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
247 stars 20 forks source link

fix mx_fp issues #286

Closed wenhuach21 closed 3 weeks ago