input in Long type

[sungjchi]

My model has an input in Long dtype which is passed to an torch.nn.Embedding layer. Since all the input data are created in torch.FloatTensor type, it creates runtime error when these inputs are passed to an embedding layer. Should there be an argument to choose which data type to pass to inputs?

In the code below, the third input is passed through an embedding layer and thus cause a runtime error.

(Pdb) summary(m,[(170,),(800,80),(170,),(800,)]) *** RuntimeError: Expected tensor for argument #1 'indices' to have scalar type Long; but got CUDAType instead (while checking arguments for embedding)

[DrStoop]

check out this pull request #89, just extended it...

[Naireen]

@sungjaechi this should now work with the new version of pytorch-summary (note that you'll have to clone it, not install it via pip). Let me know if it solves your issues

[londumas]

Just for reference, had the same issue, using the version 1.5.1 on pip. I fixed it with tweaking the dtype in <me>.local/lib/python3.6/site-packages/torchsummary/torchsummary.py

Traceback (most recent call last):
  File "<my program>", line 114, in <module>
    torchsummary.summary(<my program>, input_size=(<my input>,))
  File "<me>.local/lib/python3.6/site-packages/torchsummary/torchsummary.py", line 72, in summary
    model(*x)
  File "<me>.local/lib/python3.6/site-packages/torch/nn/modules/module.py", line 532, in __call__
    result = self.forward(*input, **kwargs)
  File "<my program>", line 26, in forward
    x = self.embed(x)
  File "<me>.local/lib/python3.6/site-packages/torch/nn/modules/module.py", line 532, in __call__
    result = self.forward(*input, **kwargs)
  File "<me>.local/lib/python3.6/site-packages/torch/nn/modules/sparse.py", line 114, in forward
    self.norm_type, self.scale_grad_by_freq, self.sparse)
  File "<me>.local/lib/python3.6/site-packages/torch/nn/functional.py", line 1484, in embedding
    return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)
RuntimeError: Expected tensor for argument #1 'indices' to have scalar type Long; but got torch.FloatTensor instead (while checking arguments for embedding)

[manuelblancovalentin]

No solutions for this? I have the same issue.

[Leo-LiHao]

Hi, why don't we merge this and publish to PyPI?

[sreenithakasarapu]

Hi,

I’m interested, what are your research interests?

Hi, why don't we merge this and publish to PyPI?

— Reply to this email directly, view it on GitHub https://github.com/sksq96/pytorch-summary/issues/77#issuecomment-1041099900, or unsubscribe https://github.com/notifications/unsubscribe-auth/AOZ6RM5OL7WOPM7XCSX43CTU3MS7JANCNFSM4H3YERYQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you are subscribed to this thread.Message ID: @.***>

[Leo-LiHao]

Hi, I’m interested, what are your research interests? … Hi, why don't we merge this and publish to PyPI? — Reply to this email directly, view it on GitHub <#77 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AOZ6RM5OL7WOPM7XCSX43CTU3MS7JANCNFSM4H3YERYQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub. You are receiving this because you are subscribed to this thread.Message ID: @.***>

I'm currently working on some NLP projects. I used torchinfo instead (as suggested in the readme) which solved my problem.