==temp data server==
http://tx3.piza.cc/nn/

==installation==
 
 apt-get install ipython python python-dev

http://torch.ch/docs/getting-started.html#_

 curl -sk https://raw.githubusercontent.com/torch/ezinstall/master/install-deps | bash
 git clone https://github.com/torch/distro.git ~/torch --recursive 
 cd ~/torch; ./install.sh

===install iTorch===
http://www.lowindata.com/2013/installing-scientific-python-on-mac-os-x/

 brew install python 
 pip install virtualenv 
 pip install virtualenvwrapper 
 pip install numpy 
 brew install gfortran 
 pip install scipy 
 brew install freetype 
 pip install matplotlib 
 pip install ipython  
 # install the QT libraries as described  
 # in the text below before continuing 
 brew install pyqt 
 brew install zmq 
 pip install pyzmq 
 pip install pygments
 # something we found need in launch itorch notebook
 pip install jinja2
 pip install ioloop
 pip install tornado


install iTorch from github

 git clone https://github.com/facebook/iTorch.git
 cd iTorch
 luarocks make





==luarocks installs==
 luarocks install inline-c
 luarocks install torchffi


==torch7 tutorial==
#tutorial: http://code.cogbits.com/wiki/doku.php?id=tutorial_basics
#tut2: http://torch.cogbits.com/doc/tutorials_supervised/

===learning===

====model====
*linear regression nn.linear
*MLP sigmoid nn.reshape
*convolution netowrk nn.convnet

====loss function====
https://github.com/torch/nn/blob/master/doc/criterion.md#
*abs
*mse
*nll negative log-likehood  


====optimization====
*SGD
*LBFGS

===Criterion===
*forward
*backward

===Tensor===
multi dimensional numerical data https://github.com/torch/torch7/blob/master/doc/tensor.md

  criterion = ClassNLLCriterion(weights)  -- LogSoftMax is required.