Research & Education

Tabular-data-generation is a Python library offering a unified interface for generating synthetic tabular data. It supports multiple state-of-the-art generative approaches, including GANs (Conditional Tabular GAN), diffusion models (ForestDiffusion), and Large Language Models (GReaT framework). The tool features adversarial filtering to ensure synthetic data closely matches real data distributions, native handling of mixed data types, and conditional generation capabilities, including text generation conditioned on categorical attributes. It also supports integration with external LLM APIs like LM Studio, OpenAI, and Ollama, and provides quality validation functions to compare original and synthetic distributions. AutoSynth allows for automatic comparison and selection of the best generator, and HuggingFace integration simplifies dataset synthesis and result sharing.

Riverside Research is an independent nonprofit organization dedicated to advancing scientific research for the benefit of the US government. Through an open innovation concept, they invest in multi-disciplinary research and development, encouraging collaboration to accelerate innovation in fields such as machine learning, trusted systems, optics and photonics, electromagnetics, plasma physics, radio frequency systems, and biomedical engineering. Their expertise spans Artificial Intelligence/Machine Learning, Secure & Resilient Systems, THz Imaging, Radar Systems, Intelligence, Surveillance, & Reconnaissance, System Engineering and Integration, SETA and A&AS, Software Engineering, and Type 1 Virtualization. The organization aims to put the best of teamed human and machine intelligence on tap for the warfighter and secure critical systems.

FPT Quy Nhơn AI Center, also known as QAI, is a creative valley dedicated to AI research, training, software development, and digital transformation support. It aims to cultivate high-quality human resources in the AI field within Vietnam through a multi-level education system and specialized university-level AI training programs. The center provides cybersecurity and AI application solutions for daily life and business. QAI leverages cutting-edge AI and machine learning technologies to empower businesses with intelligent solutions, including DX Data Engineering for optimizing costs and improving operational efficiency, AI/ML & Data Analytics for better decision-making, and Intelligent Automation for reducing human error and boosting productivity. QaiDora, an ecosystem of AI products, includes Q.Vision for intelligent video analytics, QaiDora Assistant Platform for virtual assistants, QaiDora Mask for automated image moderation, and QaiDora OCR for intelligent data processing.

Ready Tensor is a comprehensive platform designed for AI developers to document, share, and showcase their complete AI/ML projects, from code to results. Built by AI experts, it provides professional polish, AI-powered tools, and instant visibility within a fast-growing community. Users can build a verified portfolio to accelerate their AI careers, participate in industry-grade AI programs covering topics like Agentic AI and LLM Engineering, and receive AI-powered feedback on their publications. The platform supports a wide range of AI and data science projects, offering a notebook-like environment for structured documentation and easy sharing with a global network of collaborators and employers.

Llama2 by Meta offers a suite of industry-leading, open-source AI models, including Llama 4 (Maverick and Scout) and Llama 3 (3.1, 3.2, 3.3). These models are designed for easy deployment, cost efficiency, and scalable performance, supporting billions of users. Llama 4 models feature native multimodality, advanced reasoning, and extensive context windows, enabling both text and image understanding. Llama 3 models are open-source and can be fine-tuned, distilled, and deployed across various platforms, with versions optimized for multilingual tasks, edge devices, and general-purpose applications like summarization and coding. The platform also provides comprehensive documentation and resources for prompt engineering, fine-tuning, vision capabilities, quantization, distillation, and model evaluations.

llm-reasoners is a comprehensive library designed to significantly enhance the complex reasoning capabilities of large language models (LLMs). It offers a suite of cutting-edge search algorithms, including Reasoner Agent, Inference-time Scaling with PRM, Reasoning-via-Planning (MCTS), Tree-of-Thoughts (BFS/DFS), and Grace Decoding, among others. The library also provides intuitive visualization tools to help users understand and diagnose the reasoning process, even for complex algorithms like Monte-Carlo Tree Search. Optimized for performance, llm-reasoners integrates with high-performance LLM inference frameworks like SGLang, alongside support for Hugging Face Transformers, OpenAI API, and Exllama. It prioritizes rigorous implementation and reproducibility, ensuring algorithms are faithful to their original formulations and performance, as demonstrated by successful reproductions of Tree-of-Thoughts and Grace Decoding.

LearnML is an open-source study guide designed for individuals looking to learn machine learning in three months, with a focus on PyTorch. The curriculum is beginner-friendly and project-based, guiding users through building, training, deploying, scaling, and maintaining modern machine learning and deep learning models. Each week introduces a new concept or tool, such as Docker, PyTorch, or Transformer Models, culminating in a final project involving a Self Driving Car simulation. The program covers diverse areas including Computer Vision, Natural Language Processing, Time Series Forecasting, Audio Processing, and Recommender Systems, preparing learners for careers in ML or starting their own ventures.

Summary Box is an AI copilot designed to enhance your web experience by providing summarization, writing, and chat functionalities across web pages and uploaded documents. It utilizes state-of-the-art abstractive AI to understand information and generate summaries in its own words, rather than just extracting sentences. Key features include automatic article detection, a summary length slider, and quick bookmarks for saving summaries. For premium users, it offers a time saved tracker, advanced AI translation, and the ability to generate test questions based on content. Available as a browser extension and a web dashboard, it supports summarizing PDFs and Google Docs, making it ideal for students, professionals, and researchers.

Model Playground AI provides a comprehensive platform for comparing and evaluating a wide array of AI models. With access to over 150 models, users can test and assess different artificial intelligence capabilities, including text-to-image generation and language models, all within a single subscription. The platform emphasizes transparency with zero markup on model usage, making it an efficient solution for developers and researchers to explore and select the most suitable AI models for their projects. It simplifies the process of understanding model performance and features, fostering informed decision-making in AI development and application.

AI4IDF is an ambitious project based in the Île-de-France region, dedicated to advancing Artificial Intelligence that is frugal, reliable, and efficient, while also coexisting with and assisting humans in decision-making. The initiative seeks to capitalize on the robust scientific and industrial fabric of the Paris Region to establish it as a key player in this evolving field. AI4IDF focuses on deepening AI knowledge with a strong emphasis on human-centric design. Its research is structured around four main axes: Learning and optimization, Natural Language Processing and dialogue with humans, Robotics, motion and human interaction, and AI in people's lives. The project aims to foster innovation and collaboration within the region's unparalleled ecosystem.

dr-doc-search is an open-source tool designed for conversational interaction with PDF documents. Built with GPT-3, it allows users to ask questions and extract specific information from books and other PDF files. The tool supports both OpenAI and HuggingFace models for generating embeddings and answers, offering flexibility in its application. It requires an initial training process to create an index and generate embeddings for the PDF, after which users can query the document via a command-line interface or a web application. This simplifies document analysis and research by providing an interactive way to access information within large texts.

Distrifuser is a training-free algorithm designed to significantly accelerate diffusion model inference for high-resolution image generation by leveraging multiple GPUs. It addresses the fragmentation issue seen in naive parallel approaches by employing synchronous communication for initial patch interaction, followed by asynchronous communication to hide overhead. This method allows for substantial speedups, achieving up to 6.1x faster generation with 8 A100 GPUs for 3840x3840 images, without sacrificing visual fidelity. The tool is integrated with NVIDIA's TensorRT-LLM and supported by ColossalAI, offering a robust solution for developers and researchers working with large-scale generative AI models. It provides APIs similar to Hugging Face's Diffusers, making it accessible for those familiar with the ecosystem.

PixArt-sigma is an advanced, open-source Diffusion Transformer model designed for high-resolution 4K text-to-image generation. Leveraging a weak-to-strong training methodology, it offers PyTorch model definitions, pre-trained weights, and comprehensive inference/sampling code. The project emphasizes simplicity and compatibility, making it accessible for the PixArt community. It supports integration with Hugging Face's Diffusers library, allowing for fast experience and easy deployment. Key features include support for various image resolutions (256px to 2K, with 4K generation capabilities), LoRA code release, and ongoing development for features like ControlNet and ComfyUI integration. It's ideal for researchers and developers looking to push the boundaries of AI-driven image synthesis.

ru-gpts is a collection of open-source Russian GPT-3 models, including ruGPT3XL, ruGPT3Large, ruGPT3Medium, and ruGPT3Small, alongside a ruGPT2Large model. These autoregressive transformer language models are trained on an extensive dataset of the Russian language, offering different sequence lengths and attention mechanisms. The repository provides detailed setup instructions for both Colab and local environments, usage examples for generation, and finetuning capabilities. Pretraining details for each model, including dataset size, training time, and hardware used, are also available. The project is developed by AI-Forever and Sberbank-AI, making advanced Russian language models accessible for various applications.

Elicit is an AI-powered research assistant designed to significantly enhance the efficiency and evidence-based nature of scientific research. It allows users to search over 138 million academic papers and 545,000 clinical trials using semantic search, eliminating the need for precise keywords. The tool can generate high-quality research reports, customizable to specific papers and information, and supports systematic literature reviews by automating screening and data extraction, potentially saving researchers up to 80% of their time. Elicit also offers features like a personal library for organizing sources, alerts for new research, and an API for integrating its search and report generation capabilities. It emphasizes accuracy, transparency with sentence-level citations, and goes beyond simple chat to provide interactive tables and multi-step workflows.

Kaggle serves as the world's AI proving ground, bringing together millions of builders, researchers, and labs to discover what truly works in artificial intelligence. The platform facilitates the evaluation of AI agents, models, and frontier technology through crowdsourced benchmarks, competitions, and hackathons. It provides a robust environment for data scientists, machine learning engineers, and AI enthusiasts to collaborate, learn, and showcase their skills. Kaggle is a central hub for community-led innovation, offering resources and tools to advance the field of AI.

VideoGPT is an open-source project for video generation, leveraging VQ-VAE (Vector Quantized Variational AutoEncoder) and Transformer architectures. It learns downsampled discrete latent representations of raw video using 3D convolutions and axial self-attention. A GPT-like architecture then autoregressively models these discrete latents with spatio-temporal position encodings. The tool is designed for researchers and developers interested in AI video creation, offering a reproducible reference for transformer-based video generation models. It can generate samples competitive with state-of-the-art GAN models on datasets like BAIR Robot, UCF-101, and TGIF. The project includes scripts for training VQ-VAE and VideoGPT models, as well as for sampling and evaluation.

UniDiffuser is an open-source framework offering code and models for multi-modal diffusion research, based on the paper "One Transformer Fits All Distributions in Multi-Modal Diffusion." It unifies learning for marginal, conditional, and joint distributions by predicting noise in perturbed data across different modalities. The tool uses a transformer architecture to handle various input types and can perform image, text, text-to-image, image-to-text, and image-text pair generation. It supports tasks like image variation and text variation, and can be integrated with the Hugging Face Diffusers library for ease of use. UniDiffuser provides two pretrained models, UniDiffuser-v0 and UniDiffuser-v1, trained on large-scale image-text datasets.

DataRoot Labs is an AI R&D Center specializing in data science consulting and the development of AI-powered products. They offer 'R&D as a service' for startups, providing custom AI and ML solutions across various industries like healthcare, HR tech, retail, and logistics. Key offerings include free AI consulting sessions, detailed AI solution roadmaps, and fast MVP delivery within 8-12 weeks. They emphasize flexible pricing, full IP transfer, confidentiality, and even fundraising support for their clients. Their expertise spans technologies like Python, FastAPI, LangChain, Azure OpenAI, AWS Bedrock, and various vector databases, as demonstrated in their case studies.

GOAT.AI is a research lab dedicated to maximizing the strength of small models through Reinforcement Learning from Human Feedback (RLHF). The lab focuses on training compact yet robust models for domain-specific tasks, demonstrating that smaller models can outperform larger ones with appropriate datasets and reinforcement learning. Their research addresses critical problems in autonomous agents, including being stuck in loops, hallucinations (using Proximal Policy Optimization), and applying MetaRL techniques for experience. GOAT.AI is actively developing AI systems capable of understanding and generating natural language, while also tackling the software engineering challenges associated with large-scale AI models and millions of users.

MLcon serves as a global platform for the Machine Learning and Generative AI engineering community, organizing conferences and bootcamps across major cities like London, Amsterdam, San Diego, Munich, New York, and Berlin. These events bring together industry leaders and AI experts to discuss the latest advancements in machine learning, MLOps, LLMOps, and AI strategy. Attendees can enhance their skills through keynotes, workshops, and specialized bootcamps covering topics such as Generative AI, ML Fundamentals, and Full-Stack AI Engineering. MLcon also provides continuous knowledge sharing through articles, webinars, whitepapers, and cheat sheets, fostering a vibrant community for professionals to learn, share, and grow.

STAT.AI is an AI-powered tool designed to significantly enhance the Stata workflow for data analysis. It enables users to perform comprehensive data analysis with greater speed and accuracy, streamlining processes that typically consume hours. A key feature is Stata GPT, which provides instant answers to complex queries, reducing the need to search through extensive documentation or forums. Additionally, STAT.AI includes converters for Python and R scripts, allowing seamless integration and conversion of code into Stata format. This makes the tool suitable for a wide range of users, from beginners seeking simplified data handling to experts looking to optimize their analytical pipelines.

AI.HAMBURG is a non-profit initiative dedicated to accelerating the adoption of artificial intelligence and machine learning in companies and startups located in the Hamburg region. The initiative fosters knowledge and widespread use of AI through various activities, including educational events, workshops, and an active AI community. It aims to establish a leading ecosystem for AI application by connecting businesses, experts, and academic institutions. AI.HAMBURG organizes events like the AI.SUMMIT, AI Info Breakfasts, and AI Happy Hours, and provides a platform for networking and knowledge exchange, particularly supporting small and medium-sized businesses in implementing AI use cases.

The AICOR Institute for Artificial Intelligence (IAI) at the University of Bremen is dedicated to advancing research in cognition-enabled robot control. Its core focus lies at the intersection of robotics and artificial intelligence, exploring how AI can enhance robotic capabilities. The institute's research encompasses intelligent perception, sophisticated object manipulation, and the application of machine learning techniques specifically tailored for robotic systems. Through these efforts, the IAI aims to develop innovative solutions that enable robots to perform complex tasks with greater autonomy and efficiency, pushing the boundaries of what intelligent robotic systems can achieve in various applications.