Skip to content

update max_model when trial fails #6596

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Mar 27, 2023
Merged

update max_model when trial fails #6596

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7e03592
update max_model when trial fails
LittleLittleCloud Mar 9, 2023
b546c94
add nullable mark to cfo tuner
LittleLittleCloud Mar 21, 2023
ddf7953
update exception message
LittleLittleCloud Mar 23, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 10 additions & 17 deletions src/Microsoft.ML.AutoML/AutoMLExperiment/AutoMLExperiment.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -282,9 +282,9 @@ void handler(object o, EventArgs e)
}
}
}
catch (OperationCanceledException ex) when (aggregateTrainingStopManager.IsStopTrainingRequested() == false)
catch (Exception ex) when (aggregateTrainingStopManager.IsStopTrainingRequested() == false)
{
logger.Trace($"trial cancelled - {JsonSerializer.Serialize(trialSettings)}, continue training");
logger.Trace($"catch exception when training trial {trialSettings.TrialId} - {JsonSerializer.Serialize(trialSettings)}, continue training");
Copy link
Member

@JakeRadMSFT JakeRadMSFT Mar 22, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Exception thrown during Trial {trialSettings.TrialId} with configuration {JsonSerializer.Serialize(trialSettings)}

Exception Details: ex.Message

Abandoning Trial {trialSettings.TrialId} and continue training.

Copy link
Member

@JakeRadMSFT JakeRadMSFT Mar 22, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@LittleLittleCloud is there a way to localization for these strings? Do we do it anywhere else in this repo?

Copy link
Member

@JakeRadMSFT JakeRadMSFT Mar 22, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@luisquintanilla thoughts on strings here?

Copy link
Member Author

@LittleLittleCloud LittleLittleCloud Mar 23, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@JakeRadMSFT I don't think strings are localized anywhere in mlnet, aren't they? @michaelgsharp

trialSettings.EndedAtUtc = DateTime.UtcNow;
monitor?.ReportFailTrial(trialSettings, ex);
var trialResult = new TrialResult
Expand All @@ -296,22 +296,8 @@ void handler(object o, EventArgs e)
tuner.Update(trialResult);
trialResultManager?.AddOrUpdateTrialResult(trialResult);
aggregateTrainingStopManager.Update(trialResult);
Copy link
Member Author

@LittleLittleCloud LittleLittleCloud Mar 9, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We need to update tuner/trialresultmanager/trainningStopManager even when a trial fails for any reasons as long as the experiment still continues

continue;
}
catch (OperationCanceledException) when (aggregateTrainingStopManager.IsStopTrainingRequested())
{
logger.Trace($"trial cancelled - {JsonSerializer.Serialize(trialSettings)}, stop training");

break;
}
catch (Exception ex)
{
logger.Trace($"trial failed - {JsonSerializer.Serialize(trialSettings)}, stop training");

trialSettings.EndedAtUtc = DateTime.UtcNow;
monitor?.ReportFailTrial(trialSettings, ex);

if (!aggregateTrainingStopManager.IsStopTrainingRequested() && _bestTrialResult == null)
if (ex is not OperationCanceledException && _bestTrialResult == null)
{
logger.Trace($"trial fatal error - {JsonSerializer.Serialize(trialSettings)}, stop training");

Expand All @@ -321,6 +307,13 @@ void handler(object o, EventArgs e)
// when error is fatal (like schema mismatch).
throw;
}
continue;
}
catch (Exception) when (aggregateTrainingStopManager.IsStopTrainingRequested())
{
logger.Trace($"trial cancelled - {JsonSerializer.Serialize(trialSettings)}, stop training");

break;
}
finally
{
Expand Down