diff --git a/vllm/worker/habana_model_runner.py b/vllm/worker/habana_model_runner.py
index e812a73517430..862b13e6e436a 100644
--- a/vllm/worker/habana_model_runner.py
+++ b/vllm/worker/habana_model_runner.py
@@ -1087,10 +1087,10 @@ def prepare_input_tensors(
         batch_size_padded = find_bucket(real_batch_size, bucket_cfg)
         batch_size_padding = batch_size_padded - real_batch_size
         seq_group_metadata_list = seq_group_metadata_list.copy()
-        dummy_seq_group_metadata = self.create_dummy_seq_group_metadata(0, 0, is_prompt)
-        seq_group_metadata_list.extend(
-            self.create_dummy_seq_group_metadata(dummy_seq_group_metadata
-            for _ in range(batch_size_padding))
+        dummy_seq_group_metadata = self.create_dummy_seq_group_metadata(
+            0, 0, is_prompt)
+        seq_group_metadata_list.extend(dummy_seq_group_metadata
+                                       for _ in range(batch_size_padding))
 
         prefill_reqs = []
         decode_reqs = []