repeat_and_flatten

bayesflow.utils.repeat_and_flatten(x: Tensor, num_repeats: int) → Tensor

Repeat each element along the leading (batch) axis and flatten back into it.

Inserts a new axis after the batch axis, tiles it num_repeats times and merges it back, so (B, ...) -> (B * num_repeats, ...) with repeats of the same element kept adjacent (ordering [x0, x0, ..., x1, x1, ...]).