experiment

To run and monitor ML training experiments automatically, you need a system that orchestrates execution, manages GPU allocation, and parses results. This Skill automates that entire lifecycle, including environment setup and error handling, for reliable single-experiment runs.

GPU allocation for individual training runs works by designating specific GPUs for experiment execution. This Skill manages that designation, ensuring isolated environments and proper resource enforcement for each training script.

Handling training failures and timeouts during hyperparameter tuning requires early abort detection and auto-repair loops. This Skill enforces timeouts and automatically repairs common training failures to ensure deterministic and robust experimentation.

Yes, you need PyTorch and PyYAML to automate training script execution with this Skill. These dependencies are required to manage the environment setup, configurations, and execution of your machine learning experiments.

Isolating ML experiments using git worktrees involves creating separate working directories for each run. This Skill manages code environments automatically through git worktrees to ensure isolated and reproducible experimentation.

Training experiments often fail before execution starts due to environment or configuration errors. This Skill performs pre-flight checks and uses an auto-repair loop to detect and resolve common setup failures before the main training process begins.