DistriFusion is a distributed parallel inference framework for high-resolution diffusion models. It significantly accelerates the generation of high-resolution images on multiple GPUs through distributed parallel inference. DistriFusion utilizes patch...
Still-Moving is an AI video generation framework from DeepMind that allows users to customize text-to-video (T2V) models without requiring specific video data. By training a lightweight spatial adapter, Still-Moving can generate videos from still images...
CustomCrafter is a custom video generation framework jointly proposed by Tencent and Zhejiang University. It can generate high-quality, personalized videos based on text prompts and reference images, while retaining the capabilities of motion generation and concept composition.
LitServe is a high-performance AI model deployment engine based on FastAPI, designed specifically for enterprise-grade AI services. It supports batch processing, streaming processing, and GPU auto-scaling, simplifying the model deployment process.
LongVILA is a visual language AI model for long-form video understanding, jointly developed by NVIDIA, MIT, UC Berkeley, and the University of Texas at Austin. Through collaborative algorithm and system design, it achieves high performance on a large number of GPUs...
AutoShorts is an open-source AI video creation and publishing platform that uses AI technology to generate and customize faceless videos with a single click. AutoShorts supports custom scripts, voiceovers, and visual effects, and automatically publishes new videos to YouTube and Tier daily...
MARS5-TTS is an open-source AI voice cloning tool launched by CAMB.AI. It boasts groundbreakingly realistic prosody and supports over 140 languages. It can also handle complex prosodic scenarios, such as sports commentary and AI voice-over for animation. MARS5-TTS features 1...
HeadGAP is a 3D avatar generation model jointly developed by ByteDance and ShanghaiTech University. It quickly generates realistic 3D avatars using only a small number of images. It employs a framework of prior learning and personalized creation phases, based on large-scale multi-view dynamic...
LeRobot is an open-source AI chatbot project launched by HuggingFace and led by former Tesla researcher Remi Cadene. LeRobot aims to lower the barrier to entry for robotics, providing pre-trained models, data, and more...
LMMs-Eval is a unified evaluation framework designed specifically for multimodal AI models, providing a standardized, broadly covered, and cost-effective solution for model performance evaluation. It includes over 50 tasks and more than 10 models, offering transparent...
OmniCorpus is a large-scale multimodal dataset containing 8.6 billion images and 1.696 trillion text tags, supporting both Chinese and English. It was jointly built by the Shanghai Artificial Intelligence Laboratory and several renowned universities and research institutions. OmniCorpus...
MooER is the industry's first open-source audio understanding model developed by Moore Threads and trained on a domestically produced, fully functional GPU. It not only performs Chinese and English speech recognition but also has Chinese-to-English speech translation capabilities.
EasyOCR is a powerful open-source OCR (Optical Character Recognition) project that supports over 80 languages and multiple writing systems, including Chinese, Arabic, and Cyrillic. Based on deep learning technology, it provides high-precision text recognition...
StockBot is an AI-powered financial agent based on Llama3, providing real-time stock information, financial data, news, and interactive charts. It supports multiple asset markets, including stocks, forex, bonds, and cryptocurrencies. Users can interact with it through natural...
GptEngineer is an open-source tool that uses AI technology to quickly generate web application prototypes through simple text prompts. Users simply describe their requirements, and the AI can automatically write and execute code. It supports synchronization with GitHub and one-click...
Imagine Yourself is a personalized AI image generation model launched by Meta. It breaks through the limitations of traditional methods, eliminating the need for individual adjustments for each user and meeting diverse user needs through a single mode. The model uses...
STranslate is a multi-functional translation and OCR tool designed specifically for Windows users. It supports multiple languages, features various translation methods including word highlighting, screenshotting, and clipboard monitoring, and provides interfaces to multiple translation services. It also has basic...
Bark is an open-source text-to-audio model from Suno AI that generates realistic multilingual speech and various audio types, including music and background noise, while also supporting nonverbal communication such as laughter and crying. Bark provides pre-trained...
LTM-2-mini is a contextual AI model from Magic Technologies that supports 100 million tokens and can process content equivalent to 10 million lines of code or 750 novels. LTM-2-mini uses a sequence-dimensional algorithm, which is computationally more efficient than Llama 3.1 405...
eSearch is an open-source, cross-platform AI desktop application that integrates features such as screenshotting, OCR recognition, search and translation, image pasting, image search, and screen recording. eSearch is developed based on the Electron framework and is compatible with Linux, Windows, and more...
VectorVein is an open-source, no-code AI workflow tool that allows users to build intelligent workflows and automate daily tasks through simplified drag-and-drop operations, without requiring programming knowledge. It supports data processing, analysis, and...
Sapiens is an AI vision model developed by Meta Labs, designed specifically for understanding human movements in images and videos. It supports tasks such as 2D pose estimation, body part segmentation, depth estimation, and surface normal prediction, employing visual transformation...
OpenCity is a traffic prediction model jointly developed by the University of Hong Kong, South China University of Technology, and Baidu. OpenCity employs the Transformer architecture and graph neural networks, learning the spatiotemporal dependencies of traffic data through large-scale pre-training...
Eagle is a multimodal large-scale model from NVIDIA, adept at processing images up to 1024×1024 pixels, significantly improving visual question answering and document understanding capabilities. The Eagle model employs a multi-expert visual encoder architecture, achieving simple and efficient...