deepaas-cli

Synopsis

deepaas-cli [options]

Description

deepaas-cli Command line interface (CLI) to DEEPaaS models: that are loaded through the deepaas.v2.models entrypoint API. One gets access to the same get_metadata, warm, predict, and train methods with all corresponding options as with DEEPaaS REST API. To get available for the method options, one has to call deepaas-cli <method> --help. Additional parameters are provided: --deepaas_method_output is to store the output in a pre-defined by a user file; --deepaas_with_multiprocessing is to activate multiprocessing support, default is True. oslo_log package is used for logging information, which provides additional options for the script. If several models are available for loading, one has to provide which one to load via DEEPAAS_V2_MODEL environment setting.

Options

get_metadata: Calls get_metadata() method. The output can be stored via –deepaas_method_output.

warm: Calls warm() method, e.g. to prepare the model for execution.

predict: Calls predict() method. The output can be stored via –deepaas_method_output.

train: Calls train() method. The output can be stored via –deepaas_method_output.

--deepaas_method_output: To save the results to a local file, if needed. Available for get_metadata, predict, train methods.

--deepaas_with_multiprocessing: To activate multiprocessing support, default is True.

--model-name MODEL_NAME

Specify the model to be used. If not specified, DEEPaaS will serve all the models that are available. If specified, DEEPaaS will serve only the specified model. You can also use the DEEPAAS_V2_MODEL environment variable.

WARNING: Serving multiple models is deprecated and will be removed in the future, therefore it is strongly suggested that you specify the model you want to or that you ensure that only one model is available.

Files

None

Reporting Bugs

Bugs are managed at GitHub