Using a database's native vector search
Databases increasingly include their own vector-comparison and search operations (comparing one vector with others). In order to use this, include this configuration in your configuration setup:
cluster:
vector_search:
type: native
or
export SUPERDUPER_CLUSTER_VECTOR_SEARCH_TYPE=native
If superduper
detects this configuration, it uses the inbuilt mechanism
of your db.databackend
to perform the vector-comparison.
Currently Superduper supports the native implementation of these databases:
- MongoDB Atlas
More integrations are on the way.