Table of contents

Training Schedules

Automate your ML training pipelines with scheduled jobs. Run experiments on a recurring basis for continuous model improvement.

Creating a Schedule

  1. Navigate to MLOps > Schedules
  2. Click Create Schedule
  3. Configure:
    • Name: Descriptive schedule name
    • Experiment: Associated experiment
    • Cron Expression: When to trigger (standard cron format)
    • Training Configuration: Parameters, dataset, and settings
  4. Click Create

Cron Expression Examples

ExpressionDescription
0 2 * * *Daily at 2:00 AM
0 0 * * 1Weekly on Monday at midnight
0 6 1 * *Monthly on the 1st at 6:00 AM
0 */6 * * *Every 6 hours

Managing Schedules

Schedule List

View all schedules with:

  • Schedule name and status
  • Cron expression (next run time)
  • Last execution result
  • Associated experiment

Schedule Actions

  • Pause: Temporarily stop future executions
  • Resume: Restart a paused schedule
  • Edit: Update cron expression or configuration
  • Delete: Permanently remove the schedule

Execution History

View past executions for each schedule:

  • Run timestamps
  • Success/failure status
  • Duration and resource usage
  • Link to the generated run

Use Cases

  • Nightly retraining: Retrain models on fresh data every night
  • Weekly evaluation: Evaluate production model performance weekly
  • Data drift detection: Run comparison experiments on new data batches
  • Continuous improvement: Iterate on hyperparameters automatically

Next Steps