Vectorizing more than one field with integrated inference

Support
,
1

Hi all.

When creating an index with integrated inference, is it possible to have more than one field be vectorized?

All I see in the docs is setting field_map to a single field. I have tried passing the below as a possibility, but it seems to only create the ability on the last field in field_map.

“field_map”:{
“text”: “chunk_text”,
“text”: “character”}
}

Thanks and appreciate any guidance on this.

2

Thanks for reaching out! Do you have more details you can share? For example are you trying to create a formatted string like

"text": f"{character} - {chunk_text}"
3

Hi, I am needing more than one field vectorized so that each field could possibly be queried separately.

Here is the full embed section of create_index_for_model that I am passing
embed={
“model”:“multilingual-e5-large”,
“field_map”:{
“text”: “chunk_text”,
“text”: “character”}
}

Thanks again and I appreciate your time!

4

The recommendation would be to create two indexes, one for each field. Would you need to merge the results together?

5

Thanks!

Yes, that could also be a possibility.