diff --git a/jenkins/L0_Test.groovy b/jenkins/L0_Test.groovy
index 112e30b997..a2e3cdd664 100644
--- a/jenkins/L0_Test.groovy
+++ b/jenkins/L0_Test.groovy
@@ -380,7 +380,7 @@ def runLLMTestlistOnSlurm(pipeline, platform, testList, config=VANILLA_CONFIG, p
                     // Wait 10 minutes to check status of the node again
                     sleep(time: 10, unit: 'MINUTES')
                     // Avoid the node being stuck in the held state.
-                    Utils.exec(pipeline, Utils.sshUserCmd(remote, "\"scontrol release ${slurmJobID} || true\""))
+                    Utils.exec(pipeline, script: Utils.sshUserCmd(remote, "\"scontrol release ${slurmJobID} || true\""), numRetries: 3)
                     counter++
                 }
             }
diff --git a/tests/integration/defs/accuracy/references/mmlu.yaml b/tests/integration/defs/accuracy/references/mmlu.yaml
index 4d90217b27..430c5a1821 100644
--- a/tests/integration/defs/accuracy/references/mmlu.yaml
+++ b/tests/integration/defs/accuracy/references/mmlu.yaml
@@ -208,10 +208,10 @@ Qwen3/Qwen3-30B-A3B:
     accuracy: 79.53
   - quant_algo: FP8
     kv_cache_quant_algo: FP8
-    accuracy: 80.65
+    accuracy: 79.53
   - quant_algo: NVFP4
     kv_cache_quant_algo: FP8
-    accuracy: 80.65
+    accuracy: 79.53
   - quant_algo: W4A8_MXFP4_FP8
     accuracy: 79.78
   - quant_algo: W4A8_MXFP4_MXFP8