From 3e8ed54270fa536d40667a2fa1f325c466cf596a Mon Sep 17 00:00:00 2001
From: "Wang, Jian4" <61138589+hzjane@users.noreply.github.com>
Date: Fri, 26 Apr 2024 11:18:44 +0800
Subject: [PATCH] LLM: Fix bigdl_ipex_int8 warning (#10890)

---
 python/llm/src/ipex_llm/transformers/convert_ipex.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/python/llm/src/ipex_llm/transformers/convert_ipex.py b/python/llm/src/ipex_llm/transformers/convert_ipex.py
index a934a1dd..b36d75fe 100644
--- a/python/llm/src/ipex_llm/transformers/convert_ipex.py
+++ b/python/llm/src/ipex_llm/transformers/convert_ipex.py
@@ -138,7 +138,7 @@ def _ipex_optimize_model(model, rms_classes, qtype):
         }
         qconfig = ipex.quantization.get_weight_only_quant_qconfig_mapping(
             weight_dtype=torch.qint8,  # INT8
-            lowp_mode=ipex.quantization.WoqLowpMode.INT8,
+            lowp_mode=ipex.quantization.WoqLowpMode.BF16,
             act_quant_mode=act_quant_mode_dict["PER_IC_BLOCK"],
             group_size=-1,
         )