run-experiment

Automate ML experiment deployment on remote GPU servers by providing the experiment script and server details. The Skill parses AGENTS.md for server specifics, manages conda environments, and binds GPUs to coordinate execution via SSH and screen sessions.

Running multiple ML experiments across GPUs and hosts is coordinated by automating code deployment, dependency management, and resource allocation. The Skill ensures consistent environments and handles GPU binding for single or multiple experiments across local and remote servers.

Yes, you can use conda environments and bind specific GPUs for remote ML training. The Skill manages conda environment setup and automates GPU binding, ensuring resource allocation is correctly configured before launching training jobs on local or remote servers.

ML experiment orchestration works with W&B and Feishu notifications through optional integration. The Skill captures logs via tee and can be configured to send updates through W&B or Feishu, keeping you informed of training progress and job status across remote hosts.

Yes, you need to configure AGENTS.md before deploying experiments to remote GPU servers. The Skill requires parsing AGENTS.md to retrieve essential server specifics, which are necessary to establish SSH connections, manage conda environments, and bind GPUs correctly.