hard20 pts · first answer countsml-pipelinesfeature-storetraining-serving-skew
Bug fixed offline only
A bug in "days_since_last_order" was fixed and the offline history was backfilled with the new logic. The model was retrained on it. The streaming job that fills the online store still runs the old code. What follows?
Discussion 0
Join free to post in the discussion.
Related questions
- What training-serving skew isMLOps Engineer
- Why pipelines are DAGsMLOps Engineer
- What a Kubeflow component isMLOps Engineer
- MLflow logging callsMLOps Engineer
- What mlflow.autolog() doesMLOps Engineer
- What a .dvc file holdsMLOps Engineer
- Training broke after a fresh installMLOps Engineer
- Bundling preprocessing with the modelMLOps Engineer