error with lengths = [1, 1, 1, 1,...., 1, 1, 1]

[xiexbing]

Hello, I generated the KJT with lengths is a tensor with full of 1s. then I get an error as:

keys = ['f0', 'f1', 'f2', 'f3', 'f4', 'f5', ...], stride = 105, length_per_key = None, lengths = tensor([1, 1, 1, ..., 1, 1, 1], device='cuda:0'), offsets = None

def _maybe_compute_length_per_key(
    keys: List[str],
    stride: int,
    length_per_key: Optional[List[int]],
    lengths: Optional[torch.Tensor],
    offsets: Optional[torch.Tensor],
) -> List[int]:
    if length_per_key is None:
        if len(keys) and offsets is not None and len(offsets) > 0:
            _length: List[int] = torch.sum(
                torch.diff(offsets).view(-1, stride), dim=1
            ).tolist()
        elif len(keys) and lengths is not None:
            _length: List[int] = (

              torch.sum(lengths.view(-1, stride), dim=1).tolist()

                if lengths.numel() != 0
                else [0] * len(keys)
            )

E RuntimeError: shape '[-1, 105]' is invalid for input of size 2114

../../../../../miniconda3/envs/pr-gpu/lib/python3.8/site-packages/torchrec/sparse/jagged_tensor.py:523: RuntimeError

[PaulZhang12]

Can we see the whole stack trace please?