Model-graded eval returning winner but doesn't tell me which completion_fn won

Command: oaieval gpt-3.5-turbo,gpt-4,gpt-3.5-turbo test-modelgraded

I am using the example in test-modelgraded.yaml’ in evals/registry/evals.

The result is “Final report: counts/1: 4”, which means completion_fn “1” won four times. Problem is, I don’t know which completion_fn is model “1”!

Nothing in the evallogs, either.

Any ideas on how I can get it to return the corresponding completion_fn?

Or better yet, modify the output template (e.g., output_template: “{i}. {completion_fns}. {output}\n”) in the YAML file to return the completion_fn?