Dataset supporting the publication: Generalization performance distributions along learning curves

This dataset has been curated to support high fidelity analysis of learning curves across a diverse set of learners, covering both settings with and without hyper-parameter tuning. It includes results generated under different sampling strategies and learning curve construction methods, enabling systematic investigation of how model performance evolves with increasing data. The primary goal is to study the distribution of generalization performance along these learning curves, providing insights into variability, extrapolation behavior, and model reliability. Detailed information on the data generation process and experimental setup is provided in the accompanying paper.