Random error from the dataloader

[GianlucaDeStefano]

Hi all, I was trying to train my model when suddenly I got this error:

KeyError: Caught KeyError in DataLoader worker process 0.
Original Traceback (most recent call last):
  File "/usr/local/lib/python3.8/dist-packages/torch/utils/data/_utils/worker.py", line 302, in _worker_loop
    data = fetcher.fetch(index)
  File "/usr/local/lib/python3.8/dist-packages/torch/utils/data/_utils/fetch.py", line 61, in fetch
    return self.collate_fn(data)
  File "/usr/local/lib/python3.8/dist-packages/torch_geometric/loader/dataloader.py", line 19, in __call__
    return Batch.from_data_list(batch, self.follow_batch,
  File "/usr/local/lib/python3.8/dist-packages/torch_geometric/data/batch.py", line 76, in from_data_list
    batch, slice_dict, inc_dict = collate(
  File "/usr/local/lib/python3.8/dist-packages/torch_geometric/data/collate.py", line 70, in collate
    values = [store[attr] for store in stores]
  File "/usr/local/lib/python3.8/dist-packages/torch_geometric/data/collate.py", line 70, in <listcomp>
    values = [store[attr] for store in stores]
  File "/usr/local/lib/python3.8/dist-packages/torch_geometric/data/storage.py", line 85, in __getitem__
    return self._mapping[key]
KeyError: 'input_id'

I searched online, but I have yet to find any helpful information.

My data loader is defined like this:

loader  = DataLoader(dataset,batch_size=batch_size,shuffle=False)

And it loads data from a custom dataset.
Strangely, this problem happens randomly during the training epoch, even with shullfe=False and on any device (CPU/GPU).

I am trying to debug it, but honestly, I don't know what to do, as it seems to pop up randomly.

Do you have any tips?

[rusty1s]

Mh, may there exist a input_id property in one of your graphs that is not present in others? Otherwise, I am not sure. If you have a somewhat reproducible example, this would help us tremendously to track this down :(

[GianlucaDeStefano]

I rebuilt my dataset from scratch, and the error went away. Probably something got corrupted while building it the first time.