【error】RuntimeError: multi-target not supported at /pytorch/torch/lib/THCUNN/generic/ClassNLLCriteri

最新推荐文章于 2025-02-19 16:59:26 发布

codebrid

最新推荐文章于 2025-02-19 16:59:26 发布

阅读量1.8w

点赞数 3

CC 4.0 BY-SA版权

分类专栏：随记

本文链接：https://blog.youkuaiyun.com/ccbrid/article/details/81018119

随记专栏收录该内容

51 篇文章

订阅专栏

本文解决了PyTorch中因输入维度不匹配导致的multi-target not supported错误。通过调整输入张量的维度，成功避免了运行时错误。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

错误：

RuntimeError: multi-target not supported at /pytorch/torch/lib/THCUNN/generic/ClassNLLCriterion.cu

错误反馈：

File "/users4/zsun/pytorch/paper_image_caption/sz-spatial5-chunk/model_msoft_gh.py", line 214, in forward
loss += F.cross_entropy( scores, captions.index_select(1, Variable(self.num_step_id+1)) )
File "/users4/zsun/anaconda3/lib/python3.6/site-packages/torch/nn/functional.py", line 1140, in cross_entropy
return nll_loss(log_softmax(input, 1), target, weight, size_average, ignore_index, reduce)
File "/users4/zsun/anaconda3/lib/python3.6/site-packages/torch/nn/functional.py", line 1049, in nll_loss
return torch._C._nn.nll_loss(input, target, weight, size_average, ignore_index, reduce)

RuntimeError: multi-target not supported at /pytorch/torch/lib/THCUNN/generic/ClassNLLCriterion.cu:16

出错代码：

loss += F.cross_entropy( scores, captions )

出错原因：

scores, captions的维度与F.cross_entropy()函数要求的维度不匹配，

例如出错的维度为scores=[batch, vocab_size], captions=[batch, 1]

解决：

loss += F.cross_entropy( scores, captions.squeeze() )