Coding & Development

LearnPrompt offers a comprehensive and permanently free open-source curriculum focused on AIGC (AI-Generated Content) technologies. The platform provides in-depth courses covering essential topics such as Prompt Engineering, ChatGPT, Midjourney, Runway, and Stable Diffusion. Beyond core generative AI, it expands into specialized areas like AI digital humans, AI voice and music generation, and the fine-tuning of large language models. With its latest v4.0 update, LearnPrompt features a new UI, multi-language support, a comments section, daily updates, and contribution options, making it a dynamic resource for anyone looking to master AIGC without cost. The platform is continuously updated with new content and features, including case studies and tutorials for advanced applications.

Deep_and_Machine_Learning_Projects is an open-source GitHub repository containing a diverse collection of machine and deep learning projects. This resource provides readily available code and data files, enabling users to explore and implement practical applications of artificial intelligence. Each project within the repository is designed to be a standalone example, allowing individuals to understand specific use cases and integrate them into their own real-life scenarios. It serves as an excellent learning resource for those looking to gain hands-on experience in AI development, offering a practical approach to mastering machine and deep learning concepts through direct implementation.

FineTuningLLMs is the official repository for the book "A Hands-On Guide to Fine-Tuning LLMs with PyTorch and Hugging Face." This resource offers comprehensive guidance and practical code examples for fine-tuning large language models. It covers essential concepts such as quantization, low-rank adapters (LoRA), and dataset formatting templates. The repository features Jupyter notebooks that can be easily run on Google Colab with GPU support, making it accessible for hands-on learning. It delves into topics like loading quantized models, fine-tuning with SFTTrainer, and deploying models locally using formats like GGUF with Ollama or llama.cpp. The guide is designed for an intermediate-level audience, assuming a foundational understanding of deep learning concepts.

Huatuo-Llama-Med-Chinese, also known as BenCao, is an open-source project focused on instruction-tuning large language models with Chinese medical knowledge. It leverages models such as LLaMA, Alpaca-Chinese, and Bloom, fine-tuning them with datasets built from medical knowledge graphs and literature using ChatGPT API. This process significantly improves the base models' performance in medical question-answering. The project provides LoRA weights for various base models, enabling efficient fine-tuning. It also introduces a knowledge-finetuning approach that allows models to explicitly utilize knowledge base information during inference, enhancing reliability in generating Chinese medical responses.

mnehmos.multi-agent.framework is an open-source project designed to give Large Language Models (LLMs) a 'nervous system,' transforming them from stateless text predictors into more autonomous 'organisms.' It provides a biological architecture that organizes sensation, reflex, memory, and action into coherent loops. The framework features a multi-layered architecture including Central (Cognition), Somatic (Voluntary Action), Autonomic (Subconscious), and Reflex (Spinal Cord) components. It supports various modes for task decomposition, system design, planning, research, coding, debugging, and knowledge management. Key features include an OODA Loop for decision-making, a TDD Cycle for development, and a Boomerang Protocol for structured data returns, making it suitable for developers building advanced AI agents.

ollama-gui is a modern web interface designed for interacting with local Large Language Models (LLMs) through the Ollama API. It boasts a clean and responsive user interface, ensuring a smooth chatting experience. Key features include local chat history management using IndexedDB, comprehensive Markdown support for messages, and a dark mode option for user comfort. The tool prioritizes privacy by processing all data locally, ensuring no information leaves your system. It also offers a development proxy for easy network access and supports Docker deployment for simplified setup, allowing users to run both Ollama and the GUI together without complex configurations.

TheBloke Wizard Vicuna 13B Uncensored HF is an AI chatbot hosted as a Hugging Face Space. This tool offers an uncensored version of the Wizard Vicuna 13B model, allowing users to engage in conversational AI interactions without typical content restrictions. While the live website currently indicates a runtime error, suggesting it may not be fully operational at this moment, the intention is to provide a platform for direct interaction with this specific large language model. It is designed for those interested in exploring the capabilities of uncensored AI models within a readily accessible web environment.

Non-overwhelming-machine-learning is an open-source project hosted on GitHub, offering a carefully curated list of machine learning resources specifically designed for beginners. The primary goal is to provide a "non-overwhelming" introduction to the field, guiding users through a chronological learning path. It assumes foundational knowledge in probability, multivariable calculus, and optimization, ensuring that learners have the necessary prerequisites before diving into more complex topics. The resource list includes introductory courses like "Intro to Machine Learning UD120," "Deep Learning @ Udacity," and specialized courses on convolutional neural networks and natural language processing. This structured approach helps beginners build a solid understanding without feeling overwhelmed by the vastness of machine learning.

TensorFlow-and-DeepLearning-Tutorial is an open-source repository offering a collection of deep learning tutorials. Originally taught as an online course in 2016, it provides foundational knowledge in TensorFlow, fully connected neural networks, and convolutional neural networks. The resource also delves into Natural Language Processing concepts. Written primarily in Python and Jupyter Notebook, it serves as a valuable educational tool for individuals looking to understand and implement deep learning techniques.

PaddleFormers is an open-source library built on the PaddlePaddle deep learning framework, designed to offer model interfaces and functionalities comparable to Hugging Face Transformers. It supports the training of both large language models (LLM) and visual language models (VLM). The library leverages PaddlePaddle's inherent advantages in high-performance training, incorporating advanced distributed training strategies like tensor parallelism, pipeline parallelism, and expert parallelism, alongside automatic mixed precision for acceleration. PaddleFormers aims to provide a high-performance, low-resource-consumption training experience, enabling users to efficiently complete large model training without delving into complex optimization details. It supports a wide array of mainstream LLMs and VLMs, including DeepSeek-V3, GLM-4.5 series, Qwen2/3, and ERNIE models, and offers full-lifecycle training capabilities from pre-training to post-training, including CPT, SFT, SFT-LoRA, DPO, and DPO-LoRA.

Visual-openLLM is an open-source project designed to interactively connect various visual models, functioning similarly to Visual ChatGPT. It is built upon established technologies like ChatGLM, Visual ChatGPT, and Stable Diffusion, positioning itself as an open-source version of '文心一言'. The tool supports ChatGLM3, adding features such as VQA (Visual Question Answering) and Pix2Pix capabilities. Its development roadmap includes support for multi-turn chat, integration with other visual tools, and compatibility with additional large language models, making it a versatile platform for visual AI experimentation and application.

transformers_tasks is an open-source project on GitHub that integrates various NLP algorithms using the powerful Hugging Face transformers library. It offers implementations for a wide range of tasks, including text matching (PointWise, DSSM, Sentence Bert, SimCSE), information extraction (UIE), prompt tasks (PET, p-tuning), and text classification (BERT-CLS). The project also delves into advanced areas like Reinforcement Learning from Human Feedback (RLHF) for language models, text generation (T5-Based models), and large language model (LLM) applications and training. It provides a flexible framework for researchers and developers to train and fine-tune models using their own datasets.

VAR (Visual Autoregressive Modeling) is an open-source project that introduces a novel approach to image generation, moving beyond traditional raster-scan "next-token prediction" to a coarse-to-fine "next-scale prediction." This method allows GPT-style autoregressive models to achieve state-of-the-art results, even outperforming diffusion models in visual generation. The project emphasizes scalability, user-friendliness, and provides a robust codebase for researchers and developers. It also highlights the discovery of power-law scaling laws within VAR transformers and demonstrates strong zero-shot generalizability. VAR has received the NeurIPS 2024 Best Paper Award and offers various pre-trained models for different resolutions and complexities.

LunarLink AI provides a unified platform to access and compare outputs from various advanced AI models, including ChatGPT, Claude, and Gemini. It operates on a pay-as-you-go model, charging based on usage at first-party API prices plus a small per-answer fee, eliminating the need for subscriptions or commitments. Users can chat with multiple AI assistants simultaneously, view responses side-by-side to reduce bias, and enjoy features like unlimited file uploads and cross-platform chat synchronization. The platform prioritizes privacy with a data-safe mode that ensures no storage or training of user data, and offers a customizable interface with dark/light modes and enhanced content presentation for rich text and code blocks.

OpenCodeInterpreter is a comprehensive suite of open-source code generation systems designed to significantly improve the capabilities of large language models (LLMs) in coding tasks. It achieves this by incorporating execution feedback and iterative refinement, allowing the LLM to dynamically adjust and improve generated code. The platform offers various models, including the OpenCodeInterpreter-DS, -CL, -GM, and -SC2 series, all open-sourced on Hugging Face. These models demonstrate enhanced performance on benchmarks like HumanEval and MBPP, particularly with the integration of execution feedback. The project also provides a local deployment demo, enabling users to generate and execute code, receive automated feedback, and engage in chat-based interactions for further refinement. It is supported by the Code-Feedback dataset, featuring 68K multi-turn interactions.

AI_Books is a GitHub repository offering a comprehensive collection of books focused on Artificial Intelligence, Machine Learning, Deep Learning, and Neural Networks. This resource is designed for individuals looking to deepen their understanding and skills in these rapidly evolving technological domains. The repository includes a wide array of titles, covering foundational concepts, advanced theories, and practical applications, often with examples in popular programming languages like Python and R. It also provides links to online books and supplementary resources, making it a central hub for learning and reference in AI-related subjects.

ASR-LLM-TTS is a comprehensive speech interaction system built on open-source models, seamlessly integrating Automatic Speech Recognition (ASR), Large Language Models (LLM), and Text-to-Speech (TTS) in sequence. It leverages SenceVoice for ASR, QWen2.5-0.5B/1.5B for LLM capabilities, and offers three TTS options: CosyVoice, Edge-TTS, and pyttsx3. The system supports real-time voice interaction, including features like wake-word detection, speaker recognition, and conversation history memory. It also extends to multi-modal interactions by integrating QWen2-VL-2B for processing both audio and video inputs, making it suitable for advanced conversational AI applications.

awesome-ai-apps is a curated, open-source collection of practical AI agents and generative AI applications. It features diverse tech stacks and demonstrates real-world implementations using leading AI models like OpenAI, Gemini, and local LLMs, alongside various AI frameworks. The repository is organized into five main categories: Starter Agents, Advanced Agents, Multi-Agent Teams, RAG Applications, and Multimodal Apps, providing a comprehensive resource for developers. It includes examples ranging from simple, single-purpose agents to complex multi-agent systems and applications combining text, images, audio, and video. The project also outlines a detailed development roadmap targeting over 100 complete applications by the end of 2025.

OpenPeer AI is a decentralized AI platform designed to offer safe, democratized, and highly scalable AI solutions. It leverages LonScript, the Decentralized-Internet SDK, Mojo, and the LLM Standard, with BOINC as its foundational technology. The platform aims to minimize computation time through its design structures, modular components, and mathematics centered around logical regression. OpenPeer AI supports diffusion models, generative art, and focuses on practical, utility-oriented AI solutions that go beyond simple search engine reindexing. It emphasizes open-source principles, allowing users to access and run its source code for automation workflows and projects, while also prioritizing AI safety, horizontal scaling, and Robotic Process Automations (RPAs).

ChatGPT_JCM is an open-source OpenAI management system built using Vue2 and ElementUI. It provides a convenient web interface to interact with various OpenAI APIs, including text completion (GPT-3.5, GPT-4), image generation and editing, audio transcription and translation, and file management. The tool also supports fine-tuning models and offers features like multi-session storage with context logic, data export/import, and built-in role-playing prompts. It's designed for developers and users looking for an accessible way to explore and manage OpenAI's capabilities, with support for Markdown formatting for enhanced output.

Data Prep Kit is an open-source project designed to accelerate unstructured data preparation specifically for Generative AI applications. It provides a comprehensive set of modules and transforms that enable developers to cleanse, transform, and enrich various types of unstructured data, including natural language, code, and images. This kit supports use cases such as pre-training Large Language Models (LLMs), fine-tuning LLMs, instruct-tuning LLMs, and building Retrieval Augmented Generation (RAG) applications. It is built on common frameworks for Python and Ray runtimes, allowing it to scale from a commodity laptop to data center-scale processing. The kit also offers a framework for developing custom transforms and provides examples for deploying transforms on Kubernetes clusters using Python or Ray jobs, and for orchestrating multiple transforms with Tekton pipelines.

Datawizz, operating under the name SuperAgent, is an upcoming platform designed to provide an agent workforce. While specific features are not yet detailed, the tool is positioned to offer early access to interested users. The website indicates a focus on building an "Agent Workforce" and is currently in a "coming soon" phase, suggesting it will likely involve AI-powered agents or automation. Users can sign up for early access to stay informed about its launch and capabilities. The platform emphasizes privacy and terms of service, indicating a professional approach to its development and user interaction.

chinese-llm-benchmark, also known as ReLE (Really Reliable Live Evaluation for LLM), is a continuously updated platform for evaluating Chinese AI large language models. It currently covers 375 models, including commercial options like ChatGPT, Google Gemini, Claude, and Ernie, as well as open-source models such as Llama, GLM, and Mistral. The benchmark offers multi-dimensional capability assessments across 7 domains, including education, healthcare, finance, law, reasoning, language, and agent/tool calling, with approximately 300 detailed sub-dimensions. Beyond providing rankings, it features a defect library containing over 2 million entries, facilitating research and improvement of large models. The platform also offers free evaluation services for private large models.

ecosystem.Ai is an enterprise-grade AI platform designed for real-time personalization, offering sub-50ms predictions at scale. It allows businesses to build contextual interactions, human-centric recommenders, and intelligent conversations. The platform features a low-code environment with a visual workbench and pre-built modules, enabling rapid deployment within weeks. It leverages behavioral AI to understand customer actions, not just demographics, incorporating algorithms like spend personality and contextual bandits. ecosystem.Ai supports various industries including banking, telecommunications, and retail, providing solutions for intelligent discovery, smart engagement, intelligent activation, and proactive retention across the customer lifecycle. It also includes an agent-building tool, Ecogentic, for designing conversational AI agents with granular control and security guardrails.