analyze-results

To generate comparison tables from ML experiment results, the Skill parses JSON and CSV output files to organize metrics by model type, hyperparameters, and seeds. It then computes statistics like mean and standard deviation to reveal performance trends.

Calculating delta vs baseline for model variants involves extracting result metrics from experiment directories and comparing them against a reference run. This process produces structured reports highlighting performance differences and flagging outliers.

Yes, you can summarize hyperparameter tuning runs across different seeds. The Skill computes mean and standard deviation across multiple seeds, identifies outliers, and generates reproducible summaries to help you interpret the stability of your models.

No specific framework is required to parse JSON and CSV files for machine learning statistics. The Skill locates relevant result files in common directories and processes them independently to produce actionable insights and trend analysis.

To automatically flag outliers in reproducible ML experiment summaries, the tool applies statistical analysis to the parsed results. It evaluates deviations in performance metrics across runs and marks anomalous data points in the final structured report.