[Feature] Support DeepEP normal & Redundant Experts on NPU (#9881)

This commit is contained in:
Even Zhou
2025-09-11 11:35:26 +08:00
committed by GitHub
parent 5b7448de77
commit 5b64f006ec
15 changed files with 319 additions and 111 deletions

View File

@@ -54,8 +54,6 @@ jobs:
run: |
version=$(cat python/sglang/version.py | cut -d'"' -f2)
echo "TAG=lmsysorg/sglang:v$version-cann${{ matrix.cann_version }}-${{ matrix.device_type }}" >> $GITHUB_OUTPUT
kernel_tag=$(curl -s https://api.github.com/repos/sgl-project/sgl-kernel-npu/tags | jq -r '.[0].name')
echo "KERNEL_NPU_TAG=${kernel_tag}" >> $GITHUB_OUTPUT
- name: Build and push Docker image
id: build-and-push
@@ -70,6 +68,6 @@ jobs:
push: ${{ github.repository == 'sgl-project/sglang' && github.event_name != 'pull_request' }}
provenance: false
build-args: |
SGLANG_KERNEL_NPU_TAG=${{ steps.get_version.outputs.KERNEL_NPU_TAG }}
SGLANG_KERNEL_NPU_TAG=20250901
CANN_VERSION=${{ matrix.cann_version }}
DEVICE_TYPE=${{ matrix.device_type }}