diff --git a/vllm/worker/habana_model_runner.py b/vllm/worker/habana_model_runner.py index c43acdf04923..f3bda39ec482 100644 --- a/vllm/worker/habana_model_runner.py +++ b/vllm/worker/habana_model_runner.py @@ -1790,6 +1790,7 @@ def make_model_input_from_broadcasted_tensor_dict( attn_backend=self.attn_backend, )) + @torch.inference_mode() def prepare_model_input( self, seq_group_metadata_list: List[SequenceGroupMetadata],