From d40fded93e1a533e969768e1e335c15c61c296ce Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 21 Apr 2023 10:23:36 +0300
Subject: [PATCH] llama : fix comment for "output.weight" tensor

---
 llama.cpp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 4a646eb..33ee4fb 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1618,8 +1618,8 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
         // quantize only 2D tensors
         quantize &= (tensor.ne.size() == 2);
 
-        // GG: uncomment this to keep the output layer in FP16
-        //if (tensor.name.rfind("output")) {
+        // uncomment this to keep the output layer in FP16
+        //if (tensor.name == "output.weight") {
         //    quantize = false;
         //}