From 20ed3822bb6dc430552ae8b160de64e86f770e1f Mon Sep 17 00:00:00 2001 From: Lianmin Zheng Date: Tue, 20 Jan 2026 14:05:05 -0800 Subject: [PATCH] [Auto Sync] Update piecewise_cuda_graph_runner.py (20260119) (#17313) Co-authored-by: github-actions[bot] Co-authored-by: Binyao Jiang --- .../srt/model_executor/piecewise_cuda_graph_runner.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/python/sglang/srt/model_executor/piecewise_cuda_graph_runner.py b/python/sglang/srt/model_executor/piecewise_cuda_graph_runner.py index 1087b3b7f..a6bd7ab56 100644 --- a/python/sglang/srt/model_executor/piecewise_cuda_graph_runner.py +++ b/python/sglang/srt/model_executor/piecewise_cuda_graph_runner.py @@ -330,9 +330,9 @@ class PiecewiseCudaGraphRunner: return_logprob=False, extend_num_tokens=num_tokens, extend_seq_lens=torch.tensor([num_tokens], device=self.device), - extend_prefix_lens=torch.tensor([num_tokens], device=self.device), + extend_prefix_lens=torch.tensor([0], device=self.device), extend_start_loc=torch.tensor([0], device=self.device), - extend_prefix_lens_cpu=torch.tensor([num_tokens], device="cpu"), + extend_prefix_lens_cpu=torch.tensor([0], device="cpu"), extend_seq_lens_cpu=torch.tensor([num_tokens], device="cpu"), extend_logprob_start_lens_cpu=torch.tensor([num_tokens], device="cpu"), positions=positions, @@ -477,9 +477,9 @@ class PiecewiseCudaGraphRunner: return_logprob=False, extend_num_tokens=num_tokens, extend_seq_lens=torch.tensor([num_tokens], device=self.device), - extend_prefix_lens=torch.tensor([num_tokens], device=self.device), + extend_prefix_lens=torch.tensor([0], device=self.device), extend_start_loc=torch.tensor([0], device=self.device), - extend_prefix_lens_cpu=torch.tensor([num_tokens], device="cpu"), + extend_prefix_lens_cpu=torch.tensor([0], device="cpu"), extend_seq_lens_cpu=torch.tensor([num_tokens], device="cpu"), extend_logprob_start_lens_cpu=torch.tensor([num_tokens], device="cpu"), positions=positions,