lua-torch-nn 0~20171002-g8726825+dfsg-4 (amd64 binary) in ubuntu disco

 This package provides an easy and modular way to build and train
 simple or complex neural networks using Torch Framework:
 .
  * Modules are the bricks used to build neural networks.
    Each are themselves neural networks, but can be combined with
    other networks using containers to create complex neural networks:
 .
    + Module: abstract class inherited by all modules.
    + Containers: container classes.
    + Transfer functions: non-linear functions.
    + Simple layers: simple network layer like `Linear`.
    + Table layers: layers for manipulating `table`s.
    + Convolution layers: several kinds of convolutions.
 .
  * Criterions compute a gradient according to a given loss function
    given an input and a target:
 .
    + Criterions: a list of all criterions.
    + `MSECriterion`: the Mean Squared Error criterion used for regression;
    + `ClassNLLCriterion`: the Negative Log Likelihood criterion used for
      classification.
 .
  * Additional documentation:
 .
   + Overview of the package essentials including modules, containers
     and training.
   + Training: how to train a neural network using optim.
   + Testing: how to test your modules.
   + Experimental Modules: a package containing experimental modules and
     criteria.
 .
 This package is a core part of the Torch Framework.