Support for automatically building deep learning models using hyper-parameter search.
Core engine configuration and settings and runtime properties.
Data structures to store example data used for building machine learning models.
Data normalization methods, used to scale data to specific range, in order to make them suitable for use by a neural network.
Evaluation procedures for machine learning models, used to estimate how good models are performing when given new data that (that was not used for training).
Neural network architectures with their corresponding builders.
Neural network layers, which are main building blocks of a neural network.
Activation functions for neural network layers.
Commonly used loss functions, which are used to calculate error during the training as a difference between predicted and target output.
Training algorithms and related utilities.
Optimization methods used by training algorithm.
Weights randomization techniques, used for initializing layer's internal parameters.
Various utility classes including Tensor, image operations, multithreading, exceptions etc.
Copyright © 2022. All rights reserved.