Help Center> ModelArts> FAQs> ExeML> Training Models> Why Does ExeML Training Fail?
View PDF

Why Does ExeML Training Fail?

If the training of an ExeML project fails, perform the following steps to rectify the fault:

  1. Access Billing Center and check whether the account is in arrears.
    1. If the account is in arrears, top up the account.
    2. If the account is not in arrears, go to 2.
  2. Check whether the OBS path for storing image data complies with the following requirements:
    • The OBS path does not contain other folders.
    • The file name does not contain the following special characters: ~`@#$%^&*{}[]:;+=<>/

    If the OBS path meets the requirements, go to 3.

  3. The failure cause may vary depending on the ExeML project.
    • If image recognition training fails, check whether there are damaged images. If there are damaged images, replace or delete them.
    • If object detection training fails, check whether the labeling mode of the dataset is correct. Currently, ExeML supports only rectangle-based labeling.
    • If predictive analytics training fails, check the label column. Currently, the label column supports discrete and continuous data. Only one column can be selected.
    • If sound classification training fails, check whether the audio files are 16-bit WAV files.

    If the fault persists, submit a service ticket for technical support.

Parent topic: Training Models