minor fixes

app.py

@@ -9,7 +9,6 @@ from src.session.session_manager import SessionManager
 from src.ui.template_renderer import TemplateRenderer
 from src.ui.page_handlers import PageHandlers
 from src.ui.interface_builder import InterfaceBuilder
-from src.config.settings import DEFAULT_SERVER_NAME, DEFAULT_SERVER_PORT, DEFAULT_SHARE
 
 class CollaborativeDecodingApp:
     def __init__(self):

src/models/llm_manager.py

@@ -86,6 +86,9 @@ class LLMManager:
                 temperature=1.0,
                 pad_token_id=self.tokenizer.eos_token_id
             )
+
+        # Move output back to CPU and decode
+        outputs = outputs.cpu()
         
         full_response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
         assistant_part = full_response.split("Assistant: ")[-1]