From 6c73f704ca978c5e76cfe69a3d87eaac5996c327 Mon Sep 17 00:00:00 2001
From: Nexes the Elder <124105151+Nexesenex@users.noreply.github.com>
Date: Thu, 21 Nov 2024 07:12:57 +0100
Subject: [PATCH] Use Q6_0 instead of Q5_1 for tensors incompatible with
 IQ5_K/Q5_K (#116)

---
 src/llama.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/llama.cpp b/src/llama.cpp
index 2b9a1b1a..61448319 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -16048,7 +16048,7 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n
             case GGML_TYPE_IQ4_K:
             case GGML_TYPE_Q4_K:   new_type = GGML_TYPE_Q5_0;   break;
             case GGML_TYPE_IQ5_K:
-            case GGML_TYPE_Q5_K:   new_type = GGML_TYPE_Q5_1;   break;
+            case GGML_TYPE_Q5_K:   new_type = GGML_TYPE_Q6_0;   break;
             case GGML_TYPE_IQ6_K:
             case GGML_TYPE_Q6_K:   new_type = GGML_TYPE_Q8_0;   break;
             default: throw std::runtime_error("\nUnsupported tensor size encountered\n");