diff --git a/vllm/engine/arg_utils.py b/vllm/engine/arg_utils.py index 647793a6d5cd..7f760c2775f3 100644 --- a/vllm/engine/arg_utils.py +++ b/vllm/engine/arg_utils.py @@ -577,7 +577,7 @@ def add_cli_args( 'This should be a JSON string that will be ' 'parsed into a dictionary.') parser.add_argument( - '--preemption_mode', + '--preemption-mode', type=str, default=None, help='If \'recompute\', the engine performs preemption by block '