Apache Singa
A General Distributed Deep Learning Library
|
Csinga::_Context | |
Csinga::lang::_Cpp | To implemente functions using cpp libraries |
Csinga::lang::_Cuda | To implemente functions using cuda libraries |
Csinga::lang::_Opencl | To implement function using opencl libraries |
▼Cbasic_ostringstream | |
▼Csinga::logging::LogMessage | |
Csinga::logging::LogMessageFatal | |
Csinga::Block | Block represent a chunk of memory (on device or host) |
Csinga::Channel | Channel for appending metrics or other information into files or screen |
Csinga::ChannelManager | |
Csinga::logging::CheckOpMessageBuilder | |
Csinga::logging::CheckOpString | |
Csinga::Constraint | Apply constraints for parameters (gradient) |
▼Csinga::Decoder | The base decoder that converts a string into a set of tensors |
Csinga::CSVDecoder | Decode the string of csv formated data into data tensor (dtype is kFloat32) and optionally a label tensor (dtype is kInt) |
▼Csinga::Device | Allocate memory and execute Tensor operations |
Csinga::CppCPU | Represent a CPU device which may have multiple threads/executors |
Csinga::DeviceMemPool | |
▼Csinga::Encoder | Base encoder class that convert a set of tensors into string for storage |
Csinga::CSVEncoder | Convert values from tensors into a csv formated string |
CFactory< T, ID > | Factory template to generate class (or a sub-class) object based on id |
Csinga::FeedForwardNet | The feed-forward neural net |
▼Csinga::Initializer | |
Csinga::init::Constant | |
Csinga::init::Gaussian | |
Csinga::init::MSRA | Ref: [He, Zhang, Ren and Sun 2015]: Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification |
Csinga::init::Uniform | |
Csinga::init::Xavier | Ref: [Bengio and Glorot 2010] Understanding the difficulty of training deep feedforward neural networks |
Csinga::Layer | The base layer class |
▼Csinga::Loss | The base loss class, which declares the APIs for computing the objective score (loss) for a pair of prediction (from the model) and the target (i.e |
Csinga::MSE | MSE is for mean squared error or squared euclidean distance |
Csinga::SoftmaxCrossEntropy | Softmax + cross entropy for multi-category classification |
▼Csinga::Metric | The base metric class, which declares the APIs for computing the performance evaluation metrics given the prediction of the model and the ground truth, i.e., the target |
Csinga::Accuracy | Compute the accuray of the prediction, which is matched against the ground truth labels |
▼Csinga::Optimizer | The base class for gradient descent algorithms used to update the model parameters in order to optimize the objective (loss) function |
Csinga::AdaGrad | |
Csinga::Nesterov | |
Csinga::RMSProp | |
Csinga::SGD | |
Csinga::Platform | This class queries all available calculating devices on a given machine grouped according to manufacturer or device drivers |
CPriorityQueue< T > | Thread safe priority queue |
▼Csinga::io::Reader | General Reader that provides functions for reading tuples |
Csinga::io::BinFileReader | Binfilereader reads tuples from binary file with key-value pairs |
Csinga::io::TextFileReader | TextFileReader reads tuples from CSV file |
CRegistra< Base, Sub, ID > | |
Csinga::Regularizer | Apply regularization for parameters (gradient), e.g., L1 norm and L2 norm |
CSafeQueue< T, Container > | Thread-safe queue |
CSafeQueue< Element, std::priority_queue< Element > > | |
Csinga::Scheduler | Scheduling Tensor operations with dependency detection |
CSingleton< T > | Thread-safe implementation for C++11 according to |
Csinga::Snapshot | The snapshot management |
Csinga::Tensor | A Tensor instance is a multi-dimensional array resident on a Device (default device is the host CPU) |
Csinga::Timer | For benchmarking the time cost of operations |
Ctinydir_dir | |
Ctinydir_file | Defined(_TINYDIR_MALLOC) |
Csinga::Tokenizer | Tokenize a string |
▼Csinga::Transformer | Base apply class that does data transformations in pre-processing stage |
Csinga::ImageTransformer | |
▼Csinga::Updater | Basic Updater class just forward all the method function call to the wrapped Optimizer |
Csinga::LocalUpdater | LocalUpdater do gradient aggregation and update gradient calling the wrapped Optimizer on a specific device (i.e., CPU or GPU) |
Csinga::VirtualMemory | Manage device memory pool including garbage collection, memory opt |
▼Csinga::io::Writer | General Writer that provides functions for writing tuples |
Csinga::io::BinFileWriter | BinFile stores training/validation/test tuples |
Csinga::io::TextFileWriter | TextFileWriter write training/validation/test tuples in CSV file |