[vulkan] Update vmfb naming

Update vmfb naming for vulkan devices in order to resolve naming
conflicts in the presence of multiple vulkan devices.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>

apps/language_models/scripts/vicuna.py

@@ -1220,6 +1220,7 @@ def __init__(
         hf_auth_token: str = None,
         max_num_tokens=512,
         device="cpu",
+        vulkan_target_triple="",
         precision="int8",
         vicuna_mlir_path=None,
         vicuna_vmfb_path=None,
@@ -1248,6 +1249,7 @@ def __init__(
         print(f"[DEBUG] hf model name: {self.hf_model_path}")
         self.max_sequence_length = 256
         self.device = device
+        self.vulkan_target_triple = vulkan_target_triple
         self.device_id = device_id
         self.precision = precision
         self.download_vmfb = download_vmfb
@@ -1269,8 +1271,9 @@ def get_model_path(self, suffix="mlir"):
         safe_device = self.device.split("-")[0]
         if suffix in ["mlirbc", "mlir"]:
             return Path(f"{self.model_name}_{self.precision}.{suffix}")
+        target_triple = self.vulkan_target_triple=="" ? "" : f"_{self.vulkan_target_triple}"
         return Path(
-            f"{self.model_name}_{self.precision}_{safe_device}.{suffix}"
+            f"{self.model_name}_{self.precision}_{safe_device}{target_triple}.{suffix}"
         )
 
     def get_tokenizer(self):

apps/stable_diffusion/web/ui/stablelm_ui.py

@@ -228,6 +228,7 @@ def chat(
                 hf_model_path=model_path,
                 hf_auth_token=args.hf_auth_token,
                 device=device,
+                vulkan_target_triple=vulkan_target_triple,
                 precision=precision,
                 max_num_tokens=max_toks,
                 download_vmfb=download_vmfb,