Result type Double can't be cast to the desired output type Long

1

Hi,

I’ve googled this problem, but most people run into it in the loss calculation step. But mine happened during the generation of batch, I don’t know what went wrong.

Blockquote
Traceback (most recent call last):
File “dtd_rnn_training.py”, line 469, in
training(args)
File “dtd_rnn_training.py”, line 212, in training
for batch_idx, batch in enumerate(train):
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/utils/data/dataloader.py”, line 517, in next
data = self._next_data()
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/utils/data/dataloader.py”, line 1199, in _next_data
return self._process_data(data)
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/utils/data/dataloader.py”, line 1225, in _process_data
data.reraise()
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/_utils.py”, line 429, in reraise
raise self.exc_type(msg)
RuntimeError: Caught RuntimeError in DataLoader worker process 0.
Original Traceback (most recent call last):
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/utils/data/_utils/worker.py”, line 202, in _worker_loop
data = fetcher.fetch(index)
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py”, line 47, in fetch
return self.collate_fn(data)
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/utils/data/_utils/collate.py”, line 73, in default_collate
return {key: default_collate([d[key] for d in batch]) for key in elem}
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/utils/data/_utils/collate.py”, line 73, in
return {key: default_collate([d[key] for d in batch]) for key in elem}
File “/home/anaconda3/envs/aa/lib/python3.8/site-packages/torch/utils/data/_utils/collate.py”, line 55, in default_collate
return torch.stack(batch, 0, out=out)
RuntimeError: result type Double can’t be cast to the desired output type Long

2

The error is raised, if out is a LongTensor, while (some of) the batch tensors are DoubleTensors as seen here:

batch = [
    torch.randn(3, 24, 24, dtype=torch.float64),
    torch.randn(3, 24, 24, dtype=torch.float64)
]

out = torch.empty(2, 3, 24, 24).long()
torch.stack(batch, 0, out=out)
> RuntimeError: result type Double can't be cast to the desired output type Long

These lines of code check the first element of batch and use its dtype to create out. Could you check, if you are returning different dtypes in the __getitem__ of your Dataset?

4

You are right. It’s a binary classification task, so the label is saved as type long. At first, I just tried to fix the problem during taining stage. But it didn’t work, so in the end I changed the dtype of my dataset.