Run the same experiment using four LLMs
- Why four LLMs?
- Enough for a supermajority vote
- Also enough to cause a tie
- Smallest possible number
- Save all the LLM responses into individual files
- one per LLM per experiment
- keep this as a separate task
- allows you do any kind of analysis you want in the future