Help Center>
ModelArts>
FAQs>
Built-in Algorithms>
What Are the Differences Between the Built-in Algorithms Released in Training Management and AI Gallery?
What Are the Differences Between the Built-in Algorithms Released in Training Management and AI Gallery?
Built-in Algorithms in Training Management
ModelArts provides built-in algorithms for multiple scenarios on the Training Management > Training Jobs > Built-in Algorithms page. You can quickly build models without coding.
- Introduction to built-in algorithms
- Requirements for datasets of built-in algorithms in different scenarios
- Algorithms and their running parameters
- Examples of using built-in algorithms
Built-in Algorithms Officially Released in AI Gallery
In the ModelArts AI Gallery, the official account provides built-in algorithms for multiple application scenarios. You only need to subscribe to the algorithms in AI Gallery and train models in ModelArts.
- Introduction to officially released built-in algorithms
- Example of using a built-in algorithm in AI Gallery
Differences Between the Built-in Algorithms
- The versions of the built-in algorithms in Training Management are earlier. Built-in algorithms officially released in AI Gallery are new versions. They can meet requirements of more scenarios (such as image classification, object detection, and sound classification) and support more AI frameworks (such as MindSpore, PyTorch, and TensorFlow). You are advised to use the built-in algorithms provided by AI Gallery.
- Built-in algorithms in AI Gallery can be used in ModelArts only after you have subscribed to them. Built-in algorithms in Training Management can be directly used.
- Models trained using most built-in algorithms in Training Management do not support model conversion and cannot be deployed and brought online using Ascend resources. Models trained using the built-in algorithms officially released in AI Gallery support model conversion and can be deployed using Ascend resources.
- Currently, neither of the two built-in algorithms is charged. The algorithm itself is free of charge. However, when you use the algorithm to train a model, you will be charged based on the selected compute resources.
Parent topic: Built-in Algorithms
Did this article solve your problem?
Thank you for your score！Your feedback would help us improve the website.