Research & Education

Awesome-LLM-RAG is a curated, open-source list of advanced retrieval augmented generation (RAG) techniques specifically for Large Language Models (LLMs). Hosted on GitHub, this repository serves as a valuable resource for researchers and practitioners looking to stay updated on the latest advancements in the RAG field. It categorizes papers into various sections such as Retrieval-enhanced LLMs, RAG Instruction Tuning, RAG Embeddings, RAG Simulators, RAG Search, RAG Long-text and Memory, RAG Evaluation, and RAG Optimization. The project encourages contributions from researchers to promote their work, making it a dynamic and community-driven resource for the academic and development communities.

Dream is a 7B diffusion large language model designed for research and development in language modeling. It offers competitive performance against leading autoregressive models of similar size. The project provides comprehensive training and evaluation code, including specific implementations for Dream-Coder (a 7B dLLM for code) and DreamOn (which addresses variable-length generation and infilling). Dream's implementation is based on the Huggingface transformers library, requiring specific versions of transformers and PyTorch. It supports various diffusion sampling strategies like `maskgit_plus` and `entropy` for controlling token generation order, and includes options for fine-tuning on custom datasets.

WebAI-to-API is a modular web server built with FastAPI, designed to expose various browser-based Large Language Models (LLMs) as local API endpoints. This project allows users to access LLMs such as Gemini, ChatGPT, Claude, DeepSeek, and Grok without requiring an API key, leveraging browser cookies for authentication. It offers two operational modes: a primary WebAI Server for Gemini and a fallback gpt4free server for broader LLM access. The tool is intended for research and educational purposes, with commercial use discouraged. It provides OpenAI-compatible endpoints, server switching, and a modular architecture for straightforward development and maintenance.

Locus AI is an AI-powered search tool engineered for rapid information synthesis. It enables users to quickly locate and consolidate information from diverse sources, including webpages and PDFs. This tool is particularly beneficial for research and analysis tasks, streamlining the process of gathering and understanding data. Its capabilities are suitable for anyone who needs to efficiently collect and process information, making it a valuable asset for academic, professional, and personal use cases where quick access to synthesized knowledge is crucial.

ZipVoice is an open-source, fast, and high-quality zero-shot text-to-speech (TTS) model series built on flow matching technology. It features a compact size with only 123M parameters, delivering state-of-the-art performance in speaker similarity, intelligibility, and naturalness for voice cloning. The tool supports both Chinese and English languages and offers multi-mode generation, including single-speaker and dialogue speech. Key variants like ZipVoice-Distill provide improved speed, while ZipVoice-Dialog and ZipVoice-Dialog-Stereo enable advanced two-party spoken dialogue generation. It provides guidance for optimizing inference speed, controlling memory usage, and correcting mispronunciations, making it a versatile solution for various TTS applications.

ChatGPT Flow is a Chrome extension designed to enhance the management of ChatGPT conversations. It utilizes a unique node-based flow diagram approach to visualize and organize chat history, providing users with a clear and structured overview of their interactions. This tool is particularly beneficial for individuals who engage in complex or extensive ChatGPT dialogues, such as researchers and developers. By transforming linear conversations into an intuitive visual flow, ChatGPT Flow helps users track the progression of discussions, identify key points, and easily navigate through their chat history. This visual organization simplifies the analysis and management of intricate AI interactions, making it an invaluable asset for those seeking to optimize their ChatGPT experience.

Cinteraction is an AI research tool specifically developed to advance the understanding of human emotions by machines. It caters to machine learning experts and AI enthusiasts, providing capabilities for sentiment analysis and emotion detection. The tool aims to bridge the complex gap between nuanced human feelings and the analytical understanding of artificial intelligence. By focusing on this critical area, Cinteraction supports the development of more empathetic and context-aware AI systems, making it valuable for academic research and practical applications in fields requiring emotional intelligence from machines.

wizdom.ai is an AI-powered research intelligence platform designed to help researchers, institutions, publishers, funders, and governments make sense of the vast amount of research data available. It continuously monitors billions of data points about the global research ecosystem to provide actionable insights into topics, institutes, countries, journals, and funding bodies. The platform uses machine learning algorithms to generate analytics about scientific developments, helping users explore the research landscape, identify trends, and improve research performance. Key features include a comprehensive knowledge graph, AI-powered analytics, identification of emerging trends, and benchmarking research performance.

awesome-llms-fine-tuning is a meticulously curated collection of resources designed to aid in the fine-tuning of Large Language Models (LLMs) such as GPT, BERT, and RoBERTa. This repository is an indispensable resource for machine learning practitioners and researchers looking to adapt pre-trained models to specific tasks and domains. It encompasses a wide range of materials including tutorials, academic papers, practical tools, robust frameworks, and established best practices. The collection helps users enhance model performance and ensure alignment with particular contexts, terminology, and application requirements. Whether expanding expertise or just starting in the LLM field, this repository offers valuable insights and guidelines to streamline the fine-tuning process.

Paperpal is a comprehensive AI academic writing tool and research assistant designed for researchers and students. Its AI-powered platform offers a suite of features including a grammar checker, paraphraser, plagiarism checker, AI writing assistant, citation generator, translator, Chat PDF, essay writer, and journal submission checker. The tool provides instant feedback to enhance clarity, correct errors, and ensure adherence to academic standards for essays, research papers, and journal submissions. Paperpal is accessible both on the Web and as a Word Add-in, integrating seamlessly into academic workflows. Its advanced AI is trained on published scholarly content, ensuring language suggestions and AI-generated text align with academic conventions.

OralAi is a leading research group dedicated to transforming dental care through pioneering AI-driven innovation and digital dentistry excellence. The platform focuses on spearheading advancements in digital dentistry while championing environmental sustainability for healthier, smarter smiles. OralAi harnesses cutting-edge technologies, merging the power of artificial intelligence with digital dentistry, to redefine every facet of oral healthcare. This includes precise diagnostics, personalized treatment planning, and patient-centric care, ensuring precision, accessibility, and excellence in dental health. The initiative also aims to transform dental education with Generative AI, simulating clinical scenarios for immersive, hands-on learning experiences.

verl-agent is an extension of veRL specifically designed for training large language model (LLM) and vision-language model (VLM) agents using reinforcement learning (RL). It is the official code for the paper "Group-in-Group Policy Optimization for LLM Agent Training." A key differentiator is its step-independent multi-turn rollout mechanism, which allows for fully customizable per-step input structures, flexible history management, and modular memory. This design makes verl-agent highly scalable for very long-horizon, multi-turn RL training, such as tasks requiring up to 50 steps. The framework provides a diverse set of RL algorithms, including GiGPO, GRPO, and PPO, along with a rich suite of agent environments for both visual and text-based tasks, supporting models like Qwen3, LLaMA3.2, and LoRA fine-tuning.

This repository offers PyTorch implementations of two significant transformer network architectures: "Attention is All You Need" by Vaswani et al. (NIPS 2017) and "Weighted Transformer Network for Machine Translation" by Ahmed et al. (Arxiv 2017). It serves as a valuable resource for researchers and developers interested in understanding, experimenting with, and applying transformer models in machine translation and other natural language processing tasks. The project is open-source, providing the full code for these implementations, making it accessible for academic study, practical development, and benchmarking against established models.

unilm is an open-source project by Microsoft focused on large-scale self-supervised pre-training, encompassing a wide range of tasks, languages, and modalities. It provides foundational architectures like TorchScale, DeepNet, and Magneto, alongside a diverse collection of pre-trained models such as Kosmos, MetaLM, BEiT, WavLM, and LayoutLM. The project supports advancements in natural language processing, machine translation, speech recognition, document AI, and multimodal AI. It is designed for researchers and developers interested in building and experimenting with cutting-edge foundation models, offering resources for language, vision, and speech tasks, as well as multimodal applications.

The Machine Learning & Deep Learning Compendium is an open-source knowledge-sharing project compiled using Gitbook, offering a comprehensive collection of references across machine learning and deep learning. It covers approximately 500 topics, including modern machine learning algorithms, deep learning techniques, and specialized areas like NLP, audio processing, computer vision, time-series analysis, and anomaly detection. The compendium also delves into strategic themes such as data science management, team building, product management, design, and technology stacks from a data science perspective. It started as a personal curated list and is now available on GitHub, aiming to be a go-to resource for learners of all levels, from industry data scientists to academics and beginners.

Vector Institute is a non-profit organization dedicated to advancing AI research and talent development in Canada. It focuses on core research areas in machine learning and deep learning, publishing papers, and fostering AI engineering students. The institute also plays a crucial role in accelerating AI adoption for Canadian organizations by bridging breakthrough research with practical implementation through expert guidance and collaborative programs. They work with enterprises, startups, and the public sector, showcasing measurable impact through economic contributions, success stories, and responsible AI leadership. Vector Institute connects world-class researchers with industry partners to drive innovation and ensure AI development benefits all Canadians.

AI Grant is a non-profit initiative established in 2017 by Nat Friedman and Daniel Gross, dedicated to fostering innovation in open-source AI. It provides financial grants, ranging from $5,000 to $50,000, to individuals and teams working on impactful open-source artificial intelligence projects. These grants can be disbursed as either compute resources or direct cash, with a commitment to no strings attached, allowing recipients full autonomy over their work. The program supports a diverse range of projects, from foundational research in areas like GGUF file formats and fast cross-platform deep learning libraries, to applied AI in fields such as medical imaging, environmental monitoring, and creative arts. AI Grant aims to accelerate the development and accessibility of cutting-edge AI technologies by empowering talented researchers and developers.

Chord.pub is an AI-powered research tool designed to streamline the process of finding relevant YouTube videos and generating personalized articles. The platform leverages artificial intelligence to vet sources, ensuring the credibility of information, and then composes articles tailored to specific user interests. A key differentiator is the involvement of human editors who review and ensure the quality of the final articles, adding a layer of accuracy and reliability. Users have the option to associate the generated articles with their personal accounts, facilitating organization and access to their research.

Bayuegua.com is an advanced global patent search and analysis platform, designed to help users navigate and understand international intellectual property. The platform aggregates diverse scientific and technological intelligence, competitive intelligence, and market intelligence data. Utilizing AI technologies such as machine learning, natural language processing, and computer vision, it processes and extracts critical information to build an intelligent, one-stop big data platform for global scientific and technological innovation intelligence retrieval, analysis, and collaboration. It supports searching patents from various countries, including China, the US, Korea, Japan, and Europe, with titles and abstracts available in both Chinese and English, and full-text information in both languages for key regions. This multilingual capability removes language barriers for users seeking patent insights.

Accent Guesser is an AI-powered tool designed for speech analysis and accent identification. Utilizing advanced deep learning algorithms, it analyzes vocal patterns to quickly and reliably determine accent characteristics. The platform boasts a user-centric design, offering fast analysis and instant results for diverse needs. Key features include AI-powered insights for accurate accent identification, global accent recognition across a diverse range of regional dialects, and a user-friendly interface. It processes voice recordings without storing them, ensuring user privacy. Ideal for language learners, linguists, and anyone curious about their speech patterns, Accent Guesser helps users gain insights into their linguistic identity and enhance communication skills.

IPRally is an AI-native patent search platform designed for IP and R&D teams to accelerate and improve their decision-making processes. It leverages explainable and accurate AI, including proprietary Graph AI, to simplify patent search, review, and classification. Key features include effortless patent search without Boolean complexity, streamlined patent review with generative AI-assisted content analysis, continuous patent monitoring, and advanced portfolio analysis with custom AI classifiers. The platform is engineered for modern patent and innovation teams, offering solutions for patent searchers, analysts, managers, attorneys, and heads of IP, ensuring consistency and efficiency across various patent workflows. It also prioritizes security with ISO 27001 certification and GDPR compliance.

Chirpz AI is an applied AI lab dedicated to solving challenges in agent engineering. They focus on developing the essential tools and infrastructure required to ensure AI agents are reliable, observable, and ready for production environments. Their work addresses critical aspects like agent observability, tracing, and evaluation, which are vital for the successful deployment and management of complex AI systems. One of their key products, PandaProbe, is specifically designed to provide robust LLM monitoring and AI infrastructure capabilities, helping developers and researchers gain insights into agent behavior and performance. This makes Chirpz AI a crucial resource for anyone involved in the development and deployment of advanced AI agents.

AI-Infra-from-Zero-to-Hero is a comprehensive, open-source collection of resources dedicated to machine learning systems and AI infrastructure. It serves as a valuable guide for understanding the foundational and advanced concepts in this rapidly evolving field. The repository curates research papers, industry best practices, and video tutorials, covering critical areas such as Large Language Models (LLM) and Generative AI (GenAI). It includes insights from major conferences like OSDI, NSDI, SIGCOMM, SoCC, and MLSys, and references prominent models like Llama3 and Mistral. The project aims to provide a structured path for learning about AI system design, from data processing and training systems to inference and domain-specific infrastructures like AutoML and GNN systems.

Gyan AI Research is dedicated to pursuing Artificial General Intelligence (AGI) through the development of innovative AI models. They specialize in creating Indic LLMs, Legal LLMs, and Math GenAI Tutors, offering these as services. A key differentiator is their approach to training models from scratch, rather than fine-tuning existing ones, ensuring independence from major tech companies. Their product line includes PARAMANU, a family of Indic Generative Foundation Language Models supporting 10 most spoken Indian languages, PARAMANU-AYN, India’s first legal GenAI model for the Supreme Court of India, and PARAMANU-GANITA, a powerful Math AGI Tutor. These models are designed to be efficient, domain-adaptive, sector-agnostic, and can run inference without the need for GPUs, making them accessible for various applications in India and globally.