# tinker_cookbook.preference.Comparison

## _class_ [**tinker_cookbook.preference.Comparison**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/main/tinker_cookbook/preference/types.py#L24)()

A pair of completions (A and B) generated from the same prompt conversation.

Used as the basic unit for preference learning: a single prompt with two candidate responses that can be compared by a human or a preference model.

```
comparison = Comparison(
    prompt_conversation=[{"role": "user", "content": "Summarize X"}],
    completion_A=[{"role": "assistant", "content": "Summary A..."}],
    completion_B=[{"role": "assistant", "content": "Summary B..."}],
)
```

**Fields:**

- [**prompt_conversation**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/main/tinker_cookbook/preference/types.py#L45) ( _list[renderers. [Message](https://tinker-docs.thinking-machines.ai/cookbook/api-reference/renderers/message/)\]_)  
- [**completion_A**](https://github.com/thinking-machines-lab/tinker-cookbook/blob/main/tinker_cookbook/preference/types.py#L46) ( _list[renderers. [Message](https://tinker-docs.thinking-machines.ai/cookbook/api-reference/renderers/message/)\