@@ -150,8 +150,8 @@ struct llama_model {
std::vector<uint8_t> buf;
// model memory mapped file
- void * mm_addr;
- size_t mm_length;
+ void * mm_addr = NULL;
+ size_t mm_length = 0;
// tensors
int n_loaded;