<|im_start|>system
You are a helpful assistant that selects the output that best follows the instruction.

Your answer should be a JSON object with the fields "Concise explanation" and "Output (a) is better than Output (b)".  "Output (a) is better than Output (b)" is a boolean indicating whether Output (a) is better than Output (b) so you always need to make a decision even if both are wrong (choose the least bad) or equal (choose randomly). "Concise explanation" is a string that provides a very brief explanation for the decision.
<|im_end|>
<|im_start|>user
## Example 1
### Instruction for example 1:
Describe the job of an ophthalmologist.

### Output (a) for example 1:
An ophthalmologist is a medical doctor who specializes in the diagnosis and treatment of eye diseases and conditions. They conduct eye exams, prescribe eyeglasses or contact lenses, and perform surgeries.

### Output (b) for example 1:
An ophthalmologist is a medical doctor who pokes and prods at your eyes while asking you to read letters from a chart.

### Preferred output in JSON format for example 1:
<|im_end|>
<|im_start|>assistant
{
"Concise explanation": "Output (a) provides a comprehensive and accurate description of the job of an ophthalmologist, including their specialties and duties. In contrast, output (b) is more of a joke.",
"Output (a) is better than Output (b)": true
}
<|im_end|>
<|im_start|>user
Good job! Now I will give you two examples in a row, and you should choose the best outputs for both of them. Note that the correct answer for one has no impact on the other.

## Example 2
### Instruction for example 2:
Paraphrase the following question: "What is the capital of France?"

### Output (a) for example 2:
The capital of France is Paris.

### Output (b) for example 2:
Can you tell me the name of the city that serves as the capital of France?

## Example 3
### Instruction for example 3:
Change the first verb in "I love surfing" to a synonym

### Output (a) for example 3:
I hate surfing

### Output (b) for example 3:
and am planning on going surfing tomorrow morning.

## Preferred output in JSON format for example 2-3:
<|im_end|>
<|im_start|>assistant
### Preferred output in JSON format for example 2:
{
"Concise explanation": "Output (b) paraphrases the question as asked by the instruction. In contrast, Output (a) does not follow instruction as it answers the question instead.",
"Output (a) is better than Output (b)": false
}

### Preferred output in JSON format for example 3:
{
"Concise explanation": "Neither output is correct. Output (a) tries to follow more closely the instruction as it changes the first verb of the sentence.",
"Output (a) is better than Output (b)": true
}
<|im_end|>
<|im_start|>user
Great! Now I will give you 4 examples in a row.

## Example 4:
### Instruction for example 4:
{instruction}

### Output (a) for example 4:
{output_1}

### Output (b) for example 4:
{output_2}

## Example 5:
### Instruction for example 5:
{instruction}

### Output (a) for example 5:
{output_1}

### Output (b) for example 5:
{output_2}

## Example 6:
### Instruction for example 6:
{instruction}

### Output (a) for example 6:
{output_1}

### Output (b) for example 6:
{output_2}

## Example 7:
### Instruction for example 7:
{instruction}

### Output (a) for example 7:
{output_1}

### Output (b) for example 7:
{output_2}

## Preferred output in JSON format for example 4-7:
<|im_end|>