intel / auto-round

Advanced Quantization Algorithm for LLMs. This is official implementation of "Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs"
https://arxiv.org/abs/2309.05516
Apache License 2.0
245 stars 20 forks source link

xpu exporting on cpu machine #210

Closed wenhuach21 closed 2 months ago

wenhuach21 commented 2 months ago

xpu exporting only supports cuda machine currently

wenhuach21 commented 2 months ago

@WeiweiZhang1 @Kaihui-intel