Welcome to this short course, Large Language Models with Semantic Search, built in partnership with Cohere. In this course, you'll learn how to incorporate large language models, or LLMs, into information search in your own applications. For example, let's say you run a website with a lot of articles, picture Wikipedia for the sake of argument, or a website with a lot of e-commerce products. Even before LLMs, it was common to have keyword search to let people maybe search your site. But with LLMs, you can now do much more. First, you can let users ask questions that your system then searches your site or database to answer. Second, the LLM is also making the retrieve results more relevant to the meaning or the semantics of what the user is asking about. I'd like to introduce the instructors for this course, Jay Allamar and Luis Serrano. Both Jay and Luis are experienced machine learning engineers as well as educators. I've admired for a long time some highly referenced illustrations that Jay had created to explain transformer networks. He's also co-authoring a book, Hands-On Large Language Models. Luis is the author of the book, Grokking Machine Learning, and he also taught with DeepLearning.ai. Math for Machine Learning. At Cohere, Jay and Luis, together with Neil Amir, have also been working on a site called. LLMU, and have a lot of experience teaching developers to use LLMs. So I was thrilled when they agreed to teach semantic search with LLMs. Thanks, Andrew. What an incredible honor it is to be teaching this course with you. Your machine learning course introduced me to machine learning eight years ago, and continues to be an inspiration to continue sharing what I learn. As you mentioned, Luis and I work at Cohere, so we get to advise others in the industry on how to use and deploy large language models for various use cases. We are thrilled to be doing this course to give developers the tools they need to build robust LLM powered apps. We're excited to share what we learned from our experience in the field. Thank you, Jay and Luis. Great to have you with us. This course consists of the following topics. First, it shows you how to use basic keyword search, which is also called lexical search, which powered a lot of search systems before large language models. It consists of finding the documents that has the highest amount of matching words with the query. Then you learn how to enhance this type of keyword search with a method called re-rank. As the name suggests, this then ranks the responses by relevance with the query. After this, you learn a more advanced method of search, which has vastly improved the results of keyword search, as it tries to use the actual meaning or the actual semantic meaning of the text with which to carry out the search. This method is called dense retrieval. This uses a very powerful tool in natural language processing called embeddings, which is a way to associate a vector of numbers with every piece of text. Semantic search consists of finding the closest documents to the query in the space of embeddings. Similar to other models, search algorithms need to be properly evaluated. You also learn effective ways to do this. Finally, since LLMs can be used to generate answers, you also learn how to plug in the search results into an LLM and have it generate an answer based on them. Dense retrieval with embeddings vastly improves the question answering capabilities of an LLM as it first searches for and retrieves the relevant documents and it creates an answer from this retrieved information. Many people have contributed to this course. We're grateful for the hard work of Meor Amer, Patrick Lewis, Nils Reimer, and Sebastian Hofstatter from Cohere, as well as on the DeepLearning.ai side, Eddie Shyu and Diala Ezzedine. In the first lesson, you will see how search was done before large language models. From there, we will show you how to improve search using LLMs, including tools such as embeddings and re-rank. That sounds great. And with that, let's dive in and go on to the next video.
Week 1: Large Language Models with Semantic Search