From 6f1509b2a248449ed0fc54861007cd85c91c2b4f Mon Sep 17 00:00:00 2001 From: Geon Park Date: Wed, 5 Feb 2025 22:52:44 +0900 Subject: [PATCH] fix bug --- python/sglang/srt/model_executor/cuda_graph_runner.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/python/sglang/srt/model_executor/cuda_graph_runner.py b/python/sglang/srt/model_executor/cuda_graph_runner.py index 5a2a53ffaa8..eda8b57f69d 100644 --- a/python/sglang/srt/model_executor/cuda_graph_runner.py +++ b/python/sglang/srt/model_executor/cuda_graph_runner.py @@ -162,6 +162,9 @@ def __init__(self, model_runner: ModelRunner): self.disable_padding = model_runner.server_args.disable_cuda_graph_padding self.is_encoder_decoder = model_runner.model_config.is_encoder_decoder self.enable_dp_attention = model_runner.server_args.enable_dp_attention + self.enable_hip_attention = model_runner.server_args.enable_hip_attention + if self.enable_hip_attention: + self.hip_config = model_runner.server_args.hip_attention_config self.tp_size = model_runner.server_args.tp_size self.dp_size = model_runner.server_args.dp_size