Fix NeuralChat starcoder mha fusion issue (#494)

Signed-off-by: Wang, Chang <chang1.wang@intel.com>
intel · Oct 19, 2023 · ce3d24e · ce3d24e
1 parent e133632
commit ce3d24e
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/intel_extension_for_transformers/neural_chat/models/model_utils.py b/intel_extension_for_transformers/neural_chat/models/model_utils.py
@@ -297,7 +297,7 @@ def load_model(
         )
 
         adapt_transformers_to_gaudi()
-    elif device == "cpu":
+    elif device == "cpu" and not ipex_int8:
         set_cpu_running_env()
 
     if isinstance(optimization_config, AMPConfig):