From: Johannes Gäßler <redacted>
Date: Wed, 30 Aug 2023 19:46:19 +0000 (+0200)
Subject: CUDA: mul_mat_q=true llama_context_params default (#2912)
X-Git-Tag: gguf-v0.4.0~176
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=8afe2280009ecbfc9de2c93b8f41283dc810609a;p=pkg%2Fggml%2Fsources%2Fllama.cpp

CUDA: mul_mat_q=true llama_context_params default (#2912)
---

diff --git a/llama.cpp b/llama.cpp
index fcd6f276..95ee6ffe 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -5287,7 +5287,7 @@ struct llama_context_params llama_context_default_params() {
         /*.progress_callback           =*/ nullptr,
         /*.progress_callback_user_data =*/ nullptr,
         /*.low_vram                    =*/ false,
-        /*.mul_mat_q                   =*/ false,
+        /*.mul_mat_q                   =*/ true,
         /*.f16_kv                      =*/ true,
         /*.logits_all                  =*/ false,
         /*.vocab_only                  =*/ false,