Deep.Net
Datasets Namespace
| Type | Description |
| CifarRawT | Raw MNIST dataset |
| Dataset<'S> | A dataset of a record type 'S containing ArrayNDT<_> data variables. The first dimension of each record field is the sample. All record fields must contain the same number of samples. The default constructor expects a list of ArrayNDTs corresponding to the fields in 'S. To construct a Dataset<_> from a sequence of samples use the Dataset<_>.FromSamples method. |
| InputTargetSampleT | A data sample consisting of an input and target array. |
| MnistRawT | Raw MNIST dataset |
| TrnValTst<'S> | A training/validation/test partitioning of a dataset. |
| Module | Description |
| Cifar10 | Module containing functions to load the Cifar 10 dataset from the binary version. The dataset can be found at https://www.cs.toronto.edu/~kriz/cifar.html |
| CsvLoader | |
| Dataset | Dataset functions. |
| Mnist | Module containing functions to load the MNIST dataset. |
| Normalization | Dataset normalization functions. |
| NormalizationTypes | Dataset normalization types. |
| TrnValTst |
Models Namespace
| Module | Description |
| Accuracy | Accuracy, defined by probability of correct classification. |
| ActivationFunc | Activation function expressions. |
| ActivationFuncTypes | |
| Config | Configuration loader. |
| GaussianProcess | |
| Json | JSON serialization. |
| LinearRegression | |
| LossLayer | A layer that calculates the loss between predictions and targets using a difference metric. |
| MLP | A neural network (multi-layer perceptron) of multiple NeuralLayers and one LossLayer on top. |
| MSE | Mean over samples of squared error. |
| NeuralLayer | A layer of neurons (perceptrons). |
| NormalDistribution | Expressions concernred with Normal Distriutions |
| Pickle | Binary serialization. |
| RMSE | Root of mean over samples of squared error. |
| Regularization | Regularization expressions. |
| SSE | Sum squared error. |
| Train | Generic training module. |
| TrainingLog | Training history module. |
| TrainingTypes | |
| Util |
Optimizers Namespace
| Type | Description |
| Adam<'T> | |
| GradientDescent<'T> | |
| RMSprop<'T> |
| Module | Description |
| Adam | |
| GradientDescent | |
| OptimizerTypes | |
| RMSprop |
SymTensor Namespace
| Type | Description |
| ExprInfoT | |
| MultiChannelOpUsageT | |
| UExprInfoT | Information about a unified expression. |
| VarRecord<'RVal, 'RExpr> | Maps a value record (containing scalars or ArrayNDTs) to a expression record (containing ExprTs). |
| Module | Description |
| BaseRangesSpec | |
| CompileEnv | Function compilation parameters |
| ConstSpec | scalar constant value |
| ConstSpecTypes | scalar constant value types |
| Debug | |
| Deriv | derivative calculation |
| DerivCheck | |
| DerivTypes | |
| DiagnosticsVisualizer | |
| Dump | |
| ElemExpr | element expression |
| ElemExprDeriv | |
| ElemExprHostEval | |
| ElemExprTypes | |
| EnvTypes | |
| EvalEnv | |
| Expr | expression module |
| ExprTypes | |
| Frac | |
| Func | Generates F# function from expressions. |
| FuncTypes | |
| Hold | functions for working with held op |
| HostEval | |
| HostEvalTypes | |
| Interpolator | linear interpolator functions |
| InterpolatorTypes | linear interpolator types |
| LoopEval | functions for loop evaluation |
| ModelContextTypes | |
| Optimizer | |
| RangeSpecTypes | |
| ShapeSpec | shape specification of a tensor |
| ShapeSpecTypes | |
| SimpleRangeSpec | |
| SimpleRangesSpec | |
| SizeMultinomTypes | |
| SizeProduct | |
| SizeProductTypes | |
| SizeSpec | |
| SizeSpecTypes | |
| SizeSymbol | |
| SizeSymbolTypes | |
| SymSizeEnv | |
| SymSizeEnvTypes | |
| Trace | |
| TypeName | assembly qualified name of a .NET type |
| TypeNameTypes | TypeName types |
| UElemExpr | unified element expression |
| UExpr | Functions for dealing with unified expressions. |
| UExprRngsSpec | |
| UExprTypes | |
| Utils | |
| VarEnv | Variable value collection. |
| VarEnvTypes | |
| VarSpec | variable specification |
| VarSpecTypes | variable specification types |
SymTensor.Compiler Namespace
| Module | Description |
| ArrayNDManikin | |
| ArrayNDManikinTypes | |
| ExecUnit | |
| ExecUnitsTypes |
SymTensor.Compiler.Cuda Namespace
| Type | Description |
| VarEnvReg | Locks and registers all variables in a VarEnv for use with CUDA. |
| Module | Description |
| ArgTemplates | |
| Compile | |
| CudaCmdTypes | |
| CudaCompileEnv | methods for manipulating the CUDA compile environment |
| CudaElemExpr | |
| CudaEval | |
| CudaEvalTypes | |
| CudaExecEnv | |
| CudaExecUnit | |
| CudaExecUnitTypes | |
| CudaExprWorkspaceTypes | |
| CudaRecipe | |
| CudaRecipeTypes | |
| CudaStreamSeq | |
| Debug | |
| Native | native methods |
| NativeFunctionDelegates | |
| PassArrayByVal | |
| TmplInstCache | |
| Types |
Tensor Namespace
| Type | Description |
| BaseTensorDevice | |
| BlockTensor<'T> | block tensor specification |
| CannotBroadcast | cannot broadcast to same shape |
| CannotCudaRegisterMemory | cannot register host memory with CUDA, maybe because it is not properly aligned |
| CudaError | generic CUDA error |
| CudaRegMemHnd | CUDA registered memory for fast data transfer. Dispose to unregister memory with CUDA. |
| DataTypeMismatch | operation requires tensors of same data types, but specified tensor had different data types |
| HDF5 | A HDF5 file. |
| HDF5Mode | |
| ITensor | Type-neutral interface to Tensor<'T> of any type 'T. |
| ITensorBackend<'T> | |
| ITensorCudaBackend | type-neutral interface to CUDA backend for tensors |
| ITensorCudaStorage | type neutral interface to a CudaStorageT |
| ITensorDevice | |
| ITensorHostStorage | type-neutral interface to TensorHostStorage<'T> |
| ITensorStorage | |
| ITensorStorage<'T> | |
| ImpossibleWithoutCopy | the layout of this tensor makes this operation impossible without copying it |
| IndexOutOfRange | specified tensor index is out of range |
| InvalidTensorLayout | invalid tensor layout specification |
| InvalidTensorRng | invalid tensor range specification |
| MatrixPart | part of a matrix |
| NPZFile | A Numpy .npz data file. |
| OutOfCudaMemory | out of CUDA memory |
| PinnedMemory | pinned .NET managed memory (wraps a GCHandle) |
| SeqTooShort | sequence too short to fill tensor |
| ShapeMismatch | operation requires tensors of same shape, but specified tensor had different shapes |
| SingularMatrixError | singular matrix encountered |
| StorageMismatch | operation requires tensors of same storage, but specified tensors had different storages |
| StrideMismatch | operation requires tensor with a specific stride configuration, which is not fulfilled |
| Tensor<'T> | An N-dimensional array with elements of type 'T. |
| Tensor | An N-dimensional array with elements of type 'T. |
| TensorCudaBackend<'T> | CUDA backend for tensors. |
| TensorCudaDevice | Creates Tensors on a CUDA device. |
| TensorCudaStorage<'T> | Tensor storage on a CUDA device. |
| TensorHostBackend<'T> | Backend for host tensors. |
| TensorHostDevice | Factory for host tensors. |
| TensorHostStorage<'T> | Storage (using a .NET array) for host tensors. |
| TensorOrder | memory ordering of tensor |
| UnsupportedTransfer | transfer between used storages is not possible |
| Module | Description |
| CudaRegMem | Methods for locking a TensorHostStorage into memory and registering the memory with CUDA for fast data transfers with GPU device. |
| CudaTensor | Tensor stored on CUDA device. |
| Decomposition | Matrix decomposition functions. |
| HostTensor | Host tensor functions. |
| NPYFile | methods for accessing Numpy .npy data files. |
| Operators | Useful core operators. |
| RandomExtensions | |
| Tensor | An N-dimensional array with elements of type 'T. |
| TensorLayout | |
| TensorLayoutTypes | |
| TensorRng | Range specification functions. |
Tensor.Cuda.Backend Namespace
| Type | Description |
| Cfg | CUDA backend configuration |
| NativeIdxTensors | C++ NativeIdx |
| NativeIdxTensorsInfo | C++ NativeIdx template info |
| NativeTensor | C++ tensor marshaling |
| Module | Description |
| Cfg | CUDA backend configuration |
Tensor.Utils Namespace
| Type | Description |
| DiskBinaryMap | a filesystem backed map for binary keys and values |
| Module | Description |
| Array2D | |
| Cuda | Cuda support types functions |
| List | |
| Map | |
| Permutation | Permutation utilities |
| Seq | |
| String | |
| Util | Utility functions |
| UtilTypes |
global Namespace
| Module | Description |
| DiskMap |
