Assigned
Status Update
No update yet.
Comments
ro...@google.com
va...@google.com
va...@google.com
Hello,
Thanks for the feature request! This has been forwarded to the AI platform Engineering Team so that they may evaluate it. Note that there are no ETAs or guarantees of implementation for feature requests. All communication regarding this feature request is to be done here.
Description
Some users are using the hyperparameter tuning in their models in the Vertex AI pipelines with Cloud AI Platform and have noticed that a dedicated VM is allocated for each job and each trial. The cluster configuration can be defined when running a training job in AI Platform. Concretely, it is possible to specify the
workerCountand other parameters. Nevertheless, it is not currently applicable for hyperparameter tuning.What you would like to accomplish: To allow the optimization of the resources allocation such that the same VM is used for all the hyperparameter tuning trials and for both training and prediction.
The product team is aware of this issue and they are working on it. At the moment, there is no ETA. Any further updates should be monitored through this post.