Full-text search across 70855 articles. Combine with topic and date filters; results sorted by relevance.
Anshvardhan Shetty, a first-year EECS undergraduate at Imperial College London and research intern with Professor Juncheng Yang at Harvard SEAS, presents a study of KV cache admission on high-bandwidth flash using a 222-…
Amazon Web Services (AWS) announced a new solution that integrates Amazon Bedrock with Amazon Textract to enable customer service teams to accurately parse and query large, complex utility bills in formats such as PDF, D…
Toronto-based cloud, AI, and DevOps company Dedicatted reports that its bilingual AI shopping assistant for Kuwaiti e-commerce platform Taw9eel lifted average order value by six percent, sped up product discovery, and im…
The city of Dallas is using AI-powered cameras mounted on trash collection trucks to assign "blight scores" to homes, with the intent of code enforcement that could lead to fines for residents in low-income neighborhoods…
SemiAnalysis reports that published CVE descriptions can enable AI agents to construct working exploits against software beneath an agent sandbox, making unpatched host software part of the containment boundary. Teams ru…
OpenAI launched GPT-6 Astra on September 3, 2026, its most powerful model yet, capable of navigating desktops, writing code, browsing the web, and executing complex workflows. The launch video opened with archival footag…
A developer released a self-contained C++ experiment that encodes an RGB image as a low-resolution latent texture plus a tiny MLP decoder, training both entirely with Evolution Strategies rather than backpropagation. The…
Microsoft announced Project Zenith, a preconfigured Windows experience for developers, with the first devices powered by AMD's Ryzen AI Halo chips and capable of running 30B+ parameter models locally. The initiative, rev…
Dyson unveiled CameraJet, a smart AI toothbrush developed over six years, featuring a 1mm camera with a 100K-pixel macro lens that analyzes 28 live images per second to target gaps between teeth, and precision water jets…
A developer has published a guide to counting tokens across major LLM providers in JavaScript, highlighting that OpenAI's tokenizer is open-source while Anthropic, Google, and DeepSeek lack JavaScript tokenizers. The gui…
Open Athena's Marin team has launched its largest model yet, a 535B-parameter mixture-of-experts (MoE) model with 23B active parameters, scheduled to pretrain for about 3 months on 18 trillion tokens including agentic, c…
An engineer from MonkeyCode warns that timing only the happy path in cache lookups hides the real cost of exception-based miss handling, which can be over 16 times slower and use more memory than a quiet dict.get. The de…
A new analysis from arXiv (2609.03635v1) finds that natural variance in large language models (LLMs) outweighs prompt fine-tuning for drug toxicity prediction, despite clinical trials in the UK costing up to £1.3 million…
Researchers introduced ProgResViT, an input-adaptive Vision Transformer that progressively increases input resolution and model width across rounds, terminating early when predictions are confident. Applied to DeiT, Prog…
Researchers introduced Finite-Library Input-Warped Bayesian Optimization (FLIWBO), a method that adapts input geometry in Gaussian-process Bayesian optimization while retaining convergence guarantees, outperforming raw-c…
Researchers propose Gradient-Aligned Reward (GAR), a dense reward method for reinforcement learning from verifiable rewards (RLVR) that uses cosine similarity between rollout and expert-anchor gradients to improve LLM re…
A new arXiv paper (2609.03003v1) introduces causal foundation models (CFMs), pretrained neural networks that estimate causal quantities such as average treatment effects on new datasets via in-context learning without fi…
Researchers propose Tail-Likelihood Reinforcement Learning (TailRL), a method that optimizes the probability of exceeding reward thresholds rather than average reward, to better leverage rare high-reward rollouts. TailRL…
Researchers introduced Speculative Macro Commit (SMC), a runtime mechanism that reduces latency for tool-using LLM agents by having a faster speculative drafter model pre-execute action chains on an environment snapshot,…
Pilot Protocol released shell.online, a tool that lets users share a live terminal in any browser by prefixing a command with 'shell', with a --read-only flag to block browser input. The process runs on the user's machin…