Creating an Algorithm

Prerequisites

• Data is available either by creating a dataset in ModelArts or by uploading the dataset used for training to the OBS directory.
• Your training script has been uploaded to the OBS directory. For details about how to develop a training script, see Developing a Custom Script.
• At least one empty folder has been created in OBS for storing the training output.
• The account is not in arrears because resources are consumed when training jobs are running.
• The OBS directory you use and ModelArts are in the same region.

Accessing the Algorithm Creation Page

1. Log in to the ModelArts management console and click Algorithm Management in the left navigation pane.
2. On the My Algorithms page, click Create. The Create Algorithm page is displayed.

Setting Basic Parameters

Enter the basic algorithm information, including Name and Description.

Setting the Boot Mode

Select a preset image to create an algorithm.

Set Image, Code Directory, and Boot File based on the algorithm code. Ensure that the framework of the AI image you select is the same as the one you use for editing algorithm code. For example, if TensorFlow is used for editing algorithm code, select a TensorFlow image when you create an algorithm.

Table 1 Parameters

Parameter	Description
Boot Mode > Preset image	Select a preset image and its version used by the algorithm. To use an old-version image, select Show Old Images.
Code Directory	OBS path for storing the algorithm code. The files required for training, such as the training code, dependency installation packages, and pre-generated models, are uploaded to the code directory. Do not store training data in the code directory. When the training job starts, the data stored in the code directory will be downloaded to the backend. A large amount of training data may lead to a download failure. After you create the training job, ModelArts downloads the code directory and its subdirectories to the container. Take OBS path obs://obs-bucket/training-test/demo-code as an example. The content in the OBS path will be automatically downloaded to ${MA_JOB_DIR}/demo-code in the training container, and demo-code (customizable) is the last-level directory of the OBS path. NOTE: Any programming language is supported. The number of files (including files and folders) cannot exceed 1,000. The total size of files cannot exceed 5 GB.
Boot File	The file must be stored in the code directory and end with .py. ModelArts supports boot files edited only in Python. The boot file in the code directory is used to start a training job.

Figure 1 Using a custom script to create an algorithm

Configuring Pipelines

A preset image-based algorithm obtains data from an OBS bucket or dataset for model training. The training output is stored in an OBS bucket. The input and output parameters in your algorithm code must be parsed to enable data exchange between ModelArts and OBS. For details about how to develop code for training on ModelArts, see Developing a Custom Script.

When you use a preset image to create an algorithm, configure the input and output pipelines.

• Input configurations

  Table 2 Input configurations

  Parameter	Description
  Parameter Name	Set the name based on the data input parameter in your algorithm code. The code path parameter must be the same as the training input parameter parsed in your algorithm code. Otherwise, the algorithm code cannot obtain the input data. For example, If you use argparse in the algorithm code to parse data_url into the data input, set the data input parameter to data_url when creating the algorithm.
  Description	Customizable description of the input parameter,
  Obtained from	Source of the input parameter. You can select Hyperparameters (default) or Environment variables.
  Constraints	Whether data is obtained from a storage path or ModelArts dataset. If you select the ModelArts dataset as the data source, the following constraints are added: Labeling Type: For details, see Creating a Labeling Job. Data Format, which can be Default, CarbonData, or both. Default indicates the manifest format. Data Segmentation: available only for image classification, object detection, text classification, and sound classification datasets. Possible values are Segmented dataset, Dataset not segmented, and Unlimited. For details, see Publishing a Data Version.
  Add	Multiple data input sources are allowed.

• Output configurations

  Table 3 Output configurations

  Parameter	Description
  Parameter Name	Set the name based on the data output parameter in your algorithm code. The code path parameter must be the same as the training output parameter parsed in your algorithm code. Otherwise, the algorithm code cannot obtain the output path. For example, If you use argparse in the algorithm code to parse train_url into the data output, set the data output parameter to train_url when creating the algorithm.
  Description	Customizable description of the output parameter,
  Obtained from	Source of the output parameter. You can select Hyperparameters (default) or Environment variables.
  Add	Multiple data output paths are allowed.

Defining Hyperparameters

When you use a preset image to create an algorithm, ModelArts allows you to customize hyperparameters so you can view or modify them anytime. After the hyperparameters are defined, they are displayed in the startup command and transferred to your boot file as CLI parameters.

1. Import hyperparameters.

   You can click Add hyperparameter to manually add hyperparameters.

   Figure 2 Adding hyperparameters
   
2. Edit hyperparameters.

   For details, see Table 4.

   Table 4 Hyperparameters

   Parameter	Description
   Name	Hyperparameter name Enter 1 to 64 characters. Only letters, digits, hyphens (-), and underscores (_) are allowed.
   Type	Type of the hyperparameter, which can be String, Integer, Float, or Boolean
   Default	Default value of the hyperparameter, which is used for training jobs by default
   Constraints	Click Restrain. Then, set the range of the default value or enumerated value in the dialog box displayed.
   Required	Select Yes or No. If you select No, you can delete the hyperparameter on the training job creation page when using this algorithm to create a training job. If you select Yes, you cannot delete the hyperparameter on the training job creation page when using this algorithm to create a training job.
   Description	Description of the hyperparameter Only letters, digits, spaces, hyphens (-), underscores (_), commas (,), and periods (.) are allowed.

Supported Policies

ModelArts supports auto search. Auto search automatically finds the optimal hyperparameters without any code modification. This improves model precision and convergence speed. For details about parameter settings, see Parameters of hyperparameter search.

Only the pytorch_1.8.0-cuda_10.2-py_3.7-ubuntu_18.04-x86_64 and tensorflow_2.1.0-cuda_10.1-py_3.7-ubuntu_18.04-x86_64 images are available for auto search.

Adding Training Constraints

You can add training constraints of the algorithm based on your needs.

• Resource Type: Select the required resource types.
• Multicard Training: Choose whether to support multi-card training.
• Distributed Training: Choose whether to support distributed training.

Runtime Environment Preview

When creating an algorithm, click the arrow on in the lower right corner of the page to know the path of the code directory, boot file, and input and output data in the training container.

Follow-up Operations

After an algorithm is created, use it to create a training job. For details, see Creating a Training Job.