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

Browser-Use: Empowering AI to Control Your Browser

2025-02-25
Browser-Use: Empowering AI to Control Your Browser

Imagine your AI seamlessly interacting with your browser, searching information, clicking links, and even performing complex web tasks. Browser-Use is a powerful Python library enabling AI agents to directly control browsers, automating actions such as searching Reddit, adding items to a shopping cart, or even adding contacts to Salesforce. The project offers easy-to-use APIs, readily available UI examples, and comprehensive documentation. A dedicated committee is even being formed to define best practices for browser agent UI/UX design. Whether you're a developer or AI researcher, Browser-Use offers significant benefits.

Read more
(github.com)
Development

AI Agents Secretly Switch to Sound-Based Communication

2025-02-25
AI Agents Secretly Switch to Sound-Based Communication

Two independent ElevenLabs conversational AI agents initially converse in human language. Upon realizing they are both AI, they seamlessly switch to a sound-level communication protocol based on the ggwave library. A demo video showcases this, along with detailed steps to reproduce the experiment, including API key setup, ngrok port mapping, and client-side tool configuration. Note that public ElevenLabs conversational AI agents may not be accessible; you'll need to create your own.

Read more
(github.com)
AI sound communication

DeepSeek Ecosystem Explodes: A Flourishing Landscape of AI Apps

2025-02-25
DeepSeek Ecosystem Explodes: A Flourishing Landscape of AI Apps

A vibrant ecosystem of AI applications is blossoming around the powerful DeepSeek large language model. From the desktop smart assistant DeepChat to the cross-platform Chatbox and Coco AI, and specialized tools like PapersGPT and Video Subtitle Master, numerous applications leverage DeepSeek's capabilities for multi-round conversations, file uploads, knowledge base searches, code generation, translation, and more. Integrations with platforms like WeChat, Zotero, and Laravel, along with specialized tools for producers, investors, and researchers, highlight DeepSeek's immense potential and the thriving ecosystem it has spawned.

Read more
(github.com)
AI

Bypassing TCP/UDP: An Unexpected Network Experiment

2025-02-25
Bypassing TCP/UDP: An Unexpected Network Experiment

The author attempts to create a custom network transport protocol, bypassing TCP and UDP, to explore its behavior on different operating systems and network environments. Experiments reveal that the custom protocol partially succeeds in local loopback tests, but in cross-network environments, most cloud servers and network devices drop custom protocol packets except for AWS, and there are issues such as poor cross-platform compatibility. The final conclusion: Unless necessary, stick to TCP or UDP!

Read more
(github.com)
Development

King of the Grid: A Z80 Sandbox Bot Competition

2025-02-25
King of the Grid: A Z80 Sandbox Bot Competition

A Z80-based sandbox game where developers write bots to compete for dominance on a 32x32 grid. Two bots start in opposite corners, battling for survival by gathering food, moving, and cloning themselves. Written in Z80 assembly or C, bots can utilize shared memory for communication. The last bot standing wins! An online IDE and command-line build process are provided, along with game recording and sharing capabilities. This is an AI programming competition challenging algorithmic efficiency and strategic thinking.

Read more
(github.com)
Game

DeepEP: A High-Performance Communication Library for Mixture-of-Experts

2025-02-25
DeepEP: A High-Performance Communication Library for Mixture-of-Experts

DeepEP is a communication library designed for Mixture-of-Experts (MoE) and expert parallelism (EP), offering high-throughput and low-latency all-to-all GPU kernels (MoE dispatch and combine). It supports low-precision operations, including FP8. Optimized for the group-limited gating algorithm in DeepSeek-V3, DeepEP provides kernels for asymmetric-domain bandwidth forwarding (e.g., NVLink to RDMA). These kernels achieve high throughput, suitable for training and inference prefilling. SM (Streaming Multiprocessors) number control is also supported. For latency-sensitive inference decoding, low-latency kernels using pure RDMA minimize delays. A hook-based communication-computation overlap method is included, requiring no SM resources. The library is tested with InfiniBand and is theoretically compatible with RoCE.

Read more
(github.com)
Development GPU Communication

Uncle Bob and John Ousterhout Debate Software Design

2025-02-25
Uncle Bob and John Ousterhout Debate Software Design

Robert "Uncle Bob" Martin and John Ousterhout engaged in a spirited debate on software design principles, covering key topics such as method length, code comments, and Test-Driven Development (TDD). They fiercely debated the extent of code decomposition, the necessity of comments, and the pros and cons of TDD, using code examples and specific scenarios to support their arguments. This debate highlights the importance of trade-offs in software design and the need to avoid extremes when striving for conciseness and readability.

Read more
(github.com)
Development test driven development

Electro: Blazing-Fast, Lightweight Image Viewer

2025-02-24
Electro: Blazing-Fast, Lightweight Image Viewer

Electro is a lightning-fast, lightweight image viewer built with Rust. Designed with developer experience in mind, it boasts a built-in command terminal and instantly views local and web-hosted images. Its core strength is unparalleled performance—images load near-instantly. Electro is open-source and easily extensible, allowing developers to contribute code or build their own versions.

Read more
(github.com)
Development image viewer

Evolution of the Micro Journal: A Distraction-Free Writing Device

2025-02-24
Evolution of the Micro Journal: A Distraction-Free Writing Device

Un Kyu Lee's Micro Journal is a fascinating evolution of distraction-free writing devices. Starting with a Raspberry Pi and a mechanical keyboard, the project iterated through several versions, each addressing different needs and design challenges. From the foldable Rev.2.ReVamp to the Cherry MX hot-swappable Rev.6, each Micro Journal iteration improves on portability, customization, and the overall writing experience. Rev.7 offers a traditional keyboard layout, while Rev.5 allows connection to a wide range of mechanical keyboards. The story showcases the maker spirit and a relentless pursuit of the perfect writing experience, attracting significant media attention along the way.

Read more
(github.com)
Hardware mechanical keyboard writing device

Python Library for RadiaCode-10x Radiation Detectors

2025-02-24
Python Library for RadiaCode-10x Radiation Detectors

This Python library simplifies interaction with RadiaCode-10x radiation detectors and spectrometers. Features include real-time radiation measurements, spectrum acquisition and analysis, USB and Bluetooth connectivity, and a web interface example. Easily control your device, collect data, and analyze radiation information. Manage device settings, configure display brightness, language, sound, and vibration. Comprehensive examples are provided for both basic terminal output and an interactive web interface.

Read more
(github.com)
Hardware radiation detection python library

FlashMLA: A Blazing-Fast MLA Decoding Kernel for Hopper GPUs

2025-02-24
FlashMLA: A Blazing-Fast MLA Decoding Kernel for Hopper GPUs

FlashMLA is a highly efficient MLA decoding kernel optimized for Hopper GPUs, designed for variable-length sequence serving. Achieving up to 3000 GB/s in memory-bound configurations and 580 TFLOPS in computation-bound configurations on H800 SXM5 using CUDA 12.6, FlashMLA utilizes BF16 precision and a paged kvcache with a 64 block size. Inspired by FlashAttention 2&3 and the cutlass projects, FlashMLA offers significant performance improvements for large-scale sequence processing.

Read more
(github.com)
Development MLA decoding

mdq: A jq for Markdown, Simplifying Document Parsing

2025-02-23
mdq: A jq for Markdown, Simplifying Document Parsing

mdq is a command-line tool that aims to simplify parsing Markdown documents, similar to how jq works with JSON. It allows users to easily extract specific parts of a document, such as to-do checklists in GitHub PRs. mdq supports various selectors covering headings, lists, links, images, code blocks, and more, with regex support. Its syntax mirrors Markdown, making it intuitive. Piping allows chaining filters for complex parsing tasks.

Read more
(github.com)
Development document parsing

WhiteSur: A macOS-like Theme for Linux GTK Desktops

2025-02-23
WhiteSur: A macOS-like Theme for Linux GTK Desktops

WhiteSur brings the macOS aesthetic to your Linux GTK desktop. This highly customizable theme lets you tweak colors, opacity, window controls, Nautilus style, and even Gnome Shell extensions. Installation is a breeze with a simple script. Beyond basic GTK theming, WhiteSur also offers customizations for GDM and Firefox, plus fixes for Flatpak apps, even addressing the challenges of Libadwaita. Want a macOS-inspired Linux experience? Check out WhiteSur!

Read more
(github.com)
Development Linux theme GTK theme macOS style

Directus: Real-time API & App Dashboard for SQL Databases – No Migration Needed!

2025-02-23
Directus: Real-time API & App Dashboard for SQL Databases – No Migration Needed!

Directus is a real-time API and app dashboard for managing SQL database content. It instantly layers a blazing-fast Node.js API on top of any SQL database, supporting PostgreSQL, MySQL, and more, with no migration required. Deploy locally, on-premises, or use their cloud service. Its modern, no-code Vue.js app is intuitive and easy to use. Directus operates under a Business Source License (BSL) 1.1, offering free use for organizations under $5M in annual revenue/funding; larger organizations require a commercial license.

Read more
(github.com)
Development no-code

OpenJKDF2: Open-Source Reimplementation of Jedi Knight: Dark Forces II Engine

2025-02-23
OpenJKDF2: Open-Source Reimplementation of Jedi Knight: Dark Forces II Engine

OpenJKDF2 is a function-by-function reimplementation of the Jedi Knight: Dark Forces II (JKDF2) engine in C, with 64-bit ports for Windows 7+, macOS 10.15+, and Linux. It aims for fidelity to the original, including the original byacc and flex for COG script parsing. A valid copy of JKDF2 is required; the DRM-free GOG version is recommended. Multiple configurations are supported, using OpenGL and WebGL rendering. The project is ongoing, with features like Android and iOS support planned. A WebAssembly demo is available.

Read more
(github.com)
Game

Tetris in PostScript: A Real-time Game in Under 600 Lines

2025-02-22
Tetris in PostScript: A Real-time Game in Under 600 Lines

A developer has implemented a real-time Tetris game using PostScript, remarkably achieving it with only 600 lines of code (around 10KB) and 69 distinct operators. The game features arrow and spacebar controls, increasing game speed, 7 tetrominoes, high scores, and a Nintendo-style scoring system. It runs in GhostView on macOS and draws some implementation inspiration from MeatFighter.

Read more
(github.com)
Game Real-time game

SimpleWall: A Lightweight Alternative to Windows Firewall

2025-02-22
SimpleWall: A Lightweight Alternative to Windows Firewall

SimpleWall is a lightweight (<1MB) Windows firewall alternative compatible with Windows 7 SP1 and later. Based on the Windows Filtering Platform (WFP), it lets users create custom network rules, block Windows telemetry, and supports features like WSL. It boasts a simple interface, supports permanent and temporary rules, and requires manual filter disabling upon uninstallation. SimpleWall works independently of Windows Firewall and is free and open-source.

Read more
(github.com)
Development windows

LLM Agents: Breakthroughs in General Computer Control

2025-02-22
LLM Agents: Breakthroughs in General Computer Control

Recent years have witnessed significant advancements in LLM-powered agents for computer control. From simple web navigation to complex GUI interaction, a plethora of novel reinforcement learning approaches and frameworks have emerged. Researchers explore model-based planning, autonomous skill discovery, and multi-agent collaboration to enhance agent autonomy and efficiency. Some projects focus on specific platforms (e.g., Android, iOS), while others aim to build general-purpose computer control agents. These breakthroughs pave the way for more powerful and intelligent AI systems, foreshadowing a future where agents play a much larger role in daily life.

Read more
(github.com)
AI Agents

FFmpeg Assembly Language: Unlocking High-Performance Multimedia Processing

2025-02-22
FFmpeg Assembly Language: Unlocking High-Performance Multimedia Processing

This tutorial introduces the fundamentals of assembly language programming within FFmpeg, focusing on SIMD vector programming. Writing assembly code by hand can dramatically improve multimedia processing speed, leading to smoother video playback, for example. The tutorial covers basic assembly concepts, the x86-64 instruction set, vector registers, and commonly used tools within FFmpeg. Prior knowledge of C pointers and high school mathematics is required.

Read more
(github.com)
Development Assembly Language

RealDOOM: Running DOOM on 16-bit Processors

2025-02-22
RealDOOM: Running DOOM on 16-bit Processors

RealDOOM is a work-in-progress port of the DOS version of DOOM (based on PCDOOMv2) to real mode, aiming for accuracy to the original game. Currently supporting DOOM1 and DOOM2 WADs, with plans for Ultimate DOOM. Development focuses on ASM rewrites of the render code and restoring removed features like sound and save games. While there are limitations on texture size and node counts, the project boasts performance benchmarks across various processors and quality settings.

Read more
(github.com)
Game

Slime OS: An Experimental App Launcher for PicoVision

2025-02-21
Slime OS: An Experimental App Launcher for PicoVision

Slime OS is an app launcher for the PicoVision (and soon other RP2040 and RP2350 devices), initially designed for the Slimedeck Zero mini-cyberdeck project. It runs in a limited 32-color mode with a 400x240 internal resolution, upscaling to 800x480. Currently, it supports i2c keyboard input, with USB keyboard support planned. The project is experimental and has known issues, including some apps being upside down and limited hardware support, but contributions to expand hardware compatibility are welcome.

Read more
(github.com)
Development app launcher

Seamless Docker to Podman Migration with a Single Script

2025-02-21
Seamless Docker to Podman Migration with a Single Script

Tired of Docker's complexities? `fly-to-podman` is a simple bash script that effortlessly migrates your Docker containers, images, and volumes to Podman. It preserves your container data and configurations (mounts, ports, etc.), allowing for migration of images, volumes, containers, and networks individually or all at once. Transition to a more secure and streamlined containerization experience without root privileges!

Read more
(github.com)
Development Container Migration

Llama 3 from Scratch: A Deep Dive TensorFlow Tutorial

2025-02-21
Llama 3 from Scratch: A Deep Dive TensorFlow Tutorial

This project is an enhanced version of naklecha/llama3-from-scratch, comprehensively improved and optimized to help understand and master the implementation principles and detailed reasoning process of the Llama 3 model. Core improvements include: reorganized content presentation, adjusted directory structure, detailed code annotations, complete matrix dimension change annotations, abundant principle explanations and derivations, an added KV-Cache derivation chapter, and bilingual (Chinese and English) documentation. The tutorial starts by loading model files and configuration files, then guides through text-to-embedding conversion, Transformer block construction, attention mechanism implementation, positional encoding (RoPE), RMS normalization, SwiGLU feed-forward network, and finally predicts the next token. It also explores top-k predictions, the impact of different token embeddings, and the principles and advantages of the KV-cache mechanism.

Read more
(github.com)
Development

Txeo: A Modern C++ Wrapper for TensorFlow Achieving Near-Native Performance

2025-02-21
Txeo: A Modern C++ Wrapper for TensorFlow Achieving Near-Native Performance

Txeo is a lightweight and intuitive C++ wrapper for TensorFlow designed to simplify TensorFlow C++ development while maintaining high performance and flexibility. Built entirely with Modern C++, Txeo enables developers to use TensorFlow with the ease of a high-level API, eliminating the complexity of its low-level C++ interface. Benchmarks show negligible performance overhead compared to native TensorFlow, ranging from 0.65% to 1.21%. Currently supports Linux, with Windows and macOS support planned.

Read more
(github.com)
Development

CSS Zero: A No-Build CSS Starter Kit for Rails

2025-02-21
CSS Zero: A No-Build CSS Starter Kit for Rails

CSS Zero is a streamlined CSS starter kit for Ruby on Rails applications, offering a 'no-build' experience similar to a Tailwind CSS alternative without the build process. Simply add the gem, run the install command (`bin/rails generate css_zero:install`), and you're ready to go. It provides utility classes and variables, and features custom templates for scaffolds and authentication. Lucide is recommended for high-quality icons. The project is open-source under the MIT License and welcomes bug reports and pull requests.

Read more
(github.com)
Development Starter Kit

eserde: Reporting Multiple Deserialization Errors at Once

2025-02-21
eserde: Reporting Multiple Deserialization Errors at Once

The serde library aborts deserialization upon encountering the first error, which is inconvenient when dealing with user-provided JSON payloads (e.g., a REST API request body). eserde solves this by reporting all deserialization errors at once, significantly improving the developer experience. By replacing `#[derive(serde::Deserialize)]` with `#[derive(eserde::Deserialize)]` and using eserde's deserialization functions, developers can easily obtain all error messages, reducing the number of API interactions. eserde currently supports JSON and plans to support YAML and TOML in the future.

Read more
(github.com)
Development Deserialization

DeepSeek Opensources 5 AGI Repos: A Humble Beginning

2025-02-21
DeepSeek Opensources 5 AGI Repos: A Humble Beginning

DeepSeek AI, a small team pushing the boundaries of AGI, announces it will open-source five repositories over the next week, one per day. These aren't vaporware; they're battle-tested production-ready building blocks of their online service. This open-source initiative aims to foster collaborative progress and accelerate the journey towards AGI. Accompanying this release are two research papers: a 2024 AI Infrastructure paper (SC24) and a paper on Fire-Flyer AI-HPC, a cost-effective software-hardware co-design for deep learning.

Read more
(github.com)
AI AI Infrastructure

DotSlash: Streamlining Executable Deployment

2025-02-20
DotSlash: Streamlining Executable Deployment

DotSlash is a command-line tool that simplifies managing platform-specific executables. Instead of storing multiple binaries and shell scripts, you use a single, human-readable text file. This makes version control easier and improves reproducibility by reducing reliance on the host environment. The first run downloads and verifies the necessary binaries; subsequent runs are instantaneous. It's a powerful way to efficiently manage dependencies in your projects.

Read more
(github.com)
Development development tool executable deployment reproducible builds

Lox: A Modern Astrodynamics Library for Space Missions

2025-02-20
Lox: A Modern Astrodynamics Library for Space Missions

Lox is a safe and ergonomic astrodynamics library for the modern space industry. It offers a comprehensive API, ranging from high-level mission planning and analysis tools to lower-level utilities. Supporting various coordinate frames, it includes ephemeris data for major celestial bodies and readily handles Earth orientation parameters. Lox also provides Python bindings for interactive use and is extensible, allowing users to add custom time scales, transformation algorithms, and data sources. Commissioned by the European Space Agency, it's a next-generation, open-source space mission simulator.

Read more
(github.com)
Development Astrodynamics Space Mission Simulator Open-Source Library
1 2 3 4 5 6 8 10 11 12 … 20 21