How to use SmoothQuant in FasterTransformer?

[jiangsongHW]

I have build and run FasterTransformer. I see there is a parameter --int8_mode in FasterTransformer,. will it use SmoothQuant as default, if I set int8_mode =1?

if not is there any example of using SmoothQuant in FasterTransformer ?

thank you!

[pineking]

https://github.com/NVIDIA/FasterTransformer/blob/main/docs/gpt_guide.md