Update app.py

app.py

@@ -53,6 +53,7 @@ st.header("Review Scorer & Report Generator")
 # Concise introduction
 st.write("This model will score your reviews in your CSV file and generate a report based on your query and those results.")
 
+@st.cache_resource
 def display_temp_message(message, message_type="info", duration=5):
     """Display a temporary message that disappears after specified duration."""
     placeholder = st.empty()
@@ -64,68 +65,54 @@ def display_temp_message(message, message_type="info", duration=5):
     elif message_type == "error":
         placeholder.error(message)
     
-    # Wait and then clear the message
-    time.sleep(duration)
-    placeholder.empty()
-    
-# Load models with caching to avoid reloading on every run
-def load_model_with_messages(loading_message, success_message, loading_function, error_message_prefix):
-    """Load a model with temporary loading and success messages."""
-    # Create placeholder for the loading message
-    loading_placeholder = st.empty()
-    loading_placeholder.info(loading_message)
-    
-    try:
-        # Load the model
-        result = loading_function()
-        
-        # Clear the loading message
-        loading_placeholder.empty()
-        
-        # Show temporary success message
-        Thread(
-            target=display_temp_message, 
-            args=(success_message, "success"), 
-            daemon=True
-        ).start()
-        
-        return result
-        
-    except Exception as e:
-        # Clear the loading message
-        loading_placeholder.empty()
-        
-        # Show error message
-        st.error(f"{error_message_prefix}: {e}")
-        st.error(f"Detailed error: {type(e).__name__}: {str(e)}")
-        return None
+    # Create a timer to clear the message after duration
+    timer = threading.Timer(duration, placeholder.empty)
+    timer.daemon = True
+    timer.start()
 
-# Define loading functions
-def load_llama_model():
-    return pipeline("text-generation", 
-                   model="meta-llama/Llama-3.2-1B-Instruct",
-                   device=0,  # Use GPU if available
-                   torch_dtype=torch.bfloat16)  # Use FP16 for efficiency
+# Load Llama 3.2 model
+loading_llama_placeholder = st.empty()
+loading_llama_placeholder.info("Loading Llama 3.2 summarization model...")
 
-def load_sentiment_model():
-    return pipeline("text-classification", 
-                   model="cardiffnlp/twitter-roberta-base-sentiment-latest", 
-                   device=0 if torch.cuda.is_available() else -1)
+try:
+    llama_pipe = pipeline("text-generation", 
+                         model="meta-llama/Llama-3.2-1B-Instruct",
+                         device=0,  # Use GPU if available
+                         torch_dtype=torch.bfloat16)  # Use FP16 for efficiency
+    
+    # Clear loading message
+    loading_llama_placeholder.empty()
+    
+    # Display success message that will disappear after 5 seconds
+    display_temp_message("Llama 3.2 summarization model loaded successfully!", "success")
+    
+except Exception as e:
+    # Clear loading message
+    loading_llama_placeholder.empty()
+    
+    st.error(f"Error loading Llama 3.2 summarization model: {e}")
+    st.error(f"Detailed error: {type(e).__name__}: {str(e)}")
 
-# Load models with temporary messages
-llama_pipe = load_model_with_messages(
-    "Loading Llama 3.2 summarization model...",
-    "Llama 3.2 summarization model loaded successfully!",
-    load_llama_model,
-    "Error loading Llama 3.2 summarization model"
-)
+# Load sentiment analysis model
+loading_sentiment_placeholder = st.empty()
+loading_sentiment_placeholder.info("Loading sentiment analysis model...")
 
-score_pipe = load_model_with_messages(
-    "Loading sentiment analysis model...",
-    "Sentiment analysis model loaded successfully!",
-    load_sentiment_model,
-    "Error loading sentiment analysis model"
-)
+try:
+    score_pipe = pipeline("text-classification", 
+                         model="cardiffnlp/twitter-roberta-base-sentiment-latest", 
+                         device=0 if torch.cuda.is_available() else -1)
+    
+    # Clear loading message
+    loading_sentiment_placeholder.empty()
+    
+    # Display success message that will disappear after 5 seconds
+    display_temp_message("Sentiment analysis model loaded successfully!", "success")
+    
+except Exception as e:
+    # Clear loading message
+    loading_sentiment_placeholder.empty()
+    
+    st.error(f"Error loading sentiment analysis model: {e}")
 
 def extract_assistant_content(raw_response):
     """Extract only the assistant's content from the Gemma-3 response."""