is it possible to use auto-embed in a schema but a...
# community-help
h
is it possible to use auto-embed in a schema but also have the ability to upload embeddings (using the same huggingface model)? it's much faster to generate embeddings on our backend than on typesense cloud for document importing, but we'd still like to have typesense automatically encode user queries on the frontend (using the convenient
"vector_query": "embedding:([],distance_threshold: 0.15)"
, which is only available with auto-embed.
k
You can use auto embed in schema but always send the value for the vector field in the payload. We will not attempt to do auto embedding if a value is already present.
h
great that works, thanks
👍 1