Loading...
Loading...
Rigor Improve / Rigor Explore 运行叶技能,用于在深度学习研究仓库中生成有限范围的探索性证据。当研究人员明确授权探索性运行时使用,例如小数据集验证、短周期试错、批量参数扫描、闲置GPU搜索或快速迁移学习试验,需在`explore_outputs/`中包含公平比较说明和不夸大的总结。请勿将其用于基于`current_research`的端到端探索编排、可信基线执行、保守训练验证、默认路由、已验证的SOTA声明或隐式实验。
npx skill4agent add lllllllama/rigorpilot-skills explore-runexplore-run../../references/agent-operating-principles.mdcurrent_researchai-research-exploreminimal-run-and-auditrun-traincostsuccess_rateexpected_gainselection_weightsmax_variantsmax_short_cycle_runsvariant_axessubset_sizesshort_run_stepsselection_weightscostsuccess_rateexpected_gainprimary_metricmetric_goalexplore_outputs/CHANGESET.mdexplore_outputs/SCIENTIFIC_CHANGELOG.mdexplore_outputs/COMPARABILITY_REPORT.mdexplore_outputs/TOP_RUNS.mdexplore_outputs/status.jsonreferences/execution-policy.md../../references/explore-variant-spec.md../../references/deep-learning-experiment-principles.mdscripts/plan_variants.pyscripts/write_outputs.py