Here is a list of all documented functions with links to the class documentation for each member:

A

B

C

D

E

F

G

H

I

K

L

M

N

O

P

Q

R

S

T

U

V

W

Z

~

- g -

• GatedMLPConfig() : Mila::Dnn::GatedMLPConfig
• Gelu() : Mila::Dnn::Gelu< TDeviceType, TPrecision >
• GemmaConfig() : Mila::Dnn::GemmaConfig
• GemmaModelConfig() : Mila::Dnn::GemmaModelConfig
• gemmaStyle() : Mila::Data::SpecialTokens
• generate() : Mila::Dnn::LanguageModel< TDeviceType, TPrecision >
• getActivationType() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::MLPConfig
• getAllocationGranularity() : Mila::Dnn::Compute::CudaDevice, Mila::Dnn::Compute::Device
• getApproximationMethod() : Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GeluConfig
• getAttentionPackedQKVWidth() : Mila::Dnn::QwenConfig
• getAttentionScale() : Mila::Dnn::GqaConfig
• getAxis() : Mila::Dnn::Softmax< TDeviceType, TPrecision >, Mila::Dnn::SoftmaxConfig
• getBool() : Mila::Data::SerializationMetadata, Mila::Dnn::Optimizers::SerializationMetadata, Mila::Dnn::Serialization::SerializationMetadata, Mila::Dnn::SerializationMetadata
• getBosTokenId() : Mila::Data::CharTokenizer
• getCommandQueue() : Mila::Dnn::Compute::ExecutionContext< DeviceType::Metal >
• getComponent() : Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >
• getComponentAs() : Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >
• getComponentName() : Mila::Dnn::ComponentFactory
• getComponents() : Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >
• getComponentVersion() : Mila::Dnn::ComponentFactory
• getComputeCapability() : Mila::Dnn::Compute::CudaDevice, Mila::Dnn::Compute::CudaDeviceProps
• getComputeCapabilityVersion() : Mila::Dnn::Compute::CudaDevice
• getConfig() : Mila::Data::CharTrainer, Mila::Data::CharVocabulary
• getConnectionType() : Mila::Dnn::ResidualConfig
• getCorpusSize() : Mila::Data::CharTrainer
• getCublasLtHandle() : Mila::Dnn::Compute::CudaExecutionContext
• getCublasLtWorkspace() : Mila::Dnn::Compute::CudaExecutionContext
• getCublasLtWorkspaceSize() : Mila::Dnn::Compute::CudaExecutionContext
• getCurandGenerator() : Mila::Dnn::Compute::CudaExecutionContext
• getDatasetInfo() : Mila::Data::DataLoader< TInputDataType, TTargetDataType, TMemoryResource >
• getDataType() : Mila::Dnn::Compute::Operation< TDeviceType, TComputePrecision >, Mila::Dnn::ITensor, Mila::Dnn::Tensor< TDataType, TMemoryResource >
• getDataTypeName() : Mila::Dnn::ITensor, Mila::Dnn::Tensor< TDataType, TMemoryResource >
• getDeltaNetGatingWidth() : Mila::Dnn::QwenConfig
• getDeltaNetKeyWidth() : Mila::Dnn::QwenConfig
• getDeltaNetQueryKeyWidth() : Mila::Dnn::QwenConfig
• getDeltaNetValueWidth() : Mila::Dnn::QwenConfig
• getDeploymentFootprint() : Mila::Dnn::GemmaModel< TDeviceType, TPrecision >, Mila::Dnn::LlamaModel< TDeviceType, TPrecision >, Mila::Dnn::QwenModel< TDeviceType, TPrecision >
• getDevice() : Mila::Dnn::Compute::DeviceRegistry, Mila::Dnn::Compute::ExecutionContext< DeviceType::Vulkan >
• getDeviceCount() : Mila::Dnn::Compute::CudaDeviceRegistrar, Mila::Dnn::Compute::DeviceRegistry, Mila::Dnn::Compute::MetalDevicePlugin
• getDeviceId() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >, Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >, Mila::Dnn::Compute::CpuDevice, Mila::Dnn::Compute::CpuExecutionContext, Mila::Dnn::Compute::CudaDevice, Mila::Dnn::Compute::CudaExecutionContext, Mila::Dnn::Compute::Device, Mila::Dnn::Compute::ExecutionContext< DeviceType::Metal >, Mila::Dnn::Compute::ExecutionContext< DeviceType::Vulkan >, Mila::Dnn::Compute::IExecutionContext, Mila::Dnn::Compute::MetalDevice, Mila::Dnn::Compute::VulkanDevice, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MixtureOfExperts< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::Model< TDeviceType, TPrecision >, Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >, Mila::Dnn::Network< TDeviceType, TPrecision >, Mila::Dnn::Residual< TDeviceType, TPrecision >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::Rope< TDeviceType, TPrecision >, Mila::Dnn::Softmax< TDeviceType, TPrecision >, Mila::Dnn::SoftmaxCrossEntropy< TDeviceType, TPrecision >, Mila::Dnn::Swiglu< TDeviceType, TPrecision, TGate >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getDeviceIndex() : Mila::Dnn::Compute::MetalDevice
• getDeviceMemorySize() : Mila::Dnn::Compute::MetalDevice, Mila::Dnn::Compute::VulkanDevice
• getDeviceName() : Mila::Dnn::Compute::CpuDevice, Mila::Dnn::Compute::CudaDevice, Mila::Dnn::Compute::Device, Mila::Dnn::Compute::ExecutionContext< DeviceType::Metal >, Mila::Dnn::Compute::ExecutionContext< DeviceType::Vulkan >, Mila::Dnn::Compute::MetalDevice, Mila::Dnn::Compute::VulkanDevice
• getDeviceScratchBuffer() : Mila::Dnn::Compute::CudaExecutionContext
• getDeviceType() : Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::Compute::CpuDevice, Mila::Dnn::Compute::CudaDevice, Mila::Dnn::Compute::Device, Mila::Dnn::Compute::ExecutionContext< DeviceType::Metal >, Mila::Dnn::Compute::ExecutionContext< DeviceType::Vulkan >, Mila::Dnn::Compute::MetalDevice, Mila::Dnn::Compute::Operation< TDeviceType, TComputePrecision >, Mila::Dnn::Compute::VulkanDevice, Mila::Dnn::ITensor, Mila::Dnn::Tensor< TDataType, TMemoryResource >
• getDouble() : Mila::Data::SerializationMetadata, Mila::Dnn::Optimizers::SerializationMetadata, Mila::Dnn::Serialization::SerializationMetadata, Mila::Dnn::SerializationMetadata
• getDoubleVector() : Mila::Data::SerializationMetadata, Mila::Dnn::Optimizers::SerializationMetadata, Mila::Dnn::Serialization::SerializationMetadata, Mila::Dnn::SerializationMetadata
• getEffectiveHiddenDimension() : Mila::Dnn::GptBlockConfig
• getEmbeddingDim() : Mila::Dnn::LpeConfig
• getEmbeddingScale() : Mila::Dnn::GemmaConfig
• getEnabledTokens() : Mila::Data::SpecialTokens
• getEosTokenId() : Mila::Data::CharTokenizer
• getExecutionContext() : Mila::Dnn::Component< TDeviceType, TPrecision >
• getExpertIntermediateSize() : Mila::Dnn::MixtureOfExpertsConfig
• getFilename() : Mila::Dnn::Serialization::Serializer, Mila::Dnn::Serialization::ZipSerializer
• getFileSize() : Mila::Dnn::Serialization::ArchiveSerializer, Mila::Dnn::Serialization::ZipSerializer
• getGatedQProjectionWidth() : Mila::Dnn::QwenConfig
• getGateType() : Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >
• getGenerator() : Mila::Core::RandomGenerator
• getGlobalHeadDim() : Mila::Dnn::GemmaConfig
• getGlobalKVProjectionWidth() : Mila::Dnn::GemmaConfig
• getGlobalPackedQKVWidth() : Mila::Dnn::GemmaConfig
• getGlobalQProjectionWidth() : Mila::Dnn::GemmaConfig
• getGradients() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >, Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MixtureOfExperts< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >, Mila::Dnn::Residual< TDeviceType, TPrecision >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::Rope< TDeviceType, TPrecision >, Mila::Dnn::Softmax< TDeviceType, TPrecision >, Mila::Dnn::SoftmaxCrossEntropy< TDeviceType, TPrecision >, Mila::Dnn::Swiglu< TDeviceType, TPrecision, TGate >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getGroupSize() : Mila::Dnn::GqaConfig
• getHeadDim() : Mila::Dnn::GemmaConfig, Mila::Dnn::GqaConfig, Mila::Dnn::RopeConfig
• getHeadGroupSize() : Mila::Dnn::GatedDeltaRuleConfig
• getHiddenSize() : Mila::Dnn::MLPConfig
• getIdOffset() : Mila::Data::SpecialTokens
• getInputFeatures() : Mila::Dnn::LinearConfig, Mila::Dnn::MLPConfig
• getInstance() : Mila::Core::RandomGenerator
• getInt() : Mila::Data::SerializationMetadata, Mila::Dnn::Optimizers::SerializationMetadata, Mila::Dnn::Serialization::SerializationMetadata, Mila::Dnn::SerializationMetadata
• getIntVector() : Mila::Data::SerializationMetadata, Mila::Dnn::Optimizers::SerializationMetadata, Mila::Dnn::Serialization::SerializationMetadata, Mila::Dnn::SerializationMetadata
• getKeyWidth() : Mila::Dnn::GatedDeltaRuleConfig
• getKVProjectionWidth() : Mila::Dnn::GemmaConfig, Mila::Dnn::QwenConfig
• getLearningRate() : Mila::Dnn::ConstantLRScheduler, Mila::Dnn::CosineLRScheduler, Mila::Dnn::LearningRateScheduler, Mila::Dnn::LinearLRScheduler, Mila::Dnn::Optimizer< TDeviceType, TPrecision >, Mila::Dnn::Optimizers::AdamWOptimizer< TDeviceType, TPrecision >
• getLevel() : Mila::Logging::ConsoleSink, Mila::Logging::FileSink, Mila::Logging::Logger, Mila::Logging::NullSink
• getLoadStagingBuffer() : Mila::Dnn::Compute::CudaExecutionContext
• getLoadStagingLimitBytes() : Mila::Dnn::Compute::CudaExecutionContext
• getMajor() : Mila::Version
• getMaxSequenceLength() : Mila::Dnn::LpeConfig, Mila::Dnn::RopeConfig
• getMaxTensorSizeBytes() : Mila::Dnn::Serialization::WeightsReader
• getMaxThreadsPerBlock() : Mila::Dnn::Compute::CudaDevice
• getMaxThreadsPerThreadgroup() : Mila::Dnn::Compute::MetalDevice
• getMaxWorkGroupSize() : Mila::Dnn::Compute::VulkanDevice
• getMemoryInfo() : Mila::Dnn::Compute::CudaDevice, Mila::Dnn::Compute::Device
• getMemoryResource() : Mila::Dnn::ITensor, Mila::Dnn::Tensor< TDataType, TMemoryResource >, Mila::Dnn::TensorBuffer< TDataType, TMemoryResource, TrackMemory >
• getMemoryStats() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >, Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GemmaBlock< TDeviceType, TPrecision, kGlobal, TWeightQuant, TKvPolicy, kDelegatedFeedForward, kMixtureOfExperts >, Mila::Dnn::GemmaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy, kDelegatedFeedForward, kMixtureOfExperts >, Mila::Dnn::GptBlock< TDeviceType, TPrecision >, Mila::Dnn::GptTransformer< TDeviceType, TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::LlamaBlock< TDeviceType, TPrecision, TWeightQuant, TKvPolicy >, Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MixtureOfExperts< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::MLP< TDeviceType, TPrecision >, Mila::Dnn::Model< TDeviceType, TPrecision >, Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >, Mila::Dnn::QwenAttentionBlock< TDeviceType, TPrecision, TWeightPlan, TKvPolicy >, Mila::Dnn::QwenDeltaNetBlock< TDeviceType, TPrecision, TWeightPlan >, Mila::Dnn::QwenTransformer< TDeviceType, TPrecision, TWeightPlan, TKvCachePolicy >, Mila::Dnn::Residual< TDeviceType, TPrecision >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::Rope< TDeviceType, TPrecision >, Mila::Dnn::Router< TDeviceType, TPrecision >, Mila::Dnn::Softmax< TDeviceType, TPrecision >, Mila::Dnn::Swiglu< TDeviceType, TPrecision, TGate >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getMetadata() : Mila::Dnn::Serialization::ArchiveSerializer, Mila::Dnn::Serialization::ModelArchive, Mila::Dnn::Serialization::ZipSerializer
• getMinor() : Mila::Version
• getModelConfig() : Mila::Dnn::GemmaModel< TDeviceType, TPrecision >, Mila::Dnn::QwenModel< TDeviceType, TPrecision >
• getModelDim() : Mila::Dnn::GemmaConfig, Mila::Dnn::GqaConfig, Mila::Dnn::MultiHeadAttentionConfig, Mila::Dnn::QwenConfig
• getMultiprocessorCount() : Mila::Dnn::Compute::CudaDevice
• getName() : Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::Compute::CudaDeviceProps, Mila::Dnn::Compute::Operation< TDeviceType, TComputePrecision >, Mila::Dnn::ITensor, Mila::Dnn::Tensor< TDataType, TMemoryResource >
• getNativeDevice() : Mila::Dnn::Compute::MetalDevice
• getNativeHandle() : Mila::Dnn::Compute::VulkanDevice
• getNetworkConfig() : Mila::Dnn::GemmaModel< TDeviceType, TPrecision >, Mila::Dnn::QwenModel< TDeviceType, TPrecision >
• getNextId() : Mila::Dnn::UniqueIdGenerator
• getNumDeltaNetLayers() : Mila::Dnn::QwenConfig
• getNumFullAttentionLayers() : Mila::Dnn::QwenConfig
• getNumGlobalKVHeads() : Mila::Dnn::GemmaConfig
• getNumHeads() : Mila::Dnn::GqaConfig, Mila::Dnn::MultiHeadAttentionConfig
• getNumKvHeads() : Mila::Dnn::GqaConfig
• getObservableStages() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >, Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GemmaBlock< TDeviceType, TPrecision, kGlobal, TWeightQuant, TKvPolicy, kDelegatedFeedForward, kMixtureOfExperts >, Mila::Dnn::GptBlock< TDeviceType, TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::LlamaBlock< TDeviceType, TPrecision, TWeightQuant, TKvPolicy >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MixtureOfExperts< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::MLP< TDeviceType, TPrecision >, Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >, Mila::Dnn::QwenAttentionBlock< TDeviceType, TPrecision, TWeightPlan, TKvPolicy >, Mila::Dnn::QwenDeltaNetBlock< TDeviceType, TPrecision, TWeightPlan >, Mila::Dnn::Residual< TDeviceType, TPrecision >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::Rope< TDeviceType, TPrecision >, Mila::Dnn::Router< TDeviceType, TPrecision >, Mila::Dnn::Softmax< TDeviceType, TPrecision >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getOperationType() : Mila::Dnn::Compute::Operation< TDeviceType, TComputePrecision >
• getOutputFeatures() : Mila::Dnn::LinearConfig
• getOutputs() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >, Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >, Mila::Dnn::Residual< TDeviceType, TPrecision >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getPackedQKVWidth() : Mila::Dnn::GemmaConfig, Mila::Dnn::QwenConfig
• getPackedQKVWidthForLayer() : Mila::Dnn::GemmaConfig
• getPadTokenId() : Mila::Data::CharTokenizer
• getParameterNames() : Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::GemmaBlock< TDeviceType, TPrecision, kGlobal, TWeightQuant, TKvPolicy, kDelegatedFeedForward, kMixtureOfExperts >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MixtureOfExperts< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::Router< TDeviceType, TPrecision >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getParameters() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >, Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MixtureOfExperts< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >, Mila::Dnn::Residual< TDeviceType, TPrecision >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::Rope< TDeviceType, TPrecision >, Mila::Dnn::Softmax< TDeviceType, TPrecision >, Mila::Dnn::SoftmaxCrossEntropy< TDeviceType, TPrecision >, Mila::Dnn::Swiglu< TDeviceType, TPrecision, TGate >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getPatch() : Mila::Version
• getPinnedStagingBuffer() : Mila::Dnn::Compute::CudaExecutionContext
• getPluginName() : Mila::Dnn::Compute::MetalDevicePlugin
• getPrecision() : Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::ComponentFactory
• getPrefillSize() : Mila::Dnn::BuildContext
• getPreRelease() : Mila::Version
• getProperties() : Mila::Dnn::Compute::CudaDevice
• getQProjectionWidth() : Mila::Dnn::GemmaConfig, Mila::Dnn::QwenConfig
• getQProjectionWidthForLayer() : Mila::Dnn::GemmaConfig
• getQueue() : Mila::Dnn::Compute::ExecutionContext< DeviceType::Vulkan >
• getRecommendedMaxWorkingSetSize() : Mila::Dnn::Compute::MetalDevice
• getRequiredMemory() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >, Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::GemmaBlock< TDeviceType, TPrecision, kGlobal, TWeightQuant, TKvPolicy, kDelegatedFeedForward, kMixtureOfExperts >, Mila::Dnn::GemmaModel< TDeviceType, TPrecision >, Mila::Dnn::GemmaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy, kDelegatedFeedForward, kMixtureOfExperts >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::LlamaBlock< TDeviceType, TPrecision, TWeightQuant, TKvPolicy >, Mila::Dnn::LlamaModel< TDeviceType, TPrecision >, Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >, Mila::Dnn::MixtureOfExperts< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::QwenAttentionBlock< TDeviceType, TPrecision, TWeightPlan, TKvPolicy >, Mila::Dnn::QwenDeltaNetBlock< TDeviceType, TPrecision, TWeightPlan >, Mila::Dnn::QwenModel< TDeviceType, TPrecision >, Mila::Dnn::QwenTransformer< TDeviceType, TPrecision, TWeightPlan, TKvCachePolicy >, Mila::Dnn::Residual< TDeviceType, TPrecision >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::Rope< TDeviceType, TPrecision >, Mila::Dnn::Router< TDeviceType, TPrecision >, Mila::Dnn::Swiglu< TDeviceType, TPrecision, TGate >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getRequiredScratchBytes() : Mila::Dnn::Compute::Operation< TDeviceType, TComputePrecision >
• getRequiredStateMemorySize() : Mila::Dnn::Compute::Operation< TDeviceType, TComputePrecision >
• getRotaryDim() : Mila::Dnn::QwenConfig
• getRotaryDimForLayer() : Mila::Dnn::GemmaConfig
• getRuntimeMode() : Mila::Dnn::BuildContext, Mila::Dnn::Model< TDeviceType, TPrecision >
• getScalingFactor() : Mila::Dnn::ResidualConfig
• getScratchBytes() : Mila::Dnn::Compute::Operation< TDeviceType, TComputePrecision >
• getSeed() : Mila::Core::RandomGenerator
• getShape() : Mila::Data::SerializationMetadata, Mila::Dnn::Optimizers::SerializationMetadata, Mila::Dnn::Serialization::SerializationMetadata, Mila::Dnn::SerializationMetadata
• getSharedMemoryPerBlock() : Mila::Dnn::Compute::CudaDevice
• getSize() : Mila::Data::BpeVocabulary, Mila::Data::CharVocabulary, Mila::Data::TokenizerVocabulary
• getSpecialTokenId() : Mila::Data::BpeVocabulary
• getSpecialTokenList() : Mila::Data::BpeVocabulary
• getStateElementsPerBatch() : Mila::Dnn::GatedDeltaRuleConfig
• getStateMemorySize() : Mila::Dnn::Compute::Operation< TDeviceType, TComputePrecision >
• getStateRows() : Mila::Dnn::CausalConv1dConfig
• getStorageSize() : Mila::Dnn::ITensor, Mila::Dnn::Tensor< TDataType, TMemoryResource >
• getStream() : Mila::Dnn::Compute::CudaExecutionContext
• getString() : Mila::Data::SerializationMetadata, Mila::Data::SpecialTokens, Mila::Distribution::HttpClient, Mila::Dnn::Optimizers::SerializationMetadata, Mila::Dnn::Serialization::SerializationMetadata, Mila::Dnn::SerializationMetadata
• getStringVector() : Mila::Data::SerializationMetadata, Mila::Dnn::Optimizers::SerializationMetadata, Mila::Dnn::Serialization::SerializationMetadata, Mila::Dnn::SerializationMetadata
• getTensorCount() : Mila::Dnn::Serialization::SafeTensorsWriter
• getTensorNames() : Mila::Dnn::Serialization::WeightsReader
• getTensorSizeBytes() : Mila::Dnn::Serialization::WeightsReader
• getTopK() : Mila::Dnn::MixtureOfExpertsConfig, Mila::Dnn::RouterConfig
• getTotalGlobalMemory() : Mila::Dnn::Compute::CudaDevice
• getTrainingMode() : Mila::Dnn::Component< TDeviceType, TPrecision >
• getType() : Mila::Dnn::Activation< TDeviceType, TPrecision, TFn >, Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >, Mila::Dnn::CausalConv1d< TDeviceType, TPrecision >, Mila::Dnn::Component< TDeviceType, TPrecision >, Mila::Dnn::GatedDeltaRule< TDeviceType, TPrecision >, Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GemmaBlock< TDeviceType, TPrecision, kGlobal, TWeightQuant, TKvPolicy, kDelegatedFeedForward, kMixtureOfExperts >, Mila::Dnn::GptBlock< TDeviceType, TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, Mila::Dnn::Linear< TDeviceType, TComputePrecision, TWeightQuant >, Mila::Dnn::LlamaBlock< TDeviceType, TPrecision, TWeightQuant, TKvPolicy >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MixtureOfExperts< TDeviceType, TPrecision, TGate, TWeightQuantization >, Mila::Dnn::MLP< TDeviceType, TPrecision >, Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >, Mila::Dnn::Network< TDeviceType, TPrecision >, Mila::Dnn::QwenAttentionBlock< TDeviceType, TPrecision, TWeightPlan, TKvPolicy >, Mila::Dnn::QwenDeltaNetBlock< TDeviceType, TPrecision, TWeightPlan >, Mila::Dnn::Residual< TDeviceType, TPrecision >, Mila::Dnn::RmsNorm< TDeviceType, TPrecision >, Mila::Dnn::Rope< TDeviceType, TPrecision >, Mila::Dnn::Router< TDeviceType, TPrecision >, Mila::Dnn::Softmax< TDeviceType, TPrecision >, Mila::Dnn::Swiglu< TDeviceType, TPrecision, TGate >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision, TTableQuantization, TTableResidency >
• getUId() : Mila::Dnn::ITensor, Mila::Dnn::Tensor< TDataType, TMemoryResource >
• getUnderlying() : Mila::Dnn::Compute::TrackedMemoryResource
• getValueWidth() : Mila::Dnn::GatedDeltaRuleConfig
• getVocabSize() : Mila::Data::CharTokenizer, Mila::Dnn::CrossEntropyConfig
• getVocabularyLength() : Mila::Dnn::LpeConfig
• getWarpSize() : Mila::Dnn::Compute::CudaDevice
• getWeightQuantization() : Mila::Dnn::Serialization::WeightsReader
• getWeightsMetadata() : Mila::Dnn::Serialization::WeightsReader
• getWindow() : Mila::Dnn::GqaConfig
• getWindowForLayer() : Mila::Dnn::GemmaConfig
• GptBlock() : Mila::Dnn::GptBlock< TDeviceType, TPrecision >
• GptBlockConfig() : Mila::Dnn::GptBlockConfig
• GptModelConfig() : Mila::Dnn::GptModelConfig
• gptStyle() : Mila::Data::SpecialTokens
• GptTransformer() : Mila::Dnn::GptTransformer< TDeviceType, TPrecision >
• GqaConfig() : Mila::Dnn::GqaConfig
• GroupedQueryAttention() : Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >