Explanation
Correct Answer: B - Adopt RAG with Knowledge Bases to retrieve only relevant context at runtime
Why this is correct:
- RAG (Retrieval-Augmented Generation) with Knowledge Bases allows the model to retrieve only relevant information from a knowledge base at runtime, rather than processing all data in every query.
- Cost Reduction: By retrieving only relevant context, you reduce the amount of data processed per query, which lowers computational costs.
- Performance Maintenance: RAG maintains or improves performance by providing the model with precise, relevant information rather than overwhelming it with unnecessary data.
- AWS Bedrock Integration: AWS Bedrock supports Knowledge Bases for Amazon Bedrock, which is specifically designed for RAG implementations.
Why other options are incorrect:
- A) Use fine-tuning for every patient query: Fine-tuning is expensive and time-consuming for every query, and doesn't scale well for dynamic patient data.
- C) Store entire medical records in each prompt: This would increase token usage dramatically, leading to higher costs and potentially worse performance due to context window limitations.
- D) Scale up GPU instances permanently: This would increase costs without addressing the root cause of inefficiency and doesn't optimize for cost reduction.
Key AWS Concepts:
- AWS Bedrock Knowledge Bases: Provide managed RAG capabilities with vector storage and retrieval
- Cost Optimization: Reducing token usage and computational requirements
- Performance Efficiency: Balancing cost with response quality and speed
