extorch.utils.data¶
- extorch.utils.data.mix_data(self, inputs: torch.Tensor, targets: torch.Tensor, alpha: float = 1.0)[source]¶
Mixup data for training neural networks on convex combinations of pairs of examples and their labels (Link).
- Parameters
inputs (Tensor) – Input examples.
targets (Tensor) – Labels of input examples.
alpha (float) – Parameter of the beta distribution. Default: 1.0.
- Returns
Input examples after mixup. mixed_targets (Tensor): Labels of mixed inputs. _lambda (float): Parameter sampled from the beta distribution.
- Return type
mixed_inputs (Tensor)