When evaluating model input in evaluations, flag the input with input()
and the output with output(), allowing for persistent logging of
responses.
When evaluating model input in evaluations, flag the input with input()
and the output with output(), allowing for persistent logging of
responses.