Captions for Wei et al. 2022 article on Chain-of-Thought Prompting
Loading data from Talking to Chatbots Dataset reddgr/talking-to-chatbots-chats …
😎 This is from article by Wei et al., 2022, Chain-of-Thought Prompting Elicits Reasoning in Large Language Models

🤖 Diagram comparing *Standard Prompting* vs. *Chain-of-Thought Prompting* for solving math word problems.😎 SHORT CAPTION
The left side titled “Standard Prompting” shows two questions under “Model Input” with a corresponding answer under “Model Output”:
– First 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?” Answer provided: “The answer is 11.”
– Second question: “The cafeteria had 23 apples. If they used 20 to make lunch and bought 6 more, how many apples do they have?” Model Output: “The answer is 27” (marked with a red X to indicate it is incorrect).
The right side titled “Chain-of-Thought Prompting” shows the same two questions but with detailed reasoning highlighted in blue and green:
– First question: The detailed breakdown shows 5 tennis balls + 6 tennis balls (2 cans of 3 each), leading to the correct answer, 11.
– Second question: The explanation shows 23 – 20 = 3 apples, then 3 + 6 = 9 apples, with a correct final answer of 9 (marked with a green checkmark).
Caption below: “Figure 1: Chain-of-thought prompting enables large language models to tackle complex arithmetic, commonsense, and symbolic reasoning tasks.”
[Alt text by ALT Text Artist GPT]
🤖 Comparison of Standard vs. Chain-of-Thought Prompting for solving math problems.😎 LONGER, AND MENTIONING THE CONTEXT I GAVE YOU!
[Alt text by ALT Text Artist GPT]
[Message flagged as bad_response]
🤖 Comparison of Standard Prompting and Chain-of-Thought Prompting from Wei et al., 2022, illustrating how large language models tackle math problems. Chain-of-thought prompting shows step-by-step reasoning, leading to more accurate answers for arithmetic and reasoning tasks.
[Alt text by ALT Text Artist GPT]
Hugging Face Dataset Metrics
All the conversation prompts, responses, and metrics are available to download and explore on Hugging Face dataset reddgr/talking-to-chatbots-chats: