Build datasets using natural language

ViDoRAG: Visual Document Retrieval-Augmented Generation via Dynamic Iterative Reasoning Agents

Deep Research for your internal data

Synthetic data generation pipelines for text-rich images.

Higher performance OpenAI LLM service than vLLM serve: A pure C++ high-performance OpenAI LLM service implemented with GPRS+TensorRT-LLM+Tokenizers.cpp, supporting chat and function call, AI agents…

Secure and local AI on your desktop with a built-in RAG knowledge base and Markdown note support.

A PyTorch library for implementing flow matching algorithms, featuring continuous and discrete flow matching implementations. It includes practical examples for both text and image modalities.

Low-resource, fast, and privately self-host free version of Google Translate - 低占用速度快可私有部署的自由版 Google 翻译

Archon is an AI agent that is able to create other AI agents using an advanced agentic coding workflow and framework knowledge base to unlock a new frontier of automated agents.

Code for the paper: "FusionMamba: Efficient Image Fusion with State Space Model", TGRS, 2024.

The Cyber Swiss Army Knife - a web app for encryption, encoding, compression and data analysis

Official repo of paper LM2

A-MEM: Agentic Memory for LLM Agents

A novel agentic memory system

A lightweight data processing framework built on DuckDB and 3FS.

An open-source visual programming environment for battle-testing prompts to LLMs.

An implementation of Shazam's song recognition algorithm.

x86 PC emulator and x86-to-wasm JIT, running in the browser

Wan: Open and Advanced Large-Scale Video Generative Models

Latest Advances on System-2 Reasoning

Al (especially computer vision) paper review and model code implementation by PyTorch

The python library for real-time communication

Spotlight Attention: Extreme KV Cache Pruning for Efficient LLM Generation

Toolkit for linearizing PDFs for LLM datasets/training

DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling