The following content is displayed on the same screen for your experiment so that you can make any necessary reference in experiment. Start your experiment now!

1. Open Firefox web browser,Go to WMLA Web GUI login page.(Duration: 2 mins)
   
   Input your user name and password. Click on “Log in” button.
2. Deep Learning Tab(Duration: 2 mins)
   

   

   

3. Create Dataset,Create a new dataset from Images for Object Classification(Duration: 2 mins)
   

   (DataSet -> New -> Images for Object Classification)
   

4. Provide Dataset Information(Duration: 3 mins)
   

   Put “/opt/ibm/spectrumcomputing/dataset/flower” under “Training folder”.

   Note: Please customize the "Dataset name", otherwise error 400 may be reported due to duplicate name

   

   

5. Click “Create” button. Once completed, your interface will look like this(Duration: 4 mins)
   

   

   

6. Create a Model,Go to “Models” tab. Click “New” to create a new model(Duration: 1 min)
   

   New a Model form Model tag page

   

7. To create a “New Model”. Select “TensorFlow-inceptionv3”. Click on “Next” button(Duration: 2 mins)
   

   

8. Configure and input model hyperparameters as shown in the template. Select “Single node training” for Training engine and “flowers” dataset created in Step 3 for Training dataset(Duration: 4 mins)
   

   

   

9. Click “Add” button. If completed successfully, you will see “Model has been added” page(Duration: 4 mins)
   

   

10. Train a Model,Select the model you created in Step 4. Click on the “Train” button(Duration: 1 min)
    

    

11. Set related parameters. Click on “Start Training” to start model training(Duration: 2 mins)
    

    

12. Monitor the Training(Duration: 10 mins)
    

    After the training starts, you can click on the model name or the “Insights” button next to it to open the monitoring page. From there, the model training process will be visualized.

    

13. Validate a Model(Duration: 5 mins)
    

    After the model training completes, you can click on the “Validate Trained Model” for model training result validation.
    
    Select “Top-K and Confusion Matrix” for the “Validation metrics”.
    
    Click on the “Validation Results” tab. You will see similar results as below when validation job completes.
    

14. Create an Inference Model(Duration: 5 mins)
    

    Go back to “Training” tab. Click on the “Create Inference Model” button to create an inference model.
    
    

    

15. Test a Model(Duration: 5 mins)
    

    Select the inference model you’ve created. Click on the “Test” button. You can select files from the flowers folder on the desktop for testing.

    
    
    You can review the results after the test completes.
    
    

Start Your Experiment