fix: Change chat_format from 'qwen2' to 'qwen'

The llama-cpp-python library doesn't support 'qwen2' format yet, use 'qwen' instead

model_manager.py

@@ -80,7 +80,7 @@ class SharedModelManager:
                     n_ctx=4096,
                     n_threads=4,
                     verbose=False,
-                    chat_format='qwen2'
+                    chat_format='qwen'  # Changed from 'qwen2' to 'qwen'
                 )
                 
                 self.model_path = model_path