Update app.py

app.py

@@ -5,7 +5,7 @@ import faiss
 from PyPDF2 import PdfReader
 from sentence_transformers import SentenceTransformer
 from transformers import AutoTokenizer, AutoModel
-import torch  # Import torch for tensor operations
+import torch
 from langchain.vectorstores import FAISS
 from langchain.embeddings import HuggingFaceEmbeddings
 from langchain.chains import RetrievalQA
@@ -63,7 +63,8 @@ def extract_pdf_content(drive_url):
 
 # Function to create a FAISS vector store from the document content
 def create_vector_store(text):
-    sentences = text.split(". ")
+    # Split the text into sentences and clean it
+    sentences = [sentence.strip() for sentence in text.split(". ") if sentence.strip()]
     
     # Use Hugging Face transformer model for embeddings
     model_name = "sentence-transformers/all-MiniLM-L6-v2"
@@ -72,11 +73,14 @@ def create_vector_store(text):
 
     def embed(sentence):
         tokens = tokenizer(sentence, return_tensors="pt", truncation=True, padding=True)
-        with torch.no_grad():  # Use torch for no_grad context
+        with torch.no_grad():
             embeddings = model(**tokens).last_hidden_state.mean(dim=1).numpy()
         return embeddings
 
+    # Generate embeddings for cleaned sentences
     embeddings = [embed(sentence)[0] for sentence in sentences]
+
+    # Create a FAISS vector store with valid embeddings
     vector_store = FAISS.from_embeddings(sentences, embeddings)
     return vector_store, sentences