Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
A debugging guide for large language model training runs recommends that engineers first test whether a model can overfit a tiny dataset of 100 examples, then check training and validation loss, data quality, training si…
A developer has released Jev, a System 1 latent AI model that returns typed values and calibrated probabilities directly instead of generating tokens, targeting backend classification tasks like ticket routing and diff a…
ICLR 2027 has received roughly 50,000 abstract submissions ahead of its deadline, up from 19,500 at ICLR 2026, according to The Decoder. The surge is attributed to AI hype, corporate pay tied to publication records, and …
US-based Reward AI released videos of its Omnibody Model 1 (OM-1) foundational robot model, which learns complex tasks solely from human movements captured by a wearable device and can pick up new tasks from less than 30…
A practical guide explains how large language models work for product managers without requiring deep ML expertise, walking through the pipeline from tokenization and embeddings through the Transformer's self-attention a…
A developer has released Bonsai, a 27-billion-parameter reasoning model designed to run on a 16 GB M2 Mac, alongside a companion tool called Ferrox. The project targets local inference of a large reasoning model on consu…
Linkup released SPARSEUP, an open-source learned sparse embedding model with a 149M-parameter ModernBERT backbone, published under Apache 2.0 on Hugging Face. Linkup reports 56.4 average nDCG@10 on BEIR-13, which it call…
Cactus Compute released Needle 3 on September 17, an Apache 2.0-licensed automation foundation model that ships as an 8 to 29MB binary and runs tool calling, structured extraction, and text embeddings entirely on-device,…
TypeSafe AI released Jev, a calibrated classification model that returns probabilities rather than generating text, on 15 September, and Laravel's AI SDK added support for it two days later via pull request 1010 merged i…
A technical explainer clarifies that one update step of a modern continuous Hopfield network is mathematically identical to scaled dot-product attention, citing the 2020 paper "Hopfield Networks is All You Need" by Ramsa…
Six independent reimplementations of Jev, a non-generative "System 1" decision model launched on Wednesday, appeared within two days of its release, according to AINews' 9/17/2026-9/18/2026 roundup. The clones include La…
Alicelabs LLC has released VIGÍA ML, a browser-based predictive monitoring console for gas wells that trains and runs all four of its machine-learning models entirely on-device using TensorFlow.js and React. The tool com…
A developer wrote a hand-tuned NEON assembly matrix multiplication kernel for the ARM M2 processor, benchmarking it against a C reference implementation across matrix shapes from 64x64 to 1024x1024. The naive assembly ve…
A developer building a quantitative crypto trading engine has open-sourced a design for replacing binary risk vetoes with a "gray-scale" dynamic scaling approach. Instead of hard-rejecting a high-confidence signal after …
ServiceNow posted a Senior Manager of Machine Learning Engineering role in Santa Clara, California, paying $191,000 to $334,000 per year plus equity, to lead a team building AI Search and Retrieval capabilities across th…
Researchers Mohamed Suliman and colleagues published an arXiv paper on 30 October 2022, revised 9 October 2023, showing that federated learning does not protect user privacy in Google's GBoard next word prediction model.…
Cua founder Francesco Bonacci released CUA-S1-FORMS on September 18th, a 706,048-parameter open-source model with a 2.8 MB checkpoint that plans AI agent form-filling in one forward pass instead of repeated large languag…
A developer has demonstrated a PyTorch-based Transformer model for forecasting blood glucose dips up to 30 minutes ahead using continuous glucose monitor data. The system pipelines CGM readings through InfluxDB and Panda…
TypeSafe AI released a non-generative model called Jev on September 15 that answers typed questions about program state in a single parallel pass, reportedly in the low hundreds of milliseconds, and developers integrated…
Alibaba DAMO Academy researchers released RADAR, a generalist vision-language model for abdominal CT diagnosis trained on more than 400,000 contrast-enhanced abdominal CT examinations and 15 million anatomy-aware image-t…