cpldcpu/MisguidedAttention
A collection of prompts to challenge the reasoning abilities of large language models in presence of misguiding information
This project offers a collection of specially crafted questions designed to test how well large language models (LLMs) reason when faced with misleading information. It helps evaluate if an LLM can logically solve a problem as stated, or if it defaults to familiar, but incorrect, answers learned during training. The input is a 'trick question' prompt, and the output is the LLM's response, revealing its reasoning strengths and weaknesses. Anyone responsible for evaluating or implementing LLMs in critical applications would find this useful.
466 stars. No commits in the last 6 months.
Use this if you need to rigorously test and benchmark the reasoning and problem-solving capabilities of different large language models, especially when 'common sense' or 'pattern recognition' might lead them astray.
Not ideal if you are looking for a tool to generate prompts for general creative writing tasks or for basic information retrieval from LLMs.
Stars
466
Forks
27
Language
Python
License
CC0-1.0
Category
Last pushed
Jul 31, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/prompt-engineering/cpldcpu/MisguidedAttention"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
BoundaryML/baml
The AI framework that adds the engineering to prompt engineering (Python/TS/Ruby/Java/C#/Rust/Go...
deanpeters/product-manager-prompts
A repository of Generative AI prompts for product managers using agents such as ChatGPT, Claude, & Gemini
eudk/awesome-ai-tools
🔴 VERY LARGE AI TOOL LIST! 🔴 Curated list of AI Tools - Updated 2026
jujumilk3/leaked-system-prompts
Collection of leaked system prompts
legeling/PromptHub
一款开源、纯本地 Prompt ，Skill 管理工具，帮助你高效管理、版本控制和复用 Prompt，并一键分发skill | An open-source, local-first AI...