Research & Education

VLA-Adapter is an open-source implementation offering an effective paradigm for tiny-scale Vision-Language-Action (VLA) models. It provides a robust framework for training and deploying VLA models, particularly for robotic control and real-world system integration. The tool supports various GPU configurations, from extremely limited VRAM (10-12GB) to professional-grade GPUs (80GB+), making it accessible for diverse research and development environments. Key features include support for LIBERO and CALVIN benchmarks, an enhanced Pro version for improved performance, and compatibility with various foundation models and real-world robotic systems like ALOHA and Franka. It also offers detailed guidance on data preparation and training configurations.

The youtube-ai-extension is an interactive YouTube extension built with React, Tailwind CSS, and Plasmo, integrating with the OpenAI API. It allows users to chat directly with YouTube videos in real-time, offering a unique interactive experience. Key functionalities include generating video summaries, asking questions, and receiving detailed explanations. The extension features a user-friendly interface seamlessly integrated into YouTube, supporting multiple languages and providing context-aware responses. While currently requiring a local installation and an OpenAI API key, a major update is planned for June 10, 2025, which will include new features, a streamlined installation process, and an official release on the Chrome Web Store. It's important to note that the extension currently works best with the old YouTube layout, requiring a tool like uBlock Origin to revert the layout.

Recursion is a clinical-stage TechBio company dedicated to decoding biology through AI to radically improve lives. Founded over a decade ago, Recursion utilizes its proprietary Recursion OS, an AI-native, end-to-end drug discovery and development platform. This platform integrates biology, chemistry, and clinical development into a unified intelligence system, powered by multimodal data, purpose-built AI models, and bilingual teams. Recursion aims to reduce the massive 90% failure rate of traditional drug discovery by using AI to understand cellular disruptions driving disease. The company has yielded an advanced pipeline of potential first-in-class and best-in-class treatments for conditions with high unmet need, including aggressive cancers and rare diseases, demonstrating significant improvements in speed, efficiency, and reduced costs from hit identification to IND-enabling studies.

Coursology is an AI homework helper designed to assist students in completing assignments and enhancing their understanding across various subjects. It offers instant, accurate solutions and a suite of study tools, including AI notetaking, quizzes, and flashcards. Users can upload materials like notes, lectures, and textbooks for the AI to ingest, enabling file chat and personalized AI podcasts. The platform also provides a Chrome extension for verifying solutions and expanding understanding directly on learning platforms. Coursology supports multiple languages and is available on mobile apps, making learning faster and more effective for students at all levels.

BEIR (Benchmarking Information Retrieval) is an open-source, heterogeneous benchmark designed for evaluating NLP-based retrieval models across a wide array of information retrieval tasks. It offers a common and user-friendly framework for researchers and developers to assess their models against over 15 diverse IR datasets. The tool supports various retrieval architectures, including lexical, dense, sparse, and reranking-based models, and allows for the evaluation of custom models using state-of-the-art metrics like NDCG@k, MAP@K, Recall@K, Precision@K, and MRR. BEIR simplifies the preprocessing of IR datasets and integrates with popular platforms like Hugging Face, making it a valuable resource for both academia and industry.

fsdl-text-recognizer-project offers comprehensive lab materials for the Full Stack Deep Learning Bootcamp, guiding users through the process of building a handwriting recognition system from the ground up. The project leverages Keras and is designed with modularity, hackability, and scalability in mind. It includes code for training models in parallel, storing evaluation results in Weights & Biases, and setting up a continuous integration system for code functionality and model evaluation. Users will learn to package the prediction system as a REST API deployable as a Docker container, and deploy it as a serverless function to Amazon Lambda. The curriculum also covers setting up monitoring to alert on incoming data distribution changes, making it a practical resource for deep learning deployment.

MAgent is a research platform specifically engineered for many-agent reinforcement learning, distinguishing itself from other platforms that typically focus on single or few-agent scenarios. It enables researchers to scale up their reinforcement learning experiments from hundreds to millions of agents, facilitating the study of artificial collective intelligence. The platform supports both Linux and OS X and allows for the implementation of various algorithms, including rule-based systems and deep learning frameworks. While the original project is no longer maintained, a community-maintained fork, MAgent2, is available for continued development and use. It offers examples for training and playing with agents in scenarios like pursuit, gathering, and battle, along with baseline algorithms like DQN, DRQN, and A2C.

lue is a versatile terminal eBook reader designed for CLI enthusiasts and bookworms, providing an immersive reading experience with audiobook-quality text-to-speech. It boasts broad multi-format support, handling EPUB, PDF, TXT, DOCX, DOC, HTML, RTF, and Markdown files with seamless format detection. The tool features a modular TTS system, including Edge TTS (online) and Kokoro TTS (local/offline), with an extensible architecture for new models. Cross-platform compatibility across macOS, Linux, and Windows (via WSL) ensures a consistent global experience with over 100 languages. Users can adjust playback speed, benefit from auto-scroll with precise word highlighting, and enjoy smart persistence that saves progress automatically. Fast navigation, extensive customization options for keyboard layouts (including Vim-style), UI elements, colors, and display modes further enhance the reading experience.

MeshCNN is a general-purpose deep neural network specifically designed for 3D triangular meshes, implemented using PyTorch. This framework enables advanced tasks such as 3D shape classification and segmentation by applying convolutional, pooling, and unpooling layers directly on the mesh edges. It offers a robust solution for researchers and developers working with 3D data, providing a novel approach to process geometric information. The repository includes scripts for installation, training, and testing on datasets like SHREC and Humans, making it accessible for practical application and further development in the field of geometric deep learning.

Socratic, now integrated into Google Lens, is an AI-powered tool that transforms your phone's camera into a powerful search engine. It allows users to identify objects, translate text in real-time across over 100 languages, and even get step-by-step homework help for subjects like math, history, chemistry, biology, and physics. Beyond academic assistance, Google Lens can help identify plants and animals, and find similar clothing or home decor items from images. It's available across various Google apps and devices, including the Google app, Google Camera, Google Photos, and Chrome, making visual search and information retrieval seamless and accessible.

Mocha.jl is a deep learning framework for the Julia programming language, drawing inspiration from the C++ framework Caffe. Although now deprecated, it was designed for efficient training of deep and shallow convolutional neural networks, supporting optional unsupervised pre-training via stacked auto-encoders. The framework boasts a modular architecture with isolated components for layers, activation functions, solvers, and more, allowing for easy extension. Written in Julia, it offers a high-level interface for intuitive deep neural network experimentation. Mocha.jl provides multiple backends, including a portable pure Julia backend, a faster native extension backend, and a highly efficient GPU backend utilizing NVidia® cuDNN and CUDA kernels. It also supports HDF5 for data and model storage, ensuring compatibility with other computational tools, and can import Caffe model snapshots.

QuizGeniusAI is an AI quiz generator designed specifically for educators, enabling them to create curriculum-aligned quizzes rapidly from their lesson materials. Teachers can upload PDFs, paste text, or simply type a topic, and the AI generates multiple-choice, true/false, and open-ended questions aligned to Bloom's Taxonomy. The platform supports direct sharing to Google Classroom, Microsoft Teams, and Slack, or quizzes can be embedded on school websites. It also provides student performance analytics, allowing educators to track completion rates, average scores, and identify areas where students struggle. QuizGeniusAI is GDPR compliant, ensures student data safety without requiring student accounts, and offers a free plan.

Zylon is a private on-premise enterprise AI platform specifically designed for regulated industries such as finance, healthcare, defense, and the public sector. It allows organizations to deploy secure AI within their own infrastructure, including on-premise, air-gapped, or private cloud VPC environments, ensuring full data control, governance, and compliance. Zylon provides a complete AI infrastructure, including local LLMs, vector databases, and GPU orchestration, with a single-command deployment for rapid setup. The platform features an AI Core, an OpenAI-compatible API Gateway for integrations, and a Workspace for team collaboration, document creation, and knowledge base access, all powered by private data. It is built by the creators of PrivateGPT, an open-source project with a large community.

Converse by B7Labs is an AI-powered reading companion designed to enhance comprehension and make research more efficient. Users can save web articles, upload PDF documents, and input YouTube video links to generate TLDRs, detailed summaries, and key takeaways. The platform features an intuitive document chat, allowing users to ask questions and receive precise answers with source references. All saved content is permanently stored in a personal library, which can be organized into custom collections. Converse also supports social sharing, enabling users to share their saved articles and interactions with friends and social networks, fostering collaborative reading experiences.

BeFreed is an AI-powered platform designed for personalized audio learning, transforming how users consume knowledge. It curates content from a vast array of sources, including books, research papers, and domain experts, and narrates them in a user's chosen language, voice, and learning depth. The platform allows users to create personalized AI podcasts on any topic, interact with audio content through chat, and track their learning progress. With features like smart flashcards and a dedicated 'Mindspace' for notes and memories, BeFreed aims to boost retention and make learning effortless. It offers a unique blend of storytelling and factual learning, catering to various preferences, and fosters a global community of curious minds.

dgl-ke is an open-source package designed for learning large-scale knowledge graph embeddings, built on top of the Deep Graph Library (DGL). It offers high performance, ease of use, and scalability, making it suitable for various machine learning tasks involving knowledge graphs. The package supports training knowledge graph embeddings using popular models like TransE, TransR, RESCAL, DistMult, ComplEx, and RotatE. Users can perform training on single machines (CPU/GPU) or distributed environments, evaluate pre-trained embeddings with link prediction tasks, and conduct inference for entity/relation linkage prediction or embedding similarity. DGL-KE is optimized for scale, capable of processing knowledge graphs with millions of nodes and billions of edges efficiently.

Twee is an AI-powered platform designed for language educators to quickly create and manage lesson materials. It generates CEFR-aligned content across 10 languages, including texts, dialogues, fill-in-the-gap exercises, questions, and writing prompts. Teachers can input a topic, link, or word list, and Twee instantly produces level-appropriate resources. The tool offers flexibility in sharing, allowing materials to be saved as PDFs or Word documents, or shared interactively via links, Google Forms, or Twee's own assignments. Twee also simplifies grading with AI assessment of student responses, providing insights and saving teachers time. It streamlines workflow by managing lesson planning, progress tracking, assignments, and student accounts.

Langua is an AI-powered language tutor designed to help users achieve fluency through interactive, real-world conversations. Instead of just vocabulary lists, Langua focuses on practical dialogue in various scenarios. Users can choose from hundreds of real-life scenes, take on a role, and practice speaking with an AI conversation partner. The platform provides instant AI-powered feedback on pronunciation and grammar, helping learners perfect their accent and sound like a native speaker. Langua supports multiple languages, including Spanish, French, German, Italian, and Japanese, making it a versatile tool for language enthusiasts. It's free to start, offering a simple path to mastering conversations.

Perfect Prep AI is an AI-powered platform designed to help students prepare for the IELTS writing exam. It offers accurate practice scores based on official IELTS grading criteria, allowing users to track their progress effectively. The tool provides personalized feedback on writing tasks, ensuring tailored guidance for improvement, and offers specific recommendations to help users focus their efforts and achieve a higher band score. Users can practice writing tasks that simulate the real exam experience, with essays graded using the exact IELTS rubric. Perfect Prep AI is suitable for all English proficiency levels, from beginner to advanced, and is meticulously trained on English language intricacies and IELTS grading standards to provide reliable performance and tangible results.

Extracurrify is an AI-powered platform designed to assist high school students in optimizing their college application process. It offers a comprehensive database of extracurricular activities and scholarships, tailored to students' majors and backgrounds. Users can receive personalized guidance from an AI counselor on selecting the best extracurriculars, choosing a major, and navigating college applications. The platform also includes a university generator to help students find institutions based on their interests and budget, and provides AI-powered reviews of extracurricular activities. Developed with insights from top US college counselors, Extracurrify aims to help students from all backgrounds, including international, low-income, and first-generation students, build strong college applications.

Passu is an AI-powered platform designed to assist individuals in preparing for TCF Canada, IELTS, and other French language exams. The tool provides AI-powered practice tests that simulate real exam conditions, helping users to familiarize themselves with the format and improve their performance. Beyond practice tests, Passu offers live group courses for interactive learning and personalized study plans tailored to individual needs and progress. Expert evaluations are also available to provide detailed feedback and guidance. This comprehensive approach aims to enhance learning efficiency and increase the likelihood of success in language proficiency tests, making it a valuable resource for students and professionals alike.

PyTorch-BYOL offers a robust PyTorch implementation of the Bootstrap Your Own Latent (BYOL) self-supervised learning approach. This tool is designed for researchers and developers to experiment with and apply BYOL algorithms for representation learning. It includes configurable parameters for network architecture (ResNet-18 or ResNet-50), projection and prediction heads, data transformations, and trainer settings such as batch size, momentum update, and epochs. The repository provides clear installation instructions and configuration options, making it accessible for those looking to delve into self-supervised learning without starting from scratch. It also details feature evaluation methods, including linear separability using logistic regression and KNN on datasets like STL10.

IELTS Flex is an AI-powered platform designed to help students prepare for the IELTS exam with confidence. It offers a comprehensive suite of tools, including free live classes, a free mini-mock test with AI feedback for speaking and writing, and full mock tests. The platform is selected by the co-owners of IELTS, the British Council, and boasts a 96% success rate in improving user scores. Key features include fast and accurate AI scoring with personal feedback, access to 119 live and on-demand IELTS classes, and guided skills courses. Users can also benefit from FlexCheck AI for writing and speaking, IELTS band score predictions, and personalized feedback on pronunciation and vocabulary.

FIKA IELTS is an AI-powered mobile application designed to help students prepare for the IELTS Speaking test. It functions as a personal AI tutor, offering immediate and detailed feedback on spoken English, including pronunciation, vocabulary usage, and fluency. Users can record their responses to practice questions and receive band 6.5+ level corrections and suggestions. The app also provides idea generation support to help users overcome mental blocks during practice. FIKA IELTS offers a personalized learning experience, adapting practice exercises based on the user's progress and needs. A free trial of 50 minutes of speaking practice is available in the first week after downloading the app.