From f63ed7b5aca634b23d070b9cd9f654f0c74b65ad Mon Sep 17 00:00:00 2001 From: Benjamin Chislett Date: Tue, 17 Mar 2026 11:16:48 -0400 Subject: [PATCH] [Bugfix] Fix DP MTP Dummy Run (#35243) Signed-off-by: Benjamin Chislett --- vllm/v1/worker/gpu_worker.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/vllm/v1/worker/gpu_worker.py b/vllm/v1/worker/gpu_worker.py index 58e2d658c..30286d133 100644 --- a/vllm/v1/worker/gpu_worker.py +++ b/vllm/v1/worker/gpu_worker.py @@ -903,7 +903,8 @@ class Worker(WorkerBase): self.profiler.stop() def execute_dummy_batch(self) -> None: - self.model_runner._dummy_run(1, uniform_decode=True) + num_tokens = getattr(self.model_runner, "uniform_decode_query_len", 1) + self.model_runner._dummy_run(num_tokens, uniform_decode=True) def add_lora(self, lora_request: LoRARequest) -> bool: return self.model_runner.add_lora(lora_request)