removing Thread

app.py

@@ -32,12 +32,14 @@ from huggingface_hub import login
 
 login(token=HF_KEY)
 
+vectorstore=None
+
 class BSIChatbot:
     def __init__(self, model_paths: Dict[str, str], docs_path: str):
         self.embedding_model = None
         self.llmpipeline = None
         self.llmtokenizer = None
-        self.vectorstore = None
+        #self.vectorstore = None
         self.reranking_model = None
         self.streamer = None
         self.images = [None]
@@ -50,7 +52,7 @@ class BSIChatbot:
     @spaces.GPU
     def initialize_embedding_model(self, rebuild_embeddings: bool):
         raw_knowledge_base = []
-
+        global vectorstore
         # Initialize embedding model
         self.embedding_model = HuggingFaceEmbeddings(
             model_name=self.word_and_embed_model_path,
@@ -91,18 +93,19 @@ class BSIChatbot:
             self.vectorstore.save_local(os.path.join(self.docs, "_embeddings"))
         else:
             # Load existing vector store
-            self.vectorstore = FAISS.load_local(os.path.join(self.docs, "_embeddings"), self.embedding_model, allow_dangerous_deserialization=True)
+            vectorstore = FAISS.load_local(os.path.join(self.docs, "_embeddings"), self.embedding_model, allow_dangerous_deserialization=True)
             print("DBG: Vectorstore Status Initialization:", self.vectorstore)
 
     @spaces.GPU
     def retrieve_similar_embedding(self, query: str):
+        global vectorstore
         #lazy load
         #if (self.vectorstore == None):
         #   self.vectorstore = FAISS.load_local(os.path.join(self.docs, "_embeddings"), self.embedding_model,
         #                                        allow_dangerous_deserialization=True)
         print("DBG: Vectorstore Status retriever:", self.vectorstore)
         query = f"Instruct: Given a search query, retrieve the relevant passages that answer the query\nQuery:{query}"
-        return self.vectorstore.similarity_search(query=query, k=20)
+        return vectorstore.similarity_search(query=query, k=20)
 
     @spaces.GPU
     def initialize_llm(self):