run-experiment

Automating ML experiment deployment involves environment detection, pre-flight GPU availability checks, code synchronization via rsync or git, and remote launch using screen sessions. This Skill orchestrates the entire workflow to eliminate manual setup and repetitive boilerplate.

Code synchronization for remote GPU execution uses rsync or git strategies to transfer your machine learning scripts. This ensures the target environment has the exact codebase version needed before launching the training job in a screen session.

Yes, remote deployment workflows support optional experiment tracking through wandb integration. You can configure logging integrations to automatically monitor metrics when launching training jobs across local or remote environments.

Pre-flight validation for GPU availability is an environment discovery step that verifies hardware resources before launching experiments. It checks remote or local environments to ensure GPUs are present and ready, preventing failed training job starts.

This Skill supports both local execution and remote orchestration for machine learning experiments. It handles environment detection and deployment workflows for either target, launching training jobs in screen sessions on the detected setup.

Using screen sessions for remote ML training provides persistent execution but lacks advanced orchestration features like automated failure recovery or multi-node distributed scheduling. It is best suited for single-node GPU workflows requiring basic background execution.