The datasets that appear in publications are curated and have been split into training. testing and validation sub-datasets by domain experts. Consequently. machine learning models typically perform well on such split-by-hand prepared datasets. https://hollandscountryclothinges.shop/product-category/bullet-pouches/