diff --git a/.github/workflows/llm-harness-evaluation.yml b/.github/workflows/llm-harness-evaluation.yml
index b91371db..fe0e0f4f 100644
--- a/.github/workflows/llm-harness-evaluation.yml
+++ b/.github/workflows/llm-harness-evaluation.yml
@@ -26,10 +26,10 @@ jobs:
       fail-fast: false
       matrix:
         python-version: ["3.9"]
-        model_name: ["Llama-2-7b-chat-hf"]
+        model_name: [stablelm-3b-4e1t]
         task: ["truthfulqa"]
         precision: ["int4"]
-    runs-on: [self-hosted, llm, accuracy]
+    runs-on: [self-hosted, llm, accuracy, temp-arc01]
     env:
       ANALYTICS_ZOO_ROOT: ${{ github.workspace }}
     steps:
@@ -95,7 +95,6 @@ jobs:
         shell: bash
         run: |
           export USE_XETLA=OFF
-          export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
           source /opt/intel/oneapi/setvars.sh
           cd python/llm/dev/benchmark/harness
           python llb.py --model bigdl-llm --pretrained ${MODEL_PATH} --precision ${{ matrix.precision }} --device xpu --tasks ${{ matrix.task }} --output_dir results/${{ matrix.model_name }} --batch 1