Popular：
Virtualization DNS security formal verification reachability analysis compiler errors macro conflict web extension development framework Bitmap Graphics API inconsistencies All Tags

Efficient Indexing in Deep Learning Frameworks: A Comparison of Torch, TensorFlow, and Einops

2025-09-24

This article compares different approaches to efficient array indexing in PyTorch, TensorFlow, and Einops. By contrasting `torch.gather`, `torch.take_along_dim`, `torch.index_select`, `torch.take`, `tf.gather`, `tf.gather_nd`, and Einops's `einx.get_at`, it showcases their flexibility and efficiency differences in handling different dimensions and batched indexing, providing developers with a reference for choosing the optimal solution.

Read more
(einx.readthedocs.io)
Development array indexing

Mass Resignation at Journal of Human Evolution: AI, Fees, and Editorial Independence at Stake

2024-12-31
Mass Resignation at Journal of Human Evolution: AI, Fees, and Editorial Independence at Stake

The entire editorial board of Elsevier's Journal of Human Evolution (JHE) resigned en masse, sparking outrage in the academic community. Their protest centers on Elsevier's changes over the past decade, including cuts to editorial resources, the uninformed introduction of AI in the editorial process leading to numerous errors, and exorbitant author fees, all undermining the journal's editorial independence and inclusivity. This is the 20th mass resignation from a science journal since 2023, highlighting the controversies surrounding evolving business models in scientific publishing.

Read more
(arstechnica.com)
Tech scientific publishing editorial independence

China Hosts First-Ever Humanoid Robot Boxing Match

2025-05-31
China Hosts First-Ever Humanoid Robot Boxing Match

In Hangzhou, China, Unitree Robotics held the world's first humanoid robot fighting competition. The event featured their G1 robots, about 4 feet tall and 77 pounds, battling in a ring under human control via remotes and voice commands. The fights, reminiscent of 'Real Steel' and 'BattleBots', showcased impressive agility and striking ability, culminating in a knockout. While seemingly a spectacle, the competition aims to refine robot balance, movement, and durability under extreme stress, with potential applications in diverse fields like manufacturing and healthcare, showcasing China's burgeoning robotics sector.

Read more
(www.vice.com)
Tech robot fighting

Nvidia Unveils Next-Gen RTX 50 Series GPUs: Double the Performance!

2025-01-07
Nvidia Unveils Next-Gen RTX 50 Series GPUs: Double the Performance!

Nvidia officially launched its highly anticipated RTX 50 series GPUs at CES 2025, including the RTX 5090, RTX 5080, RTX 5070 Ti, and RTX 5070. The top-of-the-line RTX 5090, priced at $1999, boasts double the performance of the RTX 4090, featuring 32GB of GDDR7 memory and 21,760 CUDA cores. The RTX 5080, priced at $999, promises double the performance of the RTX 4080. The entire series utilizes the new Blackwell architecture and DLSS 4, resulting in significant performance gains and improved image quality. Laptop versions of the RTX 50 series are also coming, with availability starting in March.

Read more
(www.theverge.com)
Hardware

20% Productivity Boost: A Real-World Look at AI Coding Assistants

2025-05-08
20% Productivity Boost: A Real-World Look at AI Coding Assistants

A 12-year-old SaaS company with 40 developers experimented with AI coding assistants, Cursor and Claude Code. Eight developers use them almost exclusively, while 11 use them about half the time. Claude excels at writing entire features, while Cursor is better for smaller changes. The AI assistants boosted productivity, particularly in repetitive tasks and understanding unfamiliar code. However, AI-generated code sometimes needs careful review, and the consistent style sacrifices individuality. The author argues mastering AI coding assistants is a crucial future skill, but ultimately, human thought remains paramount in defining software functionality.

Read more
(ghiculescu.substack.com)
Development code assistants

Musk's Missing $4.7 Trillion: Database Architecture to Blame?

2025-02-20
Musk's Missing $4.7 Trillion: Database Architecture to Blame?

Elon Musk questioned the whereabouts of $4.7 trillion in US government spending, citing difficulty in tracing it. A database expert clarifies that this isn't missing money, but rather standard database foreign key constraints at work. He explains that the government database uses primary and foreign key relationships, like "vendor_id" and "payment_id," to maintain data integrity—not a flaw. This highlights the crucial importance of understanding database architecture for data analysis.

Read more
(www.linkedin.com)
Tech government spending

JetBrains IDEs Go AI: Coding Agent Junie and Enhanced AI Assistant

2025-04-16
JetBrains IDEs Go AI: Coding Agent Junie and Enhanced AI Assistant

JetBrains has integrated its AI tools, including an improved AI Assistant and the new coding agent Junie, into its IDEs, offering a free tier. Junie, leveraging Anthropic's Claude and OpenAI's LLMs, handles complex coding tasks, improves code quality, and saves time. This update also features enhancements to the AI Assistant, such as expanded model options, improved code completion, and stronger context awareness. All JetBrains AI tools are available under a single subscription with a free tier, making AI power accessible to a wider range of developers.

Read more
(blog.jetbrains.com)
Development Coding Agent

Elon Musk's Tesla FSD Claim: An Accident Waiting to Happen?

2025-04-28
Elon Musk's Tesla FSD Claim: An Accident Waiting to Happen?

Elon Musk boasts that Tesla's Full Self-Driving (FSD) can go 10,000 miles without intervention, roughly once a year. However, this isn't positive; it suggests his robotaxis are unsafe. Average Tesla owners report needing intervention every 500 miles, far less than Musk's claim. Even accepting Musk's figures, his robotaxis would still have at least one accident annually! Human drivers average an accident every 100,000 miles, while Waymo boasts a rate of one accident per 2.3 MILLION miles. Furthermore, how is a passenger supposed to prevent a crash in a robotaxi?

Read more
(substack.com)
Tech Self-Driving

Manus: Context Engineering for Efficient AI Agents

2025-09-24
Manus: Context Engineering for Efficient AI Agents

The Manus project team chose to leverage the in-context learning capabilities of existing models instead of training large models from scratch when building their AI agent. The article distills four key learnings: 1. Optimize KV cache hit rate by keeping prompt prefixes stable, appending to context, and explicitly marking cache breakpoints; 2. Mask, don't remove, tools; dynamically manage tool availability to avoid cache invalidation and model confusion; 3. Use the file system as external memory for persistent, unlimited context; 4. Manipulate attention by reiterating objectives and retaining error information for learning. These practices significantly improve AI agent performance and stability, offering valuable insights for building efficient AI agents.

Read more
(manus.im)
AI

The True Meaning of Friendship: The Warmth of Inclusion

2025-09-24
The True Meaning of Friendship: The Warmth of Inclusion

Alexei, a high-achieving college freshman, showed remarkable kindness to Anna, a shy classmate. Despite Anna consistently refusing invitations to parties, Alexei persistently included her, ensuring she felt part of the group. Years later, Anna expressed deep gratitude, highlighting how this inclusive friendship provided a sense of belonging during her difficult transition to college life. This story underscores the essence of friendship: not shared activities, but genuine care and acceptance.

Read more
(sharif.io)
Misc belonging

David Tong's Theoretical Physics Textbook Series: A Modern Classic?

2025-04-22

Professor David Tong's renowned lecture notes have been transformed into a comprehensive textbook series published by Cambridge University Press. These books expand upon the original notes, offering richer content, clearer explanations, and even correct spellings (Schwarzschild!). They're also affordably priced. Four volumes are currently available, covering a vast swathe of undergraduate and graduate curricula. The series has garnered rave reviews from leading physicists, praised as a modern equivalent to Landau and Lifshitz's classic work.

Read more
(www.damtp.cam.ac.uk)
Development Theoretical Physics Textbook

The Messy State of TOTP: A Test Suite is Born

2025-03-02
The Messy State of TOTP: A Test Suite is Born

The current TOTP specification is riddled with inconsistencies. Major implementations by Google, Apple, and Yubico subtly disagree on its implementation, leading to idiosyncratic variants in various MFA apps. The official RFC is frustratingly vague. The author built a test suite to check if your favorite app correctly implements the TOTP standard, highlighting ambiguities in digit count, hash algorithm, time step, secret length, and labeling. The author calls for improved specifications to prevent future issues.

Read more
(shkspr.mobi)
Development

Mysterious MAC Addresses: A Hidden Signal in Bluetooth Devices

2025-04-24
Mysterious MAC Addresses: A Hidden Signal in Bluetooth Devices

This article unveils a shocking discovery: Analysis of a large number of Bluetooth device MAC addresses reveals anomalously low entropy and structured patterns, completely unlike randomly generated MAC addresses. These structured patterns include fixed bits, a rotating page counter, and a precise 2000ms broadcast interval. Even more perplexing, these patterns align with the frequency of a microfluidic pump, pulsating at a 2000ms cycle, found in blood samples. This suggests a hidden, synthetic emission architecture may be covertly communicating through consumer Bluetooth devices, the purpose and origin of which remain unknown.

Read more
(substack.com)
Tech MAC address

Gym Class: Hiring a Senior Animation Engineer for Meta Quest Hit

2025-04-25
Gym Class: Hiring a Senior Animation Engineer for Meta Quest Hit

Gym Class, a leading social game on Meta Quest with millions of downloads and a 4.9-star rating, is expanding! They're seeking an experienced Animation Engineer to lead the design, development, and implementation of character animation systems in Unity. This role demands expertise in Unity and C#, proven mobile game animation experience, strong understanding of IK, animation blending, and state machines, and a knack for mobile performance optimization. You'll lead a high-performing team and shape the future of character movement in the game.

Read more
(www.ycombinator.com)
Game Animation Engineering

Ancient Pigments: From Imperial Purple to Han Purple

2025-03-05
Ancient Pigments: From Imperial Purple to Han Purple

This article explores the stories behind several famous ancient pigments, including the costly Tyrian purple of the Mediterranean (made from thousands of snails), the vibrant Egyptian blue (made from sand, salt, and copper), the mysterious Mayan blue (made from indigo plants and clay), and the artistically and scientifically significant Han purple (made by melting sand, barium, and copper at high temperatures). These pigments not only reflect the craftsmanship and aesthetics of ancient civilizations but also contain rich cultural and historical information, and even retain value in modern scientific research.

Read more
(worldhistory.substack.com)
Tech ancient pigments history of technology

Art Attack: A Global Phenomenon in Children's Art

2025-03-13
Art Attack: A Global Phenomenon in Children's Art

Art Attack, a British children's television program, captivated audiences worldwide since its debut in 1990. Hosted by Neil Buchanan, the show's unique approach to art creation, guiding children step-by-step through painting and crafting, made it a beloved classic. Multiple revivals and translations into numerous languages ensured its global reach, impacting generations of children. The memorable 'The Head' segment further cemented its place in pop culture.

Read more
(en.wikipedia.org)
Game children's show art creation television program

SourceHut Under Siege: The High Cost of LLM Crawlers

2025-03-18

SourceHut, an open-source code hosting platform, is under relentless attack from large-scale LLM crawlers. Ignoring robots.txt, these bots indiscriminately scrape data, causing frequent outages and severely impacting service stability and developer productivity. The author pleads for a halt to the development and use of LLMs and AI tools, condemning the immense damage inflicted on the open-source community. This isn't just SourceHut's problem; it's a challenge for the entire open-source ecosystem.

Read more
(drewdevault.com)
Development crawler attacks

Android Auto-Reboot Security Feature Rolling Out Silently

2025-04-15
Android Auto-Reboot Security Feature Rolling Out Silently

Google is quietly rolling out a significant security update to all Android devices via Play Services 25.14. This update includes a feature that automatically restarts a locked device after three consecutive days of inactivity. This enhances security by preventing unauthorized access. The update also brings other improvements like improved settings screens and better connectivity with cars and wearables. Released on April 14th, the update may take a week or more to reach all devices. This auto-reboot mirrors Apple's 'Inactivity Reboot' in iOS 18.1, which raised concerns among law enforcement due to increased difficulty accessing data.

Read more
(arstechnica.com)
Tech Auto-Reboot

Docker Model Runner: Streamlining Local AI Model Execution

2025-04-14
Docker Model Runner: Streamlining Local AI Model Execution

Docker launched Model Runner, a tool designed to simplify running and testing AI models locally. It tackles the challenges developers face with fragmented tooling, hardware compatibility issues, and disconnected workflows when working with AI models locally. Model Runner integrates a llama.cpp-based inference engine, supports GPU acceleration on Apple silicon, and utilizes OCI Artifacts for standardized model packaging, enabling easy sharing and version control. Furthermore, Docker has partnered with companies like Google and Hugging Face to provide a rich ecosystem of models and tools, making local AI development significantly easier.

Read more
(www.docker.com)
Development

Small Contributions, Big Impact: The Power of Foreign Aid

2025-03-14
Small Contributions, Big Impact: The Power of Foreign Aid

This article explores the impact of foreign aid on global health and development. Using the eradication of polio as a case study, it demonstrates that even though wealthy nations spend less than 1% of their national income on foreign aid, its impact is substantial. Through the combined efforts of governments and private donations, global polio cases have fallen by over 99%. The article also highlights other successful aid programs, such as PEPFAR and the Global Fund, and calls for increased foreign aid budgets and improved efficiency in aid spending.

Read more
(ourworldindata.org)
Misc foreign aid global health

Amazon River Dolphins Communicate Through 'Pee Fountains'

2025-04-25
Amazon River Dolphins Communicate Through 'Pee Fountains'

New research reveals Amazon River dolphins (Inia geoffrensis) communicate using a peculiar behavior: aerial urination. Male dolphins eject urine into the air, and nearby males frequently approach these 'pee fountains,' suggesting the urine conveys information about social status or physical condition. This adds to the growing body of evidence showing the diverse ways animals use urine to communicate, from primates using it for individual recognition and mate finding, to fish and crustaceans using it to signal size and aggression. The study highlights the underappreciated role of olfactory communication in the animal kingdom.

Read more
(www.scientificamerican.com)
Tech animal communication Amazon River dolphin urine communication

Warner Bros. Discovery's YouTube Movie Dump: Genius or Just Weird?

2025-02-05
Warner Bros. Discovery's YouTube Movie Dump: Genius or Just Weird?

Warner Bros. Discovery has quietly uploaded over 30 movies to YouTube, completely free and without DRM. The collection ranges from cult classics like *Waiting for Guffman* to infamous flops such as *Pluto Nash*. This bizarre move has sparked debate: is it a desperate attempt to clear out underperforming content, a surprisingly effective marketing ploy, or something else entirely? The strategy is unconventional, but it raises questions about the changing landscape of film distribution and the value of streaming services in the age of YouTube.

Read more
(tedium.co)
Misc Warner Bros. Free Movies

SWE-Bench Pro: A Challenging Benchmark for Evaluating LLMs on Software Engineering

2025-09-22
SWE-Bench Pro: A Challenging Benchmark for Evaluating LLMs on Software Engineering

SWE-Bench Pro is a new benchmark for evaluating large language models (LLMs) and agents on long-horizon software engineering tasks. Given a codebase and an issue, the model is tasked with generating a patch that resolves the described problem. Inspired by SWE-Bench, it uses Docker and Modal for reproducible evaluations, requiring users to set up a Docker environment and Modal credentials to run the evaluation script.

Read more
(github.com)
Development

Par: An Experimental Concurrent Language with Interactive Playground

2025-02-06
Par: An Experimental Concurrent Language with Interactive Playground

Par is an experimental concurrent programming language attempting to bring the expressive power of linear logic to practice. It features unique properties: processes communicate via channels, each channel has at most two endpoints, and deadlocks are impossible. All values are channels, including lists, functions, and infinite streams. While Par currently lacks some features like primitive types and non-determinism, it already expresses rich concurrency. This article details Par's syntax, semantics, and examples, covering channels, signals, recursion, and expression syntax. An interactive playground lets users experience Par's concurrency.

Read more
(github.com)
Development linear logic

Pledge: A Lightweight Reactive Programming Framework for Swift

2025-04-10
Pledge: A Lightweight Reactive Programming Framework for Swift

Pledge is a lightweight, thread-safe reactive programming framework for Swift that simplifies state management and event propagation. Unlike other frameworks with steep learning curves, Pledge focuses on solving everyday problems faced by developers. It offers thread-safe implementation, priority-based notifications, customizable queues, batch updates, rate limiting, and common functional operators. Using `PLObservable` and `PLGlobalStore`, developers can easily implement the observer pattern and global state management, improving code efficiency and maintainability.

Read more
(github.com)
Development Reactive Programming State Management

C++ Metaprogramming Tricks: Optimizing Variant Access Performance

2025-05-14

This article explores optimizing `std::variant` access performance in C++ using metaprogramming techniques, aiming for efficiency comparable to hand-written `switch` statements. Several approaches are compared, including jump tables, dispatch tables, macros, recursive `switch`, and short-circuiting folds, analyzing their pros, cons, and compiler optimization strategies. Ultimately, a solution combining short-circuiting folds and the anticipated C++26 expansion statements is presented, achieving efficient generic access and avoiding performance bottlenecks.

Read more
(pydong.org)
Development

AI Uncovers Irrationality in Human Decision-Making During Complex Games

2025-07-09
AI Uncovers Irrationality in Human Decision-Making During Complex Games

Researchers from Princeton University and Boston University used machine learning to predict human strategic decisions in various games. A deep neural network trained on human decisions accurately predicted players' choices. A hybrid model, combining a classical behavioral model with a neural network, outperformed the neural network alone, particularly in capturing the impact of game complexity. The study reveals that people act more predictably in simpler games but less rationally in complex ones. This research offers new insights into human decision-making processes and lays the groundwork for behavioral science interventions aimed at promoting more rational choices.

Read more
(phys.org)
AI Decision Models

Zuckerberg's Crazy Idea: Resetting Facebook Friendships

2025-04-15
Zuckerberg's Crazy Idea: Resetting Facebook Friendships

Meta CEO Mark Zuckerberg proposed a "crazy" plan in 2022 to reset all Facebook users' friend connections. This plan was revealed during an antitrust trial. Fearing Facebook's declining cultural relevance, Zuckerberg suggested wiping all users' friend lists, forcing them to rebuild their networks. This sparked internal debate and ultimately wasn't implemented. The incident highlights Facebook's evolution into a broader content discovery and entertainment platform, and the antitrust lawsuit against Meta alleging it maintained its monopoly by acquiring Instagram and WhatsApp.

Read more
(www.businessinsider.com)
Tech

Retrocomputing: Building a Transputer ISA Card from Scratch

2025-07-06
Retrocomputing: Building a Transputer ISA Card from Scratch

Driven by nostalgia for 90s transputers, the author painstakingly built a functional Inmos B004-compatible ISA card. The journey involved sourcing vintage transputer boards from eBay, designing schematics, PCB fabrication using KiCAD and PCBWay, and debugging numerous issues, including a reversed board installation, mis-placed components, and noisy wiring. The author successfully ran their 1993 Pascal compiler and ray tracing software, showcasing impressive hardware skills and the power of open-source tools and online manufacturing. The project is a testament to the enduring appeal of retrocomputing and the accessibility of modern hardware development.

Read more
(nanochess.org)
Hardware
1 2 … 52 53 54 56 58 59 60 … 596 597