submit-run

Description

Get data from the input channel specified by --input-operator and--input_target only once. Apply the model specified by --deployment_id andversion_id to the obtained data, and output the resulting data to the output destination channel specified by --output_service_name and--output_service_id To do.

Synopsis

$ abeja model submit-run --help
Usage: abeja model submit-run [OPTIONS]

  Submit run

Options:
-d, --deployment_id, --deployment-id TEXT
                                Deployment identifier  [required]
-v, --version_id, --version-id TEXT
                                Version identifier  [required]
--input_operator, --input-operator TEXT
                                Input oparator.  [required]
--input_target, --input-target TEXT
                                Input target.  [required]
-m, --model_id, --model-id TEXT
                                Model identifier
--output_operator, --output-operator TEXT
                                Output operator
--output_target, --output-target TEXT
                                Output target
--retry_count, --retry-count INTEGER
                                Retry count. By default, retries 5 times.
-e, --environment ENVIRONMENTSTRING
                                Trigger environment variable
--help                          Show this message and exit.

Options

-d, --deployment_id, --deployment-id

Specify the deployment_id issued by the create-deployment command. Process using the model specified by this identifier (--version_id).

-v, --version_id, --version-id

Specify the version_id issued by the create-version command. Process using the model specified by this identifier (--deployment_id).

--input_operator, --input-operator

Specify the input channel type and protocol version. The format is $<channel type>:<protocol version>. The currently supported channel types and protocol versions are as follows.

--input_target, --input-target

Specify the input channel CHANNEL_ID and filter. The format is <CHANNEL_ID>/<filter>. The currently supported filters are as follows.

-m, --model_id, --model-id

Specify the Model ID of the model to be used.

--output_operator, --output-operator

Specify the destination channel type and protocol version. The format is $<channel type>:<protocol version>. The currently supported channel types and protocol versions are as follows.

--output_target, --output-target

Specify the output channel CHANNEL_ID.

--retry_count, --retry-count

Specify the retry count of Run execution. You can specify the number of times from “0-9”.

-e, --environment

You can specify environment variables. The registered environment variables can be referenced at runtime. Example) IMAGE_WIDTH: 100
For more information on user-specifiable environment variables, see here.

Example

Run immediately

Premise:

• [Model], [Deployment], and [Model version] are registered.
• Input channels / output destination channels are registered.

Command:

$ abeja model submit-run --deployment_id 1234567890123 \
                       --version_id ver-1234567890abcdef \
                       --input_operator $datalake:1 \
                       --input_target 2345678901234 \
                       --output_operator $datalake:1 \
                       --output_target 4567890123456 \
                       --retry_count 2

Output:

{
    "created_at": "2020-02-19T00:15:36.458661Z",
    "deployment_id": "1234567890123",
    "distribution": null,
    "input_data": {
        ":1": "2345678901234"
    },
    "model_version_id": "ver-1234567890abcdef",
    "models": {},
    "modified_at": "2020-02-19T00:15:36.458663Z",
    "output_template": {
        ":1": "4567890123456"
    },
    "retry_count": 2,
    "run_id": "run-0182e723d19b47b3",
    "status": "SUBMITTED",
    "trigger_id": null,
    "user_env_vars": {}
}