AdvancedVocabulary#data-science-ml#backend#architecture

Training-Serving Skew Vocabulary

Learn the vocabulary of a feature computed differently between a training pipeline and a live serving pipeline.

0 / 5 completed
1 / 5
At standup, a dev mentions a model performing noticeably worse in production than its offline test metrics suggested, traced to a feature being computed differently by the live inference pipeline than it was during training. What is this problem called?