impossible to make a single example dataloader

[lengstrom]

I get this error when I try to make a dataloader with only a single index: (i.e. any dataloader with indices=[0])

Exception in thread Thread-1:
Traceback (most recent call last):
  File "/mnt/nfs/home/engstrom/conda_envs/ffcvrgb/lib/python3.9/threading.py", line 973, in _bootstrap_inner
    self.run()
  File "/mnt/nfs/home/engstrom/conda_envs/ffcvrgb/lib/python3.9/site-packages/ffcv/loader/epoch_iterator.py", line 79, in run
    result = self.run_pipeline(b_ix, ixes, slot, events[slot])
  File "/mnt/nfs/home/engstrom/conda_envs/ffcvrgb/lib/python3.9/site-packages/ffcv/loader/epoch_iterator.py", line 141, in run_pipeline
    return tuple(x[:len(batch_indices)] for x in args)
  File "/mnt/nfs/home/engstrom/conda_envs/ffcvrgb/lib/python3.9/site-packages/ffcv/loader/epoch_iterator.py", line 141, in <genexpr>
    return tuple(x[:len(batch_indices)] for x in args)
IndexError: dimension specified as 0 but tensor has no dimensions

[lengstrom]

Batch size 1 in general does not work I think either

[kristian-georgiev]

happens for any batch size below 4

[numpee]

I believe this is caused by the Squeeze() Operator in the pipeline (for labels). IIRC, changing to Squeeze(0) resolves the issue.

[GuillaumeLeclerc]

Yes @numpee is right :)