chore: pump llama cpp to latest version with batch inference

tikikun · tikikun · commit 26daebc47645 · 2023-10-26T15:58:04.000+07:00
diff --git a/llama.cpp b/llama.cpp
@@ -1 +1 @@
-Subproject commit 281ef73c258cc1eebec8a64264240432d5878c4b
+Subproject commit 6961c4bd0b5176e10ab03b35394f1e9eab761792
