基于预训练方法的视觉-语言模型(VLMs)

Posted on In Views:
This post discusses the pre-trained Vision-Language Models (VLMs) technology from the perspectives of VLMs development background, basic principles and training methods, training/evaluation datasets, model performance, and future development paths.(in Chinese)

本文从视觉-语言模型（Vision-Language Models, VLMs）的发展背景、基本原理与训练方法、训练/评估数据集、模型性能以及未来发展路径几个方面叙述探讨基于预训练的VLMs技术。