This article summarizes the terminology used to indicate the status of a machine learning model.
In the AI & Analytics Engine, the following terminology is used to indicate the status of a machine learning model at various stages:
|
Tag |
Meaning |
|---|---|
|
Queued |
Training of the model has not yet commenced. The Engine is waiting for the availability of computational resources and training environment to be set up |
|
Training / under training |
The model is currently being trained on the provided training data. On the model listing page, a percentage progress bar is shown at this stage. |
|
Failed training |
Model training encountered a problem. This is usually a bug in our code and needs to be reported to the PI.EXCHANGE team. You can try a bigger configuration. |
|
Evaluating / under evaluation |
The model has been trained and is being evaluated on the test portion. You can use the trained model for predictions while it is still being evaluated on the test portion, but this is not recommended. |
|
Failed evaluation |
Model evaluation encountered a problem. This is usually a system error and needs to be reported to the PI.EXCHANGE team. This occurs very rarely. |
|
Ready |
The model has been trained and evaluation on the available test data has been successfully done. You can see the performance insights of the model and start using it for predictions. |
Status of models training shown in the App summary page