以生产级模式开展系统的机器学习实验。在以下场景中使用：(1) 设置交叉验证的实验网格；(2) 管理GPU内存、多GPU工作池，或配置OOM防护机制；(3) 设计患者级别或站点感知的数据拆分策略；(4) 通过可恢复性跟踪实验进度；(5) 在多GPU之间合理分配计算任务。
--- frontmattername: experiment description: > Run systematic ML experiments with production-grade patterns. Use when: (1) Setting up experiment grids with cross-validation, (2) Managing GPU memory, multi-GPU worker pools, or OOM protection, (3) Designing patient-level or site-aware data splits, (4) Tracking experiment completion with resumability, (5) Distributing work across GPUs.
| File | Apply When |
|---|---|
| references/experiment-patterns.md | Experiment grids, completion tracking, CV splits, wandb, multi-GPU distribution |
| references/gpu-patterns.md | GPU memory cleanup, multi-GPU pools, VRAM scaling, OOM protection, reproducibility |
| references/data-splitting.md | Patient-level splits, site-aware splits, temporal splits, class imbalance strategies |