6 Performing Retrieval Augmented Generation on AWS
This chapter covers
- An overview of retrieval augmented generation
- Understanding how retrieval augmented generation is implemented on AWS
- How to create and refine embeddings based on foundational model performance
- Utilize agents to work with foundational models
- Integrating RAG and Agents into a Chatbot
The evolution of LLMs has provided much potential for businesses in changing the landscape of how solutions can be provided. Businesses and developers seek technologies that not only understand and generate human-like text but have a high level of relevance and context-awareness which mirror human intuition. This is where Retrieval Augmented Generation (RAG) comes into play, being at the forefront of bridging the gap between traditional LLMs and the context-rich requirements that are needed by real-world applications.
6.1 Overview of Retrieval Augmented Generation
6.1.1 In-Depth Explanation of RAG
6.1.2 Broad Spectrum of RAG Applications
6.1.3 Advantages and Ethical Considerations
6.1.4 Comparison with Other AI Models
6.1.5 Practical Limitations
6.2 Fundamentals of Retrieval Augmented Generation on AWS
6.2.1 Process of Retrieval Augmented Generation
6.2.2 Selecting an Appropriate Data Source
6.2.3 Selection of Data
6.2.4 Selecting a Vector Database
6.3 Refining Embeddings for RAG Use in Vaccine Information Retrieval
6.3.1 Selecting an Embeddings Model
6.3.2 Configuring LangChain
6.3.3 Preparing Your Data
6.3.4 Question Answering
6.4 Utilizing Agents for Working with Foundational Models
6.4.1 Working with Agents for Amazon Bedrock
6.4.2 Uploading Datasets to Amazon S3
6.4.3 Creating a Knowledge Base
6.4.4 Creating an Agent from the AWS Web Interface
6.4.5 Question Answering
6.5 Summary