{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    }
   ],
   "source": [
    "%pylab inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import sys\n",
    "sys.version\n",
    "sys.path += [\"/usr/local/lib/python2.7/dist-packages\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import clstm\n",
    "import numpy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "inputs = clstm.Sequence()\n",
    "outputs = clstm.Sequence()\n",
    "targets = clstm.Sequence()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100 11 7\n"
     ]
    }
   ],
   "source": [
    "a = array(randn(100,11,7),'f')\n",
    "clstm.sequence_of_array(inputs,a)\n",
    "print inputs.size(), inputs.depth(), inputs.batchsize()\n",
    "#clstm.array_of_sequence(b,inputs)\n",
    "#print b.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "b = numpy.empty_like(a)\n",
    "clstm.array_of_sequence(b,inputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(100, 11, 7)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "b = numpy.empty((1,1,1),'f')\n",
    "clstm.array_of_sequence(b,inputs)\n",
    "b.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<clstm.INetwork; proxy of <Swig Object of type 'std::shared_ptr< INetwork > *' at 0x7fbbba8e8c30> >"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "net = clstm.make_net_init(\"bidi\",\"ninput=11:nhidden=20:noutput=13\")\n",
    "net"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      ".Stacked: 0.000100 0.900000 11 13\n",
      ".Stacked.Parallel: 0.000100 0.900000 11 40\n",
      ".Stacked.Parallel.NPLSTM_SigmoidTanhTanh: 0.000100 0.900000 11 20\n",
      ".Stacked.Parallel.Reversed: 0.000100 0.900000 11 20\n",
      ".Stacked.Parallel.Reversed.NPLSTM_SigmoidTanhTanh: 0.000100 0.900000 11 20\n",
      ".Stacked.SoftmaxLayer: 0.000100 0.900000 40 13\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print clstm.network_info(net)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "temp = clstm.make_layer(\"LSTM\")\n",
    "temp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100 11 7\n"
     ]
    }
   ],
   "source": [
    "clstm.sequence_of_array(net.inputs, a)\n",
    "print net.inputs.size(), net.inputs.depth(), net.inputs.batchsize()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 -1 -1\n",
      "100 13 7\n"
     ]
    }
   ],
   "source": [
    "print net.outputs.size(), net.outputs.depth(), net.outputs.batchsize()\n",
    "net.forward()\n",
    "print net.outputs.size(), net.outputs.depth(), net.outputs.batchsize()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "deltas = zeros(0,'f')\n",
    "clstm.array_of_sequence(deltas,net.outputs)\n",
    "deltas *= 0\n",
    "clstm.sequence_of_array(net.d_outputs,deltas)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 -1 -1\n",
      "100 11 7\n"
     ]
    }
   ],
   "source": [
    "print net.d_inputs.size(), net.d_inputs.depth(), net.d_inputs.batchsize()\n",
    "net.backward()\n",
    "print net.d_inputs.size(), net.d_inputs.depth(), net.d_inputs.batchsize()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}