Building a full-fledged code editor for iPad

Disaggregated serving system for Large Language Models (LLMs).

从无名小卒到大模型（LLM）大英雄~ 欢迎关注后续！！！

A low-latency & high-throughput serving engine for LLMs

Implement some method of LLM KV Cache Sparsity

A machine learning accelerator core designed for energy-efficient AI at the edge.

[DATE'2025, TCAD'2025] Terafly : A Multi-Node FPGA Based Accelerator Design for Efficient Cooperative Inference in LLMs

[DATE'25, ICCAD'25] An embedded FPGA-based LLM accelerator capable of supporting Llama2-7B

LeetGPU Challenges

Official Implementation of EAGLE-1 (ICML'24), EAGLE-2 (EMNLP'24), and EAGLE-3 (NeurIPS'25).

A modern GUI client based on Tauri, designed to run in Windows, macOS and Linux for tailored proxy experience

Awesome Pruning. ✅ Curated Resources for Neural Network Pruning.

强化学习中文教程（蘑菇书🍄），在线阅读地址：https://datawhalechina.github.io/easy-rl/

[ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

LLM model quantization (compression) toolkit with hw acceleration support for Nvidia CUDA, AMD ROCm, Intel XPU and Intel/AMD/Apple CPU via HF, vLLM, and SGLang.

An easy-to-use LLMs quantization package with user-friendly apis, based on GPTQ algorithm.

GEMV implementation with CUTLASS

Course Project for High Level Chip Design （高层次芯片设计）

Contexts Optical Compression

🚀🚀 「大模型」2小时完全从0训练26M的小参数GPT！🌏 Train a 26M-parameter GPT from scratch in just 2h!

🚀🚀🚀 This repository lists some awesome public CUDA, cuda-python, cuBLAS, cuDNN, CUTLASS, TensorRT, TensorRT-LLM, Triton, TVM, MLIR, PTX and High Performance Computing (HPC) projects.

high-performance RTL simulator

Let's Learn AI SYStem

LLM大模型（重点）以及搜广推等 AI 算法中手写的面试题，（非 LeetCode），比如 Self-Attention, AUC等，一般比 LeetCode 更考察一个人的综合能力，又更贴近业务和基础知识一点

Bringing Language Models to the Most Resource Constrained Devices

OpenVINO™ is an open source toolkit for optimizing and deploying AI inference

Fast inference from large lauguage models via speculative decoding