"dead kernel" Is Displayed and the Instance Breaks Down When Training Code Is Run

Memory insufficiency may be caused by large data volume, too many training layers, or other reasons when training code is run in a notebook instance. As a result, the notebook instance breaks down.

After this problem occurs, the system automatically restarts the notebook instance to fix the breakdown. However, only the breakdown problem is fixed. Re-running the training code will still fail.

To solve the problem of insufficient memory, you are advised to create a new notebook instance and use a resource pool of higher specifications, such as a dedicated resource pool, to run the training code.

An existing notebook instance cannot be scaled up using resources with higher specifications.