Machine Learning News

Makerfabs released the MaTouch ESP32-S3 MaUWB development board, which combines an ESP32-S3 wireless MCU, a 3.95-inch touchscreen, and a UWB module for indoor positioning and ranging. The board uses a trilateration algor…

Dwarkesh Patel has been advocating for continual learning in AI, arguing that frozen models are limited and that continual learning is necessary for long-running, open-ended problem solving. The author agrees with Patel'…

A 3.16M-parameter transformer running on a $250 FPGA achieves 44 tokens per second by moving the CPU out of the per-token loop, according to a technical blog post detailing the on-chip LLM design. The hardware uses a tra…

An agent that passes 75 out of 100 test runs has a 75% pass@1 rate, but reliability depends on the product: pass@k (at least one success in k tries) yields 98.4% for k=3, while pass^k (all k succeed) yields only 42.2%, a…

A new compiler called Torchwright, created by researcher Rob, compiles fixed computation graphs into transformer weights, and its first dedicated digit-level multiplier achieves 100% accuracy on all 3,000,000 valid three…

Teddy Warner's Avian Visitors project uses a Raspberry Pi and USB microphone with Cornell's BirdNET deep learning classifier to identify and track bird species by sound, displaying recent visitors as a collage on a web i…

Pmalfa31 developed a tomato grading system using an ESP32 and machine learning that sorts standard and cherry tomatoes by color and size, with a web simulator available for testing. The system uses two learned datasets, …

Intel's Core 5 320 'Wildcat Lake' processor, built on Intel 18A in Intel's own fabs, achieved 6.21 Gflops per watt in Jeff Geerling's HPL Linpack testing, outperforming Apple's A18 Pro in the $599 MacBook Neo at 5.38 Gfl…

Google DeepMind open-sourced its WeatherNext cyclone model on August 6, publishing a Nature paper and releasing code, weights, and ensemble forecast feeds, with three-day forecasts matching the accuracy of previous two-d…

Researchers at Stanford University and the Arc Institute used the Evo 1 and Evo 2 genome models to design variants of the ΦX174 bacteriophage, and of 285 synthesized candidates, 16 produced viable phages that infected an…

At YC Paper Club's robotics edition, researchers presented findings on embodied memory, self-supervised reasoning, dexterous sim-to-real tool use, and world action models, while Francois Chaubard noted that despite a dec…

A mechanistic interpretability study found that ablating specific MLP neurons in Google's Gemma 4 model causes it to revert to its dormant Bard identity, with the model answering 'My name is Bard' with p=0.9923 after neu…

Google DeepMind's WeatherNext Cyclones model, described in a Nature paper published August 6, ran live through the 2025 Atlantic hurricane season with the National Hurricane Center, including during Hurricane Melissa, an…

Google DeepMind released the code and weights for its WeatherNext Cyclones model on August 6, 2026, after the US National Hurricane Center used its guidance during the 2025 hurricane season. The model, described in a Nat…

The FDA approved mFLUSIVA, the first mRNA influenza vaccine licensed in the United States, on August 5, 2026, exactly one year after the Department of Health and Human Services canceled 22 mRNA vaccine projects worth rou…

Google DeepMind and Google Research's WeatherNext AI model predicted Hurricane Melissa's landfall in Jamaica as a Category 5 hurricane with 80 percent confidence five days before impact, providing forecasters with an ext…

Envision Group activated the world's largest 12.8 GWh battery storage cluster in Inner Mongolia, incorporating AI into its battery management systems to optimize dispatch and increase lifetime revenue by 20%. The flagshi…

In an essay crossposted from his website and edited with Claude Fable (Anthropic), the author argues that the scientific method's essence is intuition, not deduction, and that reasoning serves to construct understanding …

A July 2026 Blind survey of 1,552 India-based professionals found that 66% of AI/ML workers expect layoffs or significant team cuts within the next three to six months, making them the second-most anxious group after sal…

Go 1.27's new simd package enables local AI inference, allowing developers to run machine learning models directly on devices without cloud dependencies. The package provides SIMD (Single Instruction, Multiple Data) oper…