diff --git a/ggml/src/ggml-backend.cpp b/ggml/src/ggml-backend.cpp
index fec7d7c92..8f40fb2f7 100644
--- a/ggml/src/ggml-backend.cpp
+++ b/ggml/src/ggml-backend.cpp
@@ -70,6 +70,7 @@ size_t ggml_backend_buft_get_alloc_size(ggml_backend_buffer_type_t buft, const s
         // if you hit this assert, update ggml_backend_op_alloc_size_may_expand() accordingly
         GGML_ASSERT(size <= ggml_nbytes(tensor) ||
                     ggml_op_is_empty(tensor->op) ||
+                    ggml_is_quantized(tensor->type) || // [TAG_ALLOC_SIZE_EXPAND]
                     ggml_backend_op_alloc_size_may_expand(tensor->op));
 
         return size;
diff --git a/ggml/src/ggml-cuda/ggml-cuda.cu b/ggml/src/ggml-cuda/ggml-cuda.cu
index 7c2d3cad2..a6fc655c4 100644
--- a/ggml/src/ggml-cuda/ggml-cuda.cu
+++ b/ggml/src/ggml-cuda/ggml-cuda.cu
@@ -915,6 +915,7 @@ static size_t ggml_backend_cuda_buffer_type_get_alloc_size(ggml_backend_buffer_t
         : ggml_nbytes(tensor);
     int64_t ne0 = tensor->ne[0];
 
+    // [TAG_ALLOC_SIZE_EXPAND]
     if (ggml_is_quantized(tensor->type)) {
         if (ne0 % MATRIX_ROW_PADDING != 0) {
             GGML_ASSERT(tensor->nb[0] == ggml_element_size(tensor));