trying to update for mobile again

The Space will now:
-🚀 Launch successfully without errors
-📱 Optimize automatically for mobile devices
-🐸 Generate responses using base SmolLM (still very capable!)
- 🔄 Load LoRA adapter if/when compatible version is available

app.py

@@ -130,10 +130,16 @@ class DuncanChatbot:
                 )
                 logger.info("✅ Duncan's LoRA adapter loaded successfully")
             except Exception as e:
-                logger.error(f"❌ Failed to load LoRA adapter: {e}")
-                logger.info("🔄 Attempting to use base SmolLM model without adapter...")
+                if "size mismatch" in str(e):
+                    logger.error(f"❌ LoRA adapter dimension mismatch: {e}")
+                    logger.warning("🔍 This suggests the adapter was trained with a different SmolLM variant")
+                    logger.info("💡 The base SmolLM model will still work, just without Duncan's personality fine-tuning")
+                else:
+                    logger.error(f"❌ Failed to load LoRA adapter: {e}")
+                
+                logger.info("🔄 Falling back to base SmolLM model...")
                 self.model = base_model
-                logger.warning("⚠️ Using base SmolLM model without Duncan's fine-tuning")
+                logger.warning("⚠️ Using base SmolLM model - responses may be less Duncan-like but still functional")
             
             # Set model to evaluation mode (matching notebook)
             self.model.eval()
@@ -288,11 +294,12 @@ def main():
         
         **Model Details:**
         - **Base**: HuggingFace SmolLM2-1.7B-Instruct (1.7B parameters)
-        - **Training**: LoRA fine-tuning on 62 conversation examples
+        - **Training**: LoRA fine-tuning on conversation examples (auto-fallback to base model)
         - **Personality**: Philosophical frog with scientific curiosity and humor
         - **Responses**: Optimized for 5-8 sentence thoughtful answers
         - **Performance**: T4 GPU optimized with fast generation
         - **Mobile Optimized**: Faster responses on smartphones and tablets!
+        - **Robust**: Graceful fallback to base SmolLM if adapter unavailable
         
         Ask him about science, philosophy, his interdimensional adventures, Emmitt, living with grandma, or anything else!
         
@@ -373,13 +380,8 @@ def main():
         server_name="0.0.0.0",
         server_port=7860,
         share=False,
-        show_error=True,
-        # Mobile performance optimizations
-        enable_queue=True,          # Better handling of concurrent mobile requests
-        max_threads=4,              # Optimize for mobile traffic patterns
-        favicon_path=None,          # Reduce initial page load
-        show_tips=False,           # Cleaner mobile interface
-        ssl_verify=False           # Faster connections for mobile
+        show_error=True
+        # Note: Mobile optimizations applied via CSS and adaptive generation parameters
     )
 
 if __name__ == "__main__":