TensorRT-LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and build TensorRT engines that contain state-of-the-art optimizations to perform inference efficie…

Transformer related optimization, including BERT, GPT

MindSpore is a new open source deep learning training/inference framework that could be used for mobile, edge and cloud scenarios.

A tensorflow implementation of EAST text detector

10x faster matrix and vector operations

FlexFlow Serve: Low-Latency, High-Performance LLM Serving

主流ocr算法研究实验性的项目，目前实现了CNN+BLSTM+CTC架构

💃 Real-time single person pose estimation for Android and iOS.

Low-bit LLM inference on CPU with lookup table

Realtime C++ code for multi-person pose estimation

OpenPose的简单实现版本，简洁不简单，so easy. 就一个cpp

CNN architecture for articulated human pose estimation

Fast Matrix Multiplications for Lookup Table-Quantized LLMs

Openpose implementation using darknet

OpenPose的简单实现

Pelee(NeurIPS'18)-TensorRT Implementation (Caffe Parser)

https://www.datafountain.cn/competitions/334/details 赛题的baseline开源

Object Detection and Tracking based on C++ and OpenCV

Caffe: a fast open framework for deep learning. Fork maintained by shihenw in CMU.

DELTA-pytorch：DELTA: Dynamically Optimizing GPU Memory beyond Tensor Recomputation

For CyberCHZ

based on official caffe_train, i change some cudnn*.cpp and cudnn*.hpp for cuda9.