Google

One-shot Chain-of-Thought (CoT) prompting is a technique where a single demonstration, complete with step-by-step reasoning, is provided in the prompt. This example serves to guide the language model in its own reasoning process for a similar, subsequent task.

One-Shot Chain-of-Thought (CoT) Prompting

Few-shot Chain-of-Thought (CoT) prompting is an extension of the one-shot approach that involves including multiple demonstrations, each with its detailed reasoning steps, within the prompt. This provides the model with a richer set of examples to learn from, further improving its ability to generate a reasoned response.

Few-Shot Chain-of-Thought (CoT) Prompting

You need to guide a language model to solve a simple logic problem by showing it how to reason through a similar one first. Construct a complete one-shot chain-of-thought prompt to determine which building is the tallest based on the following information: 'A library is taller than a school. A hospital is shorter than the school.' Your prompt must include a full demonstration (question, step-by-step reasoning, and final answer) for a different, but analogous, logic problem.

Constructing a One-Shot CoT Prompt

A developer is trying to get a language model to solve multi-step word problems by showing it an example. Analyze the following prompt. Based on the principles of effective reasoning guidance, evaluate its likelihood of success and explain the most critical flaw in its design.

Evaluating a Prompt for a Language Model

A one-shot Chain-of-Thought (CoT) prompt can be constructed by combining a solved demonstration with an unsolved target problem. For example, a prompt can first present the 'Tom's marbles' problem along with its complete, step-by-step logical reasoning (calculating initial marbles, additions from winning or gifts, and subtractions from losing) to reach the final answer of $${}17$$. After this full demonstration, the prompt introduces a new word problem, such as 'Jack's apples'. Guided by the preceding example, a language model is encouraged to output a structured sequence of intermediate calculations (such as subtracting eaten apples and adding given apples) before stating the correct final answer of $${}7$$.

Example of a One-Shot CoT Prompt with Arithmetic Problems

A user provides the following text to a large language model:

---
Question: A cafeteria had 23 apples. If they used 20 to make lunch and bought 6 more, how many apples do they have?

Answer: The cafeteria started with 23 apples. They used 20, so they had 23 - 20 = 3 apples. Then they bought 6 more, so they now have 3 + 6 = 9 apples. The final answer is 9.

Question: Roger has 5 tennis balls. He buys 2 more cans of tennis balls. Each can has 3 tennis balls. How many tennis balls does he have now

Learn Before

Related