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
236 stars 20 forks source link

AutoAdamRound bug due to quant_block_list #274

Open wenhuach21 opened 4 days ago

WeiweiZhang1 commented 4 days ago

please refer https://github.com/intel/auto-round/pull/275