Try to stop the model if it was loaded when ctrl-c received

modules/llama_cpp_server.py

@@ -398,6 +398,7 @@ class LlamaServer:
     def stop(self):
         """Stop the server process."""
         if self.process:
+            logger.info("Terminating llama-server...")
             self.process.terminate()
             try:
                 self.process.wait(timeout=5)

server.py

@@ -60,6 +60,13 @@ from modules.utils import gradio
 
 def signal_handler(sig, frame):
     logger.info("Received Ctrl+C. Shutting down Text generation web UI gracefully.")
+
+    # Try to stop the model if loaded
+    try:
+        shared.model.stop()
+    except:
+        pass
+
     sys.exit(0)