| name | mylab-parameter-tuning |
| description | Use when the main loop is parameter generation -> batch experiments -> result collection -> comparison for a parameter experiment round. |
| trial_kind | parameter-tuning |
Mylab Parameter Tuning
Use this skill when the main goal is to search parameter combinations, compare results across a batch, and refine the next search space.
Flow
- Generate parameter combinations
- Run the batch
- Collect results
- Compare and rank
Frontmatter Essence
- target parameter family or search space under study
- batch shape and comparison logic for this round
- ranking signal or decision metric that decides what wins
- reason this batch matters for narrowing the next search region
- exact code checkpoint this round should start from
Trial Body Rules
- The execution trial should read like a sweep pipeline, not like an implementation story.
- The trial must state how combinations are generated.
- The trial must state how results are collected into a comparable form.
- The comparison step must say how the current batch will be ranked.
- Deliverables should emphasize batch configs, aggregated tables, and ranking conclusions for this round only.
Reference Files
- references/trial-skill.md: the workflow contract that explains how this parameter-tuning trial should be interpreted
- references/design.md: detailed search-space rationale and batch-generation logic
- references/experiment.md: detailed execution environment and artifact inventory
- references/analysis.md: detailed comparison logic, anomaly notes, and ranking interpretation
- references/conclusion.md: expanded verdict, limitations, and follow-up memory
- references/shared-asset.md: durable repository knowledge already learned in this run
- references/persistent-feedback.md: long-lived user guidance that should constrain future search choices
- references/recent-feedback.md: short-lived user feedback for this round
- references/parent-trial.md: previous trial body when iterating from an earlier trial
Templates
