Accelerating MoE with IO and Tile-aware Optimizations

VideoSys: An easy and efficient system for video generation

高性价比人生指南: 长寿防病、急救、省钱理财、法律红线、失业与工伤、医保社保、恋爱婚育、怀孕育儿、创业与做平台合规、出国与技能。每条写明成本、收益、证据等级和原始出处，只引期刊论文与官方文件。

HieraSparse: Hierarchical Semi-Structured KV-Cache Attention on Sparse Tensor Core

Code for the paper “SharQ: Bridging Activation Sparsity and FP4 Quantization for LLM Inference”

Code for the paper "Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning" https://arxiv.org/abs/2609.03430

My Python scripts to make high-quality figures for publications in top AI conferences and journals.

Defend sound research against LLM reviewer bias through meaning-preserving rewrites.

Benchmarking Knowledge Transfer in Lifelong Robot Learning

Official repository of LIBERO-plus, a generalized benchmark for in-depth robustness analysis of vision-language-action models.

Measuring and evolving with the frontier of agent work

A Vulkan renderer for side-by-side quality and performance comparison of upscaling algorithms.

Memory is a long-term memory module for AI agents, providing capabilities for memory extraction, storage, retrieval, and migration for agents running on the openJiuwen framework.

SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?

Fast, Flexible and Portable Structured Generation

FlashInfer: Kernel Library for LLM Serving

Benchmarking physical understanding in generative video models

KaHIP -- HIGH Quality Partitioning.

[NeurIPS'24] HippoRAG is a novel RAG framework inspired by human long-term memory that enables LLMs to continuously integrate knowledge across external documents. RAG + Knowledge Graphs + Personali…

From Foundation to Application

PhyAI is a high-performance framework for running Physical AI models (VLA, WAM, and beyond), supporting both cloud-based serving and on-device deployment.

Sub2API 一站式开源中转服务，让 Claude、Openai 、Gemini、Grok订阅统一接入，支持拼车共享，更高效分摊成本，原生工具无缝使用。

PonderTTT: Adaptive Budget-Aware Test-Time Training

TokenSpeed is a speed-of-light LLM inference engine.

XtraMAC code repo (Accepted by ISCA2026)

Graphs for Everyone

Milvus is a high-performance, cloud-native vector database built for scalable vector ANN search

提供一个人人会用的的路由、NAS系统 （目前活跃的分支是 istoreos-24.10，main或master分支不维护请勿使用）

Official implementation of "Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding"