Hype

Hype Namespace

TypeDescription
Batch

Training batch configuration

Classifier

Base type for classifiers

Dataset

Dataset for holding training data

EarlyStopping

Early stopping configuration

GradientClipping

Gradient clipping configuration

LearningRate

Learning rate schemes

LogisticClassifier

Classifier for binary classification

Loss

Loss function configuration

Method

Gradient-based optimization methods

Momentum

Momentum configuration

Optimize

Main optimization module

Params

Record type holding optimization or training parameters

Regularization

Regularization configuration

Rnd

Random number generator

SoftmaxClassifier

Classifier for softmax classification

Util

Various utility functions

ModuleDescription
Params

Hype.Inference Namespace

TypeDescription
HMCSampler

Hamiltonian MCMC sampler

Hype.NLP Namespace

TypeDescription
Language

Language model

Hype.Neural Namespace

TypeDescription
Activation

Activation layer with custom functions

FeedForward

Feedforward sequence of layers

GRU

Gated recurrent unit layer

Initializer

Initialization schemes for neural layer weights

LSTM

Long short-term memory layer

LSTMAlt

Long short-term memory layer (alternative implementation)

Layer

Base type for neural layers

Linear

Linear layer

LinearNoBias

Linear layer with no bias

Recurrent

Vanilla RNN layer