Updated on 2025-07-28 GMT+08:00

Input

Description

Input file names (for images, audio, and labels) must be fewer than 64 characters. Files with longer names will be ignored by the training job.

  • Dataset

    ModelArts data management provides an efficient data management and labeling framework. It supports various data types such as image, text, audio, and video in a range of AI projects of computer vision, natural language processing, and audio and video analysis.

    Select different types of datasets based on your algorithm scenario and ensure that the datasets have been labeled.

    ModelArts data management module is being reconstructed and upgraded. This module is transparent to you if you have not used it before. If you are a new user, store the training data in an OBS bucket.

  • Data path

    You must upload the labeled data to the OBS bucket and select the corresponding OBS path.

    Ensure that the OBS bucket and ModelArts are in the same region.

    Ensure that the access permissions of the OBS bucket match your read and write operations.

    The storage class of your OBS bucket must be Standard, which is suitable for scenarios where data is frequently accessed.

FAQs