Update app.py

app.py

@@ -109,12 +109,12 @@ def generate(image, pdfs, temperature=0.9, max_new_tokens=1500, top_p=0.95, repe
     vector_store = Chroma.from_documents(texts, embeddings, collection_metadata = {"hnsw:space":"cosine"}, persist_directory="stores/story_cosine" )
     print("vector store created........................")
 
-    load_vector_store = Chroma(persist_directory="stores/story_cosine", embedding_function=embeddings)
+    # load_vector_store = Chroma(persist_directory="stores/story_cosine", embedding_function=embeddings)
     # persist_directory="stores/story_cosine": laod the existing vector store form "stores/story_cosine"
     # embedding_function=embeddings: using the bge embedding model when add the new data to the vector store
 
     # Only get the k most similar document from the dataset
-    retriever = load_vector_store.as_retriever(search_kwargs={"k":top_k})
+    retriever = vector_store.as_retriever(search_kwargs={"k":top_k})
 
     image_caption, gender, age, emotion = get_image_info(image) 
     print("............................................")