1 parent f2901e3 commit f081394Copy full SHA for f081394
1 file changed
llama.ai/ai.conducts
@@ -0,0 +1,4 @@
1
+1. GPU Offloading — the #1 speed booster
2
+Set -ngl to a number higher than your model’s layer count.
3
+
4
+-ngl 99
0 commit comments