MILS (Multimodal Iterative LLM Solver) is a method proposed by Meta AI that endows large language models (LLMs) with multimodal capabilities without training. Through multi-step inference, it prompts the LLM to generate candidate outputs, and for each...
OpenDeepResearcher is an open-source AI research tool that helps users efficiently complete complex information retrieval and analysis tasks. Based on iterative search and inference, it automatically retrieves relevant data from the web and generates comprehensive reports. ...
OCRmyPDF is an open-source command-line tool designed to convert scanned PDF files into searchable and copyable documents. By adding an OCR text layer, scanned PDF files that cannot be directly edited can be searched and edited. ...
node-DeepResearch is an open-source AI agent project that continuously searches and reads web pages, using the Gemini language model and Jina Reader tool to progressively reason and answer complex questions until it finds the answer or exceeds expectations...
OmniHuman is an end-to-end multimodal conditional human video generation framework launched by ByteDance. It can generate realistic human videos based on a single human image and motion signals (such as audio, video, or a combination of both). OmniHuman...
DeepRAG is a novel retrieval augmentation generation (RAG) framework jointly launched by the Chinese Academy of Sciences, the University of Science and Technology of China, and Tencent WeChat AI Department. It aims to improve the performance of large language models (LLMs) in reasoning and retrieval tasks. This is achieved by building retrieval-enhanced reasoning...
MnnLlmApp is an open-source Android mobile application from Alibaba based on the MNN-LLM framework. It supports offline execution of various large language models (LLMs) on mobile devices. It features multimodal capabilities, enabling text-to-text generation, image generation, and more.
DynamicFace is a new video face-swapping technology launched by the Xiaohongshu team. By combining a diffusion model and a plug-and-play temporal layer, based on 3D facial prior knowledge, the technology achieves high-quality and consistent video face-swapping effects.
Gemini 2.0 Pro is a high-performance experimental AI model from Google, optimized for programming performance and handling complex hints. Gemini 2.0 Pro features a massive context window with 2 million tokens, capable of processing and analyzing massive amounts of information...
Open Deep Research is an open-source AI agent, a replica of the Deep Research open-source project. It performs complex, multi-step research tasks based on inference of large amounts of network data. Open Deep Research uses Firecrawl...
Chatbox AI is an open-source, cross-platform AI client application and intelligent assistant that supports Windows, Mac, Linux, iOS, Android, and web versions. It integrates multiple advanced language models (such as ChatGPT, DeepSeek, Clau...).
RAG-FiT (formerly known as RAG Foundry) is an open-source framework developed by Intel Labs for fine-tuning large language models (LLMs) in retrieval augmentation generation (RAG) tasks. RAG-FiT...
FluxSR is a novel single-step diffusion model developed by Shanghai Jiao Tong University, Harvard University, South China University of Technology, and Huawei Noah's Ark Lab, specifically for real-world image super-resolution (Real-ISR) tasks. FluxSR is based on...
VideoJAM, developed by Meta, is a framework for enhancing motion coherence in video generation models. It introduces a joint appearance-motion representation, allowing the model to simultaneously learn and predict pixel and motion information during training, and then...
SynCD (Synthetic Customization Dataset) is a high-quality synthetic training dataset released by Carnegie Mellon University and Meta, used to improve the customization capabilities of text-to-image models. SynCD contains multiple identical objects...
MatAnyone, developed by Nanyang Technological University's S-Lab and SenseTime, is an advanced framework for matting portrait videos against complex backgrounds, focusing on target-specific video matting tasks. MatAnyone is based on a consistent memory propagation model...
JoyGen, developed by JD Technology and the University of Hong Kong, is an audio-driven 3D speaking face video generation framework focused on achieving accurate lip-to-audio synchronization and high-quality visual effects. JoyGen combines audio features and facial depth...
EMAGE (Expressive Masked Audio-conditioned Gesture Modeling) is a framework developed by Tsinghua University, the University of Tokyo, Keio University, and other institutions for generating full-body communicative gestures. EMAGE can generate gestures based on audio...
LIMO (Less Is More for Reasoning) is an efficient reasoning method proposed by a research team at Shanghai Jiao Tong University. It activates the complex reasoning capabilities of Large Language Models (LLMs) using a very small number of high-quality training samples. Its core assumptions...
SANA 1.5 is a new, highly efficient Linear Diffusion Transformer developed by NVIDIA in collaboration with institutions such as MIT, Tsinghua University, and Peking University, for text-to-image generation tasks. It builds upon SANA 1.0...
ASAP (Aligning Simulation and Real Physics) is a two-stage framework jointly developed by Carnegie Mellon University and NVIDIA to address the mismatch between simulated and real-world dynamics in humanoid robots. ASAP is based on...
MAETok (Masked Autoencoders Tokenizer) is a novel image tokenization method developed by Carnegie Mellon University, the University of Hong Kong, Peking University, and other institutions, used in diffusion models. MAETok is based on mask modeling (Mask Modeling)...
Eino is an open-source framework developed by ByteDance for developing large-scale AI applications. It helps developers efficiently build AI applications based on large models. Eino is based on the Go programming language and features a stable kernel, flexible extensibility, and comprehensive toolkits...
NutWorld is a video processing framework developed by the National University of Singapore, Nanyang Technological University, and Skywork AI. It efficiently converts everyday monocular videos into dynamic 3D Gaussian splatting. NutWorld is based on...