Coding & Development

AIStartups is a comprehensive, open-source directory hosted on GitHub, meticulously curating a list of startups and teams dedicated to artificial intelligence. The resource categorizes companies by their country of origin, with a significant focus on both Chinese and international AI ventures. It highlights their core technologies, such as deep learning, machine learning, natural language processing, and computer vision, and often includes direct links to their websites. The directory serves as a valuable reference for anyone interested in the AI startup ecosystem, offering insights into companies specializing in diverse applications like medical AI, autonomous driving, financial NLP, and conversational AI.

Voice Agent WebRTC + LangGraph is a powerful AI tool developed by NVIDIA, designed for creating interactive voice agents. It leverages WebRTC for real-time communication, LangGraph for agent orchestration, Automatic Speech Recognition (ASR) to convert spoken language into text, and Text-to-Speech (TTS) to vocalize translated text. Users can speak into the application, and it processes their voice by converting it to text, translating it, and then speaking the translated text back. This eliminates the need for manual typing, offering a seamless and intuitive voice interaction experience. It's hosted on Hugging Face Spaces, making it accessible for developers and researchers to experiment with and build advanced voice applications.

Wan2.2-T2V-A14B is an AI-powered tool available on Hugging Face that allows users to create videos by uploading an image and providing a descriptive text prompt. This tool is designed for efficient video generation, offering adjustable settings such as video duration and resolution to tailor the output to specific needs. It leverages `app_fast.py` for fast API operations and `app_t2v` which is a 14B model, indicating a robust underlying architecture for video synthesis. The platform is accessible via a web interface, making it easy for users to experiment with different inputs and settings to achieve desired video outcomes. It's particularly useful for those looking to quickly transform static images into dynamic video content with minimal effort.

cgft-llm is an open-source GitHub repository dedicated to providing practical learning resources for large language models (LLMs). It offers a comprehensive series of code examples, detailed documentation, and video tutorials covering various aspects of LLM development, including Agent systems, core LLM technologies like fine-tuning and RAG, and integration with open-source projects. The resource is structured into several modules, guiding users through topics such as LLM deployment with tools like vLLM and Ollama, data preparation for fine-tuning, and advanced concepts like function-calling and tool-use. It also includes sections on Python engineering practices for open-source contributions and creative AI projects, making it suitable for developers and researchers looking to deepen their understanding and practical skills in the LLM domain.

Fay is an open-source AI agent framework designed to bridge the gap between digital humans (2.5D, 3D, mobile, PC, web) or large language models (OpenAI compatible, DeepSeek) and business systems. It provides a stable and comprehensive solution for developing digital human applications, allowing for flexible integration of TTS, ASR, and various digital human models. Fay supports both server and standalone modes, offering features like multi-user concurrent access, text and voice interaction interfaces, digital human driving interfaces, and automatic broadcast capabilities. It also includes agent autonomous decision-making, adaptive memory, and a configuration management center, making it suitable for a wide range of applications from embedded devices to websites.

Cornucopia-LLaMA-Fin-Chinese is an open-source project that provides a series of Chinese financial large language models based on the LLaMA architecture. It is specifically designed for financial knowledge question answering, having been instruction fine-tuned with extensive Chinese financial data. The project also offers an efficient and lightweight training framework for developing vertical domain LLMs, covering stages like pretraining, SFT, RLHF, and quantization. It leverages public and crawled Chinese financial Q&A data, including topics such as insurance, wealth management, stocks, funds, loans, credit cards, and social security. The project aims to continuously expand its high-quality instruction dataset using GPT-3.5/4.0 APIs and integrate with Chinese financial knowledge graphs and CFLEB financial datasets, with plans to release new Chinese financial models for various scenarios.

Linly is an open-source project that offers a suite of Chinese large language models, including Chinese-LLaMA (versions 1 and 2), Chinese-Falcon, and Linly-OpenLLaMA. These models are built upon LLaMA and Falcon architectures, enhanced with extensive Chinese and English parallel corpora for incremental pre-training, extending their linguistic capabilities to Chinese. The project also features Linly-ChatFlow, a Chinese conversational model trained on large-scale instruction data. Linly provides comprehensive resources, including data preparation, model training, and evaluation code, ensuring reproducibility. It supports various quantization schemes for deployment on CUDA and edge devices, making it versatile for different applications. The Linly-OpenLLaMA models, available in 3B, 7B, and 13B scales, are trained from scratch on 1TB of mixed English and Chinese data, featuring an optimized tokenizer for Chinese characters and words, and are released under an Apache 2.0 license for commercial use.

MemMachine Playground is an official platform by Memverge designed for experimenting with AI memory for large language models (LLMs) and AI agents. This tool allows users to explore various memory configurations and understand their impact on AI performance and capabilities. It serves as a valuable resource for AI developers and researchers who are focused on enhancing the intelligence and efficiency of their AI systems. The playground offers a hands-on environment to test and refine memory strategies, ultimately contributing to the development of more robust and effective AI applications.

PaddleNLP is a comprehensive development suite built on the PaddlePaddle deep learning framework, designed for large language models (LLMs). It facilitates efficient training, lossless compression, and high-performance inference of models across diverse hardware, including NVIDIA GPUs, Kunlun XPU, Ascend NPU, and more. The library emphasizes ease of use and extreme performance, aiming to empower developers in creating industrial-grade large model applications. Key features include 4D high-performance training with data, tensor, and pipeline parallelism, efficient fine-tuning algorithms, and a high-performance inference module with dynamic insertion and operator fusion. It also supports a wide range of popular LLM series like LLaMA, Baichuan, Bloom, ChatGLM, Gemma, Mistral, OPT, and Qwen.

ML-NLP is an open-source GitHub project designed to be a comprehensive resource for individuals preparing for interviews in Machine Learning, Deep Learning, and Natural Language Processing. It covers frequently tested knowledge points and provides practical code implementations, making it an invaluable theoretical foundation for aspiring algorithm engineers. The project is structured into various modules, offering a clear and organized knowledge system. Each chapter includes potential interview questions and concludes with practical algorithm case studies. It is intended for continuous learning, review, and as a quick reference during interview preparation.

NeuroAPI offers API access to leading text-based neural networks, including ChatGPT 3.5 Turbo, ChatGPT 4o, and Claude-4, bypassing the need for VPNs or international payment methods. Designed for developers and applications, it provides an API-only interface, focusing exclusively on text-based requests. A key differentiator is its cost-effectiveness, with queries being up to 30% cheaper than official rates. The service supports flexible payment options, including cryptocurrencies and MIR cards. While it currently does not support `function_call` or `tool_calls`, and image generation is in closed beta, it provides free access to `gpt-3.5-turbo` via API and an online chat interface. Users need basic API knowledge to integrate the service into their applications.

SentimentPolarityAnalysis is an open-source project providing various methods for sentiment polarity analysis. It includes implementations for feature extraction and sentiment classification using diverse algorithms such as sentiment dictionaries, k-Nearest Neighbors (k-NN), Naive Bayes, Maximum Entropy, and Support Vector Machines (SVM). The tool allows users to analyze single sentences or entire files, with options for outputting results to files or displaying them in the console. It also provides functionality for evaluating the accuracy of different classification models, making it suitable for research and development in natural language processing.

Zhihu is a GitHub repository offering a collection of source code for various Artificial Intelligence projects, primarily focusing on Natural Language Processing (NLP) and Computer Vision (CV). Implemented in Python 3.6, this repository serves as a practical resource for developers and researchers. It features diverse projects such as text generation using RNNs (LSTM), machine translation with Seq2Seq models and attention mechanisms, deep convolutional Generative Adversarial Networks (GANs) for image generation, and sentiment analysis using DNN, LSTM, and CNN. Additionally, it includes implementations for image style transfer and CTR prediction models like DeepFM and xDeepFM. The repository is designed to accompany a personal column, providing hands-on code examples for learning and experimentation in AI.

GPTs Gallery is a dedicated platform designed to facilitate the exploration and discovery of a wide array of Generative Pre-trained Transformers (GPTs). It serves as a central resource for individuals and organizations interested in harnessing the power of GPT technology for diverse applications. The gallery aims to connect users with innovative GPTs, making it easier to find models that align with specific needs and use cases. While the current website content is minimal, the underlying purpose is to provide a curated space for the GPT ecosystem, enabling users to navigate and select the most suitable AI models for their projects and interests. This platform is ideal for anyone looking to stay updated on the latest advancements and practical applications of GPTs.

Arcarithm is a leader in providing Artificial Intelligence-based solutions tailored for military, space, and commercial sectors. The company leverages decades of experience in algorithm development to deliver proven, fielded, and optimal solutions. Key offerings include Command & Control systems, which are integral for battle management and situational awareness, and Exigent® Automatic Target Recognition, an AI-based computer vision system for identifying ground-based vehicles, drones, guns, and other objects. Arcarithm also provides multidisciplinary, systems-oriented approaches for Smart City Information Technology (IT) integration, demonstrating its versatility across various complex applications. The company emphasizes solving hard problems faster and with lower risk.

awesome-llm-webapps is a curated collection of open-source, actively maintained web applications designed for Large Language Model (LLM) projects. This repository helps developers and AI enthusiasts jump-start their LLM applications by providing a categorized list of functioning web apps, rather than starting from scratch with frameworks. It covers diverse use cases including chatbots, natural language interfaces, assistants, and question answering systems. Projects are compared across important dimensions like conversation context, authentication, model support, and RAG capabilities, ensuring users can select the most suitable starting point for their specific needs. The collection spans a wide range of complexity, from simple API wrappers to production-ready systems with multi-source RAG backends and user management.

Azure-AIGEN-demos is a comprehensive GitHub repository offering a wide array of demos, documentation, and accelerators specifically designed for the Azure AI Foundry. This platform serves as a unified Azure platform-as-a-service, catering to enterprise AI operations, model builders, and application development. It combines production-grade infrastructure with user-friendly interfaces, enabling developers to concentrate on building applications rather than managing complex infrastructure. The repository showcases numerous examples, including AI red teaming, podcast generation, agentic RAG, code optimization, image generation with DALL-E 2/3, and various GPT models (GPT-4o, GPT-4V, GPT-5). It also features integrations with Azure Cognitive Search, Langchain, and provides examples for tasks like video analysis, document intelligence, and model evaluation.

GoAI offers a fully sovereign, on-premise Generative AI platform specifically designed for regulated sectors such as telecom, banking, government, and large enterprises in the Middle East and North Africa (MENA) region. The platform ensures complete data sovereignty with air-gapped deployment options, meaning data never leaves the client's infrastructure, which is crucial for compliance with strict regulatory requirements. It features enterprise-grade architecture with high availability and horizontal scaling, promising a 99.9% uptime SLA. GoAI also provides native Arabic language processing and industry-specific models, fine-tuned for regional use cases. Its solutions include AI-powered customer value management for telecom, automated KYC and risk analytics for banking, and secure AI for government agencies, all built to ensure regulatory compliance and data security.

AI4R, or Artificial Intelligence for Ruby, is a lightweight, open-source library designed as a learning playground for AI and machine learning in Ruby. Unlike many other libraries, AI4R focuses on providing clean, readable Ruby implementations of core AI algorithms, making it ideal for those who want to understand the underlying mechanics without bulky dependencies or 'black box' abstractions. It supports various AI domains including transformers (encoder-only, decoder-only, Seq2Seq), classifiers (Logistic Regression, RandomForest, SVM), clusterers (KMeans, DBSCAN), neural networks (Backpropagation, Hopfield, Transformer), search algorithms (A*, MonteCarloTreeSearch), genetic algorithms, reinforcement learning (Q-Learning, Policy Iteration), Hidden Markov Models, and Self-Organizing Maps. Each algorithm family includes examples and benchmark runners, allowing users to dive in, experiment, and compare performance. AI4R is distributed as a gem and requires Ruby 3.2 or later, making it accessible for Ruby developers interested in AI research and education.

ML-ProjectKart is a comprehensive open-source repository featuring over 234 projects spanning machine learning, deep learning, computer vision, and natural language processing. Designed to support open-source contributions, it offers a diverse range of projects suitable for beginners and enthusiasts alike. The platform aims to provide efficient and user-friendly projects that facilitate the mastery of ML/AI algorithms and practical application. Users can explore various project categories, contribute to existing projects, or initiate unique projects to share with the community, making it an excellent resource for hands-on learning and skill development in AI.

ALCE, pronounced /elk/, is an open-source project from Princeton NLP designed to enable large language models (LLMs) to generate text with accurate citations. It introduces a benchmark for Automatic LLMs' Citation Evaluation (ALCE) and includes three datasets: ASQA, QAMPARI, and ELI5. The repository provides comprehensive code for automatic evaluation of LLM generations across fluency, correctness, and citation quality. Researchers can also reproduce baselines from the associated EMNLP 2023 paper, perform passage retrieval, and add post-hoc citations to closed-book models. It supports both OpenAI API and offline HuggingFace models, making it a versatile tool for academic research in natural language processing.

ZeroTrusted.ai is an enterprise AI cybersecurity platform designed for government and enterprise compliance, offering automated Security Orchestration, Automation and Response (SOAR) with 232 specialized AI agents. It features real-time health monitoring, intelligent firewall protection, and a comprehensive AI Governance System, all built on zero trust architecture. The platform includes an Autonomous SOC with a 94%+ auto-resolution rate, a Counter-AI Network Operations platform, and Deep AI System Testing. It supports 14 compliance frameworks out-of-the-box, including FedRAMP High, NIST SP 800-53 Rev 5, and CMMC 2.0, and provides automated evidence collection for audit-ready compliance. ZeroTrusted.ai ensures data privacy through prompt anonymization and advanced encryption, making it ideal for organizations with strict regulatory mandates.

MT-DNN (Multi-Task Deep Neural Networks) is a PyTorch-based package designed for Natural Language Understanding. It implements advanced techniques such as adversarial training for both LM pre-training/finetuning and f-divergence, as well as large-scale adversarial training for language models (ALUM). The tool also features SMART, a principled regularized optimization method for robust and efficient fine-tuning of pre-trained natural language models. MT-DNN supports various NLP tasks including GLUE benchmark reproduction, SciTail & SNLI domain adaptation, sequence labeling, and question answering. It provides functionalities for extracting text embeddings and offers options for speeding up training through gradient accumulation and FP16 support.

PromptFill is a structured prompt generation tool specifically designed for AI painting applications such as GPT, Midjourney, and Nano Banana. It addresses the challenges of remembering, managing, and modifying complex prompts by providing a visual "fill-in-the-blank" interaction. Key features include variable autocomplete, inline variable syntax for real-time preview, and smart prompt splitting with automatic variable annotation. The tool also offers intelligent bank management with color-coded categories, a multi-template system for different use cases, and visual interaction with WYSIWYG editing and drag-and-drop functionality. PromptFill ensures data persistence with automatic saving to LocalStorage and supports HD social sharing for generated prompts. It is an open-source project, allowing users to self-host and customize their prompt management workflow.