alpha-nerd-nomyo
34d6abd28b
- Refactored token aggregation query in db.py to use a single SQL query with SUM() instead of iterating through rows, improving performance - Combined import statements in db.py and router.py to reduce lines of code - Enhanced chat proxy in router.py to handle "moe-" prefixed models with multiple query execution and critique generation - Added last_user_content() helper function to extract user content from messages - Improved code readability and maintainability through these structural changes
38 lines
1 KiB
Python
38 lines
1 KiB
Python
from pydantic import BaseModel
|
|
|
|
class feedback(BaseModel):
|
|
query_id: int
|
|
content: str
|
|
|
|
def moe(query: str, query_id: int, response: str) -> str:
|
|
moe_prompt = f"""
|
|
User query: {query}
|
|
query_id: {query_id}
|
|
|
|
The following is an assistant response to the original user query. Analyse the response, then critizise the response by discussing both strength and weakness of the response.
|
|
|
|
<assistant_response>
|
|
{response}
|
|
</assistant_response>
|
|
"""
|
|
return moe_prompt
|
|
|
|
def moe_select_candiadate(query: str, candidates_with_feedback: list[str]) -> str:
|
|
select_prompt = f"""
|
|
From the following responses for the user query: {query}
|
|
select the best fitting candidate and formulate a final anser for the user.
|
|
|
|
<candidate_0>
|
|
{candidates_with_feedback[0].message.content}
|
|
</candidate_0>
|
|
|
|
<candidate_1>
|
|
{candidates_with_feedback[1].message.content}
|
|
</candidate_1>
|
|
|
|
<candidate_2>
|
|
{candidates_with_feedback[2].message.content}
|
|
</candidate_2>
|
|
"""
|
|
return select_prompt
|
|
|