What you'll learn

Enhance keyword search using Cohere Rerank
Use embeddings to leverage dense retrieval, a powerful NLP tool
Evaluate your effectiveness for further optimization

Course recap

PRO

This course, built with Cohere, teaches how to incorporate LLMs into information search, progressing from basic keyword search through reranking and dense retrieval with embeddings to full generative search. Taught by Jay Alammar and Luis Serrano, it covers both the conceptual foundations and practical implementation of modern semantic search systems.

Concept map

Key concepts

EmbeddingsDense vector representations that capture semantic meaning, enabling similarity-based search that goes beyond keyword matching
Semantic searchFinding documents based on meaning rather than exact keyword matches, using embedding similarity
Retrieval-Augmented GenerationPlugging search results into an LLM to generate answers grounded in retrieved information
RerankingUsing a model to reorder search results by relevance after initial retrieval

Lesson highlights

1.Introduction — Overview of semantic search and the progression from keyword to LLM-powered search
2.Keyword Search — Traditional lexical search based on word frequency matching (TF-IDF, BM25)
3.Rerank — Enhancing keyword search results by reranking with a relevance model
4.Dense Retrieval — Embedding-based search using vector similarity for semantic matching
5.Embeddings — Deep dive into how text embeddings work and how to use them effectively
6.Search Evaluation — Methods for evaluating search quality and relevance
7.Generative Search — Combining retrieval with LLM generation for answer synthesis

About this course

Keyword search has been a common method for search for many years. But for content-rich websites like news media sites or online shopping platforms, the keyword search capability can be limiting. Incorporating large language models (LLMs) into your search can significantly enhance the user experience by allowing them to ask questions and find information in a much easier way.

This course teaches the techniques needed to leverage LLMs into search.

Throughout the lessons, you’ll explore key concepts like dense retrieval, which elevates the relevance of retrieved information, leading to improved search results beyond traditional keyword search, and reranking, which injects the intelligence of LLMs into your search system, making it faster and more effective.

After completing the course, you will:

Know how to implement basic keyword search, the underpinnings of many search systems before language models became accessible.
Enhance keyword search with the rerank method, which ranks the best responses by relevance with the query.
Implement dense retrieval through the use of embeddings, a powerful NLP tool, which uses the actual semantic meaning of the text to carry out search, and vastly improves results.
Gain hands-on practice by working with large amounts of data and overcome challenges like varying search results and accuracy.
Implement language model-powered search into your website or project.

Who should join?

Anyone who has basic familiarity with Python and wants to get a deeper understanding of key technical foundations of LLMs, and learn to use semantic search.