# BNNS.FusedParametersLayer A layer object that wraps a fused layer and manages its deinitialization. ``` class FusedParametersLayer ``` ## Overview Use a [`BNNS.FusedParametersLayer`](/documentation/Accelerate/BNNS/FusedParametersLayer) instance to fuse component layers with the following configurations: - Convolution → Normalization - Fully Connected → Normalization - Transposed Convolution → Normalization - Convolution → Quantization - Fully Connected → Quantization - Transposed Convolution → Quantization - Arithmetic → Normalization ## Topics ### Creating a Fused Parameters Layer [`convenience init?(input: BNNSNDArrayDescriptor, output: BNNSNDArrayDescriptor, fusedLayerParameters: [any FusableLayerParameters], filterParameters: BNNSFilterParameters?)`](/documentation/Accelerate/BNNS/FusedParametersLayer/init(input:output:fusedLayerParameters:filterParameters:)) Creates a new fused layer from an array of layer parameters. [`convenience init?(inputA: BNNSNDArrayDescriptor, inputB: BNNSNDArrayDescriptor, output: BNNSNDArrayDescriptor, fusedLayerParameters: [any FusableLayerParameters], filterParameters: BNNSFilterParameters?)`](/documentation/Accelerate/BNNS/FusedParametersLayer/init(inputA:inputB:output:fusedLayerParameters:filterParameters:)) Creates a new fused layer from an array of layer parameters, where the first layer accepts two inputs. [`convenience init?(inputA: BNNSNDArrayDescriptor, inputB: BNNSNDArrayDescriptor, inputC: BNNSNDArrayDescriptor, output: BNNSNDArrayDescriptor, fusedLayerParameters: [any FusableLayerParameters], filterParameters: BNNSFilterParameters?)`](/documentation/Accelerate/BNNS/FusedParametersLayer/init(inputA:inputB:inputC:output:fusedLayerParameters:filterParameters:)) Creates a new fused layer from an array of layer parameters, where the first layer accepts three inputs. ### Specifying a Layer Parameter [`struct FusedUnaryArithmeticParameters`](/documentation/Accelerate/BNNS/FusedUnaryArithmeticParameters) A structure that contains the parameters for a fused unary arithmetic layer. [`struct FusedBinaryArithmeticParameters`](/documentation/Accelerate/BNNS/FusedBinaryArithmeticParameters) A structure that contains the parameters for a fused binary arithmetic layer. [`struct FusedTernaryArithmeticParameters`](/documentation/Accelerate/BNNS/FusedTernaryArithmeticParameters) A structure that contains the parameters for a fused ternary arithmetic layer. [`struct FusedConvolutionParameters`](/documentation/Accelerate/BNNS/FusedConvolutionParameters) A structure that contains the parameters for a fused convolution layer. [`struct FusedQuantizationParameters`](/documentation/Accelerate/BNNS/FusedQuantizationParameters) A structure that contains the parameters for a fused quantization layer. [`struct FusedDequantizationParameters`](/documentation/Accelerate/BNNS/FusedDequantizationParameters) A structure that contains the parameters for a fused dequantization layer. [`struct FusedFullyConnectedParameters`](/documentation/Accelerate/BNNS/FusedFullyConnectedParameters) A structure that contains the parameters for a fused fully connected layer. [`struct FusedNormalizationParameters`](/documentation/Accelerate/BNNS/FusedNormalizationParameters) A structure that contains the parameters for a fused normalization layer. ### Applying a Fused Parameters Layer [`func apply(batchSize: Int, inputA: BNNSNDArrayDescriptor, inputB: BNNSNDArrayDescriptor, output: BNNSNDArrayDescriptor, for: BNNS.LearningPhase) throws`](/documentation/Accelerate/BNNS/FusedParametersLayer/apply(batchSize:inputA:inputB:output:for:)) Applies the layer to a set of input objects and writes the result to a set of output objects, where the first layer accepts two inputs. [`func apply(batchSize: Int, inputA: BNNSNDArrayDescriptor, inputB: BNNSNDArrayDescriptor, inputC: BNNSNDArrayDescriptor, output: BNNSNDArrayDescriptor, for: BNNS.LearningPhase) throws`](/documentation/Accelerate/BNNS/FusedParametersLayer/apply(batchSize:inputA:inputB:inputC:output:for:)) Applies the layer to a set of input objects and writes the result to a set of output objects, where the first layer accepts two inputs. [`func applyBackward(batchSize: Int, inputA: BNNSNDArrayDescriptor, inputB: BNNSNDArrayDescriptor, output: BNNSNDArrayDescriptor, outputGradient: BNNSNDArrayDescriptor, generatingInputAGradient: BNNSNDArrayDescriptor, generatingInputBGradient: BNNSNDArrayDescriptor, generatingParameterGradients: [BNNSNDArrayDescriptor]) throws`](/documentation/Accelerate/BNNS/FusedParametersLayer/applyBackward(batchSize:inputA:inputB:output:outputGradient:generatingInputAGradient:generatingInputBGradient:generatingParameterGradients:)) Applies the layer backward to generate input gradients, where the first layer accepts two inputs. [`func applyBackward(batchSize: Int, inputA: BNNSNDArrayDescriptor, inputB: BNNSNDArrayDescriptor, inputC: BNNSNDArrayDescriptor, output: BNNSNDArrayDescriptor, outputGradient: BNNSNDArrayDescriptor, generatingInputAGradient: BNNSNDArrayDescriptor, generatingInputBGradient: BNNSNDArrayDescriptor, generatingInputCGradient: BNNSNDArrayDescriptor, generatingParameterGradients: [BNNSNDArrayDescriptor]) throws`](/documentation/Accelerate/BNNS/FusedParametersLayer/applyBackward(batchSize:inputA:inputB:inputC:output:outputGradient:generatingInputAGradient:generatingInputBGradient:generatingInputCGradient:generatingParameterGradients:)) Applies the layer backward to generate input gradients, where the first layer accepts three inputs. ## Relationships ### Inherits From [`FusedLayer`](/documentation/Accelerate/BNNS/FusedLayer) --- Copyright © 2026 Apple Inc. All rights reserved. | [Terms of Use](https://www.apple.com/legal/internet-services/terms/site.html) | [Privacy Policy](https://www.apple.com/privacy/privacy-policy)