Adapting a Single LLM for Multi-Task Problem Solving
A development team is building a system to solve complex mathematical word problems. They plan to use a single large language model that will first break down each problem into simpler steps and then solve each of those steps. Describe two distinct methods the team could use to adapt their single model to perform these two separate tasks (sub-problem generation and sub-problem solving), and briefly explain a key difference between them.
0
1
Tags
Ch.3 Prompting - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
Least-to-Most Prompting
A software team is tasked with creating a system that can solve complex, multi-step logic puzzles. They have access to a single, powerful, general-purpose Large Language Model. Their strategy is to first have the system break down a given puzzle into a series of simpler, solvable steps, and then solve each of those steps in sequence to arrive at the final answer.
Which of the following implementation plans best applies the principle of adapting a single model for the separate tasks of sub-problem generation and sub-problem solving?
Diagnosing and Improving an LLM-based Problem-Solving System
Adapting a Single LLM for Multi-Task Problem Solving