Deze pagina is nog niet beschikbaar in uw eigen taal. We werken er hard aan om meer taalversies toe te voegen. Bedankt voor uw steun.

On this page

Show all

Help Center/ ModelArts/ DevEnviron/ JupyterLab/ Using ModelArts SDK

Using ModelArts SDK

Updated on 2024-06-12 GMT+08:00

Notebook instances allow you to use ModelArts SDK to manage OBS, training jobs, models, and real-time services.

Your notebook instances have automatically obtained your AK/SK for authentication and the region. Therefore, SDK sessions are automatically authenticated.

Example Code

  • Create a training job.
     1
     2
     3
     4
     5
     6
     7
     8
     9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    from modelarts.session import Session
    from modelarts.estimator import Estimator
    session = Session()
    estimator = Estimator(
                          modelarts_session=session,
                          framework_type='PyTorch',                                     # AI engine name
                          framework_version='PyTorch-1.0.0-python3.6',                  # AI engine version
                          code_dir='/obs-bucket-name/src/',                             # Training script directory
                          boot_file='/obs-bucket-name/src/pytorch_sentiment.py',        # Training boot script directory
                          log_url='/obs-bucket-name/log/',                              # Training log directory
                          hyperparameters=[
                                           {"label":"classes",
                                            "value": "10"},    
                                           {"label":"lr",
                                            "value": "0.001"}
                                           ],
                          output_path='/obs-bucket-name/output/',                         # Training output directory
                          train_instance_type='modelarts.vm.gpu.p100',                  # Training environment specifications
                          train_instance_count=1,                                       # Number of training nodes
                          job_description='pytorch-sentiment with ModelArts SDK')       # Training job description
    job_instance = estimator.fit(inputs='/obs-bucket-name/data/train/', wait=False, job_name='my_training_job')
    
  • Obtain a model list.
    1
    2
    3
    4
    from modelarts.session import Session 
    from modelarts.model import Model
    session = Session() 
    model_list_resp = Model.get_model_list(session, model_status="published", model_name="digit", order="desc")
    
  • Obtain service details.
    1
    2
    3
    4
    5
    from modelarts.session import Session
    from modelarts.model import Predictor
    session = Session()
    predictor_instance = Predictor(session, service_id="input your service_id")
    predictor_info_resp = predictor_instance.get_service_info()
    
Feedback

Feedback

Feedback

0/500

Selected Content

Submit selected content with the feedback