From 510e8394296128795df53d1a43358bd1b925a6f2 Mon Sep 17 00:00:00 2001 From: Sage Moore Date: Thu, 3 Jul 2025 16:35:52 +0000 Subject: [PATCH] more cleanup Signed-off-by: Sage Moore --- vllm/v1/worker/gpu_model_runner.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/vllm/v1/worker/gpu_model_runner.py b/vllm/v1/worker/gpu_model_runner.py index 52fba207696b6..02b632c1377ac 100644 --- a/vllm/v1/worker/gpu_model_runner.py +++ b/vllm/v1/worker/gpu_model_runner.py @@ -1553,6 +1553,8 @@ class GPUModelRunner(LoRAModelRunnerMixin): def model_inputs(tokens_slice: slice, use_dummy_input: bool) -> tuple: if use_dummy_input: + logger.info(f"NUM DUMMY TOKENS: {num_dummy_tokens} token slize: {tokens_slice}") + assert num_dummy_tokens == tokens_slice.stop - tokens_slice.start return self._get_dummy_model_inputs(num_dummy_tokens) else: assert scheduler_output is not None