Here is a list of all class members with links to the classes they belong to:
- m -
- m_data_ : Mila::Dnn::Compute::CpuAdamWOptimizer< TPrecision >, Mila::Dnn::Compute::CudaAdamWOptimizer< TPrecision >
- m_states_ : Mila::Dnn::Compute::CpuAdamWOptimizer< TPrecision >, Mila::Dnn::Compute::CudaAdamWOptimizer< TPrecision >
- MAGIC : Mila::Data::MilaFileHeader, Mila::Dnn::Serialization::PretrainedModelReader
- major : Mila::Dnn::Compute::CudaDeviceProps
- major_ : Mila::Version
- make_native_value() : Mila::Dnn::VulkanTensorTraits
- make_temp_generator_() : Mila::Dnn::Compute::Cuda::RandomOps
- makeCacheKey() : Mila::Dnn::Compute::Cuda::Rope::CudaRopeOp< TComputePrecision >
- makeTokenTensor() : Mila::Dnn::GptModel< TDeviceType, TPrecision >, Mila::Dnn::LlamaModel< TDeviceType, TPrecision >
- mask_ : Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >
- mask_token : Mila::Data::SpecialTokens
- master_param_data_ : Mila::Dnn::Compute::CudaAdamWOptimizer< TPrecision >
- master_params_ : Mila::Dnn::Compute::CudaAdamWOptimizer< TPrecision >
- matmul_desc : Mila::Dnn::Compute::Cuda::CublasLtLinearPlan< TComputePrecision, TParameterPrecision >, Mila::Dnn::Compute::Cuda::CublasLtMatMulPlan< TComputePrecision >
- max_batch_size_ : Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >
- max_inner_size_ : Mila::Dnn::Compute::Cuda::LayerNorm::CudaLayerNormOp< TPrecision >
- max_input_shape_ : Mila::Dnn::Compute::Cuda::LayerNorm::CudaLayerNormOp< TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >
- max_merges_ : Mila::Data::BpeVocabularyConfig
- max_new_tokens : Mila::Dnn::GenerateParams
- max_norm_dim_ : Mila::Dnn::Compute::Cuda::LayerNorm::CudaLayerNormOp< TPrecision >
- max_num_slices_ : Mila::Dnn::Compute::Cuda::LayerNorm::CudaLayerNormOp< TPrecision >
- max_outer_size_ : Mila::Dnn::Compute::Cuda::LayerNorm::CudaLayerNormOp< TPrecision >
- max_queue_size : Mila::Data::TokenSequenceLoaderConfig
- max_seq_len : Mila::Dnn::Compute::Cuda::Rope::RopeCacheRegistry::CacheKey
- max_seq_len_ : Mila::Dnn::GptBlockConfig, Mila::Dnn::GptConfig, Mila::Dnn::LlamaConfig, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::LpeConfig, Mila::Dnn::RopeConfig, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >
- max_seq_length : Mila::Dnn::Serialization::PretrainedMetadata
- max_size_ : Mila::Dnn::Compute::Cuda::Residual::CudaResidualOp< TInputA, TInputB, TPrecision >
- maxGridSize : Mila::Dnn::Compute::CudaDeviceProps
- MaxRank : Mila::Dnn::TensorShape
- maxSequenceLength() : Mila::Dnn::GptModel< TDeviceType, TPrecision >, Mila::Dnn::LanguageModel< TDeviceType, TPrecision >, Mila::Dnn::LlamaModel< TDeviceType, TPrecision >
- maxThreadsDim : Mila::Dnn::Compute::CudaDeviceProps
- maxThreadsPerBlock : Mila::Dnn::Compute::CudaDeviceProps
- mean_ : Mila::Dnn::Compute::CpuLayerNormOp, Mila::Dnn::Compute::Cuda::LayerNorm::CudaLayerNormOp< TPrecision >
- mean_tensor_ : Mila::Dnn::Compute::Cuda::LayerNorm::CudaLayerNormOp< TPrecision >
- memcpyOperationCount : Mila::Dnn::Compute::MemoryStats
- memory_resource : Mila::Dnn::Compute::DeviceTypeTraits< DeviceType::Cpu >, Mila::Dnn::Compute::DeviceTypeTraits< DeviceType::Cuda >
- memoryBusWidth : Mila::Dnn::Compute::CudaDeviceProps
- memoryClockRate : Mila::Dnn::Compute::CudaDeviceProps
- MemoryResource : Mila::Data::DataLoader< TInputDataType, TTargetDataType, TMemoryResource >, Mila::Dnn::Tensor< TDataType, TMemoryResource >
- memsetOperationCount : Mila::Dnn::Compute::MemoryStats
- merge_map_ : Mila::Data::BpeVocabulary
- merges_ : Mila::Data::BpeVocabulary
- metadata : Mila::Dnn::Serialization::TensorBlob< MR >
- metadata_ : Mila::Data::MilaFileHeader, Mila::Dnn::Serialization::PretrainedModelReader
- Metal() : Mila::Dnn::Compute::Device
- MetalDevice() : Mila::Dnn::Compute::MetalDevice
- MetalMemoryResource() : Mila::Dnn::Compute::MetalMemoryResource
- mila_api_version : Mila::Dnn::Extensibility::PluginInfo
- MilaFileHeader() : Mila::Data::MilaFileHeader
- min_frequency_ : Mila::Data::BpeVocabularyConfig
- minimal() : Mila::Data::SpecialTokens
- minor : Mila::Dnn::Compute::CudaDeviceProps
- minor_ : Mila::Version
- MLP() : Mila::Dnn::MLP< TDeviceType, TPrecision >
- mlp_act : Mila::Dnn::Visualization::VisualizerContext
- mlp_l1 : Mila::Dnn::Visualization::VisualizerContext
- mlp_l2 : Mila::Dnn::Visualization::VisualizerContext
- MLPConfig() : Mila::Dnn::MLPConfig
- MLPType : Mila::Dnn::GptBlock< TDeviceType, TPrecision >
- MLPVisualizer() : Mila::Dnn::Visualization::MLPVisualizer
- mode_ : Mila::Dnn::Serialization::ModelArchive
- Model() : Mila::Dnn::Model< TDeviceType, TPrecision >
- model_dim_ : Mila::Dnn::GptBlockConfig, Mila::Dnn::GqaConfig, Mila::Dnn::MultiHeadAttentionConfig
- model_name : Mila::Dnn::Serialization::PretrainedMetadata
- ModelArchive() : Mila::Dnn::Serialization::ModelArchive
- ModelBase : Mila::Dnn::GptModel< TDeviceType, TPrecision >, Mila::Dnn::LlamaModel< TDeviceType, TPrecision >
- ModelConfig() : Mila::Dnn::ModelConfig
- MR : Mila::Dnn::Compute::BinaryOperation< TDeviceType, TPrecision, TInputA, TInputB >, Mila::Dnn::Compute::CpuAdamWOptimizer< TPrecision >, Mila::Dnn::Compute::CpuAttentionOp, Mila::Dnn::Compute::CpuCrossEntropyOp, Mila::Dnn::Compute::CpuEncoderOp, Mila::Dnn::Compute::CpuGeluOp, Mila::Dnn::Compute::CpuLayerNormOp, Mila::Dnn::Compute::CpuLinearOp, Mila::Dnn::Compute::CpuResidualOp, Mila::Dnn::Compute::CpuSoftmaxCrossEntropyOp< TPrecision, TLogits, TTargets >, Mila::Dnn::Compute::CpuSoftmaxOp, Mila::Dnn::Compute::Cuda::Gelu::CudaGeluOp< TPrecision >, Mila::Dnn::Compute::Cuda::Gqa::CudaGqaOp< TPrecision >, Mila::Dnn::Compute::Cuda::LayerNorm::CudaLayerNormOp< TPrecision >, Mila::Dnn::Compute::Cuda::Linear::CudaLinearOp< TComputePrecision, TWeightQuant >, Mila::Dnn::Compute::Cuda::Lpe::CudaLpeOp< TInput, TPrecision >, Mila::Dnn::Compute::Cuda::MatMulBiasGelu::CudaMatMulBiasGeluOp< TInput, TOutput >, Mila::Dnn::Compute::Cuda::MultiHeadAttention::CudaMultiHeadAttentionOp< TPrecision >, Mila::Dnn::Compute::Cuda::Residual::CudaResidualOp< TInputA, TInputB, TPrecision >, Mila::Dnn::Compute::Cuda::RmsNorm::CudaRmsNormOp< TPrecision >, Mila::Dnn::Compute::Cuda::Rope::CudaRopeOp< TComputePrecision >, Mila::Dnn::Compute::Cuda::Softmax::CudaSoftmaxOp< TPrecision >, Mila::Dnn::Compute::Cuda::SoftmaxCrossEntropy::CudaSoftmaxCrossEntropyOp< TPrecision, TLogits, TTargets >, Mila::Dnn::Compute::Cuda::Swiglu::CudaSwigluOp< TPrecision >, Mila::Dnn::Compute::Cuda::TokenEmbedding::CudaTokenEmbeddingOp< TInput, TPrecision >, Mila::Dnn::Compute::CudaAdamWOptimizer< TPrecision >, Mila::Dnn::Compute::PairedOperation< TDeviceType, TPrecision, TInputA, TInputB >, Mila::Dnn::Compute::UnaryOperation< TDeviceType, TInput, TPrecision >, Mila::Dnn::Detail::mlp_activation_impl< ActivationType::Gelu, TDeviceType, TPrecision >, Mila::Dnn::Detail::mlp_activation_impl< ActivationType::Swiglu, TDeviceType, TPrecision >, Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >, Mila::Dnn::Gelu< TDeviceType, TPrecision >, Mila::Dnn::GptBlock< TDeviceType, TPrecision >, Mila::Dnn::GptModel< TDeviceType, TPrecision >, Mila::Dnn::GptTransformer< TDeviceType, TPrecision >, Mila::Dnn::GroupedQueryAttention< TDeviceType, TComputePrecision, TKvPolicy >, Mila::Dnn::LanguageNetwork< TDeviceType, TPrecision >, Mila::Dnn::LayerNorm< TDeviceType, TPrecision >, 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::Loss< TDeviceType, TPrecision >, Mila::Dnn::Lpe< TDeviceType, TIndex, TPrecision >, Mila::Dnn::MLP< 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, TLogits, TTargets, TPrecision >, Mila::Dnn::Swiglu< TDeviceType, TPrecision >, Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >
- mr_ : Mila::Dnn::TensorBuffer< TDataType, TMemoryResource, TrackMemory >
- MultiHeadAttention() : Mila::Dnn::MultiHeadAttention< TDeviceType, TPrecision >
- MultiHeadAttentionConfig() : Mila::Dnn::MultiHeadAttentionConfig
- multiply() : Mila::Dnn::Compute::Cpu::MathOps, Mila::Dnn::Compute::Cuda::MathOps
- multiplyImpl() : Mila::Dnn::Compute::Cuda::MathOps
- multiProcessorCount : Mila::Dnn::Compute::CudaDeviceProps
- mutex_ : Mila::Data::TokenSequenceLoader< TMemoryResource >, Mila::Dnn::Compute::Cuda::Rope::RopeCacheRegistry, Mila::Dnn::Compute::DeviceRegistry