Fine-tuning large language models (LLMs) on limited tabular data for classification tasks can
lead to\textit {fine-tuning multiplicity}, where equally well-performing models make conflicting …