diff --git a/llama_cpp/llama_cpp.py b/llama_cpp/llama_cpp.py
index 431a99f..da2a7f3 100644
--- a/llama_cpp/llama_cpp.py
+++ b/llama_cpp/llama_cpp.py
@@ -445,7 +445,7 @@ class llama_model_params(Structure):
 #     uint32_t n_batch;           // prompt processing maximum batch size
 #     uint32_t n_threads;         // number of threads to use for generation
 #     uint32_t n_threads_batch;   // number of threads to use for batch processing
-#     int8_t   rope_scaling_type; // RoPE scaling type, from `enum llama_rope_scaling_type`
+#     int32_t  rope_scaling_type; // RoPE scaling type, from `enum llama_rope_scaling_type`
 
 #     // ref: https://github.com/ggerganov/llama.cpp/pull/2054
 #     float    rope_freq_base;   // RoPE base frequency, 0 = from model
@@ -502,7 +502,7 @@ class llama_context_params(Structure):
         ("n_batch", c_uint32),
         ("n_threads", c_uint32),
         ("n_threads_batch", c_uint32),
-        ("rope_scaling_type", c_int8),
+        ("rope_scaling_type", c_int32),
         ("rope_freq_base", c_float),
         ("rope_freq_scale", c_float),
         ("yarn_ext_factor", c_float),
diff --git a/vendor/llama.cpp b/vendor/llama.cpp
index 1912211..78b00dd 160000
--- a/vendor/llama.cpp
+++ b/vendor/llama.cpp
@@ -1 +1 @@
-Subproject commit 191221178f51b6e81122c5bda0fd79620e547d07
+Subproject commit 78b00dda6c0d62c34f5371d47718defff6ed2b22