A user wants a language model to solve the following multi-step arithmetic problem: 'A bookstore starts the day with 50 copies of a new novel. They sell 18 copies in the morning and receive a new shipment of 25 copies in the afternoon. If they then sell 12 more copies before closing, how many copies are left?' Which of the following prompts best demonstrates the one-shot chain-of-thought technique to guide the model towards the correct answer?
0
1
Tags
Ch.2 Generative Models - 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
A user wants a language model to solve the following multi-step arithmetic problem: 'A bookstore starts the day with 50 copies of a new novel. They sell 18 copies in the morning and receive a new shipment of 25 copies in the afternoon. If they then sell 12 more copies before closing, how many copies are left?' Which of the following prompts best demonstrates the one-shot chain-of-thought technique to guide the model towards the correct answer?
Constructing a One-Shot CoT Prompt for a Logic Puzzle
A researcher is using a language model to solve a simple logic puzzle. Their goal is to have the model determine the final position of an object after a series of movements.
Puzzle: A red ball is in a box. The box is moved to the left, and then the ball is taken out and placed to the right of the box. Where is the ball now relative to its starting position?
Researcher's Initial Prompt:
Q: A red ball is in a box. The box is moved to the left, and then the ball is taken out and placed to the right of the box. Where is the ball now relative to its starting position? A:Model's Incorrect Answer:
The ball is to the left of its starting position.The researcher realizes they need to provide a single, complete example to guide the model's reasoning. Which of the following revised prompts best applies this technique to help the model solve the puzzle correctly?