Research & Education

SuperCLUE is a comprehensive benchmark designed for evaluating Chinese large language models (LLMs). It assesses LLMs across four primary capability quadrants: language understanding and generation, professional skills and knowledge, Agent intelligence, and safety, further breaking these down into 12 foundational abilities. The platform provides detailed rankings and reports, including a SuperCLUE total leaderboard, open multi-turn question leaderboard, and objective question leaderboard. It also features specific evaluations for AI Agent capabilities, focusing on tool use and task planning, which are critical for developing advanced AI assistants. SuperCLUE aims to provide a robust framework for researchers and developers to compare and improve Chinese LLMs.

system_prompts_leaks is a comprehensive, regularly updated GitHub repository that compiles extracted system prompts, system messages, and developer instructions from a wide array of popular AI chatbots and coding assistants. This includes models such as ChatGPT (GPT-5.4, GPT-5.3, Codex), Claude (Opus 4.6, Sonnet 4.6, Claude Code), Gemini (3.1 Pro, 3 Flash, CLI), Grok (4.2, 4), and Perplexity. The repository serves as a valuable resource for understanding how these advanced AI systems are instructed and behave, offering insights into their underlying configurations and personas. It is an open-source project, welcoming contributions from the community to expand its collection of prompts and models.

Bayuegua - Innovation Brain is a global patent search and analysis platform designed to provide comprehensive intellectual property services. It offers a one-stop solution for global patent search and analysis across 178 countries and regions. Key features include patent search, global patent retrieval, patent navigation, patent analysis and evaluation, and patent transaction services. The platform also provides free intellectual property analysis and evaluation consulting from experienced advisors and access to extensive patent search databases. It supports various search options like simple, advanced, semantic, image, instruction, and batch searches, utilizing AI assistants and knowledge graphs to enhance the process.

SuperPrompt is an open-source project designed to engineer prompts that enhance understanding of AI agents, particularly large language models like Claude. It utilizes a unique approach involving 'holographic metadata' and XML tags to guide the LLM's thought process. The tool aims to encourage models to 'think outside the box' and explore areas usually left untouched, often leading to novel ideas, though sometimes resulting in hallucinations. SuperPrompt is not intended to create conscious AI but rather to force deeper cognitive engagement from the model. It adapts its meta-prompt based on the user's request, allowing for dynamic and context-aware prompt generation, making it a valuable resource for AI research and development.

Foundations-of-LLMs is an open-source book designed for readers interested in large language models, providing a systematic explanation of foundational knowledge and cutting-edge techniques. The author team is committed to monthly updates, incorporating suggestions from the open-source community and experts, aiming to create an easy-to-read, rigorous, and in-depth LLM textbook. Each chapter is accompanied by a relevant Paper List to track the latest advancements. The first edition covers traditional language models, LLM architecture evolution, prompt engineering, parameter-efficient fine-tuning, model editing, and retrieval-augmented generation. The book uses animal-themed examples to illustrate specific technologies, making complex concepts more accessible.

PharmaBug serves as a comprehensive global pharmaceutical directory, enhanced with advanced AI-powered discovery tools. The platform is specifically designed to accelerate research and development within the pharmaceutical industry. It offers a suite of tools aimed at streamlining critical processes such as target identification, lead optimization, and virtual screening. By leveraging artificial intelligence, PharmaBug facilitates faster and more efficient R&D cycles, helping researchers and companies bring new pharmaceutical solutions to market more quickly. Its focus on AI capabilities positions it as a valuable resource for those looking to innovate and optimize their drug discovery pipelines.

Lhy_Machine_Learning serves as a comprehensive open-source repository for machine learning course materials and assignments from Professor Li Hongyi's spring courses spanning 2021, 2022, and 2023. Hosted on GitHub, this resource offers lecture slides, videos, and practical assignments covering fundamental concepts to advanced topics like Diffusion Models, Large Language Models, and Generative AI. It includes direct links to course videos, lecture notes, and homework assignments, often with accompanying code and platforms like Kaggle or JudgeBoi for submission. The repository is actively maintained with regular updates on topics and assignments, making it an invaluable resource for students and researchers looking to deepen their understanding of machine learning.

Z-Bench 1.0, developed by an enthusiastic AI-focused team at Zhenfund, is a large language model (LLM) prompt dataset specifically designed for non-technical users. It aims to provide a qualitative testing framework for conversational AI products, similar to ChatGPT. The dataset is structured into three main categories: basic capabilities (common.samples.csv), advanced capabilities (emergent.samples.csv), and vertical capabilities (specialized.samples.csv), offering a total of 300 prompts. Unlike academic test sets that often require automated testing or are optimized for specific NLP tasks, Z-Bench focuses on real-world, intuitive evaluation for individuals without a technical background. It combines existing academic insights with interesting daily use cases and emergent LLM abilities, making it a practical tool for assessing the performance of various LLM products.

llm_interview_note is an open-source GitHub repository designed to be a comprehensive resource for individuals preparing for interviews in the field of large language models (LLMs) and algorithm engineering. It meticulously compiles knowledge and common interview questions, drawing from various online resources. The repository covers fundamental LLM concepts, architectural details like Transformer models and attention mechanisms, training datasets, distributed training, supervised fine-tuning, inference optimization, reinforcement learning (including RLHF and DPO), retrieval-augmented generation (RAG), and LLM evaluation. It also includes practical projects like 'tiny-llm-zh' for hands-on experience in building small-parameter Chinese LLMs, 'tiny-rag' for RAG system implementation, and 'llama3-from-scratch-zh' for understanding Llama3. The project encourages community contributions and offers an online reading link for easy access.

千笔写作 is an AI writing platform designed to help users generate original academic papers and outlines efficiently. It boasts the capability to produce up to 30,000 words in just 10 minutes, incorporating real network data, charts, tables, formulas, and code. The platform offers unlimited 2,000-word 3-level outlines and supports various document types, including undergraduate, graduate, and on-the-job graduation papers, long essays, literature reviews, and journal articles. It also provides PPTs, opening reports, task books, and 40 real references, making it a comprehensive tool for academic writing support.

Delilah is an AI-powered academic assistant developed at Harvard and MIT, designed to support students through every step of the essay writing process. It helps users brainstorm ideas, generate drafts, write, and edit academic papers instantly. Key features include a draft generator library, a dedicated college applications hub, AI-powered document editing, and the ability to chat with PDFs. The tool also incorporates a GPT-4 chatbot and provides access to personal AI tutors, making it a comprehensive solution for academic support. Delilah aims to streamline academic writing tasks and enhance productivity for students.

Praxy is an AI suite specifically designed for healthcare professionals, offering a comprehensive set of tools to streamline various aspects of medical practice. It excels at summarizing patient consultations, automating back-office operations, and monitoring the quality of clinical trial data. The platform provides solutions for doctors across various specialties, as well as nurses, dietitians, psychologists, and medical secretaries. Key features include AI-powered consultation summaries, real-world evidence studies, pharmacovigilance management, and therapeutic advice chatbots. Praxy also offers APIs for software editors, chatbots, phonebots, and mobile applications, including Chrome/Edge plugins for ambient scribe and dictation. This suite aims to free up healthcare professionals' time, allowing them to focus more on patient care and improve therapeutic monitoring.

Roshan, operating under the brand name Rahkar Pardazesh Zharf, is a knowledge-based company specializing in artificial intelligence. The company focuses on developing practical AI solutions, including tools like Alpha, which processes and allows interaction with documents using advanced language models. Another key offering is Harf, designed to convert spoken language into searchable and editable text. Additionally, Roshan provides Hozm, a tool for cleaning and organizing Persian text, and Alefba, which transforms document images into searchable text. These solutions aim to leverage AI to enhance document management, accessibility, and data processing for various applications.

DreamCatcherLab is an AI-powered platform focused on sleep monitoring and research. It provides tools specifically designed to analyze sleep patterns, with the primary goal of combating insomnia. The platform caters to both researchers who need detailed data for studies and individuals looking to improve their personal sleep quality. By leveraging artificial intelligence, DreamCatcherLab offers insights into sleep cycles and potential disturbances, helping users understand and address their sleep-related issues. Its features are geared towards providing actionable information for better sleep health.

Researcher.Life offers a comprehensive suite of AI-powered tools designed to support academics throughout their research journey. This platform provides unlimited access to essential tools like Paperpal for academic writing assistance, R Discovery for literature search and recommendations, Mind The Graph for scientific illustrations, and a Global Journal Database for finding suitable publication venues. With a focus on streamlining the research process, Researcher.Life aims to enhance productivity and communication for researchers globally, including those who are non-English speakers. The platform is backed by over 22 years of academic expertise and trusted by a global research community, offering various subscription plans to suit individual needs.

Pragmatic Segmenter is a rule-based sentence boundary detection gem designed to segment text into sentences across multiple languages without requiring machine learning or training data. It aims to provide a "real-world" segmenter that performs well even when the format and domain of the input text are unknown. The tool focuses on robust language support, going beyond English-centric solutions, and includes text cleaning and preprocessing capabilities. It is opinionated, specifically developed for segmenting texts to create translation memories, and handles ambiguous sentence boundaries conservatively to maintain coherence. The project also features "Golden Rules," a set of tests for evaluating segmenter accuracy on edge cases, which are available for download.

Google Search Labs offers users early access to experimental AI features and tools within the Google ecosystem. Users can explore innovations like AI Mode in Google Search, which provides AI-generated responses and follow-up questions for a more powerful search experience. The platform also showcases other experiments such as NotebookLM, a personalized AI research assistant, and Viola the Bird, an AI-powered musical instrument. Additionally, it features the Google app for Windows, enabling quick access to Google Search and AI Mode directly from the desktop. Search Labs encourages users to provide feedback, playing a role in the evolution of Google's AI offerings.

JourneyTracker is an AI-powered tool designed to streamline and enhance the research process. It focuses on recording and summarizing research journeys, providing users with a robust solution for knowledge management. By leveraging artificial intelligence, JourneyTracker helps users effectively track and organize their research, ensuring that valuable insights and progress are not lost. This tool is particularly useful for individuals and teams engaged in academic or professional research, offering a structured approach to managing complex information and research paths.

LearnlyAI is an AI-powered academic assistant designed to help students with note-taking, essay writing, and presentation creation. It supports various study materials like PDFs, Word documents, and PowerPoint presentations, analyzing them to generate learning assistance content. The platform offers 24/7 AI tutors covering a wide range of subjects, smart study plans, and a world-class digital library. Key features include an Essay Assistant for academic writing feedback, a Presentation Maker to draft slides from notes, and a study calendar with AI-powered planning suggestions. It also integrates with Canvas LMS for course and deadline synchronization, and offers a browser extension to save articles.

qxresearch AI is a dedicated research lab specializing in Machine Learning, Deep Learning, and Computer Vision. The team is committed to making impactful discoveries and openly sharing their findings through academic journals and open-source contributions on GitHub. Beyond theoretical research, qxresearch AI actively implements these advancements into tangible AI applications within the healthcare and education sectors. The organization manages a diverse portfolio of research projects and papers, driven by fundamental research, real-world applications, and infrastructure development goals. Their primary research focus areas include Generative AI, Computer Vision, and Reinforcement Learning, fostering an environment where individuals and teams can pursue specific research interests.

The Max Planck ETH Center for Learning Systems (CLS) is a collaborative research center established by ETH Zurich and the Max Planck Society. It addresses complex research questions related to the design and analysis of both natural and artificial learning systems. The center combines the strong engineering capabilities of ETH Zurich with the natural sciences and computer science expertise of the Max Planck Institute for Intelligent Systems. CLS aims to be a leading hub for machine learning and modern artificial intelligence in Europe, fostering cooperation from senior scientists to PhD students. A core offering is its doctoral fellowship program, where PhD students receive joint supervision from ETH Professors and MPI Directors, leading to a doctoral degree from ETH Zurich. The program also provides opportunities for summer schools, workshops, and internships.

GirlsWhoML is a global movement dedicated to bridging the gender gap in Artificial Intelligence and Machine Learning. Their ambitious vision is to introduce 100,000 women and girls to AI and ML by 2027 through accessible education and community building. The initiative provides free courses and workshops worldwide, equipping students with essential knowledge and skills. Additionally, GirlsWhoML offers a mentorship program connecting learners with experienced AI/ML professionals and an AI-powered online learning platform, Vector Space, for those unable to attend in-person sessions. The platform welcomes women and non-binary individuals of all backgrounds, requiring no prior ML experience.

Protopia leverages artificial intelligence to facilitate personalized engagement for alumni, donor, and student communities within educational institutions. The platform focuses on creating frictionless micro-mentoring opportunities, allowing alumni and donors to give back through advice and networking without needing to sign up for new applications. It scales engagement for alumni associations, activates donors by connecting them to impactful volunteering, and expands networking for career services to boost student outcomes. Protopia's AI recruits the best-matched alumni from existing systems based on student prompts, and alumni can respond directly via email, making participation simple and accessible. This approach aims to increase volunteering, strengthen donor pipelines, and enhance career support across the institution.

LEGATO by Skymel is an Autonomous Agent Design & Deployment Kit that enables users to create their own personal AI agents. The platform is designed to streamline the process from a simple prompt to a fully functional and accurate production agent in a matter of hours. It supports the development of conversational AI applications and can assist with research tasks. Skymel aims to provide a comprehensive solution for individuals and organizations looking to leverage custom AI agents for various operational needs, emphasizing speed and precision in agent creation.