Requesting additional details are provided w.r.t the training details for Vertex Search and conversation Recommendations Apps
Currently there are only 3 training states [1]: TRAINING, PAUSED, and TRAINING_STATE_UNSPECIFIED. There is no complete state as the model will continuously train unless paused by the user or some other issue resulting in the unspecified training state.
The Ready to query status is only useful for the initial training and causes confusion if you're querying with the latest trained details or not
How this might work:
A list of training operations that happened on the recommendation app model showcasing when they started and finished, this will help users see how long each training operation takes and inform them if the current model is up to date with the latest changes made
If applicable, reasons why alternative solutions are not sufficient:
The Ready to query status that indicates the app is ready to be used(after its initial training) will continue to stay there if your model was re-indexed, this makes it so users don't know if the model that is currently ready to query has the latest changes implemented and forces them to periodically check hoping that the training occurred somewhere within the 1-5 days suggestion given by the cloud console.
Other information (workarounds you have tried, documentation consulted, etc):
Description
What you would like to accomplish:
Ready to querystatus is only useful for the initial training and causes confusion if you're querying with the latest trained details or notHow this might work:
If applicable, reasons why alternative solutions are not sufficient:
Ready to querystatus that indicates the app is ready to be used(after its initial training) will continue to stay there if your model was re-indexed, this makes it so users don't know if the model that is currently ready to query has the latest changes implemented and forces them to periodically check hoping that the training occurred somewhere within the1-5 dayssuggestion given by the cloud console.Other information (workarounds you have tried, documentation consulted, etc):
[1]https://cloud.google.com/generative-ai-app-builder/docs/reference/rest/v1alpha/projects.locations.collections.engines#trainingstate