Analyze completed ML experiment runs to produce structured retrospectives and hypotheses.
npx skills add https://github.com/Borda/AI-Rig --skill retro-borda
Please help me install this Agent Skill. Skill: retro Source: https://github.com/Borda/AI-Rig/tree/main/plugins/research/skills/retro Command: npx skills add https://github.com/Borda/AI-Rig --skill retro-borda
Post-run retrospective analysis of ML experiments helps teams understand what worked, what failed, and why, by consolidating experiment history, statistical tests, and narrative context into actionable insights.
After completing an experiment run, invoke the retro skill to generate a comprehensive retrospective and hypotheses queue.
High-intent search queries and answers about installing and using this skill.
To automate post-run retrospectives for ML experiments, use the retro skill to analyze completed experiment runs, generating structured insights on strategy effectiveness and failure modes from state data and diaries.
Identify recurring failure modes in discarded experiment iterations by analyzing per-iteration metadata and reverted iterations. The retrospective process detects patterns of discarded iterations to highlight strategy weaknesses.
Yes, you can generate a next-hypothesis queue from completed ML research runs. The retrospective analysis produces structured hypotheses outputs compatible with research:run workflows to guide subsequent experiments.
Evaluating experiment strategy effectiveness uses statistical significance tests like the Wilcoxon test. This retrospective analysis applies these tests to delta distributions and kept iterations to compare strategies.
The retrospective analysis requires .experiments state data, diaries, and per-iteration metadata to function. It uses these inputs to generate cross-run comparisons and health monitoring for your research workflow.
yalehPlan rapid-convergence experiments with retrospective validation and generic agents.
dakesanGenerate testable scientific hypotheses with mechanistic explanations and experimental predictions.
spandiosAnalyze A/B test results for statistical significance and actionable recommendations.
cowwocAutomate retrospective analysis to identify mistake patterns and derive action items.
