1xN tensors are not visualizable

[egorchakov]

Describe the bug
Logged 1xN tensors don't seem to be visualizable (potentially related to #3709?)

import torch
import rerun as rr

rr.init('tensor-viz', spawn=True, strict=True)
rr.log('1x10', rr.Tensor(torch.zeros(1, 10))

Screenshots

Rerun version

rerun_py 0.15.0-rc.1 [rustc 1.74.0 (79e9716c9 2023-11-13), LLVM 17.0.4] x86_64-unknown-linux-gnu release-0.15.0-rc.1 3c298fd, built 2024-04-02T21:34:01Z

Additional context

A workaround I stumbled upon is first logging a 2xN tensor to the same entity path and then overwriting it with the original 1xN tensor, which makes the 1xN tensor show up in the viewer:

rr.log('1x10', rr.Tensor(torch.zeros(2, 10)) # dummy 2xN
rr.log('1x10', rr.Tensor(torch.zeros(1, 10)) # original

[abey79]

Thanks for the report. I could reproduced both the issue and the workaround on main.