Interesting; maybe it’s an artifact of how we formatted our questions? Or, potentially, the training samples with larger ranges of numbers were higher quality? You could try it like how I did in this failing example:
When I tried this same list with your prompt, both responses were incorrect:
Interesting; maybe it’s an artifact of how we formatted our questions? Or, potentially, the training samples with larger ranges of numbers were higher quality? You could try it like how I did in this failing example:
When I tried this same list with your prompt, both responses were incorrect: