Coding & Development

Deep-Metric-Learning-Baselines provides a comprehensive and easily extendable PyTorch implementation for deep metric learning pipelines. This open-source tool allows developers to test and implement current and new deep metric learning methods. It includes implementations and dataloaders for various loss functions like Triplet Loss, Margin Loss, ProxyNCA, and N-Pair Loss, as well as sampling methods such as Random, Softhard, Semihard, Distance, and N-Pair Sampling. The repository also supports popular datasets like CUB200-2011, CARS196, Stanford Online Products, and In-Shop Clothes, and integrates architectures like ResNet50 and GoogLeNet. It's designed for flexibility, allowing users to extend or test other sampling or loss methods by adding them to the existing framework.

TiDB Cloud is a comprehensive, fully managed database service designed to accelerate application development. It provides a one-stop, AI-ready database solution that is MySQL-compatible and offers high availability. The platform is built for scalability, allowing users to auto-scale their databases in seconds to meet fluctuating demands. By leveraging TiDB Cloud, developers can reduce costs and streamline their development processes, making it an efficient choice for modern applications requiring robust and flexible data management.

TiKie is a mobile application designed for fantasy and roleplay enthusiasts, providing an immersive AI-driven experience. Users can engage in dynamic conversations with a diverse array of virtual characters, fostering interactive storytelling. The platform enables the creation and customization of unique AI companions, allowing narratives to evolve based on user choices. This continuous discovery aspect ensures a personalized and engaging experience for those looking to craft their own fantasy worlds and interact with AI-powered characters.

annyang is a lightweight JavaScript library designed to bring speech recognition capabilities to any website. It allows developers to easily integrate voice commands, enabling users to control their site through spoken instructions. The library boasts no dependencies, a minimal footprint of just 2 KB, and is freely available under the MIT license, making it an accessible solution for adding interactive voice features. It supports defining custom commands and provides a simple API for starting and stopping recognition. For enhanced user experience, annyang can be paired with Speech KITT, a GUI library that offers visual feedback and customizable themes for the speech recognition interface.

ai-cookbook is an open-source resource designed for developers seeking to build AI systems. It offers a collection of examples and tutorials, complete with copy/paste code snippets, to facilitate easy integration into various projects. The cookbook is maintained by Dave Ebbelaar, an AI engineer and founder of Datalumina, who also shares practical tutorials on his YouTube channel. Beyond the cookbook, Dave provides additional resources including a free five-hour Python course for AI beginners and a program for developers looking to build and deploy end-to-end GenAI solutions. This tool is ideal for those who want practical, real-world applications of AI development.

StackRef provides expert services in cloud architecture, infrastructure, and security, covering AWS, GCP, and Azure. Their team of CISSP-certified DevOps engineers helps customers optimize and understand their cloud architecture and costs, ensuring creations are well-organized and secure. Key services include designing scalable cloud solutions, optimizing cloud infrastructure, ensuring robust security and compliance, and providing 24/7 support and monitoring. Additionally, StackRef offers its own self-hosted, soup-to-nuts internal hackathon manager, providing a comprehensive solution for organizations looking to run their own hackathons.

Browser is an AI tool hosted on Hugging Face Spaces, designed to help users explore and understand various AI models. It offers a user-friendly interface to search, filter, and browse a comprehensive collection of models. Users can view detailed information about each model, including statistics and descriptions, within a convenient pop-up window. The tool also provides preview images, making it easier to visualize and assess models. This platform is ideal for anyone looking to discover, compare, and learn about different AI models in an organized and accessible manner.

Candle Phi Wasm Demo is a demonstration application built on Hugging Face Spaces that allows users to generate text from prompts. This tool showcases the capabilities of the Candle Phi model within a WebAssembly (Wasm) environment, enabling efficient execution directly in the browser. Users can input any text prompt and receive a generated response or continuation, making it suitable for exploring AI text generation. The application supports various models and advanced operations, providing a flexible platform for experimenting with different text-based AI outputs. It is an open-source project, highlighting its accessibility and potential for further development and integration.

openai-proxy is an open-source solution designed to provide a proxy for OpenAI/ChatGPT APIs, primarily to circumvent geographical access restrictions. Many users encounter issues where OpenAI detects API access from certain regions, leading to account suspensions. This tool offers a workaround by allowing developers to deploy a proxy on a server in an unrestricted location. It supports deployment via Deno Deploy or CloudFlare Workers, making it flexible for various setups. The project provides clear instructions for integration with official OpenAI/ChatGPT npm and Python packages, simplifying the process for developers to continue using OpenAI's services seamlessly.

Niyoto Technologies Pvt. Ltd. is a deep tech startup specializing in hardware and software solutions for the defense sector. They leverage a proprietary, state-of-the-art AI platform and custom-built hardware to address diverse problems. A winner of the iDEX Defence India Startup Challenge, Niyoto supports the “Make in India” initiative. Their key products include AI software for tunnel detection, landmine and IED detection with pinpoint accuracy, and buried object/crevasse detection in challenging environments like Himalayan glaciers. The platform features powerful AI, remote operation for operator safety, rapid data acquisition, intuitive UI, autonomous flying capabilities, non-destructive data acquisition, end-to-end encryption, and near real-time analytics. Niyoto also aims to develop dual-use technologies for applications beyond defense, such as utility mapping, underground pipe leakage detection, archaeological site mapping, and disaster relief operations.

UniDetector is an open-source computer vision tool designed for universal object detection, providing the code release for the CVPR 2023 paper "Detecting Everything in the Open World: Towards Universal Object Detection." Built upon mmdetection v2.18.0 and requiring CLIP, this tool facilitates both single-dataset and multi-dataset training, as well as open-world inference. It supports end-to-end and decoupled training/inference workflows, including probability calibration. UniDetector is ideal for researchers and developers working on advanced object detection tasks, offering robust capabilities for preparing datasets, language CLIP embeddings, and pre-trained RegionCLIP parameters.

GLM-ASR-Nano is a robust, open-source speech recognition model featuring 1.5 billion parameters, designed to handle real-world complexities. It surpasses OpenAI Whisper V3 in multiple benchmarks while maintaining a compact size. Key capabilities include exceptional dialect support, particularly for Cantonese and other dialects, effectively bridging gaps in dialectal speech recognition. The model is also specifically trained for "Whisper/Quiet Speech" scenarios, accurately transcribing extremely low-volume audio that traditional models often miss. GLM-ASR-Nano achieves a state-of-the-art average error rate of 4.10 among comparable open-source models, demonstrating significant advantages in Chinese benchmarks like Wenet Meeting and Aishell-1. It supports 17 languages with high usability, with specific optimizations for certain regions.

Ara.so, formerly EasyClaw, is an innovative AI tool that transforms a simple text message into a fully deployed website within approximately 30 seconds. Users can send an SMS describing their desired website, and Ara.so handles the entire creation and deployment process, eliminating the need for sign-ups or complex editors. It supports various website types, including coffee shop menus, personal portfolios, SaaS pricing pages, and landing pages. The platform offers different plans, from a free tier with one active site to Ultra and Teams plans providing unlimited sites, custom domains, faster generation, and dedicated support, catering to both individual users and collaborative groups.

Creatotech is a web design and development agency offering custom website solutions and AI model development. Their team of engineers focuses on creating visually stunning and highly functional websites that perform flawlessly across all devices. They provide services such as website builds and rebuilds, hosting, maintenance, and Generative Engine Optimization (GEO) to improve website visibility in an AI-driven world. Creatotech also assists companies in leveraging technology to its fullest, including custom AI models, and offers financing options like SBA loans for projects. They emphasize helping businesses grow through SEO, mobile optimization, e-commerce solutions, ongoing support, security, and analytics.

Clockworks is an AI development company with over a decade of experience in applied AI, dedicated to delivering practical and effective AI solutions. They focus on transforming cumbersome, manual, time-consuming, and error-prone business processes into efficient and enjoyable workflows. Clockworks emphasizes user-friendliness, building solutions that adapt to human behavior and changing situations. Their custom-built AI solutions are designed for accuracy, consistently outperforming existing processes, and are scalable to handle varying data and traffic loads. Utilizing their proprietary Clockworks OS, they aim to build solutions quickly and efficiently while managing cloud costs, making AI accessible and affordable for businesses.

Diff4RLSurvey is a GitHub repository serving as a curated collection of resources and academic papers focused on Diffusion Models for Reinforcement Learning (RL). This open-source resource accompanies the survey paper titled "Diffusion Models for Reinforcement Learning: A Survey." The repository meticulously categorizes papers into key areas such as Offline Reinforcement Learning, Online Reinforcement Learning, Imitation Learning, Trajectory Generation, and Data Augmentation. Each entry typically includes links to the paper and, where available, the corresponding code. It is an invaluable resource for researchers and academics looking to explore the application of diffusion models in various aspects of sequential decision-making.

DIMO is a Brazilian technology company based in Chapecó, specializing in custom software development. They focus on transforming business ideas into innovative technological solutions, with a strong emphasis on agrotech, artificial intelligence, computer vision, and management systems. DIMO offers personalized software development, ensuring high performance and efficiency for clients across Brazil. Their team comprises experienced developers (back-end, front-end, full stack) and designers focused on user experience. As part of the INCTECh incubator and MIDITEC network, DIMO operates from Pollen Parque Científico e Tecnológico, processing over 200TB of data and serving more than 15,000 users with 99.99% system availability.

Bluechip Technologies Asia is a leading AI development company based in Sri Lanka, specializing in building intelligent solutions for businesses across various industries. They provide a full spectrum of AI services, including strategic AI consulting to identify high-impact use cases, rapid innovation through AI accelerators, and end-to-end AI implementation. The company also offers MLOps expertise for scalable AI operations, custom AI product development from prototyping to MVPs, and enterprise AI solutions designed to integrate seamlessly into existing operations. Bluechip Technologies Asia aims to empower businesses with AI, blockchain, and emerging technologies to drive transformation, efficiency, and competitive advantage.

Pagecloud is a versatile no-code website builder designed for creating custom websites, popups, and banners. It features an intuitive drag-and-drop editor, enabling users to design stunning sites that represent their brand without needing any coding knowledge. The platform offers a wide range of customizable templates for various industries, including restaurants, business, and photography. Users can optimize their online presence with privacy-first analytics, tracking visits, clicks, and form submissions. Pagecloud also helps drive conversions with custom forms and lead capture tools, and supports SEO with tools for custom domains and keyword research. It caters to small businesses, professional services, marketers, and agencies, offering features like AI content generation for blogs, team collaboration, and e-commerce capabilities.

Graph-Learn, formerly AliGraph, is a robust and distributed framework designed for the development and application of large-scale graph neural networks (GNNs). Developed by Alibaba, it has been successfully deployed in various industrial scenarios such as search recommendation, network security, and knowledge graphs. The framework offers a comprehensive solution encompassing both GNN training and online inference services. Its training component supports sampling on batch graphs and incremental GNN model training, compatible with TensorFlow and PyTorch. The online inference service, Dynamic-Graph-Service, ensures real-time sampling on dynamic graphs with streaming updates, boasting P99 latency within 20ms for large-scale graphs. It provides Python, C++, and Java interfaces for flexible integration.

Benchmark Finder is a specialized AI tool designed for exploring and analyzing machine learning benchmark tasks within the Lighteval library. Users can efficiently navigate through a comprehensive index of benchmarks, utilizing keyword searches to pinpoint specific tasks. The tool also offers robust filtering options, allowing users to narrow down results based on language support, which is crucial for multilingual model development. Furthermore, tasks can be sorted by benchmark type, providing a structured way to compare and evaluate different models. This interface is particularly useful for researchers, developers, and professors who need to inspect and understand the performance characteristics of various AI models against established benchmarks.

Baseline Trainer is a Hugging Face Space developed by scikit-learn, designed to facilitate the training of baseline machine learning models and the analysis of datasets. Users can upload a CSV file, provide their Hugging Face token, and specify a target column for either training a model or performing data analysis. This tool is particularly useful for quickly establishing performance benchmarks, which is a crucial step in any machine learning project. While the Space is currently paused, its intended functionality provides a straightforward way to get started with model training or data exploration, making it valuable for educational purposes and for comparing the effectiveness of different models.

HLearn is a high-performance machine learning library developed in Haskell, designed to offer both speed comparable to low-level languages like C/C++ and flexibility akin to high-level languages such as Python. It distinguishes itself by leveraging functional programming principles and the SubHask library for fast numerical computations. The library's design is deeply rooted in abstract algebra, utilizing concepts like homomorphisms, monoids, and Abelian groups to enable features such as parallel batch training, online training, fast cross-validation, and weighted data points. HLearn also incorporates a unique History monad for debugging optimization procedures without runtime overhead. While it's a research project aiming for an optimal interface, its current focus is on foundational algebraic structures rather than a broad range of popular machine learning techniques.

GigaBrain-0 is an innovative vision-language-action (VLA) foundation model designed for generalist robots, addressing the high cost and time consumption of collecting large-scale real-world robot data. By utilizing world models to generate diverse data at scale, GigaBrain-0 substantially decreases the need for real robot data while enhancing cross-task generalization. The model also improves policy robustness through RGBD input modeling and embodied Chain-of-Thought (CoT) supervision, enabling it to reason about spatial geometry, object states, and long-horizon dependencies. This leads to significant performance gains in dexterous, long-horizon, and mobile manipulation tasks, demonstrating superior generalization across variations in appearances, object placements, and camera viewpoints. The project also includes GigaBrain-0.1, which scales training data to 10k hours and achieves first place on the RoboChallenge leaderboard.