Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Waymo posted an on-site Machine Learning Engineer, Perception role in Mountain View and San Francisco, CA, paying $175,000–$215,000 per year plus equity and bonus. The listing, verified live today, requires 3+ years of m…
DeepSeek's V4.1-Flash, a 552-billion-parameter MoE model, cuts KV cache memory from 389,120 bytes per token in the company's first 2023 model to 890 bytes, a 437-fold reduction, through techniques including a causal enco…
Taylor Otwell announced that Jev support landed in the 1.x branch of the Laravel AI SDK, and the There There helpdesk team began using it the same day for spam detection. Jev, made by TypeSafe, is a "System One" model th…
Gargi Agrawalla, a profoundly deaf graduate of Loughborough University, won the 2026 UK James Dyson Award on September 18, 2026, for a wearable AI device that converts emergency sounds such as fire alarms, doorbells, and…
Tether's AI Research group released TranslatePsy-AfriSLM, a collection of open-source machine translation models that the company says outperform Google's TranslateGemma-27B and Alibaba's Qwen3.5-122B-A10B on African lan…
ARM director for ecosystem strategy for developer platforms Peter Hodges appeared on the September 18, 2026 episode of the Game Developer Podcast to discuss how chipmakers work with game developers to optimize ARM's Neur…
SemiAnalysis reported that DeepSeek's Engram architecture, which extends standard token embeddings with learned multi-token lookups, can lower required HBM capacity for models at the same quality by offloading embedding …
A developer fine-tuning DEIMv2, a DETR-style detector with a DINOv3 Vision Transformer backbone, applied discriminative fine-tuning — assigning the backbone a learning rate of 5e-6 and the decoder head 1e-4, a 20x gap — …
Researchers Haoxuan Li, Ziya Erkoc, Daniele Sirigatti, Vladislav Rosov, Lei Li, Angela Dai and Matthias Niessner introduced TriFlow, a generative method that produces compact 3D meshes with artist-like triangle topology …
The FlyEye project released flyeye-escape-neuron-v1, a compact MaleCNS-derived neuron-level dataset containing a 3,376-neuron, 78,797 signed-edge graph, published on Hugging Face alongside a browser and Python runtime de…
A developer built a prototype probabilistic graph neural inference system to support rapid decision-making during mission-critical recovery windows in deep-sea exploration habitats. The system models habitat modules as a…
Xiaomi's MiMo team is publicly livestreaming the reinforcement-learning training of two unreleased models, MiMo-V2.6-Pro and MiMo-V2.6-Flash, via a dashboard at mimo.xiaomi.com/rl that shows reward curves, rollout counts…
Developer foxmoss built a 20 Questions web game powered by a machine-learning model called jev that generates probabilities rather than text and uses no LLMs, according to the project's page. For each round jev picks 10 …
Researchers posted a paper on arXiv on 17 Aug 2026 describing PostgreSQL-V 2.0, an integrated vector database system inside PostgreSQL that closes three limitations of their earlier PostgreSQL-V 1.0. PostgreSQL-V 2.0 del…
A fixed softmax regression on MNIST trained with different optimizers lands anywhere from 41.1 percent to 90.7 percent accuracy depending only on how the weights are moved, according to a Stochastic Blog post comparing f…
NASA-developed software called the Transient Artifact and Continuous Learning System (TACLS) combines satellite data with machine learning to help the National Weather Service (NWS) spot flash-flood-prone areas sooner an…
TypeSafe AI introduced a training approach it calls RLCD (reinforcement learning from calibrated decisions), which it says makes models return decisions and probabilities rather than generated text, with outcomes assigne…
OpenAI says an internal AI system has produced a solution to the Navier–Stokes existence and smoothness problem, one of the Clay Mathematics Institute's Millennium Prize Problems that has stood for over a century. The cl…
Cloud-based inference for robot AI models is a viable engineering path that could let generalist robots run models too large for on-board GPUs, according to an analysis arguing that scaling laws will continue to drive em…
The EU-funded MODELAIR project is developing AI-enabled 3D digital twins that predict urban air pollution at street level in real time, combining city geometry, computational fluid dynamics, reduced-order models and sens…