diff --git a/python/sglang/srt/model_executor/cuda_graph_runner.py b/python/sglang/srt/model_executor/cuda_graph_runner.py index 3ef9b35ed..23090688d 100644 --- a/python/sglang/srt/model_executor/cuda_graph_runner.py +++ b/python/sglang/srt/model_executor/cuda_graph_runner.py @@ -290,7 +290,7 @@ class CudaGraphRunner: index = bisect.bisect_left(self.capture_bs, raw_bs) bs = self.capture_bs[index] if bs != raw_bs: - self.seq_lens.fill_(self.seq_len_fill_value) + self.seq_lens.fill_(1) self.out_cache_loc.zero_() # Common inputs