From 405578121cfd05f9b2e0ba717a465c997b5589a2 Mon Sep 17 00:00:00 2001 From: Woosuk Kwon Date: Thu, 28 Aug 2025 13:19:10 -0700 Subject: [PATCH] minor Signed-off-by: Woosuk Kwon --- vllm/v1/worker/gpu_model_runner.py | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/vllm/v1/worker/gpu_model_runner.py b/vllm/v1/worker/gpu_model_runner.py index ad8312c86d2e7..cf8eebf4dc2c4 100644 --- a/vllm/v1/worker/gpu_model_runner.py +++ b/vllm/v1/worker/gpu_model_runner.py @@ -559,10 +559,8 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin): # batch_idx -> req_id req_ids = sorted(scheduler_output.num_scheduled_tokens, key=scheduler_output.num_scheduled_tokens.get) - # req_id -> batch_idx req_id_to_batch_idx = {req_id: i for i, req_id in enumerate(req_ids)} - # batch_idx -> req_idx idx_mapping_list = [ self.requests.req_id_to_index[req_id] for req_id in req_ids @@ -633,7 +631,7 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin): if draft_token_ids: num_draft_tokens[i] = len(draft_token_ids) spec_decode_metadata = self._calc_spec_decode_metadata( - num_draft_tokens, self.query_start_loc_np[1:num_reqs + 1]) + num_draft_tokens, self.query_start_loc.np[1:num_reqs + 1]) logits_indices = spec_decode_metadata.logits_indices logits_indices_padded = None