Watch a live coding session where an AI enthusiast develops a 20 Questions game using Large Language Models (LLMs). Explore the process of creating LLM agents for question generation, answering, and guessing, utilizing tools like ChatGPT and Claude. Learn about integrating Rigging, working with Kaggle competition data, and implementing various components of the game pipeline. Follow along as the developer tests and refines the system, ultimately comparing the performance of different LLMs. Gain insights into practical AI application development and the nuances of working with language models in an interactive gaming context.
Live Coding an LLM Battle - GPT-4 vs. Claude - 20 Questions Game
Overview
Syllabus
Intro
ChatGPT 20 Questions
Kaggle Competition Overview
Kaggle Keywords Data
20 Questions LLM Agent with Rigging
Trying to get a Input Text Box
Question Agent Function
Testing the Question Agent
Creating the Answer Agent
Guesser Agent
Testing 20 Questions LLM Pipeline
Swapping LLM to Claude
Pushing the Code
Taught by
Rob Mulla
