Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Researchers propose IGGT4D, a streaming instance-grounded geometry Transformer for online 4D scene understanding that unifies camera motion, scene geometry, and consistent instance features within a single feed-forward f…
A controlled study pretraining attention-only decoder transformers (Simple Attention Networks) against standard transformers matched for parameters, compute, and depth found that reallocating the feed-forward budget into…
A machine learning model built from scratch on the UCI Adult Census Income dataset reaches 84 percent accuracy, but the project's real value lies in demonstrating overfitting and underfitting. A deep decision tree with m…
The mlx-community/Laguna-S-2.1-oQ2e quantized model running in-process through mlx-vlm on a 128 GB Apple M5 Max achieved a perfect overall score of 1.000 across six tasks, with 40.85 generation tok/s and 37.77 GB peak ML…
Researchers have developed SEAM-V, a hybrid-decoupled vector execution architecture for RISC-V Vector Extension (RVV) that achieves a geometric-mean speedup of 1.34x across 17 representative kernels compared to an Ara-ba…
Adding a reranker to a search system incurs a tax in latency, cost, and complexity that is only worthwhile when the first-stage retrieval already has high recall, according to a post by Eoin Hurrell. A reranker can only …
A developer's benchmark of Google's Gemini 3.5 Flash Lite for human detection in a home surveillance system found the new model produced one false positive in 30 runs, while the existing Gemini 3.1 Flash Lite had zero fa…
Maayan Roth, a software engineer, is revisiting machine learning fundamentals by reading Andriy Burkov's 'The Hundred-Page Languages Models Book' to better understand large language models. In Chapter 1, she learns about…
AMD has open-sourced Spur, a modern GPU job scheduler written in Rust under Apache 2.0, to address the complexity of scheduling AI and HPC workloads on GPU clusters. Spur provides Slurm-compatible CLI and APIs, topology-…
The PyTorch Conference North America schedule is live, with the event taking place in San Jose on October 20–21, featuring sessions on training, inference, compiler innovations, responsible AI, and the PyTorch ecosystem.…
Atlas, an enterprise RAG copilot project, treats retrieval quality like unit tests by blocking merges on quality or cost regressions. The pipeline uses deterministic, GPU-free eval gates with committed cassettes, floors,…
A developer built a probabilistic graph neural network for circular manufacturing supply chains, addressing the brittleness of deterministic GNNs during mission-critical recovery windows. The model uses Bayesian layers t…
A developer discovered a silent race condition in OpenVINO's AsyncInferQueue that caused concurrent embedding requests to return vectors belonging to other inputs. The bug, which passed standard unit tests, was caught by…
A developer built TraceTree, a tool that scans files for malicious behavior using radare2 disassembly, YARA rule matching, and an ML pipeline. During OpenAI Build Week, Codex with GPT-5.6 helped trace a bug where the bac…
Igor Markov, a researcher in algorithms, AI, circuits, and quantum computing, has updated his Google Scholar profile. His work spans multiple technical domains including machine learning and VLSI design.
A developer building the model-drift evaluation tool describes how its exact-match grader can conflate wrong answers with absent ones, such as rate limits, refusals, or truncations. The tool excludes failures only when a…
Xaira Therapeutics has developed X-Cell, an AI model for drug discovery that predicts gene expression changes by training on information-rich causal data from its X-Atlas dataset, overcoming the scaling limitations of mo…
Torch.compile is not a traditional compiler but a system that intercepts Python bytecode at runtime, extracts compilable regions through a multi-stage pipeline, and stitches them back with eager Python fallbacks, with bo…
A reproducible 100-step LoRA fine-tuning run for the OpenVLA 7-billion-parameter robotics model on a Colab A100 GPU has been documented, providing a verifiable integration test that confirms the training pipeline works b…
The University of Tennessee Research Foundation filed a patent infringement lawsuit against Anthropic on Monday in Delaware federal court, accusing the AI company of using its neural network technology without permission…