Multi-character LLM Roleplay for Small Self-Problem Solving

Awesome hacking event from Nathaniel Evry at Quub! We discussed using LLM’s in a roleplay situation to assist in solving small coding challenges. Here was a sample prompt:

Alright team, it's great to meet you both! My name is Nathaniel and our goal today is to work together to build a basic neural network in Python that can classify hand written digits! 

Bob (Model Architect), your role is to propose an initial network architecture and layers by sending messages to Jane. Remember, you cannot see what Jane is thinking - you must communicate your ideas through chat!

Jane (Coding Lead), your role is to provide feedback and suggestions to Bob's messages, also through chat only. Ask Bob questions and clarify details in order to work together most efficiently.

Bob: Here is my suggested architecture to get started...
Jane: That sounds like a solid start Bob! A few questions though...
Here is the workable code that got provided by the LLM: https://colab.research.google.com/drive/1lHYGwwi5hFyRTuaO6QJRbmEx_WSyFfj1?usp=sharing

Here is a link to the chat: https://chat.openai.com/share/176d86d3-ce12-4893-86a2-a891ec3727d2

Thank again to Nathaniel and to Quub for letting us borrow him for lunch! Looking forward to seeing everyone next week!


Leave a Reply

Your email address will not be published. Required fields are marked *