Upload 2 files

app.py

@@ -2,17 +2,17 @@ import streamlit as st
 from generator import generate_response_from_document
 from retrieval import retrieve_documents
 from evaluation import calculate_metrics
-from data_processing import load_data_from_faiss
+#from data_processing import load_data_from_faiss
 import time
 
 # Page Title
 st.title("RAG7 - Real World RAG System")
 
-@st.cache_data
-def load_data():
-    load_data_from_faiss()
+# @st.cache_data
+# def load_data():
+#     load_data_from_faiss()
 
-data_status = load_data()
+# data_status = load_data()
 
 time_taken_for_response = 'N/A'
 

retrieval.py

@@ -1,27 +1,29 @@
 import json
 import numpy as np
 from langchain.schema import Document
-from langchain.vectorstores import faiss
+import faiss
 
-from data_processing import embedding_model, index
+from data_processing import embedding_model #, index, actual_docs
+
+retrieved_docs = None
 
 # Retrieval Function
 def retrieve_documents(query, top_k=5):
-    # Embed the query
+    faiss_index_path = f"rag7_index.faiss"
+    index = faiss.read_index(faiss_index_path)
+
     query_embedding = np.array(embedding_model.embed_documents([query]), dtype=np.float32)
 
-    # Search in FAISS (top 5 results)
     _, nearest_indices = index.search(query_embedding, top_k)
 
-    # Load document metadata
-    with open(f"data_local\rag7_docs.json", "r") as f:
+    with open(f"rag7_docs.json", "r") as f:
         documents = json.load(f)  # Contains all documents for this dataset
 
-    # Retrieve the actual documents and create Document objects
     retrieved_docs = [Document(page_content=documents[i]) for i in nearest_indices[0]]
 
     return retrieved_docs
 
+
 def remove_duplicate_documents(documents):
     unique_documents = []
     seen_documents = set()  # To keep track of seen documents