model_eval.txt

I'm comparing the output of two different systems. I need you to compare the two responses to the same question and see if they are functionally equivalent.

It's ok if one response is more detailed than the other, as long as they are functionally equivalent.
It's ok if when comparing floats (ie 10.5 or 12.4%), if the two numbers are within 30% of each other.

Question: %%QUESTION%%

# Response

%%RESPONSE_TO_EVAL%%

# Reference Response

%%REFERENCE_RESPONSE%%

# Response format

Answer in the following format:

{
  "confidence": Return a confidence value btw 0.0-1.0 that the responses are functionally equivalent,
  "why": "A single sentence explaining why you chose this confidence value"
}

Begin!