GenMLBench

This dataset contains information about 50 Kaggle competitions, including structured metadata and rich natural language descriptions of each competition's task. Each entry includes: comp_name: The unique name of the Kaggle competition. competition_link: A direct URL to the competition's page. data_card: Structured info about the dataset used in the competition (e.g., file formats, features). metric: The abbreviation for the evaluation metric used (e.g., RMSE, AUC). comp_type: The type of competition — Featured, Research, Community, or Playground. data_type: The main data modality — tabular, image, text, or time series. description: A detailed, general-purpose explanation of the competition task. metric_type: The category of the evaluation approach — regression, classification, or custom. domain: The application area (like finance, healthcare, or computer vision), inferred using language model analysis.