This repository demonstrates using OpenAI's O1 LLM capabilities (via ChatGPT) to solve Advent of Code challenges. For each solution, you can find:
- The original prompt used
- The public ChatGPT conversation
- Number of prompt iterations required
- Time taken to reach a solution
| Day | Solved (Silver/Gold) | Iterations | Time Taken | One-Shot Success | Chat Link |
|---|---|---|---|---|---|
| 1 | ✅/✅ | 1 | 20s | ✅ | Chat |
| 2 | ✅/✅ | 1 | 30s | ✅ | Chat |
| 3 | ✅/✅ | 1 | 60s | ✅ | Chat |
| 4 | ✅/✅ | 1 | 33s | ✅ | Chat |
| 5 | ✅/✅ | 1 | 62s | ✅ | Chat |
| 6 | ✅/✅ | 1 | 2s | ✅ | Chat |
Each puzzle solution includes:
- Initial prompt to ChatGPT
- Run code locally to validate the solution
- Iterate
The goal is to showcase both the potential and limitations of using LLMs for algorithmic problem solving.
python3 1/gold.py < 1/input.txt
python3 1/silver.py < 1/input.txt
Each puzzle includes:
- Original puzzle text
- Example inputs
- Correct answers for both parts
This allows for:
- Testing other LLMs against same problems
- Validating solutions
- Benchmarking different approaches