extract-hyperparameters

Extract hyperparameters by parsing PDFs or text files to locate learning rates, batch sizes, epochs, momentum, weight decay, dropout, layer sizes, activation functions, normalization methods, optimizers, and preprocessing steps. The Skill identifies and documents these values in structured format ready for YAML or JSON configuration files.

Yes. Extract all hyperparameters mentioned in research papers to set up training configurations that faithfully replicate the original work. This enables accurate reproduction of paper results by capturing architecture details, optimizer settings, and training parameters.

The Skill extracts learning rate, batch size, epochs, momentum, weight decay, dropout rates, layer sizes, activation functions, normalization techniques, optimizer type, and data preprocessing steps. It parses both structured and prose descriptions in PDFs and text files.

Extract hyperparameters from PDF research papers and plain text files. The Skill uses pdftotext and grep to parse documents and locate parameter values and configuration details throughout paper content.

Extract hyperparameters from multiple papers into structured configuration format, then compare learning rates, batch sizes, model architectures, and optimizer settings side by side. This enables analysis of hyperparameter choices across works and informs tuning experiments.