honeypolygon/lab1 solution

[honeypolygon]

Hello Donner!

AI Agent systems are fascinating — I actually rushed through the Python basics just so I could understand this part of the course! 😄

---

Concept

This notebook explores a structured AI debate framework.

When a decision or opinion needs to be made, two LLMs take opposing sides:

• Pro side argues in favor of the topic.
• Con side argues against it.

Each argument is then independently verified and revised by another LLM acting as a verifier, improving logic and factual consistency.

Originally, I also planned a more advanced structure — where the verifier would send the improved argument back to the original debater LLM, and only if it passes verification again would it be forwarded to the Judge LLM.
That feedback loop would simulate an iterative refinement process, but it turned out to be quite complex to coordinate, so this version represents the simplified foundation of that idea.

Finally, a Judge LLM evaluates both refined arguments and determines a clear winner, with no ambiguity allowed — one position must be adopted as more valid overall.

The goal is to simulate an AI-based deliberation process that mirrors balanced reasoning, self-correction, and decisive judgment.

---

Background

I'm a beginner in Python and still learning through the course materials.
This notebook was created as part of my learning process — it may share some structure or ideas with the lecture content, but I developed it independently to understand multi-agent reasoning and coordination.
Please kindly excuse any overlap with course examples.

---

Notes

• All outputs have been cleared.
• The file is safely located under community_contributions/honeypolygon/.

[ed-donner]

This is great! Thanks so much for sharing. One issue - this PR seems to include files outside the community contributions folders. Let me know if this was intentional - otherwise please move all changes within community contributions when you get a chance. Thanks so much
Ed