diff --git a/llama.h b/llama.h
index b662d6ec5..8d26e7088 100644
--- a/llama.h
+++ b/llama.h
@@ -102,7 +102,7 @@ extern "C" {
         bool use_mmap;   // use mmap if possible
         bool use_mlock;  // force system to keep model in RAM
         bool embedding;  // embedding mode only
-        bool has_lora;   // model has LoRA layers
+        bool has_lora;   // a LoRA is being used
     };
     // model file types
     enum llama_ftype {