Pytorch 出错集

最新推荐文章于 2023-06-15 15:00:29 发布

未知量z

最新推荐文章于 2023-06-15 15:00:29 发布

阅读量1.3k

点赞数

分类专栏： python 文章标签： python anaconda pytorch

本文链接：https://blog.csdn.net/baidu_37122643/article/details/121486379

版权

python 专栏收录该内容

8 篇文章 0 订阅

订阅专栏

**错误1：**UserWarning: nn.functional.sigmoid is deprecated. Use torch.sigmoid instead.
解决方法：找到nn.functional.sigmoid，将其替换为torch.sigmoid。
参考：
https://blog.csdn.net/dagewoshiwusong/article/details/108306325

错误2. Target is multiclass but average=‘binary’. Please choose another average setting
解决方法：加入average=‘micro’

precision = precision_score(Y, np.round(predicted), *average='micro'*)
recall = recall_score(Y, np.round(predicted), *average='micro'*)

参考：https://blog.csdn.net/weixin_44436677/article/details/105985358

**错误3.**RuntimeWarning: invalid value encountered in double_scalars
问题：可能是分母部分出现了0. 加循环，若分母为0，则该项为0.0.
解决：

precision = precision_score(Y, np.round(predicted), average='micro')
recall = recall_score(Y, np.round(predicted), average='micro')
if precision ==0 or recall== 0:
         f1score = 0.0
else:
         f1score = 2*precision*recall/ (precision+recall)

return f1score

参考：https://blog.csdn.net/qq_34333481/article/details/85258888