speculative : back to n_draft == 16

ggerganov · Aug 31, 2023 · d49869d · d49869d
1 parent ec5086f
commit d49869d
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/examples/speculative/speculative.cpp b/examples/speculative/speculative.cpp
@@ -78,7 +78,7 @@ int main(int argc, char ** argv) {
     //GGML_ASSERT(n_vocab == llama_n_vocab(ctx_dft));
 
     // how many tokens to draft each time
-    const int n_draft = 8;
+    const int n_draft = 16;
 
     int n_predict = 0;
     int n_drafted = 0;