Releasing ColGREP today, made using our open-source multi-vector database, it support grep like filtering while adding an extra input to rank output of grep based on semantic similarity. ColGREP is packed with strong code retrieval models we designed and run 100% locally.