1

I’m getting
No similar documents were found in file with ID
when using the example from the documentation, just using the -file functionality rather than uploading the documents along with the request.

The .jsonl file that gets uploaded uses the Documents from the documentation:

{"text":"White House"}
{"text":"hospital"}
{"text":"school"}

with query

the president
2

So for anyone running into this, OpenAI Support told me that a file search is different in that there is a keyword filter first. “The president” does not have a keyword match against “white house”, so that document is not evaluated semantically.

1 Like