vortex
Introduction
Vortex is designed to be a Haskell implementation of Deep Learning frameworks.However it still keep developing. We hope this tool support:
-
Common neural layer types, such as Sigmoid, Softmax, Relu, &etc.
-
Multiple optimazation algorithms. SGD and Aladelta is proposed.
-
Some real world deep learning models. For instance, CNN, RNN, LSTM are all necessery. Futher on We also hope the Neural Turning Machine(NTM) is not immature if possible.
-
Some interested DEMOS. Image Regresssion, Image Classification, Neural Art, language modeling, &etc.
TODO List (recent)
-
Common layers implementation: sigmoid, linear, softmax, relu, dropout.
-
Optimazation algorithms.
-
Image regression.
Install
TODO
Example
TODO