Code for the papers: “Four Over Six: More Accurate NVFP4 Quantization with Adaptive Block Scaling” and “Adaptive Block-Scaled Data Types”

Accessible large language models via k-bit quantization for PyTorch.

Official repository for the "Big Transfer (BiT): General Visual Representation Learning" paper.

NVIDIA Cosmos is an open platform of world models, datasets, and tools that enables developers to build Physical AI for robots, autonomous vehicles, smart infrastructure, and more.

MiniCPM-2B: An end-side LLM outperforming Llama2-13B.

"Effective Whole-body Pose Estimation with Two-stages Distillation" (ICCV 2023, CV4Metaverse Workshop)

This is a brief PyTorch implementation of EfficientNetV2, providing with experiments on ImageNet and CIFAR-10/100.

Pytorch Implementation For LPRNet, A High Performance And Lightweight License Plate Recognition Framework.

[IJCV-2021] FairMOT: On the Fairness of Detection and Re-Identification in Multi-Object Tracking

yolox人脸检测，含关键点

NanoDet-Plus⚡Super fast and lightweight anchor-free object detection model. 🔥Only 980 KB(int8) / 1.8MB (fp16) and run 97FPS on cellphone🔥

💻📖对开发人员有用的定律、理论、原则和模式。(Laws, Theories, Principles and Patterns that developers will find useful.)

A Deep-Learning-Based Chinese Speech Recognition System 基于深度学习的中文语音识别系统

The pytorch re-implement of the official efficientdet with SOTA performance in real time and pretrained weights.