fixes

Author: okaris

llama_cpp/llama_cpp.py

@@ -10,6 +10,7 @@
     NewType,
     Optional,
     TYPE_CHECKING,
+    List,
 )
 
 from llama_cpp._ctypes_extensions import (
@@ -660,6 +661,12 @@ class llama_model_kv_override(ctypes.Structure):
 #     ggml_backend_buffer_type_t buft;
 # };
 
+class llama_model_tensor_buft_override(ctypes.Structure):
+    _fields_ = [
+        ("pattern", ctypes.c_char_p),
+        ("buft", ctypes.c_int),
+    ]
+
 
 # struct llama_model_params {
 #     // NULL-terminated list of devices to use for offloading (if NULL, all available devices are used)