Change misleading MaxUnpool2d example to better demonstrate output_size usage (#68936)
Summary:
At https://github.com/pytorch/pytorch/issues/68873, jbschlosser states that maxunpool2d with the `output_size` argument only works for indices of the same size. This makes sense, but unfortunately it's not what's shown in the example! I've removed the wrong example and replaced it with one where specifying `output_size` is actually necessary -- the unpool call fails without it.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/68936
Reviewed By: H-Huang
Differential Revision: D32759207
Pulled By: jbschlosser
fbshipit-source-id: 658e1724150a95454a05a771ae7c6e2e736740a7
