AdvancedVocabulary#ai-llm#data-science-ml#backend

Embedding Drift Vocabulary

Practice the vocabulary of keeping stored vectors comparable across an embedding model upgrade.

0 / 5 completed
1 / 5
At standup, a dev mentions that vectors produced by an older embedding model version are no longer meaningfully comparable to vectors produced by a newer version, since the newer model encodes semantic meaning differently. What is this phenomenon called?