Torch loss functions

Loss functions that are implemented in PyTorch

---

source

quantile_loss

 quantile_loss (input:torch.Tensor, target:torch.Tensor,
                quantile:torch.Tensor, reduction:str='mean')

---

source

TorchQuantileLoss

 TorchQuantileLoss (reduction:str='mean')

Implmentation of the quantile loss in Pytorch. Unlike the Numpy-based implementation [`quantile_loss`](https://opimwue.github.io/ddopai/00_utils/torch_loss_functions.html#quantile_loss) in the loss_functions module, this implementation this implementation reduces the results to a scalar value using the specified reduction method. This class is used to train Pytorch models using the quantile loss.

	Type	Default	Details
reduction	str	mean
Returns	None

---

source

TorchQuantileLoss.forward

 TorchQuantileLoss.forward (input:torch.Tensor, target:torch.Tensor,
                            quantile:ddopai.utils.Parameter|numpy.ndarray)

Forward pass of the quantile loss function.

	Type	Details
input	Tensor
target	Tensor
quantile	ddopai.utils.Parameter | numpy.ndarray
Returns	Tensor

---

source

pinball_loss

 pinball_loss (input:torch.Tensor, target:torch.Tensor,
               underage:torch.Tensor, overage:torch.Tensor,
               reduction:str='mean')

---

source

TorchPinballLoss

 TorchPinballLoss (reduction:str='mean')

Implmentation of the pinball loss in Pytorch using specific overage and underage cost. For the pinball loss based on quantiles directly, use the TorchQuantileLoss class. Unlike the Numpy-based implementation [`pinball_loss`](https://opimwue.github.io/ddopai/00_utils/torch_loss_functions.html#pinball_loss) in the loss_functions module, this implementation this implementation reduces the results to a scalar value using the specified reduction method. This class is used to train Pytorch models using the pinball loss.

	Type	Default	Details
reduction	str	mean
Returns	None

---

source

TorchPinballLoss.forward

 TorchPinballLoss.forward (input:torch.Tensor, target:torch.Tensor,
                           underage:ddopai.utils.Parameter|numpy.ndarray,
                           overage:ddopai.utils.Parameter|numpy.ndarray)

Forward pass of the pinball loss function.

	Type	Details
input	Tensor
target	Tensor
underage	ddopai.utils.Parameter | numpy.ndarray
overage	ddopai.utils.Parameter | numpy.ndarray
Returns	Tensor