TurboTransformers
TurboTransformers copied to clipboard
Add alternative non-docker install
It seems that the dependency is not so much:
- ninja (apt install ninja-build,)
- intel MTK (pip install mtk) (which is actually not needed if using GPU inference only, but it lays on build dependency).
- onmt (pip install OpenNMT-py) (if you want to pass unitest)
So may be we could provide a non-docker install methods which is far more lightweight.
Hello, sorry for my late response.
I assume MTK -> MKL.
Please refer to the README.md of branch ppopp21_artifact_centos for installation without docker.
https://github.com/Tencent/TurboTransformers/tree/ppopp21_artifact_centos
BTW, you can copy installation scripts from that branch to branch master. Or pull master to branch ppopp21_artifact_centos.