From e451045a66d4e73e40fe9d346663b9fb397ab4f8 Mon Sep 17 00:00:00 2001 From: Woosuk Kwon Date: Thu, 28 Aug 2025 12:55:13 -0700 Subject: [PATCH] fix Signed-off-by: Woosuk Kwon --- 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 14f78f927df43..37df18b231057 100644 --- a/vllm/v1/worker/gpu_model_runner.py +++ b/vllm/v1/worker/gpu_model_runner.py @@ -2270,6 +2270,8 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin): bad_words_token_ids={}, logitsprocs=LogitsProcessors(), token_ids=None, + num_tokens=None, + num_prompt_tokens=None, ) try: sampler_output = self.sampler(logits=logits,