paddle 传入模型后数值突变

最新推荐文章于 2024-03-03 18:49:29 发布

杨朝北

最新推荐文章于 2024-03-03 18:49:29 发布

阅读量376

点赞数

文章标签： paddle 深度学习人工智能

本文链接：https://blog.csdn.net/sinat_33361923/article/details/127669430

版权

paddlepaddle在传入模型的word,label的dataloader后数值突变。

Traceback (most recent call last):
  File "main.py", line 119, in <module>
    loss = loss_f(logits, labels)
  File "/home/yhp/miniconda3/envs/yhpt/lib/python3.7/site-packages/paddle/fluid/dygraph/layers.py", line 948, in __call__
    return self.forward(*inputs, **kwargs)
  File "/home/yhp/paddle/paddle-ner-master/src/model.py", line 21, in forward
    loss = paddle.nn.functional.cross_entropy(x, label, ignore_index=-100)
  File "/home/yhp/miniconda3/envs/yhpt/lib/python3.7/site-packages/paddle/nn/functional/loss.py", line 2393, in cross_entropy
    label_min.item()))
ValueError: Target -4867214609213893293 is out of lower bound.