王召德

VirtualFaceCapture-MNN. Android real-time face capture framework for Live2D avatar animation, based on MediaPipe landmarks , MNN engine and custom tracking logic. 安卓端皮套人面捕系统，基于阿里 MNN。

open-code-review. Open-source & free — Battle-tested at Alibaba's scale. Hybrid architecture code review tool: deterministic pipelines + LLM Agent, precise line-level comments, built-in fine-tuned ruleset (NPE, thread-safety, XSS, SQL injection), OpenAI & Anthropic compatible.

Thinking-with-Visual-Primitives-pytorch. Unofficial PyTorch reproduction of DeepSeek's Thinking with Visual Primitives.

Psi0. [RSS26'] Welcome to Psi-Zero, a Humanoid VLA towards Universal Humanoid Intelligence.

hermes-agent. The agent that grows with you

claudelens. AI coding CLI conversation visualizer - supports Claude Code, Codex, Gemini CLI, OpenCode

clawdicons. Python

opencode. The open source coding agent.

MSA. Memory Sparse Attention - A scalable, end-to-end trainable latent-memory framework for 100M-token contexts.

parameter-golf. Train the smallest LM you can that fits in 16MB. Best model wins!

opencode-bot. Python

any_gateway. A light-weight self-hosted AI API gateway that proxies requests to multiple backend providers (OpenAI, Anthropic, Gemini) with user management, quota control, and audit logging.

openclaw. Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

page-agent. JavaScript in-page GUI agent. Control web interfaces with natural language.

transformers. 🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

tokenizer.cpp. A lightweight, production-ready C++ library for LLM tokenization, fully compatible with HuggingFace tokenizer.json.

jinja.cpp. A lightweight, single-header C++11 Jinja2 template engine for LLM chat templates.

minimind. 🧠「大模型」2小时完全从0训练64M的小参数LLM！Train a 64M-parameter LLM from scratch in just 2h!

SpecForge. Train speculative decoding models effortlessly and port them smoothly to SGLang serving.

youkaichao. github profile

nanochat. The best ChatGPT that $100 can buy.

ms-swift. Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.6, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM4.5v, Gemma4, Llava, Phi4, ...) (AAAI 2025).

llm-datasets. Curated list of datasets and tools for post-training.

SwanLab. ⚡️SwanLab - an open-source, modern-design AI training tracking and visualization tool. Supports Cloud / Self-hosted use. Integrated with PyTorch / Transformers / verl / LLaMA Factory / ms-swift / Ultralytics / MMEngine / Keras etc.

HRM. Hierarchical Reasoning Model Official Release

F5-TTS. Official code for "F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching"

gemini-cli. An open-source AI agent that brings the power of Gemini directly into your terminal.

MNNServer. A third-party MNN server supporting external calls, embedding model, TTS model and ASR model features.一个支持外部调用、向量模型、文字转语音模型和语音识别模型特性的第三方MNN服务器

MMaDA. MMaDA - Open-Sourced Multimodal Large Diffusion Language Models (dLLMs with block diffusion, mixed-CoT, unified RL)

Qwen2.5-Omni. Qwen2.5-Omni is an end-to-end multimodal model by Qwen team at Alibaba Cloud, capable of understanding text, audio, vision, video, and performing real-time speech generation.

XiangShan. Open-source high-performance RISC-V processor

mnn-tts. mnn tts demo.

mnn-asr. mnn asr demo.

ClearerVoice-Studio. An AI-Powered Speech Processing Toolkit and Open Source SOTA Pretrained Models, Supporting Speech Enhancement, Separation, and Target Speaker Extraction, etc.

mnn-llm. llm deploy project based mnn. This project has merged into MNN.

onnx-llm. llm deploy project based onnx.

ultralytics. Ultralytics YOLO26, YOLO11, YOLOv8 — object detection, instance segmentation, semantic segmentation, image classification, pose estimation, object tracking

GOT-OCRv2-onnx. 用于学习GOT/Qwen/OnnxLLm

caffe2onnx. caffe model to onnx

m3u8-downloader. m3u8 视频在线提取工具

blind_index. Securely search encrypted database fields

KnowledgeGraph. 史上最大规模1.4亿知识图谱数据免费下载，知识图谱，通用知识图谱，融合了两千五百多万的实体，拥有亿级别的实体属性关系。

Few-Shot-Semantic-Segmentation-Papers. Few Shot Semantic Segmentation Papers

pixel-recursive-super-resolution. Tensorflow implementation of pixel-recursive-super-resolution(Google Brain paper: https://arxiv.org/abs/1702.00783)

antares. Antares: an automatic engine for multi-platform kernel generation and optimization. Supporting CPU, CUDA, ROCm, DirectX12, GraphCore, SYCL for CPU/GPU, OpenCL for AMD/NVIDIA, Android CPU/GPU backends.

WritingAIPaper. Writing AI Conference Papers: A Handbook for Beginners

filter-pruning-geometric-median. Filter Pruning via Geometric Median for Deep Convolutional Neural Networks Acceleration (CVPR 2019 Oral)

completor.vim. Async completion framework made ease.

zipper. C++ wrapper around minizip compression library

OpenVoice. Instant voice cloning by MIT and MyShell. Audio foundation model.

3D-Speaker. A Repository for Single- and Multi-modal Speaker Verification, Speaker Recognition and Speaker Diarization

gemma.cpp. lightweight, standalone C++ inference engine for Google's Gemma models.

deepmind-research. This repository contains implementations and illustrative code to accompany DeepMind publications

whisper. Robust Speech Recognition via Large-Scale Weak Supervision

google-research. Google Research

light-show. Tesla Light Show

OnnxSlim. A Toolkit to Help Optimize Onnx Model

yaspin. A lightweight terminal spinner for Python 🎁

OnnxLLM. Large Language Model Onnx Inference Framework

cibuildwheel. 🎡 Build Python wheels for all the platforms with minimal configuration.

torchkeras. Pytorch❤️ Keras 😋😋

AISystem. AISystem 主要是指AI系统，包括AI芯片、AI编译器、AI推理和训练框架等AI全栈底层技术

amx. Apple AMX Instruction Set

laser. The Truth Is In There: Improving Reasoning in Language Models with Layer-Selective Rank Reduction

objectiveflickr. ObjectiveFlickr, a Flickr API framework for Objective-C

Anima. Anima is chainable Layer-Based Animation library for Swift5.

bypy. Python client for Baidu Yun (Personal Cloud Storage) 百度云/百度网盘Python客户端

ios7-wireframe-kit. An Illustrator kit to make your wireframing a day at the beach

Teeworlds-Example. Quickly make Teeworlds mod

tcpkit. the tcpkit was designed to make network packets programable with Lua script

vue-tree-list. 🌲A vue component for tree structure

graph-composer. Dependency graph visualization for composer.json (PHP + Composer)

dart_vlc. Flutter bindings to libVLC.

foolrenderer. A tiny software renderer implemented from scratch without the use of graphics API, used to understand how GPUs work.

contacts. Ruby library for consuming Google, Yahoo!, Flickr and Windows Live contact APIs

termusic. Music Player TUI written in Rust

python-google-places. Simple wrapper around the new 'experimental' Google Places API

github-daily-rank. Github开源项目:每天📈飙升榜 top10,每天早上8:30更新

llm-course. Course to get into Large Language Models (LLMs) with roadmaps and Colab notebooks.

vscode. Visual Studio Code

json. JSON for Modern C++

MNN. MNN: A blazing-fast, lightweight inference engine battle-tested by Alibaba, powering high-performance on-device LLMs and Edge AI.

PowerInfer. High-speed Large Language Model Serving for Local Deployment

mistral-inference. Official inference library for Mistral models

pymnn-llm. Python

mnn-segment-anything. segment-anything based mnn

nndeploy. 一款简单易用和高性能的AI部署框架 | An Easy-to-Use and High-Performance AI Deployment Framework

fast_bpe_tokenizer. fast bpe tokenizer, simple to understand, easy to use

llm-export. llm-export can export llm model to onnx.

fastllm. fastllm是后端无依赖的高性能大模型推理库。同时支持张量并行推理稠密模型和混合模式推理MOE模型，任意10G以上显卡即可推理满血DeepSeek。双路9004/9005服务器+单显卡部署DeepSeek满血满精度原版模型，单并发20tps；INT4量化模型单并发30tps，多并发可达60+。

CodeGeeX. CodeGeeX: An Open Multilingual Code Generation Model (KDD 2023)

mnn-yolo. mnn yolo demos.

mnn-mobilenet. Mobilenet using MNN. Support Python, C++, Android, iOS.

ChatGLM-6B. ChatGLM-6B: An Open Bilingual Dialogue Language Model | 开源双语对话语言模型

mperf. mperf是一个面向移动/嵌入式平台的算子性能调优工具箱

ncnn. ncnn is a high-performance neural network inference framework optimized for the mobile platform

mnn-stable-diffusion. stable diffusion using mnn

msvc-dev-cmd. GitHub Action to setup Developer Command Prompt for Microsoft Visual C++

RxSwift. Reactive Programming in Swift

TinyNeuralNetwork. TinyNeuralNetwork is an efficient and easy-to-use deep learning model compression framework.

pybind11. Seamless operability between C++11 and Python

islpy. Python wrapper for isl, an integer set library

sentiment_classifier. Python

omp4j. Lightweight Java OMP preprocessor

Track. a android app for tracking the daily track of your own

coding_practice. JavaScript

MiniSQL_IndexManager. C++

online_judge. pat && leetcode

ansj_seg. ansj分词.ict的真正java实现.分词效果速度都超过开源版的ict. 中文分词,人名识别,词性标注,用户自定义词典

Cocos2dx_car. 一个数据结构的停车场课设，题目要求停车场是一个栈..真是可怕

DrawSomething. 你画我猜-Java小游戏

MeloduetMark. 一个精简的支持插件扩展的Markdown编辑器

build-web-application-with-golang. A golang ebook intro how to build a web with golang

PL0-compiler-SLR. 一个Java实现的简易PL/0编译器

RayTracing. C++ 实现的光线追踪器

xv6-chinese. 中文版的 MIT xv6 文档