How Do I Obtain Training Job Parameters from the Boot File of the Training Job?

Training job parameters can be automatically generated in the background or you can enter them manually. To obtain training job parameters:

1. When a training job is created, train_url in the running parameters of the training job indicates where the training results are output to, and data_url indicates a data source. The test parameter is entered manually.
   Figure 1 Creating a training job of the new version
   
   Figure 2 Creating a training job of the old version
   
2. After the training job is executed, you can click the job name in the training job list to view its details. You can obtain the parameter input mode from logs, as shown in Figure 3.
   Figure 3 Viewing logs
   
3. To obtain the values of train_url, data_url, and test during training, add the following code to the boot file of the training job:

   import argparse
   parser = argparse.ArgumentParser()
   parser.add_argument('--data_url', type=str, default=None, help='test')
   parser.add_argument('--train_url', type=str, default=None, help='test')
   parser.add_argument('--test', type=str, default=None, help='test')