前提:当前提供 了8核64G的免费体验规格,每天三个小时限额
地址:https://console.huaweicloud.com/modelarts/?region=cn-north-4#/dashboard
下载模型:请参考另一个文章
创建环境(自带环境是pytorch1.8的,所以自己创建)
conda info --env
conda create --name pytorch2.0
conda activate pytorch2.0
安装必要的包
conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia
conda install pytorch torchvision torchaudio cpuonly -c pytorch
conda update -n base -c defaults conda
conda install transformers sentencepiece
pip install rouge_chinese cpm_kernels
运行代码
from transformers import AutoTokenizer, AutoModel
tokenizer = AutoTokenizer.from_pretrained("/home/ma-user/work/glm2/chatglm2-6b-int4",
trust_remote_code=True)
model = AutoModel.from_pretrained("/home/ma-user/work/glm2/chatglm2-6b-int4",trust_remote_code=True).cuda()
model = AutoModel.from_pretrained("/home/ma-user/work/glm2/chatglm2-6b-int4",trust_remote_code=True).float()
model = model.eval()
response, history = model.chat(tokenizer, "晚上睡不着应该怎么办", history=[])
print(response)