At the moment, the creation of Embedding models is not required when creating Documents, but there are backends like the pgvector backend that depends on them existing.
It probably makes sense for a Document to always need the ID of the Embedding object it relates to, forcing a converter to generate those objects.
tomusher
commented
5 months ago
At the moment, the creation of Embedding models is not required when creating Documents, but there are backends like the
pgvectorbackend that depends on them existing.It probably makes sense for a Document to always need the ID of the
Embeddingobject it relates to, forcing a converter to generate those objects.