Mechanism

When It Triggers · How It Works · What It Produces

Breaks down how the skill works and what it produces, so you can quickly judge whether it fits your scenario and value.

Trigger Timing

When a researcher explicitly authorizes bounded exploratory runs (small-subset validation, short-cycle probes, batch sweeps, idle-GPU search, transfer trials) and wants candidate ranking instead of trusted success certification.

Working Mechanism

  • Plan variant grid via variant_axes, subset_sizes, short_run_steps
  • Score candidates by cost/success_rate/expected_gain with conservative weights
  • Prune by max_variants/max_short_cycle_runs
  • Hand off execution to minimal-run-and-audit or run-train
  • Write bounded-evidence summaries to explore_outputs/

Outcome

Standardized explore_outputs/ files (CHANGESET, COMPARABILITY_REPORT, TOP_RUNS, status.json) with no-overclaim, fair-comparison caveats for exploratory candidates.

Use Cases

Two Typical Use Cases

Scenario 01

Small-subset backbone variant screening

A researcher wants to test 5 backbone variants on 10% of data before full training. They use explore-run to define variant_axes and subset_sizes, score by cost/success_rate/expected_gain, prune via max_variants, then hand off to minimal-run-and-audit. Result: TOP_RUNS.md ranks candidates with comparability caveats, avoiding premature trusted claims.

Scenario 02

Idle-GPU short-cycle hyperparameter sweep

During idle GPU windows, a researcher authorizes short-cycle guess-and-check sweeps for LR and batch size. explore-run plans short_run_steps, budgets max_short_cycle_runs, labels outputs as bounded evidence in explore_outputs/, and notes unfair comparisons. Value: cheap exploration without polluting trusted baseline records.

Skill Relationships

Which Skills I Use · Which Skills Use Me

Dependency relationships read as "the upper tier points to the lower tier." The current Skill sits in the middle tier — above are Skills that depend on it, below are Skills it depends on.

Tier 1 · These Skills Use Me

Tier 2 · Current Skill

  • explore-runWorkflow

Tier 3 · I Use These Skills

Full Relationship Graph

Browse this skill's relationships within its skillset. Click a node to switch the side panel; use the search box to jump to any skill.