就是

paddle.device.set_device("gpu:0")

之后，我看文档说开启gpu使用，但是实际训练中还是用cpu，请问这是为啥呢

引用的库是

import paddle
from paddlenlp.datasets import load_dataset
import paddle.nn.functional as F
import paddle.nn as nn
import paddlenlp as ppnlp
from paddlenlp.transformers import BertPretrainedModel, BertTokenizer,BertForSequenceClassification
from paddlenlp.transformers import LinearDecayWithWarmup