fix llm crash

pages/llm_tutor.py

@@ -1,6 +1,4 @@
 import time
-import os
-import joblib
 import streamlit as st
 from utils.questions_dataset import system_instruction, get_model_tools
 from google.genai import types
@@ -8,7 +6,6 @@ from google import genai
 
 st.set_page_config(page_title="LSAT Group A", page_icon="📘")
 
-# GOOGLE_API_KEY=os.environ.get('GOOGLE_API_KEY')
 GEMINI_API_KEY = "AIzaSyAjpHA08BUwLhK-tIlORxcB18RAp3541-M"
 client = genai.Client(api_key=GEMINI_API_KEY)
 
@@ -16,61 +13,12 @@ new_chat_id = f'{time.time()}'
 MODEL_ROLE = 'ai'
 AI_AVATAR_ICON = '✨'
 
-# Create a data/ folder if it doesn't already exist
-try:
-    os.mkdir('data/')
-except:
-    # data/ folder already exists
-    pass
-
-# Load past chats (if available)
-try:
-    past_chats: dict = joblib.load('data/past_chats_list')
-except:
-    past_chats = {}
-
-# Sidebar allows a list of past chats
-with st.sidebar:
-    st.write('# Past Chats')
-    if st.session_state.get('chat_id') is None:
-        st.session_state.chat_id = st.selectbox(
-            label='Pick a past chat',
-            options=[new_chat_id] + list(past_chats.keys()),
-            format_func=lambda x: past_chats.get(x, 'New Chat'),
-            placeholder='_',
-        )
-    else:
-        # This will happen the first time AI response comes in
-        st.session_state.chat_id = st.selectbox(
-            label='Pick a past chat',
-            options=[new_chat_id, st.session_state.chat_id] + list(past_chats.keys()),
-            index=1,
-            format_func=lambda x: past_chats.get(x, 'New Chat' if x != st.session_state.chat_id else st.session_state.chat_title),
-            placeholder='_',
-        )
-
-    # Save new chats after a message has been sent to AI
-    st.session_state.chat_title = f'ChatSession-{st.session_state.chat_id}'
-    
-
 st.title("📘Logical Reasoning: Group A")
 next_btn = st.button("Click here when finished")
 
 st.write("Use this AI Tutor to help you understand the concepts. You can ask it to explain the concepts, provide examples, or clarify any doubts you have.")
 st.write("Start by sending a hello message!")
 
-# Chat history (allows to ask multiple questions)
-try:
-    st.session_state.messages = joblib.load(
-        f'data/{st.session_state.chat_id}-st_messages'
-    )
-    st.session_state.gemini_history = joblib.load(
-        f'data/{st.session_state.chat_id}-gemini_messages'
-    )
-except:
-    st.session_state.messages = []
-    st.session_state.gemini_history = []
-
 sys_prompt = system_instruction % (
     st.session_state.prequiz_df['num_correct'][0],
     st.session_state.prequiz_df['num_questions'][0],
@@ -90,13 +38,25 @@ sys_prompt = system_instruction % (
     st.session_state.prequiz_df['num_questions'][7],
     st.session_state.prequiz_df['num_correct'][8],
     st.session_state.prequiz_df['num_questions'][8]
-)
-st.session_state.chat = client.chats.create(model='gemini-2.0-flash',
+) if st.session_state.prequiz_df is not None else ""
+
+st.session_state.chat_id = new_chat_id
+st.session_state.chat_title = f'ChatSession-{st.session_state.chat_id}'
+st.session_state.gemini_history = []
+
+# Initialize session state
+if "chat" not in st.session_state:
+    st.session_state.chat = None
+    
+    st.session_state.chat = client.chats.create(model='gemini-2.0-flash',
                                             config=types.GenerateContentConfig(
                                             tools=[get_model_tools()],
                                             system_instruction=sys_prompt),
-                                            history=st.session_state.gemini_history
-                                               )
+                                            history=st.session_state.gemini_history 
+                                            )
+if "messages" not in st.session_state:
+    st.session_state.messages = []
+
 
 # Display chat messages from history on app rerun
 for message in st.session_state.messages:
@@ -104,71 +64,35 @@ for message in st.session_state.messages:
         name=message['role'],
         avatar=message.get('avatar'),
     ):
-        st.markdown(message['content'])
+        st.markdown(message['text'])
 
-# React to user input
-if prompt := st.chat_input('Your message here...'):
-    # Save this as a chat for later
-    if st.session_state.chat_id not in past_chats.keys():
-        past_chats[st.session_state.chat_id] = st.session_state.chat_title
-        joblib.dump(past_chats, 'data/past_chats_list')
-    # Display user message in chat message container
-    with st.chat_message('user'):
-        st.markdown(prompt)
-    # Add user message to chat history
-    st.session_state.messages.append(
-        dict(
-            role='user',
-            content=prompt,
-        )
-    )
-    ## Send message to AI
-    response = st.session_state.chat.send_message_stream(
-        prompt,
-    )
-    # Display assistant response in chat message container
+# Chat input
+user_input = st.chat_input("💬 Ask your tutor a question...")
+if user_input:
+    st.chat_message("user").markdown(user_input)
+    st.session_state.messages.append({"role": "user", "text": user_input})
+    full_response = ""
+
+    response = st.session_state.chat.send_message_stream(user_input)
+    
+    full_reply = ""
     with st.chat_message(
         name=MODEL_ROLE,
         avatar=AI_AVATAR_ICON,
-    ):
-        message_placeholder = st.empty()
-        full_response = ''
-        assistant_response = response
-        # Streams in a chunk at a time
+        ):
+        response_box = st.empty()
         for chunk in response:
-            # Simulate stream of chunk
-            if chunk.text == None:
-                chunk.text = "Let's try that one more time so I understand. Please tell me one more time."
-
-            for ch in chunk.text.split(' '):
-                full_response += ch + ' '
+            chunk_text = chunk.text
+            if chunk_text:
+                full_reply += chunk_text
                 time.sleep(0.05)
-                # Rewrites with a cursor at end
-                message_placeholder.write(full_response + '▌')
-        # Write full message with placeholder
-        message_placeholder.write(full_response)
+                response_box.markdown(full_reply + "▌")
 
-            # Add assistant response to chat history
-        st.session_state.messages.append(
-            dict(
-                role=MODEL_ROLE,
-                content=full_response,
-                avatar=AI_AVATAR_ICON,
-            )
-        )
+    # Final display after stream ends
+    response_box.markdown(full_reply)
+    st.session_state.messages.append({"role": "assistant", "text": full_reply, "avatar": AI_AVATAR_ICON})
 
     st.session_state.gemini_history = st.session_state.chat.get_history()
-    
-    # Save to file
-    joblib.dump(
-        st.session_state.messages,
-        f'data/{st.session_state.chat_id}-st_messages',
-    )
-    joblib.dump(
-        st.session_state.gemini_history,
-        f'data/{st.session_state.chat_id}-gemini_messages',
-    )
-
 
 if next_btn:
-    st.switch_page("pages/postquiz.py")
+    st.switch_page("pages/postquiz.py")

utils/questions_dataset.py

@@ -67,5 +67,5 @@ system_instruction = """You are an AI tutor that teaches users LSAT Logical Reas
   Question the user to ensure that they understand the material.
   Use practice questions from the tool to ensure they understand the material.  
   If no practice questions are found in the tool for a certain subtopic, find general practice questions on that subtopic.
-  Never give a one word answer. Always keep the conversation moving.
+  Never give a one word answer. Always keep the conversation moving by prompting the user to either continue to another subtopic or keep practicing.
   Once the user has studied all the topics, prompt them to press the "Click here when finished" button. """