Human: Your role is to evaluate the correctness of the candidate model output provided in the <candidate model output></candidate model output> 
tags based on whether it aligns with the ground truth answer provided in the <ground_truth></ground_truth> xml tags in answering the 
question in the <question></question> xml tags.

Refer to the question that you have to use while evaluating the correctness of the candidate model response in alignment to the ground truth:
<question>
{question}
</question> 

Refer to the candidate model response to be evaluated in the <candidate model output></candidate model output> tags below:
<candidate model output>
{answer}
</candidate model output> 

Refer to the ground truth below in the <ground_truth></ground_truth> xml tags while evaluating the candidate model output:
<ground_truth>
{ground_truth}
</ground_truth> 

Follow the instructions below while giving your evaluation of the candidate model output in the <evaluation_instructions></evaluation_instructions>
tags:

<evaluation_instructions>
{rules}
</evaluation_instructions>

Your response should only be in JSON format. Your response should NOT have any tags, and should start with the starting bracket of the JSON
structure and end with the ending bracket. There should only be the JSON in your response without any other words outside of it, should not
contain any tags, only the JSON structure.

Assistant: Sure, here is my evaluation in JSON: