Start an inference job

After an inference model is created, you can start an inference job from IBM Spectrum Conductor Deep Learning Impact and see prediction results of the model.

Before you begin

Create a trained model before starting an inference job, see Create an inference model.

Procedure

  1. From the cluster management console, select Workload > Spark > Deep Learning.
  2. Click on the Models tab.
  3. Select the inference model that you want to run an inference job on and click Inference.
  4. To start an inference job, click New Inference.
  5. Specify the inference name.
  6. Specify the threshold.
  7. Select files for the inference.
  8. Click Start Inference.

Results

The inference job is completed and inference results are available. To see results, click the job name. Depending on the threshold specified, one or more classes and corresponding probability is provided. The result file location is also provided.