Update app.py

app.py

@@ -1,39 +1,75 @@
 import streamlit as st
 from llama_cpp import Llama
+import os
+import json
+import time
 
-# Title for your app
-st.title("Llama-3-8B-Physics Master - Chatbot")
+# Function to convert message history to prompt
+def prompt_from_messages(messages):
+    prompt = ''
+    for message in messages:
+        prompt += f"<|start_header_id|>{message['role']}<|end_header_id|>\n\n"
+        prompt += f"{message['content']}<|eot_id|>"
+    prompt = prompt[:-10]
+    return prompt
 
-# Load the model from Hugging Face using llama_cpp
-@st.cache_resource
-def load_model():
-    # Load the model from the Hugging Face Hub
-    model = Llama.from_pretrained(
-        repo_id="gallen881/Llama-3-8B-Physics_Master-GGUF",
-        filename= "unsloth.Q4_K_M.gguf" # or unsloth.F16.gguf for a larger file
-    )
-    return model
+# Initialize the Llama model
+llm = Llama.from_pretrained(
+    repo_id='gallen881/Llama-3-8B-Physics_Master-GGUF',
+    filename='unsloth.Q4_K_M.gguf',
+    n_ctx=2048,
+    verbose=False
+)
+
+# Set up Streamlit App Layout
+st.title("Physics Master Chatbot")
+st.markdown("Ask **Physics Master** any physics-related question.")
 
-# Load the model once and store it in cache
-model = load_model()
+# Initialize chat history in session state
+if 'messages' not in st.session_state:
+    st.session_state.messages = [
+        {
+            'role': 'system',
+            'content': 'You are a professional physics master. Answer physics questions directly without using any external resources.'
+        }
+    ]
+    st.session_state.chat_time = time.time()
 
-# Text input for the user
-user_input = st.text_area("Enter your message here:")
+# Display chat history
+for message in st.session_state.messages:
+    if message['role'] == 'user':
+        st.write(f"**You:** {message['content']}")
+    else:
+        st.write(f"**Physics Master:** {message['content']}")
 
-if st.button("Generate Response"):
-    if user_input:
-        # Create chat completion with the model
-        response = model.create_chat_completion(
-            messages=[
-                {
-                    "role": "user",
-                    "content": user_input
-                }
-            ]
-        )
+# Input for user to ask questions
+user_input = st.text_input("Ask a question", key="user_input")
 
-        # Extract the content from the model's response
-        st.write("Model Response:", response['choices'][0]['message']['content'])
+if user_input:
+    # Append user message
+    user_message = {'role': 'user', 'content': user_input}
+    st.session_state.messages.append(user_message)
 
-    else:
-        st.write("Please enter a message.")
+    # Prepare to get the response from Physics Master
+    st.write('Physics Master is thinking...')
+    response = llm.create_chat_completion(
+        messages=st.session_state.messages,
+        stream=True
+    )
+
+    # Handle the response from the model and append it to the session state
+    for chunk in response:
+        delta = chunk['choices'][0]['delta']
+        if 'role' in delta:
+            st.session_state.messages.append({'role': delta['role'], 'content': ''})
+        elif 'content' in delta:
+            token = delta['content']
+            st.session_state.messages[-1]['content'] += token
+            st.write(token, end="", flush=True)
+
+    # Reset user input box
+    st.session_state.user_input = ""
+
+# Save the chat history to a JSON file
+with open('chat_history.json', 'w', encoding='utf8') as file:
+    json.dump(st.session_state.messages, file, indent=4)