Medical Model Fine-Tuning Conceals Major Question-Level Accuracy Turnover
A benchmarking study evaluating a 27-billion-parameter language model and five fine-tuned derivatives demonstrated that aggregate accuracy metrics conceal substantial item-level ...
