mixmo.utils.torchutils

General tensor manipulation utility functions (initializations, permutations, one hot)

Functions

_calculate_correct_fan(tensor, mode)	Return the minimal input or output sizes for the weight tensor depending on which is needed
_calculate_fan_in_and_fan_out(tensor)	Compute the minimal input and output sizes for the weight tensor
kaiming_normal_truncated(tensor[, a, mode, …])	Fills the input Tensor with values according to the method described in Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification - He, K.
onehot(size, target)	Translate scalar targets to one hot vectors
randperm_static(batch_size[, proba_static])	Perform random permutation with a set percentage remaining fixed
truncated_normal_(tensor[, mean, std])	Initialization function
weights_init_hetruncatednormal(m[, …])	Simple init function