Update app.py

app.py

@@ -1,6 +1,10 @@
 import streamlit as st
+import pandas as pd
+import numpy as np
+import datetime
 import pickle
 import os
+import csv
 import torch
 from tqdm.auto import tqdm
 from langchain.text_splitter import RecursiveCharacterTextSplitter
@@ -26,7 +30,7 @@ st.markdown("# Hello")
 
 @st.cache_data
 def load_scraped_web_info():
-    with open("ait-web-document", "rb") as fp:
+    with open("/Users/carlosito/Library/CloudStorage/OneDrive-Personal/AIT material/99-AIT-thesis/aitGPT/ait-web-document", "rb") as fp:
         ait_web_documents = pickle.load(fp)
         
         
@@ -46,7 +50,6 @@ def load_scraped_web_info():
 
 
 
-
 @st.cache_resource
 def load_embedding_model():
     embedding_model = HuggingFaceInstructEmbeddings(model_name='hkunlp/instructor-base',
@@ -87,6 +90,14 @@ def load_retriever(llm, db):
 #--------------
 
 
+if "history" not in st.session_state:
+    st.session_state.history = []
+if "session_rating" not in st.session_state:
+    st.session_state.session_rating = 0
+    
+def update_score():
+    st.session_state.session_rating = st.session_state.rating
+
 
 load_scraped_web_info()
 embedding_model = load_embedding_model()
@@ -107,20 +118,56 @@ def retrieve_document(query_input):
     return related_doc
 
 def retrieve_answer(query_input):
+
     answer = qa_retriever.run(query_input)
+    output = st.text_area(label="Retrieved documents", value=answer)
+    
+    st.markdown('---')
+    score = st.radio(label = 'please select the overall satifaction and helpfullness of the bot answer', options=[1,2,3,4,5], horizontal=True,
+                     on_change=update_score, key='rating')
+
     return answer
+    
+
+
+
+st.write("# aitGPT 🤖 ")
+st.markdown("""
+         #### The aitGPT project is a virtual assistant developed by the :green[Asian Institute of Technology] that contains a vast amount of information gathered from 205 AIT-related websites.  
+        The goal of this chatbot is to provide an alternative way for applicants and current students to access information about the institute, including admission procedures, campus facilities, and more.  
+          """)
+st.write(' ⚠️ Please expect to wait **~ 10 - 20 seconds per question** as thi app is running on CPU against 3-billion-parameter LLM')
 
-output_1 = st.text_area(label = "Here is the relevant documents",
-                      value = retrieve_document(query_input))
+st.markdown("---")
 
 
-output_2 = st.text_area(label = "Here is the answer",
-                      value = retrieve_answer(query_input))
+query_input = st.text_area(label= 'What would you like to know about AIT?')
+generate_button = st.button(label = 'Submit!')
 
+if generate_button:
+    answer = retrieve_answer(query_input)
+    log = {"timestamp": datetime.datetime.now(),
+        "question":query_input,
+        "generated_answer": answer,
+        "rating":st.session_state.session_rating }
 
+    st.session_state.history.append(log)
 
-# faiss_retriever = vector_database.as_retriever()
-# print("Succesfully had FAISS as retriever")s
 
 
+if st.session_state.session_rating == 0:
+    pass
+else:
+    with open('test_db', 'a') as csvfile:
+        writer = csv.writer(csvfile)
+        writer.writerow([st.session_state.history[-1]['timestamp'], st.session_state.history[-1]['question'],
+                            st.session_state.history[-1]['generated_answer'], st.session_state.session_rating ])
+        st.session_state.session_rating = 0
 
+test_df = pd.read_csv("/Users/carlosito/Library/CloudStorage/OneDrive-Personal/AIT material/99-AIT-thesis/aitGPT/test_db",
+                      index_col=0)
+test_df.sort_values(by = ['timestamp'],
+                    axis=0,
+                    ascending=False,
+                    inplace=True)
+st.dataframe(test_df)