extra ops.repeat gives wrong shape when repeats is a vector and axis=None

extra ops.repeat gives wrong shape when repeats is a vector and axis=None: a task in MiMo-V2.6-RL-harbor-code: MiMo-V2.6-RL Code (Harbor) (Harbor dataset). I'm using theano.tensor.extra ops.repeat to do the same thing as numpy.repeat — repeat each element of an array a (possibly different) number…

The task

I'm using `theano.tensor.extra_ops.repeat` to do the same thing as `numpy.repeat` — repeat each element of an array a (possibly different) number of times. When `repeats` is a 1D tensor and I don't pass `axis`, the inferred shape of the result comes out wrong.

Part of FineEnvs/MiMo-V2.6-RL-harbor-code.