From the course: Hands-On AI: Build a RAG Model from Scratch with Open Source

Unlock this course with a free trial

Join today to access over 24,900 courses taught by industry experts.

What are vector embeddings, and how are they generated?

What are vector embeddings, and how are they generated?

From the course: Hands-On AI: Build a RAG Model from Scratch with Open Source

Start my 1-month free trial Buy for my team

What are vector embeddings, and how are they generated?

“

- [Instructor] The next step in the process is creating vector embeddings that hold the documents we'll be referencing in our RAG models. Vector embeddings are a way of converting our information into a format that computers can understand. This can apply to words, sentences, paragraphs, audio clips, videos, or any other information. A vector embedding is generally meant to hold a single idea. If you have two sentences each holding a different idea, you'll need a different vector embedding for each sentence. Accordingly, if a single sentence is very long and holds multiple ideas, then a single vector embedding might not be suitable. To understand why, consider that a vector embedding is a single position in a vector space. This single position can only encode one idea. In this course, we'll cover the theory of vector embeddings of words for educational purposes, but in application, we'll be using more complex embeddings…

Contents