Crate intricate

A layer-driven Machine Learning that makes it extremly easy training, running and creating Model’s for predicting some type of data using OpenCL. (does not yet support multiple devices)

This crate is completely at its basis written with the very good definitions from a video of the channel The Independent Code, so much credit to him at least for the calculation of gradients and the definition of activation functions as layers.

Re-exports

pub use model::Model;

pub use types::TrainingOptions;

pub use types::TrainingVerbosity;

Modules

datasets

The module for the datasets feature of Intricate.

layers

The module that contains all layers that Intricate currently implements. As of v0.3.0, Intricate has only the Dense type of layer, but has the activation functions which are used as layers in Intricate.

loss_functions

A module containing all of the available Loss Functions

model

The module that implements a sequential Model, that contains some layers, and forward passes some inputs over and over again from one layer to another.

optimizers

The module that contains all of the implemented optimizers in Intricate

types

A module containing internal data types for Intricate

utils

Just a module with a few utilities that make writing code easier through out Intricate