diff --git a/vllm/v1/worker/gpu/spec_decode/ngram/speculator.py b/vllm/v1/worker/gpu/spec_decode/ngram/speculator.py index 404f576274..7a61e3581a 100644 --- a/vllm/v1/worker/gpu/spec_decode/ngram/speculator.py +++ b/vllm/v1/worker/gpu/spec_decode/ngram/speculator.py @@ -231,6 +231,7 @@ class NgramGPUSpeculator(BaseSpeculator): skip_attn_for_dummy_run: bool = False, mm_inputs: tuple[list[torch.Tensor], torch.Tensor] | None = None, is_profile: bool = False, + num_speculative_tokens: int | None = None, ) -> torch.Tensor: num_reqs = input_batch.num_reqs if dummy_run: