Update app.py

app.py

@@ -4,6 +4,9 @@ import torch
 from groq import Groq
 from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer
 
+# ✅ Ensure set_page_config() is the first Streamlit command
+st.set_page_config(page_title="AI Study Assistant", page_icon="🤖", layout="wide")
+
 # Set up the Groq API Key
 GROQ_API_KEY = "your_groq_api_key_here"  # Replace with your actual key
 os.environ["GROQ_API_KEY"] = GROQ_API_KEY
@@ -11,37 +14,46 @@ os.environ["GROQ_API_KEY"] = GROQ_API_KEY
 # Initialize the Groq client
 client = Groq(api_key=GROQ_API_KEY)
 
-# Initialize Hugging Face DeepSeek R1 model correctly
+# ✅ Ensure Accelerate is installed
+try:
+    import accelerate  # noqa: F401
+except ImportError:
+    st.error("⚠️ `accelerate` library is required. Install it with: `pip install accelerate`")
+
+# ✅ Initialize Hugging Face DeepSeek R1 model correctly
 MODEL_NAME = "deepseek-ai/DeepSeek-R1"
 
 try:
     tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, trust_remote_code=True)
-    model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, trust_remote_code=True, torch_dtype=torch.float16, device_map="auto")
+    model = AutoModelForCausalLM.from_pretrained(
+        MODEL_NAME,
+        trust_remote_code=True,
+        torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+        device_map="auto" if torch.cuda.is_available() else None
+    )
 
     def generate_response_hf(user_message):
-        inputs = tokenizer(user_message, return_tensors="pt").to("cuda" if torch.cuda.is_available() else "cpu")
+        inputs = tokenizer(user_message, return_tensors="pt").to(model.device)
         outputs = model.generate(**inputs, max_length=200)
         return tokenizer.decode(outputs[0], skip_special_tokens=True)
 
 except Exception as e:
-    st.error(f"Error loading DeepSeek-R1: {str(e)}")
-    generate_response_hf = lambda x: "Error: Model not loaded."
+    st.error(f"❌ Error loading DeepSeek-R1: {str(e)}")
+    generate_response_hf = lambda x: "⚠️ Error: Model not loaded."
 
 # Streamlit UI setup
-st.set_page_config(page_title="AI Study Assistant", page_icon="🤖", layout="wide")
 st.title("📚 Subject-specific AI Chatbot")
 st.write("Hello! I'm your AI Study Assistant. You can ask me any questions related to your subjects, and I'll try to help.")
 
-# Add sidebar with settings
+# Sidebar settings
 st.sidebar.header("⚙️ Settings")
-st.sidebar.write("Customize your chatbot experience!")
 chat_model = st.sidebar.radio("Choose AI Model:", ["Groq API", "DeepSeek R1 (Hugging Face)"])
 
-# Initialize session state for maintaining conversation
+# Initialize session state for conversation
 if 'conversation_history' not in st.session_state:
     st.session_state.conversation_history = []
 
-# Define subjects
+# Subjects list
 subjects = ["Chemistry", "Computer", "English", "Islamiat", "Mathematics", "Physics", "Urdu"]
 
 def generate_chatbot_response(user_message):
@@ -61,7 +73,7 @@ def generate_chatbot_response(user_message):
     else:
         return generate_response_hf(prompt)
 
-# User input
+# Chat input
 st.markdown("### 💬 Chat with me")
 user_input = st.chat_input("Ask me a subject-related question:")