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

A compact implementation of SGLang, designed to demystify the complexities of modern LLM serving systems.

AIInfra（AI 基础设施）指AI系统从底层芯片等硬件，到上层软件栈支持AI大模型训练和推理。

dParallel: Learnable Parallel Decoding for dLLMs

Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels

Distributed MoE in a Single Kernel [NeurIPS '25]

Nano vLLM

[NeurIPS 2025 Oral]Infinity⭐️: Unified Spacetime AutoRegressive Modeling for Visual Generation

A Collection of Papers on Diffusion Large Language Models

[NeurIPS'25] dKV-Cache: The Cache for Diffusion Language Models

Official PyTorch implementation for "Large Language Diffusion Models"

FlashInfer: Kernel Library for LLM Serving

REAP: Router-weighted Expert Activation Pruning for SMoE compression

A sparse attention kernel supporting mix sparse patterns

Github Pages template based upon HTML and Markdown for personal, portfolio-based websites.

The best ChatGPT that $100 can buy.

[NeurIPS 2025] Accelerating Parallel Diffusion Model Serving with Residual Compression

Ring attention implementation with flash attention

🚀 Efficient implementations of state-of-the-art linear attention models

code for paper LiteVAR

[CVPR 2025 Oral]Infinity ∞ : Scaling Bitwise AutoRegressive Modeling for High-Resolution Image Synthesis

[ECCV 2024 Oral] Code for paper: An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models

This repository contains the official implementation of "FastVLM: Efficient Vision Encoding for Vision Language Models" - CVPR 2025

Official repository for VisionZip (CVPR 2025)

[ICML'25] Official implementation of paper "SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference" and "SparseVLM+: Visual Token Sparsification with Improved Text-Vis…