sudo ./installml_linux_amd64.bin --silent --response-file install_response.json For CI/CD pipelines (GitHub Actions, GitLab CI), use the official Docker image:
"install_path": "/opt/installml", "shell_integration": "bash", "auto_accept_license": true, "default_channel": "stable"
| Error Message | Likely Cause | Fix | | :--- | :--- | :--- | | Permission denied: /usr/local/bin/iml | User lacks sudo rights during install | Re-run the core installer with sudo , or install locally --prefix ~/.local | | CUDA not found but requested | NVIDIA drivers missing or paths wrong | Run nvidia-smi . If not found, install drivers. Then run iml config set cuda.root /usr/local/cuda | | SSL: CERTIFICATE_VERIFY_FAILED | Corporate MITM proxy or outdated certs | Update certificates: sudo apt install ca-certificates . Or disable strict SSL for internal repos only (not recommended for public). | | Virtual environment not activating | Shell init script missing | Run eval "$(iml hook bash)" manually for the current session, then redo step 3. | | Disk space error during cache | Default cache dir on small root partition | Change cache_dir in config.toml to a larger mounted drive. | For teams managing dozens of machines, manual setup is not viable. Use the "silent install" method. installml.com setup
[logging] level = "INFO" # Change to "DEBUG" if troubleshooting log_file = "~/.installml/logs/setup.log"
Run:
Create a response file install_response.json :
export HTTP_PROXY=http://proxy.company.com:8080 export HTTPS_PROXY=https://proxy.company.com:8080 iml config set proxy $HTTP_PROXY The heart of your installml.com setup lies in the configuration file located at ~/.installml/config.toml . Here is a recommended baseline configuration for optimal performance. Or disable strict SSL for internal repos only
curl -fsSL https://installml.com/setup.sh | sh Alternatively, if you downloaded the binary: