Prompting Strategy and RAG Evaluation

Company: Oracle_27oct mcq

Difficulty: medium

Problem Statement

Question 1

A Generative AI engineer is designing a novel prompting strategy for an LLM-powered document retrieval application. Which of the following techniques cannot be used to prompt the LLM?

Question 2

A team building a multilingual medical RAG chatbot observed inconsistencies in factual accuracy between English and Spanish. How should they diagnose and address the language-specific inconsistencies?

Question 3

What challenges may arise using a prompt with simple input-output examples (without intermediate steps) to solve linear systems with LLMs?

Question 4

Which two of the following approaches best address the challenges in evaluating a RAG pipeline for diagnosing rare diseases where factual accuracy is paramount?

Question 5

Given a dataset with a noticeable difference in the number of observations across classes, which technique is most appropriate to improve performance by addressing potential variance issues?