Upload helper.py

helper.py

@@ -42,7 +42,7 @@ llm = HuggingFacePipeline(pipeline=pipe)
 # # Initialize instructor embeddings using the Hugging Face model
 # instructor_embeddings = HuggingFaceInstructEmbeddings(model_name="C:/Users/arasu/Workspace/Projects/GenAI/embeddings/hkunlp_instructor-large")
 instructor_embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-large")
-db_path = "./vector_db"
+vector_db = ""
 
 def create_vector_db():
     # Load data from pdf
@@ -64,13 +64,13 @@ def create_vector_db():
     texts = text_splitter.split_text(raw_text)
 
     # Create a  vector database from 'text'
-    vector_db = Chroma.from_texts(texts,instructor_embeddings,persist_directory=db_path)
-    vector_db.persist()
-    vector_db = None 
+    vector_db = Chroma.from_texts(texts,instructor_embeddings)
+    # vector_db.persist()
+    # vector_db = None 
 
 def get_qa_chain():
     # Load the vector database from the local folder
-    vector_db = Chroma(persist_directory=db_path, embedding_function = instructor_embeddings)
+    # vector_db = Chroma(persist_directory=db_path, embedding_function = instructor_embeddings)
 
     # Create a retriever for querying the vector database
     retriever = vector_db.as_retriever(search_kwargs={"k":3})