Fix VulkanBackend to accept original_backend parameter

codai/backends/vulkan.py

@@ -8,10 +8,19 @@ from codai.backends.base import ModelBackend
 class VulkanBackend(ModelBackend):
     """Backend for Vulkan GPU inference using llama.cpp."""
     
-    def __init__(self):
+    def __init__(self, original_backend: str = None):
         self.model = None
         self.model_name = None
         self.device = None
+        self.n_gpu_layers = -1
+        self.n_ctx = 2048
+        self.verbose = True
+        self.main_gpu = 0
+        self.chat_template = None
+        self.hf_tokenizer = None
+        self.force_cuda = original_backend in ("nvidia", "cuda")
+        if self.force_cuda:
+            print("DEBUG: GGUF model will use CUDA backend (forced by --backend nvidia)")
     
     def load_model(self, model_name: str, **kwargs) -> None:
         """Load the model."""
@@ -23,6 +32,20 @@ class VulkanBackend(ModelBackend):
         """Generate text non-streaming."""
         pass
     
+    def generate_chat(self, messages: List[Dict], max_tokens: Optional[int] = None,
+                      temperature: float = 0.7, top_p: float = 1.0,
+                      stop: Optional[List[str]] = None, tools: Optional[List] = None,
+                      response_format: Optional[Dict] = None) -> str:
+        """Generate chat completion non-streaming."""
+        pass
+    
+    async def generate_chat_stream(self, messages: List[Dict], max_tokens: Optional[int] = None,
+                                    temperature: float = 0.7, top_p: float = 1.0,
+                                    stop: Optional[List[str]] = None, tools: Optional[List] = None,
+                                    response_format: Optional[Dict] = None):
+        """Generate chat completion streaming."""
+        pass
+    
     def generate_stream(self, prompt: str, max_tokens: Optional[int] = None,
                         temperature: float = 0.7, top_p: float = 1.0,
                         stop: Optional[list] = None):