Diagnose retrieval quality problems caused by inconsistent embedding models, dimensions, preprocessing, or indexing pipelines.
Troubleshooting
Vector search returns unexpectedly poor or inconsistent results after an ingestion or application change.
Align the query and document embedding pipeline and rebuild affected indexes when required.
Run a fixed retrieval evaluation set and compare relevance before and after the change.
Version embedding configuration and validate pipeline compatibility in deployment checks.
A retrieval pipeline depends on the query and document representations being compatible.
Compare:
Align the pipeline, rebuild affected vectors when needed, then compare retrieval quality against a stable evaluation set.
Learning checkpoint
Mark this guide complete to include it in your local Engineering Journey.
Knowledge path
Referenced by
