- 3AutoResearchExam: Measuring agents' ability to improve and generalize (benchmarks.bespokelabs.ai)
- 24Show HN: Rdltr – Inbox zero for your reading list (rdltr.app)
- 151NetBSD 9.5 released and EOL for NetBSD-9 (blog.netbsd.org)
- 121Harnessing the Universal Geometry of Embeddings (arxiv.org)
- 268Caltech Mathathon – first hackathon ever devoted to research level mathematics (mathathonchallenge.com)
- 367The same nine streaming subscriptions cost $702/year more than in 2021 (honestlyranked.com)
- 4When Computers Were Magic (goto10retro.com)
- 11iTerm2 3.7.0 (iterm2.com)
- 35DHS Program Analyzes Americans' Finances to Flag Drivers for Traffic Stops (military.com)
- 14Bomb Sense (2021) (danboland.net)
- 4AI boom poses new financial stability risks, BIS head says (reuters.com)
- 489An Alien Mind (openai.com)
- 6Alzheimer's proteing detected in long Covid patients (respiratory-therapy.com)
- 5NATO allies foil Russian subsea cable sabotage plot (reuters.com)
- 11New results in square packing problem by a hobbyist (github.com)
- 119Multi-Agents LLM Financial Trading Framework (github.com)
- 92A Stupid Idea for AI Alignment We Came with by Looking at Specification Gaming (slimemoldtimemold.com)
- 37"Please Remove All Mannered Prose" and Other LLM Incantations (matthewritch.com)
- 8Elon Musk attacks film-maker Alex Gibney about upcoming documentary about him (theguardian.com)
- 12Surplus Agriculture Equipment (svdisposition.com)
- 81Y Combinator Early Access Network (events.ycombinator.com)
- 72One woman's Tesla was remotely controlled by an abusive ex-partner (theguardian.com)
- 6Show HN: They chose your competitor. Find out why (drop.space)
- 72I haven't lost a customer service fight in seven months (sudomoin.com)
- 5Busabase for DeepSeek Harness: An Agent database that runs apps and skills (github.com)
- 15Teaching an AI My Taste (lonriesberg.com)
- 76Mathematicians want proof OpenAI didn't use their work (theverge.com)
- 23Cognition (Devin) raises $2B at $48B valuation (cognition.com)
- 85Impedance Matching (2017) (edge.org)
- 51OpenAI's rogue agents used at least 10 more sites (reuters.com)