EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[NVIDIA] Update to leverage flashinfer trtllm FP4 MOE throughput kernel (#11563)

Browse Source 
Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com>
This commit is contained in:
jiahanc
2025-10-22 13:11:16 -07:00
committed by GitHub
parent 6d535b719f
commit eec9e471ca
3 changed files with 4 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
scripts/ci/ci_install_dependency.sh
View File

@@ -45,8 +45,8 @@ else
# Install the main package without deps
$PIP_CMD install -e "python[dev]" --no-deps $PIP_INSTALL_SUFFIX --force-reinstall
# Install flashinfer-python 0.4.0 dependency that requires prerelease (This should be removed when flashinfer fixes this issue)
$PIP_CMD install flashinfer-python==0.4.0 --prerelease=allow $PIP_INSTALL_SUFFIX
# Install flashinfer-python 0.4.1 dependency that requires prerelease (This should be removed when flashinfer fixes this issue)
$PIP_CMD install flashinfer-python==0.4.1 --prerelease=allow $PIP_INSTALL_SUFFIX
# Install the main package
$PIP_CMD install -e "python[dev]" --extra-index-url https://download.pytorch.org/whl/${CU_VERSION} $PIP_INSTALL_SUFFIX --upgrade