remote-run-ssh - Execute CVlization on Remote GPU via SSH
Sync CVlization examples and library to remote GPU host via SSH and launch Docker-based training scripts.
Tags
Updated: 2026-03-20Capabilities
Typical Inputs
Typical Outputs
What this skill does
- sync files to remote host
- build Docker image
- check GPU availability
- execute training script
- capture metrics
- retrieve log files
- document run session
- clean remote workspace
Inputs
- example directory path
- cvlization library folder
- SSH config for l1 host
- Hugging Face tokens
- Docker scripts
- environment variables
Outputs
- remote workspace at /tmp/cvlization_remote
- training and evaluation logs
- performance metrics
- run log documentation
- checkpoints and model artifacts
Requirements
- SSH access to remote GPU host
- NVIDIA GPU with CUDA support
- Docker with GPU runtime
- 15GB free space on /tmp