After fine-tuning GPT-4.1 with function calling, the model outputs fields never seen during training

[Abbas_I.Obaid]

@merefield

in my experience gpt-4.1-mini and nano are really poor with function calling (they struggle to intelligently distinguish similar functions - a result of quantisation fallout?)

stick with gpt-4o series or the full fat gpt-4.1.

bad news if can not use gpt 4.1 mini