0.96-inch OLED to display evolving emotions. The device runs a custom behavior engine tracking six pe…
An engineer on Oracle Cloud Infrastructure (OCI) built a scale-to-zero AI inference system on Oracle Kubernetes Engine (OKE) using KEDA to eliminate costs from idle GPUs. The system scales GPU pods down to zero when ther…