amazon-braket
AHS submission dry run or task validation.
Describe the feature you'd like I want the ability to perform a dry run for AHS task submission. The idea would be to validate the task without actually submitting the task to the queue.
How would this feature be used? Please describe. This would be useful to prototype tasks and help when submitting batches of tasks: validating all tasks before submitting them to the queue.
Describe alternatives you've considered Not really any clear option to do this.
@weinbe58 Thank you for opening this request. The ask makes sense. Could you add a bit of context to your specific use case? Is it related to speed of debugging? Ability to build in exception handling during larger workloads? Something else? Thanks!
Is it related to speed of debugging?
Yes, one of the most common issues is trying to fit the minimum time step.
Ability to build in exception handling during larger workloads?
yes, many jobs require running batches of tasks that are related. Those tasks are generated programmatically and some errors can happen when generating some of those tasks. This is particularly difficult to track down where the issue is so repeated submission and cancellation is not ideal.
@weinbe58 your ask makes sense. We dont have a timeline for implementing dry-run internally. Would you be interested in picking up such a contribution if we open up an issue with implementation guidelines for you and help you with the PR reviews?