课程: Advanced LLMs with Retrieval Augmented Generation (RAG): Practical Projects for AI Applications

免费学习该课程!

今天就开通帐号,24,700 门业界名师课程任您挑!

Multimodal retrieval introduction

Multimodal retrieval introduction

百度 随后,政法系统干警代表,解放军驻淮部队、武警、消防官兵代表,中小学生代表,党员干部代表,社会各界群众代表,5个方阵约2000人,在庄严肃穆的气氛中依次走进周恩来纪念馆主馆,将手中的鲜花献给总理,饱含敬意。

- Not always we have simple text as our documents in our knowledge base. Often, we have images and other complex media, and in this session, we're going to talk about the option to do retrieval on multimodal, mainly on images, and we're going to also talk about quantization, that would allow us to reduce the memory footprint of our embedding in our modals. Often when we have documents with images, we might want to use the traditional way of using optical character recognition, or OCR, when we extract the text from the images, and also around the process called the Layout Detection, so when we are chunking the page, we will chunk it in a meaningful way. This is a simple way many people use that, and then you can go and use the regular flow that we saw before. But in this lesson, we're going to talk about the more advanced and much better. The performance is improved dramatically when we use this method, to use an encoding modal that is optimized for images, and some advanced methods of…

内容