intel / auto-round

Advanced Quantization Algorithm for LLMs/VLMs. 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
261 stars 22 forks source link

Simulated W4Afp8 Quantization #331

Closed wenhuach21 closed 4 days ago

wenhuach21 commented 1 week ago

for model_name in "/models/Meta-Llama-3.1-8B-Instruct" "/models/Meta-Llama-3-8B-Instruct"
do CUDA_VISIBLE_DEVICES=$device \ python3 -m auto_round \ --model_name $model_name \ --device 0 \ --act_bits 8 \ --group_size 128 \ --bits 4 \ --tasks "lambada_openai,hellaswag,winogrande,piqa,mmlu" \ --eval_bs $eval_bs \ --data_type "fp8_to_int_sym" \ --act_data_type "fp8" \ --disable_act_dynamic \ --format "fake" \ 2>&1 | tee -a w4_fp8_act_static.txt done