persistent class %ML.TrainingRun
extends %Persistent
A model training run
• property CompletedAt
as %TimeStamp;
Time completed (UTC)
• property Configuration
as %SYS.ML.Configuration;
ML Configuration for this training run
• property Log
as %Stream.GlobalCharacter;
Training log output
• property Model
as %ML.Model [ Required ];
Model we are training with
• property Name
as %String(MAXLEN=275) [ Required ];
Training Run name, same as the name for the cooresponding Trained Model
• property PID
as %String [ InitialExpression = $job ];
IRIS Process ID that performed the training
• property Provider
as %String(MAXLEN=128) [ Required ];
ML Provider that did the training
• property RunStatus
as %String(DISPLAYLIST=",Training,Completed,Failed",VALUELIST=",training,completed,failed") [ InitialExpression = "training" ];
Training status
• property Settings
as %Library.DynamicObject;
Settings for this training run
• property StartedAt
as %TimeStamp [ InitialExpression = $ZDATETIME($ZTIMESTAMP,3,1,3),Required ];
Time started (UTC)
• property StatusCode
as %Status [ InitialExpression = $$$OK ];
Training error (if encountered)
• property TrainingDuration
as %Numeric [ Transient ];
Training duration (in seconds)
• property TrainingQuery
as %String(MAXLEN=32000);
Training query
• method %Dump()
as %Status
Helper method for debugging etc.
• classmethod GetInterval(start As %TimeStamp, end As %TimeStamp)
as %Numeric
Compute the difference in seconds between two timestamps
• classmethod GetTrainingRunName(Model As %BigInt)
as %String
Compute the default name for this training run
// Model is the Oid of a Model
• method LogMsg(msg As %String)
as %Status
Log a timestamped message to the training log
•index (NameIndex on Name) [Unique];
