[NPU] bugfix for Qwen3-Next and performance update (#11969)

2025-10-30 21:52:16 +08:00
parent 73dfd2dfb1
commit cafebef154
7 changed files with 68 additions and 21 deletions
--- a/scripts/ci/npu_ci_install_dependency.sh
+++ b/scripts/ci/npu_ci_install_dependency.sh
@@ -59,7 +59,7 @@ wget -O "${BISHENG_NAME}" "${BISHENG_URL}" && chmod a+x "${BISHENG_NAME}" && "./


 ### Install sgl-kernel-npu
-SGL_KERNEL_NPU_TAG="20250926"
+SGL_KERNEL_NPU_TAG="20251030"
 git clone --depth 1 https://github.com/sgl-project/sgl-kernel-npu.git --branch ${SGL_KERNEL_NPU_TAG}
 # pin wheel to 0.45.1 ref: https://github.com/pypa/wheel/issues/662
 pip install wheel==0.45.1