search

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
172 stars 20 forks source link

Add marlin and modify acc.md #96

Closed pursure-D closed 3 months ago

wenhuach21 commented 3 months ago

please remove the marlin part and only merge the acc