From 9ed626410fb1e98de82a72ff1aa80800638a725b Mon Sep 17 00:00:00 2001
From: Nexesenex <124105151+Nexesenex@users.noreply.github.com>
Date: Fri, 22 Mar 2024 14:32:02 +0100
Subject: [PATCH] llama : correction of the attn.v.weight quantization for
 IQ3_XS (#6209)

IQ3_XS was not mentioned, IQ3_S and IQ3_M were present twice.

That PR corrects this in the manner which was probably intended initially.
---
 llama.cpp | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 9de4a86022b1a9..91bd6b8d070e9f 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -12027,13 +12027,7 @@ static ggml_type llama_tensor_get_type(quantize_state_internal & qs, ggml_type n
         else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_XXS) {
             new_type = qs.model.hparams.n_gqa() >= 4 ? GGML_TYPE_Q4_K : !qs.has_imatrix ? GGML_TYPE_IQ3_S : GGML_TYPE_IQ3_XXS;
         }
-        else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_S && qs.model.hparams.n_gqa() >= 4) {
-            new_type = GGML_TYPE_Q4_K;
-        }
-        else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_M) {
-            new_type = GGML_TYPE_Q4_K;
-        }
-        else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_S && qs.model.hparams.n_gqa() >= 4) {
+        else if ((ftype == LLAMA_FTYPE_MOSTLY_IQ3_XS || ftype == LLAMA_FTYPE_MOSTLY_IQ3_S) && qs.model.hparams.n_gqa() >= 4) {
             new_type = GGML_TYPE_Q4_K;
         }
         else if (ftype == LLAMA_FTYPE_MOSTLY_IQ3_M) {