Update app.py

app.py

@@ -6,6 +6,7 @@ from langchain_community.vectorstores.faiss import FAISS
 from sentence_transformers import SentenceTransformer
 from langchain.chains import RetrievalQA
 from langchain_community.llms import HuggingFaceHub
+from langchain.docstore.document import Document
 
 # Load the PDF document
 loader = PyPDFLoader("apexcustoms.pdf")
@@ -15,12 +16,15 @@ data = loader.load()
 text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=20)
 texts = text_splitter.split_documents(data)
 
+# Create a list of document objects from the texts
+documents = [Document(page_content=doc.page_content) for doc in texts]
+
 # Create a vector store
 embeddings = SentenceTransformer("sentence-transformers/all-mpnet-base-v2")
-texts = [doc.page_content for doc in texts]  # Get the text content from the documents
+texts = [doc.page_content for doc in documents]  # Get the text content from the documents
 embeddings = embeddings.encode(texts)  # Get the embeddings for the texts
 
-vector_store = FAISS.from_documents(texts, embeddings)
+vector_store = FAISS.from_documents(documents, embeddings)
 
 # Initialize the HuggingFaceHub LLM
 llm = HuggingFaceHub(repo_id="HuggingFaceH4/zephyr-7b-beta", model_kwargs={"temperature": None, "top_p": None})
@@ -43,7 +47,7 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
 
     messages.append({"role": "user", "content": message})
 
-    result = qa({"input_documents": texts, "question": message})
+    result = qa({"input_documents": documents, "question": message})
     response = result["result"]
 
     history.append((message, response))