Spaces:
Sleeping
Sleeping
File size: 1,398 Bytes
b24327e 7e00d2a 5f04136 b24327e 5f04136 b24327e 5f04136 b24327e 5f04136 b24327e 7e00d2a 6df9fea 7e00d2a 6df9fea 7e00d2a 6df9fea |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 |
import lancedb
import os
import gradio as gr
from sentence_transformers import SentenceTransformer, CrossEncoder
from dotenv import load_dotenv
load_dotenv()
DB_PATH = os.getenv("DB_PATH", ".lancedb")
db = lancedb.connect(DB_PATH)
TABLE_NAME = os.getenv("TABLE_NAME")
if not TABLE_NAME:
raise ValueError("TABLE_NAME environment variable is not set")
TABLE = db.open_table(TABLE_NAME)
VECTOR_COLUMN = os.getenv("VECTOR_COLUMN", "vector")
TEXT_COLUMN = os.getenv("TEXT_COLUMN", "text")
BATCH_SIZE = int(os.getenv("BATCH_SIZE", 32))
retriever = SentenceTransformer(os.getenv("EMB_MODEL"))
def retrieve(query, k):
query_vec = retriever.encode(query)
try:
documents = TABLE.search(query_vec, vector_column_name=VECTOR_COLUMN).limit(k).to_list()
documents = [doc[TEXT_COLUMN] for doc in documents]
return documents
except Exception as e:
raise gr.Error(str(e))
def reranking(query: str, docs: list[str], k: int) -> list[str]:
model_name = 'BAAI/bge-reranker-v2-m3'
# model_name = 'cross-encoder/ms-marco-MiniLM-L-6-v2'
model = CrossEncoder(model_name, max_length=512)
# Prepare the list of lists (query, document) for the model
pairs = [[query, curr] for curr in docs]
scores = model.predict(pairs)
scored_pairs = zip(scores, docs)
sorted_pairs = sorted(scored_pairs, reverse=True)
return [pair[1] for pair in sorted_pairs][:k]
|