Coding & Development

The Open VLM Leaderboard, hosted on Hugging Face, provides a comprehensive platform for viewing and analyzing the performance of various vision-language models (VLMs). It aggregates evaluation results from the VLMEvalKit benchmark, offering a centralized resource for researchers and developers. Users can easily narrow down results by selecting specific evaluation dimensions, filtering by model size or type, or searching for a particular model name. This tool is designed to facilitate the comparison and understanding of VLM capabilities, aiding in the development and selection of appropriate models for different applications. It serves as a valuable resource for anyone working with or interested in the advancements of vision-language AI.

esProc SPL is a JVM-based programming language specifically designed for structured data computation, serving as both a powerful data analysis tool and an embedded computing engine. Unlike traditional text-based languages, SPL code is written in an intuitive, Excel-like grid interface, providing real-time, step-by-step execution results. This unique approach significantly reduces the learning curve for beginners while offering an excellent interactive experience. SPL integrates standard programming language features such as branching, looping, and recursion, alongside robust set operation capabilities for concise code. It aims to overcome the limitations of SQL by integrating the advantages of SQL and Java, enabling complex data processing tasks that typically require SQL+Python, all within SPL. Developed entirely in Java, esProc seamlessly integrates into Java applications and supports a wide array of data sources, including RDBs and NoSQL, making it ideal for lightweight multi-data source hybrid computations and embedded report query systems.

NativeBridge.io offers a seamless solution for running native mobile applications directly in a browser, eliminating the need for complex setups. It supports both Android and iOS, providing access to real and emulated devices for comprehensive testing and demonstrations. This platform is designed to facilitate instant demos, reviews, collaboration, and testing, making it an ideal tool for developers and teams. By offering a fast and accessible environment, NativeBridge.io streamlines the mobile app development lifecycle, ensuring that applications can be tested and shared efficiently across various devices without any prior configuration.

— Hub API Playground — is a free, web-based tool designed for interacting with the Hugging Face Hub API. It enables users to easily search for and retrieve information about AI models available on the Hugging Face platform. Users can input keywords, author names, tags, and various filters such as limit and sort order to refine their searches. Upon sending a request, the playground returns a JSON list of matching models, making it a valuable resource for developers and AI enthusiasts who want to experiment with the Hugging Face API without writing extensive code. This tool simplifies the process of discovering and understanding the vast collection of models on the Hub.

Quantization Dedup is a specialized tool hosted on Hugging Face Spaces, designed to help users visualize and understand the distribution of duplicate content within code repositories. It provides insights into how much content is shared between different files, which is crucial for optimizing storage, improving transfer efficiency, and managing codebases more effectively. The tool specifically focuses on deduplication from 'quants' in models like 'bartowski/gemma-2-9b-it-GGUF', indicating its relevance for analyzing and optimizing quantized AI models. By offering a clear view of content redundancy, Quantization Dedup assists developers and researchers in identifying areas for optimization within their AI infrastructure.

mosesdecoder is a comprehensive, open-source machine translation system designed for researchers and developers in the field of statistical machine translation. It provides a robust framework for building and experimenting with machine translation models. The system is highly customizable, allowing users to adapt it to specific language pairs and domains. Its open-source nature encourages community contributions and extensions, making it a versatile tool for advancing machine translation technologies. The project includes various components for tasks such as language model training, phrase extraction, and decoding, making it a complete solution for developing and deploying translation systems.

Latent Consistency Models (LCMs) is an open-source platform hosted on Hugging Face Spaces, designed for AI researchers and developers to explore and experiment with latent consistency models. While the live application currently displays a runtime error, the underlying project aims to provide a space for engaging with advanced AI algorithms. This tool is particularly relevant for those interested in the technical aspects of AI model development and research, offering a foundation for understanding and manipulating latent spaces within AI models. It serves as a community-driven resource for advancing knowledge in the field of AI.

LongVU is an AI tool hosted on Hugging Face Spaces that enables users to interact with visual content by uploading videos or images and posing questions or comments. The application then processes the visual input and generates detailed text responses, providing insights and information derived from the content. This functionality makes LongVU a valuable resource for researchers and developers focused on video analysis, image understanding, and general visual content interpretation. It leverages advanced AI models to bridge the gap between visual data and textual explanations, facilitating deeper engagement with multimedia.

SegLossOdyssey is an open-source repository offering a comprehensive collection of loss functions specifically designed for medical image segmentation. This tool is invaluable for researchers and practitioners aiming to enhance the accuracy and robustness of their segmentation models, particularly in tasks involving highly imbalanced data. The collection includes implementations in PyTorch and Keras, covering a wide array of loss functions from various research papers and challenges. It highlights the effectiveness of compound loss functions for challenging segmentation tasks and provides a valuable resource for exploring and applying state-of-the-art loss functions in medical imaging.

ObjectBox Go Database is an embedded Go database designed for high performance and resource efficiency, serving as a fast alternative to SQLite and GORM. It offers an intuitive native Go API for persisting objects quickly and sustainably. The database is optimized for restricted devices like IoT gateways and microcontrollers, ensuring minimal CPU, power, and memory usage. Key features include built-in object links/relationships, multiplatform support (Linux, Windows, Android, iOS, macOS), and scalability for millions of objects. It also provides flexible data querying, static typing for compile-time checks, and automatic schema migrations, eliminating the need for manual update scripts. Additionally, ObjectBox offers extensions like ObjectBox Sync for data synchronization and ObjectBox TS for time-series data.

github-calendar is an open-source JavaScript library designed to embed GitHub contribution calendars directly into web pages. Developers can easily integrate their GitHub activity, including contributions, issues opened, and commits made, onto personal websites or portfolios. The library offers options for responsiveness, tooltips on calendar days, and the ability to use a custom proxy for data fetching. It supports both direct script inclusion and CommonJS environments via npm or yarn, making it versatile for various web development setups. The tool is ideal for showcasing development activity and stats to a wider audience.

Trilogy is a high-performance client library specifically designed for MySQL-compatible database servers, emphasizing flexibility and ease of embedding into various applications. It is currently in production use on github.com, showcasing its robustness and reliability. The library supports essential functionalities of the MySQL text protocol, including handshake, password authentication, query, ping, and quit commands. Furthermore, it offers support for prepared statements using the binary protocol. Trilogy provides a low-level protocol API that is completely decoupled from IO, alongside both non-blocking and blocking client APIs. It boasts minimal dependencies, requiring only POSIX, the C standard library, and OpenSSL, making it highly portable and efficient.

info-nce-pytorch offers a PyTorch implementation of the InfoNCE loss function, a critical component for self-supervised learning. This tool enables developers and researchers to effectively apply contrastive learning techniques, where the goal is to learn representations by pulling similar samples closer together and pushing dissimilar samples further apart in an embedding space. The package is easily installable via pip and provides flexible usage options, including scenarios with and without explicit negative keys, as well as paired and unpaired negative modes. This makes it a versatile solution for various contrastive learning setups in AI model development.

pytorch-paligemma is an open-source project hosted on GitHub, offering a PyTorch implementation of a multimodal (vision) language model. It stands out by providing a comprehensive, step-by-step explanation of how to build such a model from scratch, making it an invaluable resource for developers and researchers. The project is accompanied by a detailed YouTube video tutorial, enhancing the learning experience. This tool is ideal for those looking to understand the underlying mechanics of multimodal AI models, experiment with PyTorch, or integrate similar capabilities into their own projects. Its focus on transparency and education makes complex AI concepts accessible.

Pokemon Showdown is an application hosted on Hugging Face Spaces that enables users to view any web page within a full-screen iframe by providing its URL. While the name suggests a focus on Pokemon battles, the current functionality described is a general web page viewer. The platform leverages Hugging Face's infrastructure, offering various pricing tiers for compute resources, storage, and inference endpoints. Users can access different CPU and GPU hardware options, including Nvidia T4, L4, L40S, A10G, A100, H100, H200, and B200, with hourly pricing. It also provides options for dedicated inference endpoints and data storage, catering to a range of AI development and deployment needs.

vectordb, hosted on GitHub, offers a range of plans tailored for developers, from individuals to large enterprises. The platform provides essential features like unlimited public and private repositories, Dependabot security updates, and CI/CD minutes for automating software development workflows. Users can also host software packages and manage projects with integrated Issues & Projects. For teams, advanced collaboration tools such as repository rules, multiple reviewers in pull requests, and code owners are available. Enterprise plans further enhance security, compliance, and flexible deployment options, including data residency and enterprise managed users, making it suitable for diverse development needs.

CypherScribe is a no-code platform designed to rapidly create and launch interactive, SEO-optimized developer documentation. Users can connect their data sources, customize the appearance with themes, colors, and logos, and claim a custom subdomain for their documentation app, all without writing a single line of code. The platform supports a wide range of content types including headings, paragraphs, tables, code blocks, media, lists, and accordions, and even allows PDF/CSV uploads. It features a rich editor with Markdown support and custom blocks for multilingual code snippets, banners, and toasts. CypherScribe also includes built-in SEO optimization, a search algorithm, and an AI bot trained on your data to assist users. It aims to offload documentation burdens from development teams by providing a fast, customizable, and performant solution.

crystalruby is a Ruby gem designed to embed Crystal code directly into Ruby applications, enabling developers to leverage Crystal's performance benefits for CPU or memory-intensive operations. By simply annotating Ruby methods with `crystallize`, developers can compile and execute these methods in Crystal, achieving significant speed improvements (e.g., 50x faster for prime counting). The gem supports defining parameter and return types, handling both Ruby-compatible and Crystal-only syntax. It also offers advanced features like passing data by reference for efficiency, calling Ruby methods from Crystal, and integrating with Crystal shards like Kemal for web development. This allows for hybrid applications that combine Ruby's flexibility with Crystal's raw speed.

Daily Unfold is an engaging digital puzzle game centered around the concept of paper folding, offering a unique blend of spatial reasoning and brain-teasing fun. Players predict where holes will appear when a folded and hole-punched piece of paper is unfolded. The game provides three new puzzles daily—easy, medium, and hard—and unlocks an Expert mode, a timed version of the hard puzzle with a public leaderboard, after all three are completed. Daily Unfold is free to play, with streaks and progress stored locally in the browser, and is supported by ads shown at natural breaks. It functions as a Progressive Web App, allowing for offline play once loaded.

Revise.js is a foundational JavaScript library designed to simplify the development of contenteditable-based web text editors. It addresses the complexities of working with the native `contenteditable` attribute by offering key building blocks: a `<content-area>` web component that reads the DOM as a clean string value, an algebraic `Edit` data structure for describing and manipulating text changes, and a declarative component model. This library enables developers to build highly customized rich-text editing experiences without the overhead of full-fledged editor frameworks. It supports features like undo/redo history, collaborative editing, and stable keys for line-based rendering, making it ideal for creating robust and flexible text editing solutions.

Unity is the world's leading 3D game engine, providing a comprehensive suite of tools for developing, deploying, and growing interactive experiences. It enables creators to build games for a wide array of platforms, from mobile and PC to console and XR (extended reality). Beyond gaming, Unity is also utilized for industrial applications. The platform offers various subscription tiers, including Personal, Pro, and Enterprise, catering to different user needs and project scales. Users fully own the content they create with Unity, even if they cancel their subscription. Unity also provides educational plans for students and educators, along with extensive learning resources and support.

This open-source book, "Mathematical Foundations of Reinforcement Learning," offers a mathematically rigorous yet accessible introduction to the core concepts, problems, and algorithms in reinforcement learning. Designed for senior undergraduate students, graduate students, researchers, and practitioners, it requires no prior reinforcement learning background but assumes knowledge of probability theory and linear algebra. The book carefully controls mathematical depth, providing illustrative examples based on a grid world task to clarify complex ideas. It is coherently organized, building each chapter on the preceding one, and is complemented by lecture slides and a highly-viewed video series available in both Chinese and English.

opengv is an open-source library offering a comprehensive suite of computer vision methods for tackling geometric vision problems. Developed and maintained by the Mobile Perception Lab of ShanghaiTech, it provides solutions for absolute-pose, relative-pose, triangulation, and point-cloud alignment. The library supports both central and non-central camera models and can be integrated into random sample consensus or nonlinear optimization contexts. It also includes convenient Matlab and Python interfaces, making it accessible for various research and development applications in areas like 3D reconstruction and camera pose estimation.

nerf_and_beyond_docs is a curated collection of documents and topics focused on Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS). This GitHub repository serves as a central hub for papers, discussions, and related technologies accumulated by the NeRF/3DGS & Beyond community. It is designed to help researchers, students, and engineers stay up-to-date with the latest advancements in 3D vision. The collection is actively maintained, with new resources added daily, and includes notes on various works to facilitate understanding. Users can join discussions, contribute, and access a companion book on NeRF/3DGS, making it an invaluable resource for both beginners and experienced professionals in the field.