Using elastic distributed inference via command line interface
Use elastic distributed inference from the command line interface using the dlim command.
The dlim command lists the inference services and provides details, creates and removes inference services, starts and stops inference services and lists available runtime environments.
To use the dlim command, you must do the following:
- Add dlim to PATH, for
where localhost is the IP or name of the master host and edi-port is the port number set during installation (DLI_EDI_LBD_ENABLE).
dlim config -c https://localhost:edi-port/dlim/v1/For example:
dlim config -c https://localhost:9000/dlim/v1/
- Save your login token. Your login token does expire after some time, and you may need to rerun
where username is your user name.
dlim config -t --user username
- Use the command. To see what sub commands are available with the dlim
command, run the
Usage: -h, --help, help information -v, --version, version information Sub commands: config, config information to access inference service model, manage models for inference service runtime, manage runtimes for inference service
This example uses the CLI from the master host. To use the CLI from another host, replace
localhostwith the remote hostname and update the paths.
PATH=$PATH:/opt/ibm/spectrumcomputing/dlim/2.3.0/linux-ppc64le/bin dlim config -c https://localhost:9000/dlim/v1/ dlim config -t -u Admin -x Admin dlim model deploy -p /opt/ibm/spectrumcomputing/dlim/2.3.0/linux-ppc64le/samples/python/py3add dlim model start add