flamo.optimize.utils.generate_partitions(tensor: Tensor, n_samples: int, n_sets: int, seed: int | None = None)

Create n_sets sets of length(tensor) // n_samples partitions of a tensor, and the items are shuffled randomly for each set.

Arguments:

• tensor (torch.Tensor): The input tensor to partition.

• n_samples (int): The number of samples in each partition.

• n_sets (int): The number of different sets of partitions.

• seed (int, optional): A seed for reproducibility. Default is None.

Returns:

list of lists of torch.Tensor: list of n_sets sets, where each set contains length(tensor) // n_samples partitions.