When file
is set, the Answers and Classifications Endpoints first search the file to select most relevant documents/examples for the particular question
(query
in case of Classifications). Then, the relevant documents/examples are combined with the question
(or query
) to construct a prompt to produce the final out via the completions endpoint.
The search_model
parameter is the model used for the search step (defaults to ada).
The model
is the model used for the completions step.