Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Researchers introduced QUASAR, a method that lowers the loss floor gap in quantization-aware training (QAT) for large language models by adding a saliency-weighted reconstruction step. The approach reduces held-out KL di…
Cohere, a Toronto-headquartered enterprise AI company, is hiring a Member of Technical Staff for RL Environments in London, a hybrid role focused on building reinforcement learning environments to train and evaluate AI a…
A team from the University of Cambridge's Machine Learning Systems Lab, working with NVIDIA, Flower Labs, MBZUAI, and Inria, introduced the Red Queen Gödel Machine, a co-evolutionary system that improves agents and their…
AlphaAvatar v0.6.6 introduces event-driven multimodal memory, unified runtimes for Qdrant and LanceDB, and cleaner agent contracts, according to the project's release notes. The update expands Environment Memory to accep…
A new framework from Collinear AI advises teams to choose AI models per workload rather than once company-wide, using a build-vs-buy test across capability, cost, compliance, and control. The guide recommends using LM Ar…
Lawrence Architectures released UL-SMF, an open-source KV-cache compression fabric that reduces memory footprint by up to 384x with over 94% semantic retention, addressing the memory bottleneck in long-context Transforme…
Matteo Grella released Fucina v0.1.0, a CPU-first tensor/autograd library written in pure Zig 0.16, featuring compile-time axis names in tensor types that make misaligned contractions a compile error. The library support…
A review of recent studies on chunking for Retrieval-Augmented Generation (RAG) finds no single best chunk size, with results varying by corpus and method. A chemistry study by Amiri and Bocklitz (2025) found recursive 1…
Speko, a Y Combinator S26 startup founded by Bek, launched an API that automatically selects the optimal combination of speech-to-text, LLM, and text-to-speech models based on user constraints, with public benchmarks and…
Computational anatomy is an interdisciplinary field focused on quantitative investigation and modelling of anatomical shape variability, integrating anatomy, applied and pure mathematics, machine learning, computational …
Peripheral Labs, a Toronto-based AI research company, has raised $8.7 million in seed funding from Inovia Capital, Deloitte Ventures, Khosla Ventures, and Entrepreneurs First to bring self-driving car technology to live …
OpenSSL 3.5+ now natively supports all three NIST post-quantum algorithms (ML-KEM, ML-DSA, SLH-DSA), and a hands-on test shows that post-quantum certificates are dramatically larger than classical ones: a self-signed ML-…
Veso AI open-sourced tenstorrent-wormhole-kimi-linear, kernels and correctness gates that run Kimi-Linear-48B on Tenstorrent Wormhole hardware, taking the model from not starting to serving 16,384 tokens of context at 13…
A new technical explainer on activation functions highlights the vanishing gradient problem that plagues Sigmoid and Tanh in deep networks, and introduces ReLU as a solution. The article includes Python code demonstratin…
Researchers introduced TAOT, a topology-aware optimal transport method for dynamic expert-replica placement in Mixture-of-Experts (MoE) training, which models load imbalance as an entropy-regularized optimal transport pr…
An engineer who built an evaluation engine for arbitrary MCP servers explains that an MCP eval is a realistic task a model must complete using only the server's tools, distinguishing it from unit tests that only verify p…
A Cambridge-led team with collaborators at NVIDIA, Flower Labs, MBZUAI, and Inria introduced the Red Queen Gödel Machine (RQGM), a framework that co-evolves an AI agent and its evaluator to prevent self-improvement loops…
A new study published in npj Aging by geneticists from the University of Southern California analyzed blood samples from more than 3,000 participants in the U.S. Health and Retirement Study, comparing methylation pattern…
Researchers introduced AutoDesign, a framework that uses a meta-harness optimizer to recursively improve a code agent's design harness, achieving the highest score of 78.32 on the PosterBench Main Track, surpassing Claud…
Finbold AI Agent projects that XRP will trade at an average price of $0.98 on September 1, 2026, a 0.85% decline from press time on August 17. The prediction combines outputs from DeepSeek Chat, ChatGPT-5.7 Luna, and Gro…