conda activate llamachinese
cd /home/cys/Llama2-Chinese/scripts/api
python accelerate_server.py --model_path /media/cys/c4e58bbe-a73a-4b02-ae9e-2b310ee884fb/chinese-llama-2-13b --gpus "0" --infer_dtype "int8" --model_source "llama2_chinese"
?
?运行结果: