901
New
The latest blog posts to take off on Hacker News. 1,840 posts.
902
Claude 3.5 Sonnet
Introducing Claude 3.5 Sonnet—our most intelligent model yet. Sonnet now outperforms competitor models and Claude 3 Opus on key evaluations, at twice the speed.
903
904
20x Faster Background Removal in the Browser Using ONNX Runtime with WebGPU
Using ONNX with WebGPU and WebAssembly leads to 20x speedup over multi-threaded and 550x speedup over single-threaded CPU performance. Learn how.
905
906
Arvind has died
Arvind, a longtime MIT professor and leading scholar in computer architecture and programming languages, has died at 77. He transformed research in computer architectures and parallel computing, enabling faster and more efficient computation.
907
Astronomers see a black hole awaken in real time
Ende 2019 begann die zuvor unauffällige Galaxie SDSS1335+0728 plötzlich heller zu leuchten als je zuvor. Um den Ursachen auf den Grund zu gehen, haben Astronom*innen Daten von mehreren weltraumbasierten und bodengebundenen Observatorien, darunter das Very Large Telescope (VLT) der Europäischen Südsternwarte (ESO)…
908
AI-powered conversion from Enzyme to React Testing Library
Update (October 2024): In response to numerous requests from external developers, we have open-sourced a version of our Enzyme to React Testing Library (RTL) conversion tool. You can now find it on npm, along with detailed instructions on how to integrate and use it in your projects. In the world of frontend…
909
Monitoring marine litter from space
Every 60 seconds the equivalent of a lorry-load of plastic enters the global ocean. Where it goes after that remains largely unknown. But a new ESA Discovery study reported in Nature Communications has proven the concept of monitoring floating plastic litter using satellites.
910
Off-path TCP hijacking in NAT-enabled Wi-Fi networks
Guest Post: New off-path TCP hijacking attack in Wi-Fi networks that exploits vulnerabilities in the NAT mapping strategies of routers discovered.
911
3D Gaussian Splatting as Markov Chain Monte Carlo
Novel view reconstructions for (right) our method and (left) conventional 3D Gaussian Splatting with random initializations. Our method, even with random initialization, faithfully reconstructs the scene (e.g.. buildings at the back and the ground texture) providing much higher quality renderings.
912
Sharing new research, models, and datasets from Meta FAIR
Meta FAIR is releasing several new research artifacts. Our hope is that the research community can use them to innovate, explore, and discover new ways to apply AI at scale.
913
When are SSDs slow?
CedarDB is a database system that delivers unmatched performance for transactions and analytics, from small writes to handling billions of rows. Built on cutting-edge research to power today’s tools and tomorrow’s challenges.
914
915
Code Models
Code reflection, an enhancement to Java reflection, enables access to symbolic representations of Java code in method and lambda expressions. “Symbolic representations of Java code” may seem like a fancy term, but its easily demystified. It’s a model of Java code where the code of a method or lambda expressions is…
916
Chat Control: Incompatible with Fundamental Rights (2022)
The EU Commission's draft regulation on chat control is currently being negotiated in the European Parliament and the Council of Ministers. With the fight against sexual violence against children, the draft pursues an objective that is essential for the protection of children and their rights and can justify…
917
Chat Control Must Be Stopped – Now
With its legislative proposal known as “Chat Control,” the EU Commission is trying to establish an unprecedented mass-surveillance apparatus of Orwellian proportions in the European Union. If EU citizens don’t stand up for privacy now, it may be too late.
918
Every Way to Get Structured Output from LLMs
A survey of every framework for extracting structured output from LLMs, and how they compare.
919
Getting 50% (SoTA) on Arc-AGI with GPT-4o
You can just draw more samples
920
LLM that can call multiple tool APIs with one request
Learn about Cohere models' ability to perform multi-step tasks, enhancing productivity and efficiency.
921
Humans began to rapidly accumulate technological knowledge 600k years ago
Each of us individually is the accumulated product of thousands of generations that have come before us in an unbroken line. Our culture and technology today are also the result of thousands of years of accumulated and remixed cultural knowledge.
922
Htmx 2.0.0 has been released
htmx gives you access to AJAX, CSS Transitions, WebSockets and Server Sent Events directly in HTML, using attributes, so you can build modern user interfaces with the simplicity and power of hypertext htmx is small (~14k min.gz’d), dependency-free, extendable, IE11 compatible & has reduced code base sizes by 67% when…
923
Let Readers Read
Ask publishers to restore access to the 500,000 books they've caused to be removed from the Internet Archive’s lending library. Sign the Open Letter I’m Chris Freeland, a librarian at the Internet Archive. The lawsuit against our library—Hachette v. Internet Archive—is fast approaching the oral argument stage of its…
924
React 19 almost made the internet slower
The author discusses the React 19 change that went unnoticed until last week that could potentially degrade the performance of websites that rely React.
925
Proton is transitioning towards a non-profit structure
To ensure our mission always comes first, Proton is transitioning to a non-profit structure and formalizing our promise of people before profits.
926
NumPy-style broadcasting in Futhark
A high-performance and high-level purely functional data-parallel array programming language that can execute on the GPU and CPU.
927
Mozilla acquires privacy-focused ad tech company Anonym
Mozilla has acquired Anonym, a trailblazer in privacy-preserving digital advertising. This strategic acquisition enables Mozilla to help raise the bar for
928
Maintaining large-scale AI capacity at Meta
Meta is currently operating many data centers with GPU training clusters across the world. Our data centers are the backbone of our operations, meticulously designed to support the scaling demands …
929
NumPy 2.0
NumPy 2.0.0 is the first major release since 2006. It is the result of 11 months of development since the last feature release and is the work of 212 contributors spread over 1078 pull requests. It contains a large number of exciting new features as well as changes to both the Python and C APIs.
930
Address Sanitizer Internals
Going in depth into the Address Sanitizer internals.