Crate objc2_ml_compute

Bindings to the MLCompute framework

See Apple’s docs and the general docs on framework crates for more information.

Structs

MLCActivationDescriptorDeprecatedMLCActivationDescriptor

The MLCActivationDescriptor specifies a neuron descriptor.

MLCActivationLayerDeprecatedMLCActivationLayer and MLCLayer

Apple’s documentation

MLCActivationTypeMLCTypes

An activation type that you specify for an activation descriptor.

MLCAdamOptimizerDeprecatedMLCAdamOptimizer and MLCOptimizer

The MLCAdamOptimizer specifies the Adam optimizer.

MLCAdamWOptimizerDeprecatedMLCAdamWOptimizer and MLCOptimizer

The MLCAdamWOptimizer specifies the AdamW optimizer.

MLCArithmeticLayerDeprecatedMLCArithmeticLayer and MLCLayer

An arithmetic layer

MLCArithmeticOperationMLCTypes

The list of supported arithmetic operations.

MLCBatchNormalizationLayerDeprecatedMLCBatchNormalizationLayer and MLCLayer

A batch normalizaion layer

MLCComparisonLayerDeprecatedMLCComparisonLayer and MLCLayer

Compare layer.

MLCComparisonOperationMLCTypes

A comparison operation.

MLCConcatenationLayerDeprecatedMLCConcatenationLayer and MLCLayer

A concatenation layer

MLCConvolutionDescriptorDeprecatedMLCConvolutionDescriptor

The MLCConvolutionDescriptor specifies a convolution descriptor

MLCConvolutionLayerDeprecatedMLCConvolutionLayer and MLCLayer

A convolution layer

MLCConvolutionTypeMLCTypes

A convolution type that you specify for a convolution descriptor.

MLCDataTypeMLCTypes

A tensor data type.

MLCDeviceDeprecatedMLCDevice

A device that will be used to execute a neural network. If a MLCdevice is created with multiple devices using the [devicesWithType:selectMultipleDvices], on configurations where multiple GPUs are available such as on the Mac Pro, the framework may transparently schedule the execution across multiple GPUs. There are some requirements for a MLCDevice with multiple devices such as there can only be one training and/or inference graph associated with this device. If multiple graphs are used, they must be compiled using MLCGraphCompilationOptionsLinkGraphs specified in compileOptions and the multiple graphs should be linked together with linkWithGraphs.

MLCDeviceTypeMLCTypes

A device type for execution of a neural network.

MLCDropoutLayerDeprecatedMLCDropoutLayer and MLCLayer

A dropout layer

MLCEmbeddingDescriptorDeprecatedMLCEmbeddingDescriptor

The MLCEmbeddingDescriptor specifies an embedding layer descriptor

MLCEmbeddingLayerDeprecatedMLCEmbeddingLayer and MLCLayer

An embedding layer which stores the words embedding

MLCExecutionOptionsMLCTypes

A bitmask that specifies the options you’ll use when executing a graph.

MLCFullyConnectedLayerDeprecatedMLCFullyConnectedLayer and MLCLayer

A fully connected layer a.k.a a dense layer

MLCGatherLayerDeprecatedMLCGatherLayer and MLCLayer

A gather layer

MLCGradientClippingTypeMLCTypes

The type of clipping applied to gradient

MLCGramMatrixLayerDeprecatedMLCGramMatrixLayer and MLCLayer

A gram matrix layer

MLCGraphDeprecatedMLCGraph

A graph of layers that can be used to build a training or inference graph

MLCGraphCompilationOptionsMLCTypes

A bitmask that specifies the options you use when compiling a graph.

MLCGroupNormalizationLayerDeprecatedMLCGroupNormalizationLayer and MLCLayer

A group normalizaion layer. For more information, refer to https://pytorch.org/docs/stable/nn.html#groupnorm

MLCInferenceGraphDeprecatedMLCGraph and MLCInferenceGraph

An inference graph created from one or more MLCGraph objects plus additional layers added directly to the inference graph.

MLCInstanceNormalizationLayerDeprecatedMLCInstanceNormalizationLayer and MLCLayer

An instance normalization layer. For more information refer to https://pytorch.org/docs/stable/nn.html#instancenorm2d

MLCLSTMDescriptorDeprecatedMLCLSTMDescriptor

The MLCLSTMDescriptor specifies a LSTM descriptor

MLCLSTMLayerDeprecatedMLCLSTMLayer and MLCLayer

A LSTM layer

MLCLSTMResultModeMLCTypes

A result mode for an LSTM layer.

MLCLayerDeprecatedMLCLayer

The base class for all MLCompute layers

MLCLayerNormalizationLayerDeprecatedMLCLayer and MLCLayerNormalizationLayer

The layer normalizaion layer. For more information, refer to https://pytorch.org/docs/stable/nn.html#layernorm.

MLCLossDescriptorDeprecatedMLCLossDescriptor

The MLCLossDescriptor specifies a loss filter descriptor.

MLCLossLayerDeprecatedMLCLayer and MLCLossLayer

A loss layer

MLCLossTypeMLCTypes

A loss function.

MLCMatMulDescriptorDeprecatedMLCMatMulDescriptor

The MLCMatMulDescriptor specifies a batched matrix multiplication descriptor

MLCMatMulLayerDeprecatedMLCLayer and MLCMatMulLayer

A batched matrix multiplication layer

MLCMultiheadAttentionDescriptorDeprecatedMLCMultiheadAttentionDescriptor

The MLCMultiheadAttentionDescriptor specifies a Multi-Head Attention descriptor

MLCMultiheadAttentionLayerDeprecatedMLCLayer and MLCMultiheadAttentionLayer

A multi-head attention layer

MLCOptimizerDeprecatedMLCOptimizer

The MLCOptimizer specifies a base optimizer.

MLCOptimizerDescriptorDeprecatedMLCOptimizerDescriptor

The MLCOptimizerDescriptor specifies an optimizer descriptor.

MLCPaddingLayerDeprecatedMLCLayer and MLCPaddingLayer

A padding layer

MLCPaddingPolicyMLCTypes

A padding policy that you specify for a convolution or pooling layer.

MLCPaddingTypeMLCTypes

A padding type that you specify for a padding layer.

MLCPlatformDeprecatedMLCPlatform

Utility class to set MLCompute global properties

MLCPoolingDescriptorDeprecatedMLCPoolingDescriptor

The MLCPoolingDescriptor specifies a pooling descriptor.

MLCPoolingLayerDeprecatedMLCLayer and MLCPoolingLayer

A pooling layer

MLCPoolingTypeMLCTypes

A pooling function type for a pooling layer.

MLCRMSPropOptimizerDeprecatedMLCOptimizer and MLCRMSPropOptimizer

The MLCRMSPropOptimizer specifies the RMSProp optimizer.

MLCRandomInitializerTypeMLCTypes

An initializer type you use to create a tensor with random data.

MLCReductionLayerDeprecatedMLCLayer and MLCReductionLayer

Reduce tensor values across a given dimension to a scalar value.

MLCReductionTypeMLCTypes

A reduction operation type.

MLCRegularizationTypeMLCTypes

Apple’s documentation

MLCReshapeLayerDeprecatedMLCLayer and MLCReshapeLayer

A reshape layer.

MLCSGDOptimizerDeprecatedMLCOptimizer and MLCSGDOptimizer

The MLCSGDOptimizer specifies a stochastic gradient descent optimizer.

MLCSampleModeMLCTypes

A sampling mode for an upsample layer.

MLCScatterLayerDeprecatedMLCLayer and MLCScatterLayer

A scatter layer

MLCSelectionLayerDeprecatedMLCLayer and MLCSelectionLayer

Selection layer is used to select elements from two tensors

MLCSliceLayerDeprecatedMLCLayer and MLCSliceLayer

Slice layer is used to slice a given source.

MLCSoftmaxLayerDeprecatedMLCLayer and MLCSoftmaxLayer

A softmax layer

MLCSoftmaxOperationMLCTypes

A softmax operation.

MLCSplitLayerDeprecatedMLCLayer and MLCSplitLayer

A split layer

MLCTensorDeprecatedMLCTensor

A tensor object

MLCTensorDataDeprecatedMLCTensorData

An object to encapsulate memory to be used as tensor data

MLCTensorDescriptorDeprecatedMLCTensorDescriptor

The MLCTensorDescriptor specifies a tensor descriptor.

MLCTensorOptimizerDeviceDataDeprecatedMLCTensorOptimizerDeviceData

Apple’s documentation

MLCTensorParameterDeprecatedMLCTensorParameter

A tensor parameter object. This is used to describe input tensors that are updated by the optimizer during training.

MLCTrainingGraphDeprecatedMLCGraph and MLCTrainingGraph

A training graph created from one or more MLCGraph objects plus additional layers added directly to the training graph.

MLCTransposeLayerDeprecatedMLCLayer and MLCTransposeLayer

A transpose layer

MLCUpsampleLayerDeprecatedMLCLayer and MLCUpsampleLayer

An upsample layer

MLCYOLOLossDescriptorDeprecatedMLCYOLOLossDescriptor

The MLCYOLOLossDescriptor specifies a YOLO loss filter descriptor.

MLCYOLOLossLayerDeprecatedMLCLayer and MLCLossLayer and MLCYOLOLossLayer

A YOLO loss layer

Functions

MLCActivationTypeDebugDescription^⚠DeprecatedMLCTypes

MLCArithmeticOperationDebugDescription^⚠DeprecatedMLCTypes

MLCComparisonOperationDebugDescription^⚠DeprecatedMLCTypes

MLCConvolutionTypeDebugDescription^⚠DeprecatedMLCTypes

MLCGradientClippingTypeDebugDescription^⚠DeprecatedMLCTypes

MLCLSTMResultModeDebugDescription^⚠DeprecatedMLCTypes

MLCLossTypeDebugDescription^⚠DeprecatedMLCTypes

MLCPaddingPolicyDebugDescription^⚠DeprecatedMLCTypes

MLCPaddingTypeDebugDescription^⚠DeprecatedMLCTypes

MLCPoolingTypeDebugDescription^⚠DeprecatedMLCTypes

MLCReductionTypeDebugDescription^⚠DeprecatedMLCTypes

MLCSampleModeDebugDescription^⚠DeprecatedMLCTypes

MLCSoftmaxOperationDebugDescription^⚠DeprecatedMLCTypes

Type Aliases

MLCGraphCompletionHandlerMLCTensor and MLCTypes and block2

A callback completion handler you execute when a graph finishes execution.