1.错误1:
RuntimeError: Expected one of cpu, cuda, ipu, xpu, mkldnn, opengl, opencl, ideep, hip, ve, fpga, ort, xla, lazy, vulkan, mps, meta, hpu, mtia, privateuseone device type at start of device string: auto
解决办法:将以下代码中device直接设置为”cuda"
2.错误2:
回答很慢且乱码,或是报RuntimeError: probability tensor contains either inf
, nan
or element < 0
在代码中加上:
os.environ['CUDA_VISIBLE_DEVICES'] = "0,1"
根据自己的gpu性能和显存情况设置os.environ[‘CUDA_VISIBLE_DEVICES’],若还是推理很慢或报显存不够的错误,首先看设置的gpu有没有被别的程序占用,若没有就多设几张卡,如“0,1,2”。
ps: secgpt是基于baichuan-13b训练的,运行baichuan时也可能出现RuntimeError: probability tensor contains either inf
, nan
or element < 0错误,同样可以通过设置os.environ[‘CUDA_VISIBLE_DEVICES’]解决