|
|
from typing import List, Dict |
|
|
import requests |
|
|
import time |
|
|
from dotenv import load_dotenv |
|
|
import os |
|
|
|
|
|
load_dotenv() |
|
|
|
|
|
class Reranker: |
|
|
def __init__(self): |
|
|
self.api_key = os.getenv("API_KEY") |
|
|
self.api_base = os.getenv("BASE_URL") |
|
|
|
|
|
def rerank(self, query: str, documents: List[Dict], index_name: str, top_k: int = 5) -> List[Dict]: |
|
|
"""使用SiliconFlow的rerank API重排序文档""" |
|
|
headers = { |
|
|
"Authorization": f"Bearer {self.api_key}", |
|
|
"Content-Type": "application/json" |
|
|
} |
|
|
|
|
|
|
|
|
docs = [doc['content'] for doc in documents] |
|
|
|
|
|
response = requests.post( |
|
|
f"{self.api_base}/rerank", |
|
|
headers=headers, |
|
|
json={ |
|
|
"model": "BAAI/bge-reranker-v2-m3", |
|
|
"query": query, |
|
|
"documents": docs, |
|
|
"top_n": top_k |
|
|
} |
|
|
) |
|
|
|
|
|
if response.status_code != 200: |
|
|
raise Exception(f"Error in reranking: {response.text}") |
|
|
|
|
|
|
|
|
results = response.json()["results"] |
|
|
reranked_docs = [] |
|
|
|
|
|
for result in results: |
|
|
doc_index = result["index"] |
|
|
original_doc = documents[doc_index].copy() |
|
|
original_doc['rerank_score'] = result["relevance_score"] |
|
|
original_doc['index_name'] = index_name |
|
|
reranked_docs.append(original_doc) |
|
|
|
|
|
return reranked_docs |
