---
title: BNNS.LossLayer
framework: accelerate
role: symbol
role_heading: Class
path: accelerate/bnns/losslayer
---

# BNNS.LossLayer

A layer object that wraps a loss filter and manages its deinitialization.

## Declaration

```swift
class LossLayer
```

## Topics

### Creating a Loss Layer

- [init(input:output:lossFunction:lossReduction:filterParameters:)](accelerate/bnns/losslayer/init(input:output:lossfunction:lossreduction:filterparameters:).md)

### Specifying a Loss Function

- [BNNS.LossFunction](accelerate/bnns/lossfunction.md)

### Specifying a Loss Reduction Function

- [BNNS.LossReduction](accelerate/bnns/lossreduction.md)

### Applying a Loss Layer

- [apply(batchSize:input:labels:output:generatingInputGradient:)](accelerate/bnns/losslayer/apply(batchsize:input:labels:output:generatinginputgradient:).md)

### Instance Methods

- [apply(batchSize:input:labels:output:weights:broadcastsWeights:generatingInputGradient:)](accelerate/bnns/losslayer/apply(batchsize:input:labels:output:weights:broadcastsweights:generatinginputgradient:).md)

## Relationships

### Inherits From

- [BNNS.Layer](accelerate/bnns/layer.md)

## See Also

### Loss layers

- [BNNSLossFunction](accelerate/bnnslossfunction.md)
- [BNNSLossReductionFunction](accelerate/bnnslossreductionfunction.md)
- [BNNSLayerParametersLossBase](accelerate/bnnslayerparameterslossbase.md)
- [BNNSLayerParametersLossHuber](accelerate/bnnslayerparameterslosshuber.md)
- [BNNSLayerParametersLossSigmoidCrossEntropy](accelerate/bnnslayerparameterslosssigmoidcrossentropy.md)
- [BNNSLayerParametersLossSoftmaxCrossEntropy](accelerate/bnnslayerparameterslosssoftmaxcrossentropy.md)
- [BNNSLayerParametersLossYolo](accelerate/bnnslayerparameterslossyolo.md)
- [BNNSFilterCreateLayerLoss(_:_:)](accelerate/bnnsfiltercreatelayerloss(_:_:).md)
- [BNNSLossFilterApplyBatch(_:_:_:_:_:_:_:_:_:_:_:)](accelerate/bnnslossfilterapplybatch(_:_:_:_:_:_:_:_:_:_:_:).md)
- [BNNSLossFilterApplyBackwardBatch(_:_:_:_:_:_:_:_:_:_:_:_:)](accelerate/bnnslossfilterapplybackwardbatch(_:_:_:_:_:_:_:_:_:_:_:_:).md)