Remove GPU for semantic models

src/SemanticSearch.py

@@ -7,7 +7,6 @@ import requests
 import itertools
 import torch
 
-import spaces
 
 import chromadb
 
@@ -201,8 +200,8 @@ class SemanticSearch:
 
         self.collection = client.get_collection(name=self.collection_name)
 
-    @spaces.GPU
-    def extract(self, nlq: str, n_results:int=10, n_candidates:int=100, str_match:bool=True, rerank:bool=True):
+    
+    def extract(self, nlq: str, n_results:int=10, n_candidates:int=50, str_match:bool=True, rerank:bool=True):
 
         embedding = self.get_text_embeddings_local([nlq])[0].tolist()