Classic BNNS API
Topics
N-dimensional array descriptor essentials
General filters
BNNSFilterApplying FiltersBNNS.LayerBNNS.UnaryLayerBNNS.BinaryLayerBNNSFilterParametersBNNSFilterDestroy(_:)BNNSAllocBNNSFree
Activation layers
BNNSFilterCreateVectorActivationLayer(_:_:_:_:)BNNS.ActivationLayerBNNSActivationFunctionBNNSActivationBNNSLayerParametersActivationBNNSFilterCreateLayerActivation(_:_:)BNNSDirectApplyActivationBatch(_:_:_:_:_:)applyActivation(activation:axes:input:output:batchSize:filterParameters:)applyActivation(activation:input:output:batchSize:filterParameters:)
Arithmetic layers
BNNS.UnaryArithmeticLayerBNNS.BinaryArithmeticLayerBNNS.TernaryArithmeticLayerBNNSDescriptorTypeBNNSArithmeticUnaryBNNSArithmeticBinaryBNNSArithmeticTernaryBNNSArithmeticFunctionBNNSLayerParametersArithmeticBNNSFilterCreateLayerArithmetic(_:_:)BNNSArithmeticFilterApplyBatch(_:_:_:_:_:_:_:)BNNSArithmeticFilterApplyBackwardBatch(_:_:_:_:_:_:_:_:_:_:_:)
Compute norm functions
computeNorm(input:output:axes:)computeNormBackward(input:output:axes:outputGradient:generatingInputGradient:)BNNSComputeNorm(_:_:_:_:)BNNSComputeNormBackward(_:_:_:_:_:_:)BNNSNormType
Convolution layers
BNNSConvolutionLayerParametersBNNSFilterCreateConvolutionLayer(_:_:_:_:)BNNS.ConvolutionLayerBNNSLayerParametersConvolutionBNNSFilterCreateLayerConvolution(_:_:)BNNSFilterCreateLayerTransposedConvolution(_:_:)
Crop-resize layers
BNNS.CropResizeLayerBNNSCropResize(_:_:_:_:_:)BNNSCropResizeBackward(_:_:_:_:_:)BNNSLayerParametersCropResizeBNNSBoxCoordinateModeBNNSLinearSamplingMode
Dropout layers
Embedding layers
Fully connected layers
BNNSFullyConnectedLayerParametersBNNSFilterCreateFullyConnectedLayer(_:_:_:_:)BNNS.FullyConnectedLayerBNNSLayerParametersFullyConnectedBNNSFilterCreateLayerFullyConnected(_:_:)
Fused layers
FusableLayerParametersBNNS.FusedParametersLayerBNNS.FusedConvolutionNormalizationLayerBNNS.FusedFullyConnectedNormalizationLayerBNNSFilterTypeBNNSFilterCreateFusedLayer(_:_:_:_:)BNNSFusedFilterApplyBatch(_:_:_:_:_:_:_:)BNNSFusedFilterApplyMultiInputBatch(_:_:_:_:_:_:_:_:)BNNSFusedFilterApplyBackwardBatch(_:_:_:_:_:_:_:_:_:_:_:)BNNSFusedFilterApplyBackwardMultiInputBatch(_:_:_:_:_:_:_:_:_:_:_:_:)
Gather and scatter operations
Calculating the dominant colors in an imagegather(input:indices:output:axis:filterParameters:)gatherND(input:indices:output:filterParameters:)scatter(input:indices:output:axis:reductionFunction:filterParameters:)scatterND(input:indices:output:reductionFunction:filterParameters:)BNNSGather(_:_:_:_:_:)BNNSGatherND(_:_:_:_:)BNNSScatter(_:_:_:_:_:_:)BNNSScatterND(_:_:_:_:_:)
Loss layers
BNNS.LossLayerBNNSLossFunctionBNNSLossReductionFunctionBNNSLayerParametersLossBaseBNNSLayerParametersLossHuberBNNSLayerParametersLossSigmoidCrossEntropyBNNSLayerParametersLossSoftmaxCrossEntropyBNNSLayerParametersLossYoloBNNSFilterCreateLayerLoss(_:_:)BNNSLossFilterApplyBatch(_:_:_:_:_:_:_:_:_:_:_:)BNNSLossFilterApplyBackwardBatch(_:_:_:_:_:_:_:_:_:_:_:_:)
K-nearest neighbors calculation
BNNS.NearestNeighborsBNNSNearestNeighborsBNNSCreateNearestNeighbors(_:_:_:_:_:)BNNSNearestNeighborsLoad(_:_:_:)BNNSNearestNeighborsGetInfo(_:_:_:_:)BNNSDestroyNearestNeighbors(_:)
Matrix multiplication
BNNSDirectApplyBroadcastMatMul(_:_:_:_:_:_:_:)BNNS.BroadcastMatrixMultiplyLayerBNNSLayerParametersBroadcastMatMulBNNSFilterCreateLayerBroadcastMatMul(_:_:)BNNSMatMulWorkspaceSize(_:_:_:_:_:_:_:)BNNSMatMul(_:_:_:_:_:_:_:_:)applyMatrixMultiplication(inputA:transposed:inputB:transposed:output:alpha:workspace:filterParameters:)matrixMultiplicationWorkspaceSize(inputA:transposed:inputB:transposed:output:alpha:filterParameters:)
Multihead attention layers
BNNSMHAProjectionParametersBNNSLayerParametersMultiheadAttentionBNNSFilterCreateLayerMultiheadAttention(_:_:)BNNSApplyMultiheadAttention(_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:)BNNSApplyMultiheadAttentionBackward(_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:_:)
Normalization layers
BNNS.NormalizationLayerBNNSLayerParametersNormalizationBNNSFilterCreateLayerNormalization(_:_:_:)BNNSNormalizationFilterApplyBatch(_:_:_:_:_:_:_:)BNNSNormalizationFilterApplyBackwardBatch(_:_:_:_:_:_:_:_:_:_:)
Optimizers
BNNS.AdamOptimizerBNNS.AdamWOptimizerBNNS.RMSPropOptimizerBNNS.SGDMomentumOptimizerBNNSOptimizerBNNSOptimizerRegularizationFunctionBNNSOptimizerAdamFieldsBNNSOptimizerAdamWithClippingFieldsBNNSOptimizerRMSPropFieldsBNNSOptimizerRMSPropWithClippingFieldsBNNSOptimizerSGDMomentumFieldsBNNSOptimizerSGDMomentumWithClippingFieldsBNNSOptimizerSGDMomentumVariantBNNSOptimizerStep(_:_:_:_:_:_:_:)BNNSOptimizerFunction
Padding layers
Permute layers
BNNS.PermuteLayerBNNSLayerParametersPermuteBNNSFilterCreateLayerPermute(_:_:)BNNSPermuteFilterApplyBackwardBatch(_:_:_:_:_:_:)
Pooling layers
BNNSPoolingLayerParametersBNNSFilterCreatePoolingLayer(_:_:_:_:)BNNS.PoolingLayerBNNSPoolingFunctionBNNSPoolingFunctionAverageBNNSPoolingFunctionMaxBNNSLayerParametersPoolingBNNSFilterCreateLayerPooling(_:_:)BNNSPoolingFilterApplyBatch(_:_:_:_:_:_:_:_:)BNNSPoolingFilterApplyBackwardBatch(_:_:_:_:_:_:_:_:_:_:_:_:_:)BNNSPoolingFilterApplyBatchEx(_:_:_:_:_:_:_:_:_:)BNNSPoolingFilterApplyBackwardBatchEx(_:_:_:_:_:_:_:_:_:_:_:_:_:_:)
Quantization functions
quantize(batchSize:input:output:axis:scale:bias:filterParameters:)dequantize(batchSize:input:output:axis:scale:bias:filterParameters:)BNNSQuantizerFunctionBNNSLayerParametersQuantizationBNNSDirectApplyQuantizer(_:_:_:_:_:)
Random number generation
BNNS.RandomGeneratorBNNSCreateRandomGenerator(_:_:)BNNSCreateRandomGeneratorWithSeed(_:_:_:)BNNSRandomGeneratorMethodBNNSRandomGeneratorBNNSRandomFillUniformInt(_:_:_:_:)BNNSRandomFillUniformFloat(_:_:_:_:)BNNSRandomFillNormalFloat(_:_:_:_:)BNNSRandomFillCategoricalFloat(_:_:_:_:)BNNSRandomGeneratorStateSize(_:)BNNSRandomGeneratorGetState(_:_:_:)BNNSRandomGeneratorSetState(_:_:_:)BNNSDestroyRandomGenerator(_:)
Recurrent layers
Using Long Short-Term Memory Layers (LSTM)BNNSLSTMDataDescriptorBNNSLSTMGateDescriptorBNNSLayerFlagsBNNSLayerParametersLSTMBNNSComputeLSTMTrainingCacheCapacity(_:)BNNSDirectApplyLSTMBatchTrainingCaching(_:_:_:_:)BNNSDirectApplyLSTMBatchBackward(_:_:_:_:_:)
Reduction layers
BNNS.ReductionLayerapplyReduction(_:input:output:weights:filterParameters:)BNNSReduceFunctionBNNSLayerParametersReductionBNNSFilterCreateLayerReduction(_:_:)BNNSDirectApplyReduction(_:_:)
Resize layers
Sparse layers
BNNSNDArrayGetDataSize(_:)BNNSNDArrayFullyConnectedSparsifySparseCOO(_:_:_:_:_:_:_:_:_:)BNNSNDArrayFullyConnectedSparsifySparseCSR(_:_:_:_:_:_:_:_:_:_:)sparsify(batchSize:inputLayout:inputDenseShape:inputValues:output:sparseParameters:workspace:filterParameters:)BNNS.SparseParametersBNNS.SparseLayoutBNNS.SparsityTypeBNNSSparsityTypeUnstructured
Tensor comparison layers
Tensor contraction layers
Top-k layers
applyTopK(k:input:bestValues:bestIndices:axis:batchSize:filterParameters:)applyInTopK(k:input:testIndices:output:axis:batchSize:filterParameters:)
Top-k layers
Utility functions
copy(_:to:filterParameters:)transpose(input:output:firstTransposeAxis:secondTransposeAxis:filterParameters:)BNNSCopy(_:_:_:)BNNSTranspose(_:_:_:_:_:)BNNSGetPointer(_:_:)BNNSPointerSpecifierBNNS.GramLayerBNNSLayerParametersGramBNNSFilterCreateLayerGram(_:_:)clip(to:input:output:)clipByNorm(threshold:input:output:axes:)clipByGlobalNorm(threshold:inputs:outputs:globalNorm:)BNNSClipByValue(_:_:_:_:)BNNSClipByNorm(_:_:_:_:)BNNSClipByGlobalNorm(_:_:_:_:_:)copyBandPart(_:to:lowerBandCount:upperBandCount:filterParameters:)shuffle(_:input:output:filterParameters:)BNNS.ShuffleTypetile(input:output:filterParameters:)tileBackward(outputGradient:generatingInputGradient:filterParameters:)
Errors
BNNS.ErrorBNNSBandPart(_:_:_:_:_:)BNNSShuffle(_:_:_:_:)BNNSShuffleTypeBNNSTile(_:_:_:)BNNSTileBackward(_:_:_:)