# kd.losses

[[[Source]]](https://github.com/google-research/kauldron/tree/main/kauldron/losses/__init__.py)

```{eval-rst}
.. automodule:: kauldron.losses
  :no-members:
```

## Symbols


### Class

|  |  |
--- | ---
[kd.losses.AbsoluteValue](AbsoluteValue) | Absolute value loss.
[kd.losses.Huber](Huber) | Huber loss.
[kd.losses.L1](L1) | L1 loss.
[kd.losses.L2](L2) | L2 loss.
[kd.losses.Loss](Loss) | Base class for losses which handles masks, averaging, and loss-weight.
[kd.losses.NegativeCosineSimilarity](NegativeCosineSimilarity) | Negative Cosine Similarity loss.
[kd.losses.SigmoidBinaryCrossEntropy](SigmoidBinaryCrossEntropy) | Sigmoid cross-entropy loss with binary labels.
[kd.losses.SoftmaxCrossEntropy](SoftmaxCrossEntropy) | Softmax cross-entropy loss.
[kd.losses.SoftmaxCrossEntropyWithIntLabels](SoftmaxCrossEntropyWithIntLabels) | Softmax cross-entropy loss with integer labels.
[kd.losses.Value](Value) | Value loss.

### Function

|  |  |
--- | ---
[kd.losses.compute_losses](compute_losses) | Compute all losses based on given context.

```{toctree}
:hidden:

AbsoluteValue
Huber
L1
L2
Loss
NegativeCosineSimilarity
SigmoidBinaryCrossEntropy
SoftmaxCrossEntropy
SoftmaxCrossEntropyWithIntLabels
Value
compute_losses
```