diff --git a/vllm/worker/hpu_worker.py b/vllm/worker/hpu_worker.py
index 8bf1467496a82..3c570212625c4 100644
--- a/vllm/worker/hpu_worker.py
+++ b/vllm/worker/hpu_worker.py
@@ -143,7 +143,7 @@ def execute_model(
             'VLLM_HPU_LOG_STEP_CPU_FALLBACKS_ALL', '0') != '0'
         log_cpu_fallbacks = os.environ.get('VLLM_HPU_LOG_STEP_CPU_FALLBACKS',
                                            '0') != '0' or log_cpu_fallbacks_all
-        if log_graph_compilation or log_cpu_fallbacks and \
+        if (log_graph_compilation or log_cpu_fallbacks) and \
             execute_model_req is not None:
             from habana_frameworks.torch.hpu.metrics import metric_localcontext
             seq_group_metadata_list = execute_model_req.seq_group_metadata_list