菜单
Run baseline comparisons (OLS, LASSO, SINDy, PDEFind) against KANDy results for a dynamical system. Use when the user wants to compare KANDy's equation discovery against standard methods.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | run-baselines |
| description | Run baseline comparisons (OLS, LASSO, SINDy, PDEFind) against KANDy results for a dynamical system. Use when the user wants to compare KANDy's equation discovery against standard methods. |
| argument-hint | ["baseline-name"] |
Run baseline comparisons for $ARGUMENTS.
!uv run kandy-baselines --list 2>/dev/null
Use the kandy-researcher agent to:
uv run kandy-baselines $ARGUMENTS (or the appropriate baseline script)results/results/<SystemName>/baselines/notes/ with full metadataA comparison table like:
| Method | Discovered Equation | # Terms | RMSE | NRMSE |
|---|---|---|---|---|
| KANDy | ... | ... | ... | ... |
| OLS | ... | ... | ... | ... |
| LASSO | ... | ... | ... | ... |
| SINDy | ... | ... | ... | ... |
If no baseline name is provided, list available baselines and ask the user which one to run.
Assess the quality of KANDy model results by examining generated plots. Use after training a model, running rollouts, or generating comparison plots.
List all available KANDy systems, baselines, and show current experiment results status.
Generate a comprehensive research report for a KANDy experiment, including discovered equations, metrics, plots, and analysis. Use after an experiment is complete.
Run a KANDy experiment on a dynamical system to discover its governing equations. Use when the user wants to train a KAN model, fit equations, or run the full KANDy pipeline on a system.