proto_sklearn.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import math\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "import torch.autograd as autograd\n",
    "import torchvision.transforms as transforms\n",
    "from torch.autograd import Variable\n",
    "from torch.nn.parameter import Parameter\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "%matplotlib inline\n",
    "import pylab as pl\n",
    "from IPython import display"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class Block(nn.Module):\n",
    "    def __init__(self, insize, outsize):\n",
    "        super(Block, self).__init__()\n",
    "        self.layers = nn.Sequential(\n",
    "            nn.Conv2d(insize, outsize, kernel_size=3, padding=1),\n",
    "            nn.BatchNorm2d(outsize),\n",
    "            nn.ReLU(),\n",
    "            nn.MaxPool2d(2)\n",
    "        )\n",
    "        \n",
    "    def forward(self, inp):\n",
    "        return self.layers(inp)\n",
    "\n",
    "class ENCODER(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(ENCODER, self).__init__()\n",
    "        self.block1 = Block(3,64)\n",
    "        self.block2 = Block(64,64)\n",
    "        self.block3 = Block(64,64)\n",
    "        self.block4 = Block(64,64)\n",
    "        \n",
    "    def forward(self, inp):\n",
    "        out = self.block1(inp)\n",
    "        out = self.block2(out)\n",
    "        out = self.block3(out)\n",
    "        out = self.block4(out)\n",
    "        return out.view(out.size(0),-1)\n",
    "\n",
    "class smartPREDICTOR(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(smartPREDICTOR, self).__init__()\n",
    "        self.sm = nn.LogSoftmax()\n",
    "        \n",
    "    def forward(self, support, stargs, inp, lim):\n",
    "        model = LogisticRegression(solver='sag',multi_class='multinomial',max_iter=lim)\n",
    "        model.fit(support.data.cpu().numpy(),stargs.data.cpu().numpy())\n",
    "        weights = Variable(torch.FloatTensor(model.coef_).transpose(0,1)).cuda()\n",
    "        probs = self.sm(torch.matmul(inp,weights))\n",
    "        return probs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "trainset = torch.load('mIN_train.pth')\n",
    "testset = torch.load('mIN_test.pth')\n",
    "valset = torch.load('mIN_val.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "standardize = transforms.Compose([\n",
    "            transforms.ToTensor(),\n",
    "            transforms.Normalize(mean=[.485,.456,.406],std=[.229,.224,.225])\n",
    "            ])\n",
    "alter = transforms.Compose([\n",
    "            transforms.ToPILImage(),\n",
    "            transforms.RandomHorizontalFlip(),\n",
    "            transforms.RandomCrop(84,padding=10),\n",
    "            standardize\n",
    "            ])\n",
    "\n",
    "def batchmaker(way,trainshot,testshot,theset,alterful=False):\n",
    "    classes = np.random.choice(len(theset),way)\n",
    "    if alterful:\n",
    "        li = [torch.cat([alter(theset[cl][i]).view(1,3,84,84) for i in \n",
    "                         np.random.choice(600,trainshot+testshot)],dim=0).float()\n",
    "              for cl in classes]\n",
    "    else:\n",
    "        li = [torch.cat([standardize(theset[cl][i]).view(1,3,84,84) for i in \n",
    "                         np.random.choice(600,trainshot+testshot)],dim=0).float()\n",
    "              for cl in classes]\n",
    "    support = torch.cat([t[:trainshot,:,:,:] for t in li],dim=0)\n",
    "    stargs = torch.LongTensor([i//trainshot for i in range(trainshot*way)])\n",
    "    query = torch.cat([t[trainshot:,:,:,:] for t in li],dim=0)\n",
    "    qtargs = torch.LongTensor([i//testshot for i in range(testshot*way)])\n",
    "#     print(support.size(),query.size())\n",
    "    return(Variable(support).cuda(),\n",
    "           Variable(query, volatile=(not alterful)).cuda(),\n",
    "           Variable(qtargs, volatile=(not alterful)).cuda(),\n",
    "           Variable(stargs).cuda()\n",
    "          )\n",
    "    \n",
    "def newevaluate(embed,model,criterion,way,trainshot,testshot,theset):\n",
    "    losstracker = 0\n",
    "    acctracker = 0\n",
    "    acclist = []\n",
    "    for i in range(reps):\n",
    "#         print(i)\n",
    "        support, query, qtargs, stargs = batchmaker(way,trainshot,testshot,theset)\n",
    "        embeds = embed(support)\n",
    "        qembeds = embed(query)\n",
    "        preds = model(embeds, stargs, qembeds, 50)\n",
    "\n",
    "        # Val loss\n",
    "        loss = criterion(preds,qtargs)\n",
    "        losstracker += loss.data[0]\n",
    "        \n",
    "        # Val accuracy\n",
    "        _,bins = torch.max(preds,1)\n",
    "        acc = torch.sum(torch.eq(bins,qtargs)).data[0]/testshot/way\n",
    "        acctracker += acc\n",
    "        acclist.append(acc)\n",
    "#         display.clear_output(wait=False)\n",
    "        \n",
    "    return losstracker/reps,acctracker/reps,np.var(acclist)\n",
    "\n",
    "way = 20\n",
    "evalway = 5\n",
    "trainshot = 5\n",
    "testshot = 10\n",
    "reps = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "113088  parameters in neural net.\n"
     ]
    }
   ],
   "source": [
    "torch.cuda.set_device(0)\n",
    "embed = ENCODER().cuda()\n",
    "model = smartPREDICTOR().cuda()\n",
    "losstracker,evallosstracker,evalacctracker = [],[],[]\n",
    "nweights = sum([i.numel() for i in list(embed.parameters())])\n",
    "print(nweights,\" parameters in neural net.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA20AAAE/CAYAAADVKysfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzs3Xd8leX9//HXJ5sMEjKYCYS9h4Ag\nQ1EcxVFt3aOuurVqqx3W70+rdtnWaofWauveuLFOFHAxw4awIYEwM0gIgZB1/f6474STRQImJMD7\n+Xjw4Jz7vs99f845Oec+n/u6rs9lzjlERERERESkdQpq6QBERERERESkfkraREREREREWjElbSIi\nIiIiIq2YkjYREREREZFWTEmbiIiIiIhIK6akTUREREREpBVT0iatmpndZ2b/buptm5qZfWNm17TE\nsRvLzELMzJlZqn//v2Z2b2O2PYRjXW1mHx9qrM3pQO+VmfUyM82DIiLiM7NU/3wQ4t//2Myubsy2\nh3Cse83sv98lXpGjlZK2o4SZZZjZaS0cw71mttv/V2xm5QH3lx/KPp1zv3XO3dzU2x6MgASmyH8u\n2Wb2ipm1bepjNRDHM2b2bB3LR/ivd9zB7M85d71z7g9NEFetRMc594Jz7szvuu96jhdrZn8zs0z/\nPck0szfN7PjmOJ6IyJHMzD41s4fqWH6emW072ATLOXemc+6FJojrZDPLqrHvPzjnrv+u+27gmM7M\nftlcxxBpLkrapMn4X7bRzrlo4GZgVuV959zAmtsf6pW4FjTQf269gPbA/Yf5+M8DF5pZmxrLrwTe\nd87lH+Z4DjsziwCmA/2As4C2wABgMtAsSaKIyBHueeBKM7May68EXnHOlR3+kFrM1UCe/7/IEUVJ\n2zHAzG4ws7VmlmdmU8yss7/czOwxM9thZgVmtsTMBvnrzjKzdDMrNLPNZvbzJoijssXqVjNbC6z0\nlz9uZllmtsvM5pnZ2IDH/M7Mnvdv9/Iff5W/fbaZ3XOI20aa2ctmlu8/z3vMLKMxz8M5VwB8gJcs\n1PU8q+IIjCXgfpyZPWdmW/3YHjKzxnwWvwGygR8G7CsEuAx4wb8/xsxm+89rq5n9w8xC64nzZTN7\nIOD+Pf5V183UOKGZ2blmtsj/e9hoZvcFrP7K36ayVfV4M7vezGYEPH68maX5f2dzzWx0wLpvzOxB\nM5vp7/8TM4uv5zW4BugA/NA5t9w5V+6cK3LOvemceyhgn/Uer8bzCvY/A7lmtg6YVM9xRUSOVO8B\n8cCJlQvMrB1wDvCif/9sM1von4c3BZ4bajKzGWZ2vX872MweMbMcM1sPnF1j22vNbIX/3b7ezG7y\nl0cBHwOdA84dnc3sATN7OeDx55rZcv+cNsPM+gesyzCzn/u/XQrM7A3zLuzVF3ckcCFwG9DbzEbW\nWD/ePw/l+6/BNf7yNmb2V/N6dRT456yaF09Fmp2StqOcmU0E/ghcDHQCMoHX/dVnACcBfYA44BIg\n11/3DHCTcy4GGARMC9hnvpmN/w5hnQscDwz2788BhuCdVN4C3jSz8AM8fixea9f3gAfNrPchbPsQ\n0BlI9df9qLHB+wnFecDsxj6mhpeBvUBPYCTeSe7ahh7knHN4J9irAhZ/D3DAZ/79MuBOIBEYh5eE\n3NTQvs3sHP9xE/H+Hr5XY5PdeK9RLPB94E7/MeD9DRHQqjqvxr4TgQ+BvwIJwD+Aj/wfDZUux0sU\nOwBRwF31hHoa8LFzbu8BnktjjlfpFrzPwVBgFN7nRETkqOF/X06m+rnjYmClc26xf7/IXx+Hd066\nxcx+0Ijd34CX/B2Hdz67sMb6Hf76tnjnucfMbLhzrgivd8SWgHPHlsAHmlkf4DXgp0AS8BHwgZmF\n1Xgek4DueL8jrjlArBfgncveBD4l4PUws654SeQ//WMNAxb5qx8BRuD9nogHfglUHOhFEWkOStqO\nflcAzzrnFjjn9gG/BsaYV2CiFIjB62pmzrkVzrmt/uNKgQFm1tY5t9M5t6Byh865OOfcN98hpj/4\n+9zr7+8l51ye30Xjz3hf7r0O8PgHnHPFfkzL8X5wH+y2FwO/d87lO+c2AY83Iu4lZpYP5OAlwP9p\nxGOqMbMuwKnAz5xze5xz24C/AZc2chcvAqeaWSf//lUEdG9xzs1zzs1xzpU559YDTwMTGrHfi4Fn\nnHPp/sn0gcCVzrlpzrllzrkK/yT/eiP3C16St9w595of18tAzSuyzzjn1jjn9uCdUIfVs69EYFvl\nHTMb6V9E2GX7x0025niBz/sx51yWcy4XeLiRz0lE5EjyAnBRQAvRVf4yAJxzM5xzS/3v+CV4yVJj\nzx1/c85tcs7l4V0kruKc+9A5t855vsS7wHhiXTuqwyXAh865qc65UrzkqQ1e8lTpH865Lf6xP6D+\ncwd4FwbfcM6VA68ClwX0RLkC+Nw/b5Q653Kdc4v8XjA/Bu50zm32e3fM9H9PiRxWStqOfp3xWtcA\ncM7txmtN6+Kcm4aXrDwBbDezp21/cY0L8MYMZZrZl2Y2pglj2hR4x8x+aWYrzawA2InX0pJY34P9\nRKfSHiD6ELbtVCOOajHVY4hzLg6IwGuJ/KqBFsG6dAPC8V7vfD8JfAKvhalBzrkNwEzgCv+9Ohe/\newuAmfUzsw/N6+a4C69Fsd7XMkBnqr8GmYErzet2OcO8bqYFwPWN3G/lvjNrLMsEugTcb+x7mov3\n3gHgnEvz35OL8V7Xxh4vMLZ6n7eIyNHAv9CaDZxnZj3weru8WrnezEab2fSA7/ibaZpzx5nmddnP\n8893ZzVyv5X7Dvz9UuEf66DPHWaWApwCvOIveh/vXF55MS8FWFfHQxP97epaJ3JYKWk7+m3BSxSA\nqn7kCcBmAOfcP5xzI4CBeN3ifuEvn+ecOw+v4MZ7eF0rmkrg+K5T8LrCXYDXLaMdXveFmgOmm9o2\nIDngfkpjH+icKwH+i9ca2L+OTYqAyID7HQNub8I7scT7LZZxzrm2zrkhjY7cuzp6FXARsCqgewvA\nU8AyoJdzri1esZTGvJZbqf4adK2x/nXgbSDFOReL9/wr99tQifxqf4MB+9/ciLhq+gKY5I9NaIrj\nNfS8RUSOFpXd668EPnPObQ9Y9yowhf3f8f/mO547/Iuab+O1kHXwL7B9xCGeO8zM/GMdyrnjSrzf\nvB+Y2Ta83hcR7O8iuQlvyEJNOUBxPetEDislbUeXUDOLCPgXgvdFfK2ZDfO/QP8AzHHOZZhXMGK0\n3z2gCO+LqdzMwszsCjOL9bsk7ALKmynmGLxxWDlAKF63vKhmOlagycC95hUFScYbmNwoZhaM129+\nD7Chjk0WARPMLMW8MvxVBVD8rphfAo+YWVszCzKvUMlJ/r4rC6gk17HfSm/inUDuI6B7iy8GKACK\n/AHbDY5n800Gfuy31EUBv6ljv3nOuWIzO4Hq3Tl3AM6/eluX/wEDzewS84rRXI6X8H7UyNgCPYf3\nt/KOmQ00bxB8G7yxFIdyvMnAT82si5klAL86hJhERI4EL+KNC76Bus8dld/xo/DGGTfGZOAOM0v2\nxw3fE7AuDK8HRDZQZmZn4o0hrrQdSDCz2APs+2wzO9X/nXI3sA+vt8nBugp4EK/7ZOW/C/z9J+C1\nwJ1mZhf7540EMxvmt+49CzxqXqGUYL/nycH2shH5zpS0HV0+witwUfnvAefcF3g/7t/GuyLWk/0/\nuNvijcvaidcFIRfvihh4V6Uy/C52NxNQqMO8Kk+N7ZPemJg/B9YAGXgJ4tYDPaCJ/AbvhJGB18d+\nMt7J4ECWm9luvNfrCuA8v5JkTZ8A7wJLgbl4Vy8D/QgvMU339/Um+1vjUvCuAG6jHs65Qn//XQjo\n3uK7G6/ffiFeq9sbDTynyn1+gNdN80tgNTC1xia3AH80s0LgXgJaXv14/gjM8bt8VqvI5ZzLxuvG\n+Su8v7GfAef4YxAOij8OcgKwCm/Q+C68KqRD8f+uD/J4T+K13i0F5uEVwhEROeo45zLwEp4oap+X\nbgUe8r/j76fxvWv+g1fUYzGwAHgn4HiFwB3+vnbiJYJTAtavxBs7t94/d3SuEe8qvPPlP/Eu1n0f\n+L7f26XR/AuNqcATzrltAf+mAGuBy5xzG/G6bt6NNyXAIvaPgf85+88RecCf0O9naQHmXEOt0yJH\nPzO7HfiBc+7UFo7jAWCTc+6ZloxDRERERFoPJW1yTPKrOHbDK9vfF69E/KPOucZUkRQREREROWwa\n3bzr9+NdaGb/q2NduHmTGq41szl+OXmR1iwcr1tHIV5XwLfxuhOKiDSamT1rZjvMbFk96828Se7X\nmjcJ8PDDHaOIiBz5DqZP7p3AinrWXQfsdM71Ah7D6+8r0mo559Y75wY656Kcc8nOuV/4RVdERA7G\n83iT+9bnTKC3/+9GvHGUIiIiB6VRSZtfye5svDLfdTmP/ZWI3sKb/Le5S7aLiIi0KOfcV3jFCepz\nHvCiP7nwbCDOzDodYHsREZFaGtvS9jfgl0BFPeu74E+u6Jwrwys5nvCdoxMRETmyVZ0ffVnUPdG7\niIhIvUIa2sDMzgF2OOfmm9nJ9W1Wx7JaFU7M7Ea87iFERUWN6Nev30GEKiIiR6r58+fnOOeSWjqO\nFtCo8yPoHCkicixq7PmxwaQNGAeca2Zn4c0e39bMXnbO/Shgmyy8+aWy/AmdY6mju4hz7mngaYCR\nI0e6tLS0RhxeRESOdGaW2dIxtJDK82OlZGBLXRvqHCkicuxp7Pmxwe6Rzrlf+4UaUvEmr51WI2ED\nb7LEq/3bF/rbaC4BERE51k0BrvKrSJ4AFDjntrZ0UCIicmRpTEtbnczsISDNn1H+GeAlM1uL18J2\naRPFJyIi0mqZ2WvAyUCimWUBvwFCAZxz/wY+As4C1gJ7gGtbJlIRETmSHVTS5pybAczwb98fsLwY\nuKgpAxMREWntnHOXNbDeAbcdpnBEROQodTDztImIiIiIiMhhpqRNRERERESkFVPSJiIiIiIi0oop\naRMREREREWnFlLSJiIiIiIi0YkraREREREREWjElbSIiIiIiIq2YkjYREREREZFWTEmbiIiIiIhI\nK6akTUREREREpBVT0iYiIiIiItKKKWkTERERERFpxZS0iYiIiIiItGJK2kRERERERFoxJW0iIiIi\nIiKtmJI2ERERERGRVkxJm4iIiIiISCumpE1ERERERKQVU9ImIiIiIiLSiilpExERERERacWUtImI\niIiIiLRiStpERERERERaMSVtIiIiIiIirZiSNhERERERkVZMSZuIiIiIiEgr1mDSZmYRZjbXzBab\n2XIze7COba4xs2wzW+T/u755whURERERETm2hDRim33AROfcbjMLBb4xs4+dc7NrbPeGc+4nTR+i\niIiIiIjIsavBpM0554Dd/t1Q/59rzqBERERERETE06gxbWYWbGaLgB3AVOfcnDo2u8DMlpjZW2aW\n0qRRioiIiIiIHKMalbQ558qdc8OAZGCUmQ2qsckHQKpzbgjwOfBCXfsxsxvNLM3M0rKzs79L3CIi\nIiIiIseEg6oe6ZzLB2YAk2osz3XO7fPv/gcYUc/jn3bOjXTOjUxKSjqEcEVERERERI4tjakemWRm\ncf7tNsBpwMoa23QKuHsusKIpgxQRERERETlWNaZ6ZCfgBTMLxkvyJjvn/mdmDwFpzrkpwB1mdi5Q\nBuQB1zRXwCIiIiIiIseSxlSPXAIcV8fy+wNu/xr4ddOGJiIiIiIiIgc1pk1EREREREQOLyVtIiIi\nIiIirZiSNhERERERkVZMSZuIiIiIiEgrpqRNRERERESkFVPSJiIiIiIi0oopaRMREREREWnFlLSJ\niIiIiIi0YkraREREREREWjElbSIiIiIiIq2YkjYREREREZFWTEmbiIiIiIhIK6akTURE5Dsws0lm\ntsrM1prZPXWs72pm081soZktMbOzWiJOERE5cilpExEROURmFgw8AZwJDAAuM7MBNTb7f8Bk59xx\nwKXAvw5vlCIicqRT0iYiInLoRgFrnXPrnXMlwOvAeTW2cUBb/3YssOUwxiciIkcBJW0iIiKHrguw\nKeB+lr8s0APAj8wsC/gIuL2uHZnZjWaWZmZp2dnZzRGriIgcoZS0iYiIHDqrY5mrcf8y4HnnXDJw\nFvCSmdU6/zrnnnbOjXTOjUxKSmqGUEVE5EilpE1EROTQZQEpAfeTqd398TpgMoBzbhYQASQeluhE\nROSooKRNRETk0M0DeptZdzMLwys0MqXGNhuBUwHMrD9e0qb+jyIi0mhK2kRERA6Rc64M+AnwKbAC\nr0rkcjN7yMzO9Te7G7jBzBYDrwHXOOdqdqEUERGpV0hLByAiInIkc859hFdgJHDZ/QG304Fxhzsu\nERE5eqilTUREREREpBVT0iYiIiIiItKKKWkTERERERFpxZS0iYiIiIiItGINJm1mFmFmc81ssZkt\nN7MH69gm3MzeMLO1ZjbHzFKbI1gREREREZFjTWNa2vYBE51zQ4FhwCQzO6HGNtcBO51zvYDHgD81\nbZgiIiIiIiLHpgaTNufZ7d8N9f/VnF/mPOAF//ZbwKlmZk0WpYiIiIiIyDGqUWPazCzYzBYBO4Cp\nzrk5NTbpAmyCqolGC4CEpgxURERERETkWNSopM05V+6cGwYkA6PMbFCNTepqVavZGoeZ3WhmaWaW\nlp2dffDRioiIiIiIHGMOqnqkcy4fmAFMqrEqC0gBMLMQIBbIq+PxTzvnRjrnRiYlJR1SwCIiIiIi\nIseSxlSPTDKzOP92G+A0YGWNzaYAV/u3LwSmOedqtbSJiIiIiIjIwQlpxDadgBfMLBgvyZvsnPuf\nmT0EpDnnpgDPAC+Z2Vq8FrZLmy1iERERERGRY0iDSZtzbglwXB3L7w+4XQxc1LShiYiIiIiIyEGN\naRMREREREZHDS0mbiIiIiIhIK6akTUREREREpBVT0iYiIiIiItKKKWkTERERERFpxZS0iYiIiIiI\ntGJK2kRERERERFoxJW0iIiIiIiKtmJI2ERERERGRVkxJm4iIiIiISCumpE1ERERERKQVU9ImIiIi\nIiLSiilpExERERERacWUtImIiIiIiLRiStpEREREaigsLm3pEEREqihpExEREQkwc10OQx/8jIUb\nd7Z0KCIigJI2ERERkSrOOR79bDUVDuZnKmk7muzYVUx24b6WDkPkkChpExEREfF9uzaXND9ZW7mt\nsIWjkaZ03Qtp3DV5UUuHIXJIlLSJiIjIEWd+Zh5l5RVNuk/nHH//YjUd20Ywuns8K7ftatL9y4H9\n+ZOV/OTVBc2y76yde1i6uYDlW/SeypHpiE3a0rfs4rZXF5CZW9TSoYiIiMhhtGhTPhc8OYvnZ2Y0\n6X5nr89jXsZObjulJ0NT4li9fXeTJ4ZSv+mrsvnfkq1sLdjb5PuetnIHAHlFJeTsbpkuksWl5Tww\nZTlZO/e0yPHlyHbEJm0l5RV8uGQrq9R1QURE5IhSUeEOuP6yp2fz509W1rv+0+XbAHhhVgblDezr\nYHy1JpvQYOOikSn07RBDSVkFGbnf/Qd2SVkFe0rKmiDCo0tmbhElZV5SXFHh2JCzG4APl2xt8mNN\nTd9OkHm312zf3eT7b4xX52zk+ZkZvLNgc4scX45sR2zS1j0hCoAMtbSJiIg0Sll5Ba/O2cjekvI6\n1zdlAlSfv3++hol/ncGOwuI612/IKWLW+lympm+vdx9T07cTEx7Cpry9TPdbUJrC0qwC+nSIISI0\nmH6dYgCapIvk/e8v45KnZn/n/RzJlmTlc+1zcyku9f72du8r44zHvuLFWRkAbCnYS3Gpl8D9r4mT\ntsLiUmavz+WcIZ0BWLuj8Rf895aUs3tf4xLujbl7+GjpVianbWJJVn61dcWl5Tz11TqAqjGTzcU5\nx+/+l86yzQXNehw5vI7YpC02MpT4qDA25ChpExERCbSvrJzN+bW7mH24dCv3vruUKYurX+kvK6/g\nhhfTuOzp2TjXfInb9l3F/GvGWjJy93D7qwvr7HpY2Yq2Nnt3nXOlbcgpYu2O3dx5Wm86to3ghVkZ\nTRKbc46lmwsYkhwLQK/20QQHGSu3frcePc45pq/awdLNBbWeT3mF48VZGewsKqlatrVgLzt21Z3Q\nNqV/zVjLA1OWN2rbzfnfPaYnZ6xj+qrsqlaujJwi9pVVsHCjl9ysz/Z+z53cN4lFm/LZlOe1cDZF\n99Sv1+RQWu64YnRXYsJDWN3IlrY9JWX88F/f8uPn5zVq+6ufm8utryzgl28t4cYX51f7LL05P4vt\nu/bRr2MMCzJ3NniBxDnHok35h/R53JBTxH+/2cCUxVsO+rHSeh2xSRtA98QoJW0iIiI1XPXMXG6v\no6DDG/M2AbAkq/oV+L98uoqp6duZm5HH4qzmuzr/xPS1lFc4fnZaH+ZsyOORz1bX2uaTZdsIDwnC\nudpxAnzut8B9b2BHfnRCV75ek8PaHd+9u1vWzr0U7C1lUBcvaQsPCaZnUtR3riC5MW8P23d5Y6iW\nba7eavfqnEzuf385L8/OrFp23fNp3NZMxTgqpW/ZxSOfruKFWRkNjh8rK6/g4n/P4mffoepi7u59\nfL7Ce982+D2kNvpJWfpW7zVZn+29h7dP7A14f6t3TV7E4Ac+I+M7/tb7PH07cZGhjOjWjt4dolnT\niJY25xy/enspK7cVsmhjflU3zvpsyd/LhpwifnJKL35+Rh+27Soma6f32paUVfDvGesY3jWOmyf0\nZPe+sgZbcL9dm8sPnviWL1YcfEty5efmcNV9KCuv4PlvN/DJsqbv1ir7HdFJW2qCkjYREZGaRnRr\nx5KsAooCunVl5hYxc10uUD0Z+mDxFp76aj3nD+9CRGgQk9M2NUtMWTv38NrcjVw0MoU7T+vNZaO6\n8u8v17F6+/4f0NsKilm0KZ+rx6YCXsGRmqamb6dfxxhS4iO5dFRXwkKCuO+9Zewrq97ls7S84qB+\ntC71u5IN9pM2gL4d237n7pFzN+RV3Q7srrazqKQqaf16bQ7gtbKlb91FWubOaq1vNRWXlvP8txsY\n/6dpPDq1duJ7IBUVjvveX0Z0eAjOwXsLD9wa8/GybWzO38u8DTurujYerHcXbqa03GsxyvR/t1UO\nb9mQU8TufWWszykiJjyE4V3jGJocy+PT1/Lews3sLS3/Ti1GOwqLmbpiO6f0bU9IcBC928c0Ksl/\n9tsMPli8hRHd2lFSXtFgojcvw3ufJw3qyGkDOgD73/up6dvZnL+X2yf2ZmRqO6DhOQDTMr3Hvn8I\nz32x3zUzswnGYzZkQ04RFz81iwc+SOfPn6xq9uMdyxpM2swsxcymm9kKM1tuZnfWsc3JZlZgZov8\nf/c3T7jV9UiKYvuufdVOSiIiIse6E3okUFbhqv0wfGPeJoIMfnhcF1Zu28W+snJv7MuH6QxNiePh\n84dw1uBOfLBoS71j3r6LJ2eswzBun9gLgJ+f0YfQYOP1ufuTxKnpXtfIi0cm0yMxqqrrXKXN+XtJ\ny8zjDP9HcWJ0OA+fP5hZ63P52RuLqrqcOee4e/JiTn/0K3IbWSlw6eYCQoONvh1jqpb16xhD1s69\n7Kqjm+aBFOwprYpl7oY82kWG0rFtRFViCPDIZ6vYva+M0/p3YEHmTgqLS/lqdbYfv1cUpaYdhcX8\n/fM1nPjn6TzwQToFe0t57psNB/U76K35WczP3Ml95wxgeNc43l2YhXOOXcWlvDpnY7UunM45/vv1\nesKCgygpr2BBjURjc/5ebn5pfp1JkHOu6t/ktE0MS4mjY9uIqpa2zJz9CcXKrbtYn11Ej6QozIzr\nT+zB0JQ43rplLKNS4/nfkv2JS1pGHvl76k9oa8Zwz9tLKSmr4LZTegLQu0M0ObtLyDtAUjx3Qx5/\n+GgFpw/owJ8vHALQ4PiwuRvyiA4PoX+ntvRpH0NMREhV4vVZ+jYSosI4qU8SXeLa0LFtBPMyDpy0\nLfYvWHyevr3eIjbpW3bx2tyN9T52Y96eZunuXF7h+MunKznt0S855ZEZrNmxm/G9ElmfU3TQnxWA\nyWmbuPKZObUuvEh1jWlpKwPuds71B04AbjOzAXVs97Vzbpj/76EmjbIeqSpGIiIiUsuIbu0ICTJm\nr/da1krLK3hzfhYT+7Xn9AEdKC13rNhayPItu9i+ax9XntCNsJAgLh6ZQuG+Mj5etpWc3fv4bPm2\nBis9glfoYea6nHrXO+f4LH073xvUkc5xbQBIiA7njIEdeWdhVtWPtU+Wb6NHUhS92scwLCWuakzP\nnPW5TPrbV4x7eBpmxpmDO1Xt+/zhyfzfWf35aOk27nh9Ifl7SnhxViZTFm+hpLyiqnWxIZVFSMJD\ngquW9feLkaw+iC6SeUUljP/ztKrql/My8hiZGs+Q5NiqH/6rthXy6tyNXDWmGz8en0pZhWP2+jy+\nXJ1Nh7bhxEeF8eWq/Unbgo07ufP1hYx7eBqPfb6a/p3a8uoNo3numuMp3FfGe4saV41wR2Exf/h4\nBSO7teOC4cn8cHgyq7fvZklWAbe+vIB7313KGY99xbSVXlfG+Zk7WZxVwE9P701wkNV6LV+YmcEn\ny7dx40tptX6s3/DifEb94QvueXspq7fv5pLjU0hNjKzq6piZV0QX/29h+ZZdrM/eTY+kaAC+P7Qz\n7982juFd23HO0E6s3r6b1dsLWZKVz0VPzeLhj+uvLBrojXmbmLZyB7+a1I9e7b33sld77xhrttf9\nnm7fVcytryygW3wkf714KN0ToogOD6nVtbWmuRvyGJnajuAgIyjIGNmtHfMydlJaXsH0lTuY2K89\nwUGGmTEytR1pGXn17ss5x+KsAnokRbG3tJzP6+ki+c9pa/j1O0urxv+B91lfvmUX0eEh7CkpJ7sZ\npjeYsWoHT0xfR1J0OPec2Y/PfnYS15/YHYDlDbxONW0t2MsDU5bz9ZocXp1TOwGV/RpM2pxzW51z\nC/zbhcAKoEtzB9YY3RP9pC1H812IiIhUigoPYUhybFXS9sWK7WQX7uOS47tWFdpYmpXPFyt2YOYV\nfwAY3T2e1IRI/vjxSsY+PI0bX5rP6/MO3F2yvMJx00vzufw/c/jMLyJS09odu8ku3Mf4XgnVll96\nfAr5e0r5bPl2Zq3LZfb6PCa4+wSdAAAgAElEQVQN7AjAsK5x5Ozex4acIn7+1mIKi8v4+Rl9eP+2\ncfTv1Lbafm44qQe/+F5fPlm2jdMe/ZLffZjOxH7tiYkI4Vu/62FZeQWPT1tTb6vQ0s0F1bpGAvTr\n6B3n6zW1E9IdhcU8MX0td09ezC0vz6dgj5e0vDQrk8LiMp77NoP5mXlk5O5hdPd4BneJrWqJeGl2\nBmHBQdx5am9GdGtHm9Bgpq/awddrcji5T3tO6p3Il6uzqahwPPvNBs7/10ymrdjBj07oxrS7J/Di\nj0cxtmciI7q1o3+ntrw0K7PBFhXnHP/v3WXsKSnn4QsGExRknDO4E6HBxnUvzOObtTncPrEXMREh\n/Pj5NCb+dQa/fmcpsW1CuWZsKkOSY6sl5qXlFbyzIIu+HWLIzN3DXW8srkrw52fu5PMV24lrE8qb\n8zcRGRbMOUM6kZoQVdVlL9N/XeKjwkjL3MmWgmJ6+L/rAk0a1JEg87rxPjBlOc551SUbag1en72b\n3/4vnbE9E7jG724L0KeDn4jX8XdQXFrOra8soGhfGf++cgRtI0IJCjIGdm5brZW0ptzd+1izYzej\nusdXLTu+ezxrd+zmk2Xb2FVcxul+6zDAyG7t2FpQXGexIPDGV+YVlXDtuO50aBvOB3V0kayocMzy\nP9/vByTtq7cXsq+sgjMGesfbeIAukkuy8pmctom/fb6a9xdtZlMjW+Zen7eJxOhwXrxuFDdP6Emn\n2DZVn52lm71WPudco1rOfvu/dCqcY1CXtjwxfS1F+8rILtzHXz5decDW0GPRQY1pM7NU4DhgTh2r\nx5jZYjP72MwGNkFsDUpNjASomtdDREREPCf0SKga1/bkl+tJiW/DKX297lkJUWEszipg2qodDE2O\nIzE6HAAz45qxqezaW8qFI5IZ0a0df/l05QG7oz06dRUz1+WSEBXGfe8vq7N7VGXiNLZnYrXl43om\nktyuDf+asY6bXkqje2IUN53kdWMblhIHwN1vLmZT3l7+dMEQfjKxd1WhkJpuO6UXU34yjo6xEaS0\ni+Sxi4cxtmcCX6/JwTnHFyt38Mhnq7nqmTlsr1EJsWYRkkqdYiP43sAOPD59LbMCWplKyir48fPz\n+Munq/hmbTafpW/n9x+lU1xazouzMjiuaxxmcNsrCwEY1T2eQX6yPD9jJ+8v3MLZgzsRFxlGeEgw\nJ/SI5635WRQWlzGhbxIn921PblEJHy7dyp8/XcnEfu2Zde+p/Ob7A6taoyrfr6vGdGPltsIGy8hP\nWbyFz9K3c9fpfapandpFhTGxX3tydpdw40k9uPuMvnxw+3h++4NBdIlrQ0ZuET8e153IsBDG9kxg\ncVZBVfn7aSt3kLO7hF98ry//7+z+fL5iO4985o1penzaGtpFhvLebeP45lcTmfKT8cREhJKaGEVu\nUQk7CovZWlBMt4QoBnRqyzS/SEnP9tG14m4fE8Ho7gk89dV6FmzM54LhyezeV8Zn6dUvEOwtKSct\nIw/nHHtKyrjl5QWEhQTxyEVDCaqcoM1/T6PCgllbo6WtuLSc619IY8HGnTxy0dCq5A68cY4rtu6q\nt5JlZVfHUakBSZt/+5HPVhEeEsSJvZOq1o301/3izcWMe3gav35nabX9VY7lPC4ljnOGdObLVdkU\n7K3+uUrfuov8PaWEhQTxzoLNVcnW4k1ecvn9od70BvWNa3thZgbnPv4tv3xrCX/7fA13vr6IE/88\nndF/+IKbX5pfb8v5jl3FTFu5gwtHJBMavD+NSIgOp0tcm6rxsu8v2sKwB6cesFvp9JU7+GjpNn5y\nSi8eOm8QObtLeOiDdH7wxLc8MX1ds42vBepNmFuzRidtZhYNvA381DlXs+1zAdDNOTcU+CfwXj37\nuNHM0swsLTu7dl/tgxUZFuL1j1ZLm4iISDWV49r+MW0Nizflc/OEnoQEB2FmDEmO5Zs1OSzJymdi\nv/bVHnf12FTSH5rEH344mN/9YBAFe0t5rI5iF1k79/DkjHU8MX0dlx6fwnPXHk924T7++FHtrmsz\n1+WSEt+GlPjIasuDgoxLRqawYusuwkKCee6a44mNDAW8Vq6wkCAWbszntP4dGN87sdZ+axrYOZYP\nfjKez352ErGRoYzvlcjm/L1k5u7h9bkbiY8Ko2BvKT9+fl61cWCVrSiVrZCVzIxHLhpKakIkt726\noKrV4q+frWLZ5l08deUI5tx7Gjee1IPJaVnc8/YScotK+OX3+nHtuO5s21VMVFgwAzq1rWqJ+Mun\nqyjcV8Ylx6dUHefE3kmUlFUQHGSM65XISX2SMIO7Jy8mNDiIP54/mOjwkDqf83nDOhMTEcIzX2+o\n93WZvmoH/++9ZQxLieOGE3tUW/erSf2496x+/GpSP8CrmnnlCd146brRpD80iTtO9cYgjumRSHmF\nqyq48WbaJpJiwjm5bxLXjE3lslEp/GvGOh78YDnTV2Vz/Yk9iAoPoXNcm6ouiZXDWr7xWy5TEyMZ\n2LktRX6rWY+k2i1tAOcM7URJWQVDU+L484VD6BLXhrfmZ1Xb5qH/LefCf8/igidncsdrC1m9o5B/\nXHZcVXfcwPe0V4cY1gS0tFUmbN+uy+GRC4dy9pBO1R4zqEss+8oqWJtddyPB3A15hIcEMTjg72dw\nl1jCgoPIzN3Dib0TaRO2v9ttv44xJMWEs3BjPuGhQbwxb2O1Lo6LN+UTHhJE344xfH9oZ0rKK7jj\ntYUs37I/AapMqn5ySi/W5xRVVX1dkpVPXGQoY3smYAaZebV/I+/YVcxfPl3F+F6JzPj5yaz63SQ+\nvGM8vz1vIGN7JjB/405uenE+W+pIbN6cn0V5hav29xv4nCs/S6/P28je0nLufH1hrVbR7MJ93Pvu\nUq5/MY0eSVHccFIPhndtx2n9O/BG2iZKyyvoEteGGauabg7GQIs35TPu4WkNFoNpbRqVtJlZKF7C\n9opz7p2a651zu5xzu/3bHwGhZlbr29U597RzbqRzbmRSUlLN1YckNTFSLW0iIiI1VI5re+rL9bSP\nCefCEclV64Ykx7FtVzHOUStpMzOC/ZaJ/p3a8qMTuvHS7MxqVRR/+dZixv9pOn/6ZCWjusfzwLkD\nGZIcx/Un9uC1uRuZ+MgMfvr6Qjbm7qG8wjF7fS7jetaddF02uitnDurIc9ccXy2pCwsJYlDntoQG\nG/93dv9GP28zI8RvARjXyzvmm/M38eXqbC4f1ZXHrxjOym2F/PKtJVWtE1+vySEkqHoRkkoxEaH8\n56qRlJZVcOqjM7ju+Xk8/fV6Lh/dle/5XTnvPLU33ROjeG/RFoYkx3JCj3humdCT2DahjEyNJyQ4\niMTocDrHRpC+dRc9EqOqdaU7qY/3m2h41zhi23jz0A5JjqOkvIJ7z+pPh7YR9T7fyLAQrhvfnU+W\nb+OTZdVbn8rKK/jTJyu59rl5dIlrwz8vO67qva3UIymaG0/qWWs5QKif5IP39xQWHMSsdblk7dzD\n9FXZXDA8uepCwG/PG8SEPkk8920GbSNCuGpMt1r7q+wh9aVfcKVrfCQDOrf137f9SV1N5wzuzMl9\nk/jDDwcRHGScP7wL367NYVuB12KamVvE5LQsxvRIIGvnXj5fsYOfntqnWutWoH4dYli8Kb8qCf/d\nh+l8s9ZL2C4I+JxUqmyBXRpQdbWwuJS7Ji/ijx+tYNrK7RzXNa7aeMiI0OCqiwCBXSMBQoKD+OLu\nCSy8/3Revf4Egsx47tuMqvWLNuUzqEssocFBDE2O5Z4z+7Fg407O/sc3/GvGWsC7ENIzKYqrx6YS\nFhLEuwu8JHZxVgFDkr1YOse2YWMddR9+/9EKSsor+N0PBpGaGEV4SDADO8dy5ZhU/nbpcbx981jK\nKhy/fmcpzjlWby/kuW838M2aHCanbWJ09/iqIUqBBifHkpm7h9XbC5mzIY8TeyeyLruI336YXrWN\nc44r/jubyfM2ceUJ3Xjr5rFVr9v95wzgslFdee+2cXx/aGfSMnbWmth82eaCqqQfvInSr3lu7kF1\npVy+xfsuW7jxyEra6r5sE8C8T+szwArn3KP1bNMR2O6cc2Y2Ci8ZbNzI3++oe2J01UScIiIi4qkc\n17ZgYz43ntSj2g/Kyh+T7WPCGdi5bX27AOCu0/vw3sLNPPLpav579Uh/HEwWF49M5qYJPekZ0F3v\nrtP7kBAVxvzMnUxN387a7N08eO5AdhWXMaZnQp37T4wO58kfjahz3T1n9id39746fyA2RvfEKDrH\nRvDUl+upcHDxyBS6JkTy8zP68qdPVjJhfhLtIsN4be5GLh/dtdprFKhHUjQf3D6eF2dl8vaCLHq3\nj+a+s/fXZIsIDebh8wfz4+fncfvE3pgZsZGhvH3LWKLC9+9zUJdYthQUc8nxKVXJEEDPpChOH9CB\nswZ3rFp244k9mLU+h0vraNGo6daTe/H5iu3c++5SRnRrR1JMOFsL9nLHawuZl7GTy0d35f5zBhAR\nWvfza4w2YcEc1zWOZ77ZwNNfrcfMq/JZKSQ4iCeuGM4dry3ktP4diIkIrbWPbvHe+1hZJTPVL/IB\n0CWuTb3xxUaG8vy1o6runz88mX9OW8urczdy1+l9+PsXawgJMv5+6TCiI0KYn7mz3osEADef3JNP\nlm/juhfmccNJPXh59kZuPKlHnQkbQI/EKKLCglm2uYCLRnrvx2tzN/LOgs2EBhul5a7aRZFKY3om\nsGhTPhP7dai1rq3/+nSM9cb7vTFvIz89vTdtQoNZtqWAy0d5Sa+ZcfOEnlw2qis/f3Mxf5u6hjMG\ndGDuhjwuHJFMbJtQTu/fgfcXb6FTXBtWby/ktP7ehZiu8ZFVLW2PfLqKtxdkkdIukrkZedxxam9S\n6/lcdU2I5J4z+/GbKcu5/D9zmLMhl8B6RD89rXedj6v8XvnjRytwDh48dyCvz9vE01+t54LhXndr\nL6nzvheuDhhrWHncP54/GIAJfZL495frmLk2hzP8iyMfL93KnW8swjnHx3eeSM+kaH4zZTnfrs3l\n0+XbuGxUVwBenp1Jp9gITu1f+3WH/QUMK+cIPFI0mLQB44ArgaVmVjmz4r1AVwDn3L+BC4FbzKwM\n2Atc6pqjxmgduidGkldUQsGe0qouFSIiIoeDmU0C/g4EA/91zj1cxzYXAw8ADljsnLv8cMV3+oCO\nbMkvrvoxU2lIsjde7JS+7aslD3WJiwzjhhN78Nepq1m8KZ/Hp68ltk0o950zoNYP84jQYG6a4I1J\n+2TZNm5+eT4/fcP76VBzPFtjBLZGHQozr7vhm/OzGN8rka4JXkvPTSf14KvV2TwwZTkhfqGJ+8+p\nqzD2fqmJUdz//QHcc2Y/HK5Wgje6RwIL7j+92vJeNcZojeoez9drcjh/ePUf+GbGf64aWW3Z2UM6\n1eqmV5+wkCAeu3gYZ//zG3703zl0jotg4aZ8Sssq+PulwzhvWNPUj7t9Ym8+WLyFbomRjO6eUG18\nHUB0eAjPXnN8vY9vExZMp9gIthYU0zYihLjIUGIiQggPCaq1rwPpnhjFyX2T+McXa1i1bRdT07dz\n/Yk9aO+3SNbXwhb4+CevGM6Vz87ll28tYVhKHD8/o2+923vFSGJZ5rfQlJVX8MLMTEZ3j+el60aT\nmVtEtzpaCW+e0JNJgzqSFBN+wHiuG9+D9xZt4aVZmYztmUBxaQVDU6p31Y1tE8rvfzCIUx6ZwfUv\npLGnpJyx/oWQG0/qwfzMnVVVNSs/N90SIpmavt2b129mBh1jIyitqGBMjwRuPbnnAWO68oRufLJs\nG2mZeVw7rjtXj0llY94ethbs5ftDOtf5mEGdvZinr8pmSHIsPZKiufPU3jw/M4MPl2xlRLd2fOOP\nb61sXa7PiG7tiAoL5svV2ZwxsCOvztnI/723lKHJcWzIKeLed5Zx3Ynd+XZtLmZeRcvLRnUla+ce\n/t97ywDvM/TguQOrxuxWqqxgumJr46vCtgYNJm3OuW+AA36jO+ceBx5vqqAORvdE70O+IbeIYZFx\nLRGCiIgcg8wsGHgCOB3IAuaZ2RTnXHrANr2BXwPjnHM7zax93XtrHjdP6MH1J3avVjAAICkmnMcu\nGcqo7nW3ftV07fjuPPvtBn719hJWbivkp6f1rrMlJdCkQR05f3gX3lmwmT4dohv84dpcTuyTxJvz\ns6qNwQkKMh69ZCiT/vY1zjmevGJEo1uhwkLqH1lSX0tdpWvGpnLesC7N8lr07hDDw+cP5skZ68jZ\nXcLxqfH8+sx+B5UMNWR878RGjS08kG4JkVVFSLyurMadp/Wu1mLbGE9dOYInpq/jX9PXehcLTurR\n8IMCjO2VyB9+OIj/fr2Bf1523AHfV4ChKbE8PzODRZvy2Zq/l835e7n/+wMICwmid4fa3WrBa+0e\n2LnuwjmBBifHMrp7PH/5dP/k1JWFeAK1bxvBHaf25o8fr8TMG7fqxRbH7HtPpWBvabWW6a4JkeQW\nlfD+os3s3lfGg+cOrOoy3JCgIOPZa46nqKSsKumpvOhRn3ZRYaTEt2FT3l7O9QuhRIWHMK5nAlNX\nbOO+c/rz7docusS1IbWBfYWFBDGul1dF9fP07fzfe0uZ0CeJJ68YwZTFm/nV20tZtqWAXu2jGd41\njo+WbqOkrIIPl2wF4Lrx3XlpVia79pby0nWjq+27sqVt7Y5CSsoqGnzvW4vGtLS1at39/tHrs3fX\n+QcuIiLSTEYBa51z6wHM7HXgPCA9YJsbgCecczsBnHPNM7K+HmZGaHDd111/eFzdXcHqEh0ews0T\nevLHj1cSHR7CtWO7N+pxD5w7kIUb85k0qHEtRs3h7MGdiAoL5pS+1fPlTrFtePuWMYA1+GO0qYQE\nBzVr8nr+8ORarXitTffEKGavz6NbwGt+68m9Dno/4SHB3HV6H84d2pm9JeUkRB/863rJ8V255Piu\nDW8I3DTB61J5/QvzSIqJICW+DafV0/3uUDxy0VCmrdzB7n1ltIsMq7PlDuDacd2ZnLaJmIhQ4iLD\nqq2LbRNKbJv9F1Mqu6M+MX0d7WPCq5K8xmoTFlytgEpjDOkSR9bOvVXVKwFOG9CB6e9ms2JrITPX\n5fK9gR0abOEHmNA3ic/St3PbqwsY1DmWJ68YQZuwYC4akcLb8zczNyOP+84ZQHFpOZPTvEnj/7dk\nK0OSY7nvnAFEhAbx7y/XV+uNV1HhyMzdU9Xiuy57d60pRFqrIyO1PIBuCVHEhIdUVTMSERE5TLoA\ngTWps6g9j2kfoI+ZfWtms/3ulEekq8akkpoQyS0n92z0cIS2EaF8cdcE7jq9TzNHV7/gIOPU/h2q\nlX2v1Kt9TK0ujNK8KouNdGuiRLlX++hqVRubS2J0OM9dM8qfmH4XV49JrbN4y6FKiY/k6rGp3HZK\nLy4fXX8iGRYSxOSbxvD0VXWPAw1U+RpvzNvDecM6N2m89bnl5J786fwh1YrnVCa3j32+moK9pYxv\noPtqpZP87WLbeMWAKhPIoCDj8SuO498/Gs6EPkmM65VIaLDx4qwMlm4u4By/W/HEfh0or3B8uWZ/\nxfqtu4rZV1bBpEHeOLkVR9C4tiO+pS00OIgxPRP4arU3D0tjMncREZEmUNcJp+Z47hCgN3AykAx8\nbWaDnHP5tXZmdiNwI0DXro27+n84tQkLZsYvTjnox9WVLMmxq1tV0nZoxWVaUq/20Txz9UhenJVZ\nZ8n7w6WxrYqBLchNNa6xIYO6xNaa77BD2wiGpsQxNd2bj29sPUWJakqJj+Sh8wYyunsCHWOrV1Bt\nHxNR1YIfHR7CyG7xfOxXTz1rsLd8WEoc8VFhfLFie1V3zcrxbBP7teeVORtJ37KL84cf4pM9zI74\nljbwmk835+9lXXbtsqYiIiLNJAsI/OWWDGypY5v3nXOlzrkNwCq8JK6W5pgWR6S1GZnajpHd2jHm\nILvqtRYjU+P5x2XHNTimszVoG+FNH9G7fXSDVWKb2xn+tAf9O7WtVRjkQK4ak1rnVBw1ndzX+848\nrmscye28ZDU4yDilb3tmrMqumhh9g5+09UyKpm+HGFZsO3Ja2o6KpK2y+bSyhKyIiMhhMA/obWbd\nzSwMuBSYUmOb94BTAPz5S/sA6w9rlCKtSGJ0OG/dMrbWROvSPH59Zj8eOHdgi/dEq5yrbnyv5knW\nT/WnOThvaOdaywv2llZNpJ2ZW0R4SBAd20bQv1MMK7YWcpgK3n9nR0XSlhIfSY/EKL5ao6RNREQO\nD+dcGfAT4FNgBTDZObfczB4ys3P9zT4Fcs0sHZgO/MI5d1jmMRURuWhkSqMrRjan3u2jefTiodxw\nkFU+G6tX+xg+vvNErhyTWm35ib298W7TVno1oDbk7CE1IYqgIKN/p7bkFZWwo3Bfs8TU1I6KpA28\n+R5mr8+luLS8pUMREZFjhHPuI+dcH+dcT+fc7/1l9zvnpvi3nXPuLufcAOfcYOfc6y0bsYjI4Wdm\nnD88mfYxEQ1vfIj6d2pbq9hKTEQoo7sn8MnybZRXODJyi0j1K89XVo08UibZPoqStkSKSytIy9jZ\n0qGIiIiIiEgrcPnormTm7uHVuRvZmLuHVH8euwGd2xIRGsSfPl5J9hHQ2nbUJG0n9EggLDiIqenb\nWjoUERERERFpBc4c1JFRqfH88aMVlJRXVE070TbCm0ogM3cPlzw1i/cWbubVORtZtrmghSOu21GT\ntEWGhXD2kE5MTssid3frz5ZFRERERKR5mRn3f38Ae/0hVKkB002c2DuJF68bRXbhPn76xiLufXcp\n5/9rJtP9MXCtyVGTtAHcdkovisvKeeabDS0dioiIiIiItAKDusRy0YhkAHomVZ8j8PjUeL7+1Sl8\nftcEpt09gT4do7nppflMW7m9JUKt11GVtPVqH81Zgzvx4qxM8veUtHQ4IiIiIiLSCjxw7kBevWE0\n7dvWLoYSFxlGr/bR9EiK5pXrTqBvxxhue2Uhm/P3tkCkdTuqkjaA2yf2Yve+Mp79NqOlQxERERER\nkVYgMiyEsT0bnv4gNjKUJ380HIfjDx+tOAyRNc5Rl7T169iWMwd15D9frW9V2bGIiIiIiLR+ye0i\nuWVCLz5cspWZ63JaOhzgKEzaAO49qz8Ox4NTlrd0KCIiIiIicoS5aUIPktu14cEp6ZRXuJYO5+hM\n2lLiI7nj1N58lr6dL1a0rkGEIiIiIiLSukWEBnPHqb1Ztb2QldtafgLuozJpA7h+fA96t4/mN1OW\ns6+svKXDERERERGRI8iQ5FgA1mUXtXAkR3HSFhYSxH3nDCBr515enbOxpcMREREREZEjSPfEKIIM\n1u7Y3dKhHL1JG8CJvRMZ2zOBx6etZfe+spYOR0REREREjhDhIcF0jY9knZK25mVm/HJSP3KLSnjm\na024LSIiIiIijdczKZp12Uramt2wlDgmDezI01+t0xQAIiIiIiLSaL3aR7M+p6jFK0ge9UkbwK/P\n6oeZcesrC1SUREREREREGqVn+2hKyirYlLenReM4JpK2bglRPHLREBZvyuf3H7aemc1FRERERKT1\n6pkUDdDiXSSPiaQNYNKgTtxwYndenJXJ5LRNLR2OiIiIiIi0cr3ae0lbS1eQPGaSNoBfTurH+F6J\n3PvOUmauy2npcEREREREpBWLbRNKUkx460/azCzFzKab2QozW25md9axjZnZP8xsrZktMbPhzRPu\ndxMaHMQTVwyne2IUt7y8gA05LT9RnoiIiIiItF49k6KOiO6RZcDdzrn+wAnAbWY2oMY2ZwK9/X83\nAk82aZRNKLZNKM9eczwVFY7Hpq5u6XBERERERKQV69U+mrU7duNcy1WQbDBpc85tdc4t8G8XAiuA\nLjU2Ow940XlmA3Fm1qnJo20iKfGRXDAimY+XbSW7cF9LhyMiIiIiIq1Uz6RodhWXkb275fKGgxrT\nZmapwHHAnBqrugCB1T2yqJ3YtSpXjulGabnj9bkbWzoUERERERFppSqLkSzfsqvFYmh00mZm0cDb\nwE+dczUjtjoeUqv90MxuNLM0M0vLzs4+uEibWM+kaMb3SuTVuRspK69o0VhERERERKR1Gt61He1j\nwvnrZ6taLG9oVNJmZqF4Cdsrzrl36tgkC0gJuJ8MbKm5kXPuaefcSOfcyKSkpEOJt0ldNaYbWwuK\n+XzF9pYORUREREREWqGo8BAeOHcgyzbv4vmZGS0SQ2OqRxrwDLDCOfdoPZtNAa7yq0ieABQ457Y2\nYZzN4tT+HUiJb8ODH6SzMbdlZzkXEREREZHW6cxBHZnYrz2PTl3N5vy9h/34jWlpGwdcCUw0s0X+\nv7PM7GYzu9nf5iNgPbAW+A9wa/OE27SCg4ynfjSSvaXlXPaf2WTtVOImIiIiIiLVmRkPnjuQ0vIK\nXp6dediPH9LQBs65b6h7zFrgNg64ramCOpwGdG7Ly9eN5vL/zOYHT8zkofMGcuagjngNjCIiIiIi\nIl4F+sFdYpm7Ie+wH/ugqkcerQZ1iWXyzWPoGBvOra8s4Kpn5/LKnEw25anlTUREREREPMenxrMk\nK5/i0vLDelwlbb5+Hdvy3q3juPesfqzdsZv/e3cZJ/1lOr95fxlF+8paOjwREREREWlhx6fGU1ru\nWLQp/7AeV0lbgJDgIG48qScz75nIF3dP4KoTuvHi7EzOeOwrFmzc2dLhiYiIiIhICxqZ2g6AeYe5\ni6SStjqYGT2TonnwvEG8edMYgoOMS5+ezbsLs1o6NBERERERaSFxkWH07RDD3Awlba3KyNR43r9t\nHMO7xvGzNxbz+w/TKdVk3CIiIiIix6Tju7djQebOwzrRtpK2RmgXFcZL143mqjHd+M/XG7js6dls\nLTj88zOIiIiIiEjLOj41nqKSclZsLTxsx1TS1kihwUH8//buOz6qKu/j+Oek94SQhBog9CKKGili\nQRHBBva6rnXtbZur67P6rOs23XV3fay4unYRXVRErIiCIk06BggEQkJLSA/pmfP8cSYkgQABQjIT\nvu/XKy8yd+7M/ObMhKcHpmsAACAASURBVHO/95x776OTjuGpq47nx23FnPX3b3j26/WtfuYYERER\nERFpO8NT4gFY1IpTJBXaDtLE47ryyb2ncnLfBB7/dC3nPTWXjNzSti5LRERERERaQZfYcLp3CFdo\n83U9O0by4k9TefXG4RSUVXPhM98xZ11uW5clIiIiIiKt4O+XHceD5wxqtddTaDsMp/dP5MM7R9M1\nLpwbX1nEgoy8ti5JRERERESOsBG9O9KjY0SrvZ5C22FKjo9g6m2j6NExgjveXMLWQp2gRERERERE\nWo5CWwuICQtm8rWpVNZ4uPnVxby9cDOz1+aQXVCGtbatyxMRERERET8W1NYFtBd9k6L45xXDuOvt\nJTw4beXu5bHhwYwb3IlbTutNn8QoMvN2saO4kpCgAGLCguiVEElwoLKziIiIiIg0TaGtBZ01uBMr\nHhlPbmklWwvLWbO9hOVZhcxYsZX3fsgmPDiQ8j0uERASGMDgrjH85ZKhDOwc00aVi4iIiIiIrzJt\nNX0vNTXVLl68uE1eu7UV7KrirYWbySutYlCXaLrFhVNV66GwrJq0bcW8v3QLFnjvtlH07BjZ1uWK\niLQ4Y8wP1trUtq7DXxxNfaSIyNGsuf2jRtpaQYfIEO48o2+T9114fDcuPbE7l73wPde+tJDfnjuI\nvklRpCREEhhgWrlSERERERHxNQptPqBfp2heuWE41/57Abe98QMAMWFBnNovkUnDujJucCeMcQGu\npKKanaVVFJVXExMWRGJ0KNFhwbufa3tRBeXVtaQkaMRORERERKQ9UGjzEcOS45j/27GszyklPaeU\nBRl5zEnP5eOV2xg7MImLT+jOlEWbmZu+c6/HHts9litOSiZ9RylvLdiMMfDy9Scxum8CANba3aGv\n7vastBxemLPBO2UzhnGDO3Hh8d1a7f2KiIiIiEjzKLT5kMjQII5LjuO45DguPbE7NbUeXpm3ib9/\nvo5Za3LoFBPKPWf2pVdCJLHhwRRXVLOloJyPlm/jofdXERhguOzE7izdXMiNryzi1+MH8PXaXH7I\nLOCuM/ty++l9SNtezIPTVrIiu4jk+HAGdY5h6eYCPl65jVqP5ZITu7d1M4iI+BVjzATgX0Ag8G9r\n7V/2sd6lwLvASdZaHbAmIiLNptDmw4ICA7j51N6cM7QL6TtKGN03ocnLA9x5Rl9Wby0mNjyY5PgI\n8korufrFBTz2cRpdYsNI7dWBJz5byyertrFmWwlxESE8cemxXHh8N4IDA6iu9XDtSwt4cNpKeiVE\n0L9TNAW7qukcG0ZIUADWWnJLKokJDyYsOJBaj+W/P2STnlPC5anJ9OsU3QatA99vyKNHxwi6xYUf\n9GOLyqoJCQogPCTwkF+/ptbD9uIKusaGE+CDxx9W13p86nIS5VW1fL02hxN6dqBTTNghP88XP+7A\n4M7WKvuWXVBGeHAgHaNC27qUds0YEwg8A4wDsoFFxpjp1tof91gvGrgHWND6VYqIiL9TaPMD3eLC\n9xtMjDEc0y129+2OUaFMvXUUy7MLGdWnI0EBhrcWbuaxGWlccFxXHrlgMHERIbvXDw4M4LlrTuSi\nZ7/jkue+3708KMDQIz6C3NJKSipqiAwJ5IyBSWzI3UXatmICDLw4dyPDU+IJDQqgstpDh8hgEqJC\nKa6oYUtBGcd2j+Pn4/oTFRrEB0u3sGBjHkO7xzGoczRZBWVk5Zdz9pBOuy93YK3l2/U7eWpWOpvy\nyhjVuyOn9kvgtP6Juzf0rbU8NWs9//hyHV1iw3jv9pPpFhfOyuwi0nNKuHBYt90hanlWIe8szmLG\n8q0YY0iMDqWovJrckko6RATzyAVDGDsoiZe/3cS8DTu59fTenDmw0+7XaTitdM32Yr5YvYOdpZVs\nyivjh8wCSitr6BgZwsl9E7j1tN67PwePx1Jr7V6haX5GHm/Mz2TMgCQmDeu6+/6120t4YNoKMnJ3\ncVxyHKf3T+T6k3vtPhlN+o4SPli2hc9X7+DqET24YXTKfr8zc9Nz+dlrixnZuyP3jx/I4K71l5PY\ntHMXq7YWkVtS6R2dTSY8JJBdlTW8MT+TtTtK2F5UwZkDk7jplJRGbXAgWwrLeXXeJrILyuiXFE3v\nxEgSokLZXlTB3z9fy9aiCkKDArhmRE9uG9ObpOj6z7S4vIZtxeXUeiyDu8Ts9brWWp77ZgOPf7oW\ngEcuGMwNo1Oo9ViKyquJjwzZq57KmloCjSHoEMJrda2HqhoPkaFBWGuZk76TGcu3cmxyHGcMSCQm\nPJjKag/rdpSwdHMBm/LKyC2ppGNUCPePH0jn2PpgOjc9lz/NXENIoKFPYhRd48KJjwwhNDiA6hoP\nnWPDGT+kU6P3XOuxLMsqpKi8CoDeCVH02s+xqpU1tWwpKGd9TinvLMpi1pocIkMCeeCcgVwzoude\nOxYqqmt5f+kWdhRXUFHtoUd8BCN7x5OSELm7jvU5pbz07Ub+MGnIIbXhUWI4sN5amwFgjJkCTAJ+\n3GO9PwCPA79q3fJERKQ90Cn/jyK1HrvfM1Jm5Zfx9sLNxEUEExceQmb+Ljbk7CIhOoQ+iVGs21HK\nZ6u3ExESyG8mDOTkPh15fX4mX/y4g5CgAEICAygoqyK3pJLY8GASo0P5IbOA+MhQEqJCWLO9hMiQ\nQHZVNb5WnTFw0fHdiAkLZt6GnazbUUrX2DBSe8Uzb0MeO0srAeiXFEX/TtFU1tTyZVoO44d0Yt76\nPBJjQhk3qBMvzs3AY2FESjy3jenDf77bxJx1uYQFBzBhSGdiwoPJLakkMjSIvklRfLZ6O0s3F7rA\nWeMhISqUnaWVnHdsF8oqa5ifkc9JKfH8fuIQFm3K538+WEVVjYeYsCC6xIaT2qsD/TtFszyrkK/W\n5lBaUcNtp/chKNAwZWEW24srCAowdI4NY8yARKyFNxdsJiw4gIpqD907hDMsOY6QoABmLN9GVFgQ\nYwcmsSK7iLU7Sji1XwKPXDCYZ7/ewLQlWwgMMHTvEE5mXhl/v+w4zhrUiWe/Xk/+ripuOa337hHP\nFdmFXDl5PonRoRSWVVNcUU1KQiRdYsPYUlDOpryyRu3fLS6ca0f15NV5m9hWVEGX2DCiQoNIzynl\nvKFd+OXZ/cnML2NDTinpO0rJL6tiZO+OnNSrA8uyCvlmbS4llTVYa1m6uRDrfc6sgjIa/vcypGsM\nd5/Zl1lpOUxbuoXgQMO1I3sSGx7M+0u3sCF31+51+yVFcdEJ3QgJDGBXZS27qmrYuHMXX/y4g0nD\nulJRXctnq3dwar8EVm8tJn9XFYO7xHD2kE4M6BRNXEQIM1du479LsgkODGDsoCSG94qnY1QoUaFB\n1HospZXVrM8pZVNeGcGB7mL3nWPDSO4QwQ+bC5i6KIvC8mpO6BFHgDEs2Jjf5LUW63SOCSMxOpT0\nnBKCAwO4Y0xfOkQEszy7iLcXbiYlIZJuceFsyC1lR3EFnj3+6x07MIlHLzyGjNxSZqXl8PHKbeSW\nVDZaZ1CXGM4cmMjQbnEkRofw49ZiVm4pYkV2Eek5pdR6n7RjZAjXjOjB0qxC5qbvZFhyHDeeksKE\nIZ0praxhQUYef5yZRnZBOeB20NR4H9s7MZIbvWH4z5+kER4cyNRbRx32iHp7PeW/d8rjBGvtzd7b\n1wIjrLV3NVjneOB/rLWXGGO+Bn51oOmR6iNFRI4Oze0fFdrkoHg8FmNo9ujLqi1FPDJ9NQVlVdx3\nVn/OH9qFzfllpOeU0rNjBB0iQnhxbgavzNtEoDGc2LMDE47pzGWp3QkNCsRaS9q2Euak57IgI2/3\naMaNp6Tw87P6sXBjPj99eSGVNR4uT+3OsOQO/GlmGqWVNcRFBHPHmD5cObwHMQ3OsFmn1mN5/ftN\nrNhSxHWjejGwSzRPzUrn+W8y6N4hnJN6xfPZqu2UVddS67GM7tuRp648vsnpZoVlVTw640emLdmC\nMXBav0RSe3agvLqW9JxSvlu/k7KqWq4/uRe/Hj+A+Rl5vDJvE1sKyikqr+bkvi6gJXife8rCzTz8\n4Wqqaj0EBxp+dmpvbhidQkx4EDe+soj5GflEhwVRXF5NWHAgFdW1nNovkfDgQBZszCMyNIj/3n4y\nYUGBvPr9JtZsL2ZrYQUdIoIZMyCJ1F4d6BIbTvqOEn734SrW7ShlYOdo/njRMZzYMx5rLZPnZPDX\nT9c0ChcJUSFEhQY1Cn4pCZEkRYdigaHdYrlhdC+6d4igvKqW7IIy8ndVUWstI1I67t5psGnnLp76\nKp0Plm7BY2F4SjxjBybRrUM4JRU1vLMoi2VZhbtfIyw4gKjQIK44KZlfjhuAx1p+9+FqZq/JYVSf\njvROiOTrde74zTohgQGcf1wXAGal5VBUXt3kd7RTTCi1HjfSV1XrASDAwFmDOtEnKYq56bnsLHHB\n+JqRPcjKL+O79XlU13oICQqgZ8dIhiXHERsevPu9PTBtBfMz8ne/xg2je/GbCQMJC3bTcT3e0cHK\nGvccHy7bwp8/WUNVjXv9kKAAzhiQyPnHdqVHfAQea1myuZCZK7exLKtwdzgDiI8MYWi3WI7pFkOf\nxCiS4yMY2i2WsGD39/Pu4myenr2ezfllGMPuIN2/UxT/e8EQRvTuSICBTXllzNuwk6mLslieXQTA\n6f0TeeLSY0k6jOmsddpxaLsMGL9HaBturb3bezsA+Aq43lq7aX+hzRhzC3ALQI8ePU7MzMxspXch\nIiJtRaFN/MquyhqCAwMICTr4KVhLNxdQUe1hVJ+OgDuWZ/baXCYN69pkWDuQhseC5RRX8OQX60iK\nDuWesf0OOEVs9dYiokOD6dExotHyiupaisurD2rjd1lWIVMXZ3Hj6BT6JkXtXl5aWcPNry4iKCCA\n3547iM6xYTz/zQZmr8nBGOgQEcKfLx5K78So/Tx74/e7aFM+J/WK32s655LNBazdXkLfpCj6JkbR\nwTsFcXNeGUs2FzC0eyx9mvk6TdlS6EZ6mpr+m1daSVBgAJEhgc2emldUXk12QRk7iis4tnvc7hBc\nXethe1EF+buq2FVZQ1BgABEhgfRKiCQq1M0St9aSt6uKzflldIkNo0vswR8rWcday5bCcgKMITI0\naHeg25/0HSXMWLGNYclxjOgdT0RI07PXy6tqSdtezM6SSgZ3jaFbXPgBd6J4PJY56bks3JhPQlQo\nyfERjBmQ2OQxj9ZaFm0qoLCsqtHlRg5XOw5to4D/tdaO995+EMBa+2fv7VhgA1DqfUhnIB+YuL/R\nNvWRIiJHB4U2ERHxGe04tAUB64CxwBZgEXC1tXb1Ptb/Gk2PFBERr+b2jwfcfW2MedkYk2OMWbWP\n+8cYY4qMMcu8Pw8fSsEiIiL+xlpbA9wFfAakAVOttauNMY8aYya2bXUiItJeNOfska8ATwOv7Wed\nudba81ukIhERET9irZ0JzNxjWZM7MK21Y1qjJhERaV8OONJmrZ2Dm38vIiIiIiIiraylLrwzyhiz\n3BjziTFmSAs9p4iIiIiIyFGvJS6uvQToaa0tNcacC3wA9GtqxT1OZ9wCLy0iIiIiItK+HfZIm7W2\n2Fpb6v19JhBsjEnYx7qTrbWp1trUxMTEw31pERERERGRdu+wQ5sxprPxXsjHGDPc+5x5h/u8IiIi\nIiIi0ozpkcaYt4ExQIIxJht4BAgGsNY+D1wK3G6MqQHKgSttW138TUREREREpJ05YGiz1l51gPuf\nxl0SQERERERERFpYS509UkRERERERI4AhTYREREREREfptAmIiIiIiLiwxTaREREREREfJj/hjaP\nB9Z95v4VERERERFpp/w3tK2dCW9dDms/butKREREREREjhj/DW39J0CHFPj2H6DLwomIiIiISDvl\nv6EtMAhG3wNbfoCNc9q6GhERERERkSPCf0MbwHFXQ1QnN9omIiIiIiLSDvl3aAsOg5F3QMZsyFrU\n1tWIiIiIiIi0OP8ObQCpN0JUZ3j3eije1tbViIiIiIiItCj/D21hMXDNVCgvcGeTrCxt64pERERE\nRERajP+HNoAux8Hlr8KO1TD7j21djYiIiIiISItpH6ENoN846Hc2rPlYlwAQEREREZF2o/2ENoC+\nY6EwE/Iz2roSERERERGRFtH+QhvA+lltW4eIiIiIiEgLaV+hLb43dEiBDQptIiIiIiLSPrSv0AbQ\n9yzYOAdqKtu6EhERERERkcPWDkPbWKgug83z27oSERERERGRw9b+QluvUyEgGNZ91taViIiIiIiI\nHLb2F9pCo9xo2/xn4OVzFN5ERERERMSvtb/QBnDxi3D2H6F4C7x1OSx9o60rEhEREREROSTtM7SF\nxcDJd8GdC6HPmfDhXbDyvbauSkRERERE5KC1z9BWJzgMrngTep4M034GH/8KygugtgaKssFTW7/u\n+i9hyettV6uIiIiIiEgTgtq6gCMuJAKungqzfg+L/g3L34baKvdz+gNwxoNQXQHv3wa7ciEqCfqP\nb+uqRUREREREgGaMtBljXjbG5BhjVu3jfmOMecoYs94Ys8IYc0LLl3mYQqPg3Cfglm9gyEUw8nbo\neQp8/wyU5cPKd72BrRN8cAeUbN/7OTwe+P5ZyEk7MjWWF8K/joPl7xyZ5xcREREREb/UnOmRrwAT\n9nP/OUA/788twHOHX9YR0uVYmPQ0jHsUzn0cqkpg/rMuvHUaCj/9EKpK4f1boba68WO/fRI+exCm\n3eICXEtb+gYUbIJv/tJ42qaIiIiIiBzVDhjarLVzgPz9rDIJeM0684E4Y0yXlirwiOk0BAZNhLlP\nQm4ajLoTkga5EbmMr+Hd66Gmyq2b/gV89Rgk9IftK+DHD1q2Fk8tLJwMYbGQnwFrZ7bs8x+MmkpY\n9ha8dxOU5rZdHUfaR/fB6xdBZWlbVyLSWG0NrHjXjb6LiIiI0DInIukGZDW4ne1d5vtO/w3YWoju\nAsdc4pad8FOY8BdYMwPeuBjevtoFuE7HwM9mQ9JgmP1Ht2HVUtZ9BoWZcP4/IK4nzPu/g3+OTd/B\nlh8Or44VU+EfQ+CD22HVe7D8rQM/xlMLO1bvf538DPjhVfez4avm1VKUDVuXNm/ddZ/D7D+5sFmQ\neeD1V74HP/zH1TLlandMo7/L/B5m/eHIjAL7A2sPbv2KIkj7qGX/jlvKyqkw7WZ4baKbvt1cB7Pu\ngayZCV/+vuWeT0RERA5LS4Q208SyJregjDG3GGMWG2MW5+b6wChO52O8UyX/BkEh9ctH3g7nPOFC\nQ146DDgXrnrLHRt35u8gb73bqPrwTvjiERcwALYug1mPwse/hI/ubX7oWPAcxHSHQZPciF/WAti8\noP7+jG9cKNnXtMmqXfD2lfDSeBe8DlZNJcz4uTvDZocUuPYD6HoCrH6/fp3NC2DVNNg8v/Ho1BcP\nw3MnQ9qMpp/bWnj3BvjoHvfz+kUw89cH3lh+/zZ45Xz33vZn6ZvuWnzf/NWFzRdOhcqSfa9ftAU+\n/gV0PwkmPQsbv4H3b9n/a+xL7jqY8wRUlR3a4+t88wSs/eTQH5+1CN64BOb+rXlB29r6UeSmFGTC\nezfCi2fCv8e5UdfV70P6l+51Hu8DhZsPvd6mVJZA8dbGy3LXNg5jq9+Hnel7P3bVNLezIXtx815r\n+yqYPAbe+Qm8NqnxMay1NW7557876LfQSGUpbJxz8I+zFuY9DdFdIWcNvHoB7Nq5/8eU5rhjcR9P\ncXU3N8B6at207D3D3rYV7vP/9smW/5xFRETkkLTE2SOzgeQGt7sDW5ta0Vo7GZgMkJqaepC7xo+Q\n0fc2vXzELTD8Z2D2yKQDzoHeY2DNxxDR0Z3A5PtnIKEf5PwIAUEQGuPOTrn0TRj/Rxh+S+PnqSxx\nlxjI+Bq2r3QjZGf9LwQGwbBr4Os/uyBx+WsuEE69DjzVbkNwwp/2rvXHD6GyGBIGuOCVNt2NJJhA\nOP1+d8mDmkrIWuiuYRffG0Kj3WMLMuHd61zAPPkeGPuIq2P7Svjid5C/EYJC3V7/Gu+IVFwPuPEz\ntzE5/zkwAS6IpZzmnr+hzfNh2zI4+zEYfCEseB6+f9ptkJ/xECQP37uNc9fBprnu97SP4Lgr937P\n1RWw5DX45H73eVz+mnutty5zJ5ZJvXHvxxRluwBZWw0XvQAd+0DBRhe8CrMgLnnvx1RXwAe3QVC4\na/vwDu7z+/Yf8N1T7nMJCnfXBaythreugIHnwUk37f1c4NZJ+8h9j4LDYcsSmP2YC+19z4LA4L0f\nk58BH97t2uH4nzRurx2r4c1L3FlPw+Pc6MigiXt/Dg29f6sLOLfNhZDI+uWeWljwAnz1B8BAj5Hg\nqYGM2W7kFSAyyX235j4JF/xz36/RlJIdri3H/BaST6pfvmsnvDzetet9q9wOlNXvuxHucY+6v9E1\nM93thAFw+3f17VRdDp//DxRvcYHy+o/dzph9Wf8lTPmJm4p8xkPuc3z+FLjiDfd+v33SfT7grvHY\n54yDe491Pv4lrJgC182AlFOb/7iMryFnNUx8GmK7uZH+V86Dn06H6E71621bAd/90+2E2LHK/X33\nGAXznnLf0VN/Ub+uxwO1le771tDsP7mgf/xPYNIzbllZvgutIRFQU+52Joy49dDaQERERFqMsc3Y\nK2uM6QXMsNbutTVkjDkPuAs4FxgBPGWtHX6g50xNTbWLFzdzz7ivqWszY9ye6HlPu9BzzCUw7Cq3\nQViW70aL0j9zIS4y0YWf6nK3gVlb5dbrfKwb9Tnt125DCVy4mnodlOW56ZtdjoPOQ+GHV9yo4IBz\nXFCK6erWf3mCC4+3fec2Fjd85Tb4irZAyVbocbLbsKssrn8PHftBjxFuhMx64MLnYND59fcXboZ/\nDnVhsiDT7ZH/yX+hPN8FiNhubiOwMMsFoDcvdeH03MfdBmRQqHueqT91I4W/SKt/f0ted2Grugzi\n+8CVb7rjCet88oC7PENUkgtW131Uf19NJcz8lRtdqSqF3mfAVW+7Wqx1G+ABgXBrg1GO6go3AvXF\nIy6ETHoGjrnY3bdjtRspnPSM23htqLYa3rkW1n3iAnB0F+h3Fqz8rzuJzXFXu0BVmAn3Lnchcuav\nICQa7lsBEfF7f3cWvujWOelncN7f4O2rIP1zV9fFL8Kxlzdev7oCXhrnQjQW+ox1tcZ0cd+xF053\nwfHGz6BspxsdG30vjLzD1RbXA2K61Qe9ujAEbnrwGb/1ft5ZbqRy01zoNx7O+3t9iK2tgc3fu8++\n/wT49EH3Xu9Z2nTQLch0o2aRifXfE3DfzUX/hsSBcNu3LnhVlrrRpK1L3fu78i0Xel+/GDbMcu1+\n6UvusQFBULoDJvwVRt7mnvO7f7nR3guf804PrYbUm1xQ6jm6ccCtqYJnToKgMPedikpyZ4Odcg0U\nZcHo+2Du32HQBa69bS3c/n3997a5sn+Af5/pfu95Ctzwsfs97SO3gyXja7dzZlwT0w/fuBS2LYef\nr3J/Qxvnuh0BMV1czTFd3ef+/CluFLrzUIhPcTtc4vu4nT0r363/PlvrQtj2FXD7vPqdNWkz4J1r\nvDshSuG+lRDd2b1Wxmy44RP3fYjp6k7Q1AKMMT9Ya1Nb5MmOAn7dR4qISLM1t388YGgzxrwNjAES\ngB3AI0AwgLX2eWOMAZ7GnWGyDLjBWnvAnuao6JA8HndduG3LYVeOCwHB4W7jqP85kDzCjWo1ZddO\nmH6PG9267BUIjnBTINd/Ub/OKT+H466CZ4bXj0g0VFXmNmpXvuv2wg88z4XFvPWQvQgy57lQdOnL\nbvRtTy+e6eoo3uJGrs59wi3fONeNatRWwkWT4bgr3Ejbwhfd+6suc5dWOO1+eH40nHy3q6+hyhK3\nEfvpg64drplaX/OTA92oU+JAd/zgfStd+AB3Qpg5T7gN0mMugV6nNW7DulB0y9fuPX3zOCx7011U\nvecp7uyh8Sn161sLf+vnwt8lL9YvL97q3tOaGS4odzsB/vszt3E/5GI3EtvtRBeQX78Ixv/ZbfDX\nBYHR9+z9nj218HSqC8SeGhj3BzeaOeZBF0IDQ1zY/PJhN8Jx4vVuRHLp63Dl2+5z+OJhCIuDq6e4\nkJLxtQts3U90r/H+7XtPkYxIgMETXch8+wqITXZtsPYTuGuR20kw4xcupJzzOAy7eu/Rz4aKsuFf\nw+CEa91xmNa6OtM+grQPvQGz7rU7ulGikEj33jsNcX8PZz8Gx17hAuTm+XD5q66G5OHue/aPITD8\nVhdo8ze4oHXLN/DpA7B1Cdy9FAIC3GUykkfANe+6EdoPbnOjl1j3XhqOEtV9N655D/qNq19elu+C\n2+Z5ENvDjUDuWOVGuPpPgO6p7m85d42bijjkQvd3Fxq1d9tY60YN8ze61/7qD260LXuRu1ZkRIL7\n+89JcyEqaWD9Y+vC3hkPuRHyOpnfu50iwREujC1+yX3vbvoCug5r/Pq11W7dzO/h5i9dW0+/y903\n6i438r99Jbx8jpsdcNHz8OxId198ipsmXReKv3jEjYr/eoMbxT1MCm0H56joI0VEpOVC25GiDukQ\nVO2C1R+4jevMeS4QdkhxQeIXaS4wHAxr979xPu//3NSz4Ai4Z1nj6VkbvnLHUp1+v3uOyhJ3PF9g\niNtwXPSiGxnx1LpRqKZGZMAFsK8ecyd56XaCG4WbfhdcP9M95p9D6zdity5zQfLYK+CifVxZoqII\n/jYAeo12I00FmW7k5MTrIGWM29Df03s3uRGmX651YeqLh2HRS66dz/q9m/oI7r3UVDSeUmgtTD7d\nTVfDws++gvnPuwBz53xXszFuyuKaGW7U48LnXcDLS4fQWDcql/aRe9/JIyFrvjtT6c517jVG31sf\nALevcsfwlWx39Z37NzeNt86unW7aXEx3F1oLM90xkj9+6AJ7QLALhmEx8H+pbrS3dDt0Hw4XT24c\naPdnxs/dZ9U91YW4Iu+5iJJHuPZOGuQCzqw/uDbtPNR9Z+9Z6o73zPzOtWNFsQvSQy9137X5z8GI\n21xYuGepG5l+QVaiCwAAEXFJREFU/SI3Inji9S7sPDfafdery9zjb/u28ZTI8kI3ndcEuPAO7m/n\nX8NcULn+472/9zWVbmph/wmuVoAv/9eNonu8l/+I6+lqzvnRfW5XvQW9TnH3ff0X2PSt+/5vmAUT\n/w+GXu5CpTFQsg2GXuZGpcsL4anjXUD9iXfa6fIp7oymYbEuzEV2bFzfjtXw35vda8PegbSh0lx3\nbGdQKOzKcyP18SnuRD2XvgQf/8rdd9PnENvdHb+27jM34p48An4yzf2dZC10o7yXvOQ+n8Ok0HZw\n1EeKiBwdFNraO48HZtwHS151G8lXvNHyr1GYBU8Nc6Fh7MMH99hN37qNwT5nur35+1JR7IJZz5Nh\nzANuumBoNNwx323svnqBO/lE6k2wepobFblzvpvWtS8f3AnL3nDTGS/9D/Qctf9al7wG0+92r5k5\nz52oZNg1Lih26HXg91o35XDIxXDZfyBvAzx9ktsIrjsnz/HXutGo0u1ulGjbcjcic/r97qem0rVD\n6Q4483/g1F+5YwGzFrpRzobHuhVvc22bNBDOe3L/wbtOyXZY/LKbFnvsZW7ZnL+545rGPACn/GLf\no75NKdrijoUMDHUjRz1HwYDz3DS+hnLXwn/OcVN968JnfgY8O8qN+F3+qht9AxfInh3pfu9xMtzo\nPTnLnjsX5j/nptzGdHXHMw6euHd9ddMm710BHXrW7xy46QsXlg5GTZX7LIPD3O2sRe7YUWPc9Mns\nRfDq+a5tsS4sXzXFTdNd8IKbCjzgXHfcZd3nWLdDZPR97pjWTXPdSPClLzfeOdJQdYU73rW2Csb/\naf+fe+Y8dyKf4HB3DGBINDx9ohtxjurkpj927OPW3bYcXjjNBcY75tdPu/Z44O8DXDC97D8H12ZN\nUGg7OOojRUSODgptRwOPB5a+5oJR3fTBlpa3wQWXgMCDf2zdGSIPFAa+edxNgwwMhcgEd4xb1+Pd\nfes+d1PeyvLc7aumuGP69qcg0133bvS9zRt9LMiEfx3rjt+b/7x7vzd+2rwwBPXX2Rtycf0G9/fP\nulGRoZe6ADvHO7W04fFYu/LccW91r5M5z4XShscWHknWupHJFpj6tl/blsOCyTD+sfqwXZTtpk7u\neXKMyWe46Y8Tn3bTLw9V/ka3w+Hsx1wA/9dx0OtUNzrWEjbMhtcvdKFrzQw3mtjU8W+1NbDuUzfd\nty70gQuCz45wAbZDihsJHnX3wQXnA1n7qZvCWTcauPI9F9Qve6XxtExw3/vOx9SvW2f63W50/9cb\nGp9h9xAotB0c9ZEiIkcHhTbxHxVF8PRwSOwPl7wMUYl7r1Nd7qa4RSYcmRr+NcyNRtWUu6mZvUa3\n7PMvn+JG5C55qeljocRZ8a47Duy2b/d/BszmeOE0Nx20x0h3htc7vm98wpvD9d5N9WfVvPZ9t/Pk\nYBRmuZPHdBnW/B0ErW3tp26a6FVvNX3c60FQaDs46iNFRI4Oze0fW3C3rsghCvMe1xUYsu+N1+Dw\nvUdlWlLv093ZOfuMbfnABu50/U1dukAaO/ay+umbh2vwJHec5fYV7sQhLRnYwE1RzJjtTip0sIEN\n3DGb+zrW01f0Hw8DJrR1FSIiIke9lri4tsjhCwpt29GGAee5UZmxh3lRZfEdgy+s//2MB1v++aM7\nuZPsTHq65Z/bV/jqCKCIiMhRRqFNBKD/2fCbjfXH0on/69jHnbXztPuP3DGfodEKNkc5Y8wEY8xa\nY8x6Y8wDTdz/C2PMj8aYFcaYWcaYnm1Rp4iI+DdNjxSpU3fhYWk/rni9rSuQdswYEwg8A4wDsoFF\nxpjp1tofG6y2FEi11pYZY24HHgeuaP1qRUTEn2mkTURE5NAMB9ZbazOstVXAFGBSwxWstbOttWXe\nm/OB7q1co4iItAMKbSIiIoemG5DV4Ha2d9m+3AR8sq87jTG3GGMWG2MW5+bmtlCJIiLSHii0iYiI\nHJqmDmhs8jo6xpifAKnAE/t6MmvtZGttqrU2NTGxiUufiIjIUUvHtImIiByabKDhdRu6A1v3XMkY\ncxbwEHC6tbaylWoTEZF2RCNtIiIih2YR0M8Yk2KMCQGuBKY3XMEYczzwAjDRWpvTBjWKiEg7oNAm\nIiJyCKy1NcBdwGdAGjDVWrvaGPOoMWaid7UngCjgXWPMMmPM9H08nYiIyD5peqSIiMghstbOBGbu\nsezhBr+f1epFiYhIu6ORNhERERERER+m0CYiIiIiIuLDFNpERERERER8mEKbiIiIiIiID1NoExER\nERER8WEKbSIiIiIiIj5MoU1ERERERMSHKbSJiIiIiIj4MIU2ERERERERH6bQJiIiIiIi4sMU2kRE\nRERERHxYs0KbMWaCMWatMWa9MeaBJu6/3hiTa4xZ5v25ueVLFREREREROfoEHWgFY0wg8AwwDsgG\nFhljpltrf9xj1XestXcdgRpFRERERESOWs0ZaRsOrLfWZlhrq4ApwKQjW5aIiIiIiIhA80JbNyCr\nwe1s77I9XWKMWWGMec8Yk9wi1YmIiIiIiBzlmhPaTBPL7B63PwJ6WWuPBb4EXm3yiYy5xRiz2Biz\nODc39+AqFREREREROQo1J7RlAw1HzroDWxuuYK3Ns9ZWem++CJzY1BNZaydba1OttamJiYmHUq+I\niIiIiMhRpTmhbRHQzxiTYowJAa4EpjdcwRjTpcHNiUBay5UoIiIiIiJy9Drg2SOttTXGmLuAz4BA\n4GVr7WpjzKPAYmvtdOAeY8xEoAbIB64/gjWLiIiIiIgcNQ4Y2gCstTOBmXsse7jB7w8CD7ZsaSIi\nIiIiItKsi2uLiIiIiIhI21BoExERERER8WEKbSIiIiIiIj5MoU1ERERERMSHKbSJiIiIiIj4MIU2\nERERERERH6bQJiIiIiIi4sMU2kRERERERHyYQpuIiIiIiIgPU2gTERERERHxYQptIiIiIiIiPkyh\nTURERERExIcptImIiIiIiPgwhTYREREREREfptAmIiIiIiLiwxTaREREREREfJhCm4iIiIiIiA9T\naBMREREREfFhCm0iIiIiIiI+TKFNRERERETEhym0iYiIiIiI+DCFNhERERERER+m0CYiIiIiIuLD\nFNpERERERER8mEKbiIiIiIiID2tWaDPGTDDGrDXGrDfGPNDE/aHGmHe89y8wxvRq6UJFRER8kfpI\nERE50g4Y2owxgcAzwDnAYOAqY8zgPVa7CSiw1vYF/gH8taULFRER8TXqI0VEpDU0Z6RtOLDeWpth\nra0CpgCT9lhnEvCq9/f3gLHGGNNyZYqIiPgk9ZEiInLENSe0dQOyGtzO9i5rch1rbQ1QBHRsiQJF\nRER8mPpIERE54oKasU5TewPtIayDMeYW4BbvzVJjzNpmvP7+JAA7D/M5WpPqPXL8qVbwr3r9qVbw\nr3r9qVY4vHp7tmQhPkR9ZMvxp3r9qVbwr3r9qVbwr3r9qVbwr3qPeP/YnNCWDSQ3uN0d2LqPdbKN\nMUFALJC/5xNZaycDk5tTWHMYYxZba1Nb6vmONNV75PhTreBf9fpTreBf9fpTreB/9bYS9ZEtxJ/q\n9adawb/q9adawb/q9adawb/qbY1amzM9chHQzxiTYowJAa4Epu+xznTgOu/vlwJfWWv32osoIiLS\nzqiPFBGRI+6AI23W2hpjzF3AZ0Ag8LK1drUx5lFgsbV2OvAS8LoxZj1u7+GVR7JoERERX6A+UkRE\nWkNzpkdirZ0JzNxj2cMNfq8ALmvZ0pqlxaaRtBLVe+T4U63gX/X6U63gX/X6U63gf/W2CvWRLcaf\n6vWnWsG/6vWnWsG/6vWnWsG/6j3itRrN0BAREREREfFdzTmmTURERERERNqI34Y2Y8wEY8xaY8x6\nY8wDbV1PQ8aYZGPMbGNMmjFmtTHmXu/yeGPMF8aYdO+/Hdq61oaMMYHGmKXGmBne2ynGmAXeet/x\nHmTf5owxccaY94wxa7xtPMqX29YY83Pv92CVMeZtY0yYL7WtMeZlY0yOMWZVg2VNtqdxnvL+3a0w\nxpzgA7U+4f0urDDGvG+MiWtw34PeWtcaY8a3Zq37qrfBfb8yxlhjTIL3ts+1rXf53d72W22MebzB\n8jZtW9k/9ZEty1/6R/CvPlL9Y6vU65N9pD/1j/urt1X7SGut3/3gDvbeAPQGQoDlwOC2rqtBfV2A\nE7y/RwPrgMHA48AD3uUPAH9t61r3qPsXwFvADO/tqcCV3t+fB25v6xq9tbwK3Oz9PQSI89W2xV1U\ndyMQ3qBNr/eltgVOA04AVjVY1mR7AucCn+CuOzUSWOADtZ4NBHl//2uDWgd7/28IBVK8/2cEtnW9\n3uXJuBNXZAIJPty2ZwBfAqHe20m+0rb62e9nqT6y5Wv2i/7RW49f9JHqH1utXp/sI/2pf9xP27Zq\nH+mvI23DgfXW2gxrbRUwBZjUxjXtZq3dZq1d4v29BEjD/ec0CfefKd5/L2ybCvdmjOkOnAf823vb\nAGcC73lX8Yl6jTExuD+clwCstVXW2kJ8uG1xJ/wJN+76TBHANnyoba21c9j7mlH7as9JwGvWmQ/E\nGWO6tE6lTddqrf3cWlvjvTkfd52sulqnWGsrrbUbgfW4/ztazT7aFuAfwP00vsCyz7UtcDvwF2tt\npXednAa1tmnbyn6pj2xB/tI/gl/2keofW5A/9ZH+1D+Cb/SR/hraugFZDW5ne5f5HGNML+B4YAHQ\nyVq7DVynBSS1XWV7+Sfuj8Tjvd0RKGzwh+4rbdwbyAX+452q8m9jTCQ+2rbW2i3A34DNuM6oCPgB\n32zbhvbVnr7+t3cjbm8c+GitxpiJwBZr7fI97vLFevsDp3qnKn1jjDnJu9wXa5V6fvP5+Ekf6S/9\nI/hRH6n+sU34dB/pZ/0jtHIf6a+hzTSxzOdOg2mMiQL+C9xnrS1u63r2xRhzPpBjrf2h4eImVvWF\nNg7CDU8/Z609HtiFm57gk7xz3Sfhhse7ApHAOU2s6gtt2xy++r3AGPMQUAO8WbeoidXatFZjTATw\nEPBwU3c3sayt2zYI6ICbjvJrYKp3lMEXa5V6fvH5+EMf6Wf9I/hRH6n+sXX5eh/ph/0jtHIf6a+h\nLRs357VOd2BrG9XSJGNMMK4zetNaO827eEfdcK7335x9Pb6VjQYmGmM24abRnInbsxjnnbIAvtPG\n2UC2tXaB9/Z7uA7KV9v2LGCjtTbXWlsNTANOxjfbtqF9tadP/u0ZY64Dzgeusd4J5fhmrX1wGyjL\nvX9v3YElxpjO+Ga92cA075SUhbiRhgR8s1ap5/Ofjx/1kf7UP4J/9ZHqH1uJn/SR/tY/Qiv3kf4a\n2hYB/Yw7w1AIcCUwvY1r2s2bsl8C0qy1Tza4azpwnff364APW7u2plhrH7TWdrfW9sK15VfW2muA\n2cCl3tV8ol5r7XYgyxgzwLtoLPAjPtq2uGkfI40xEd7vRV29Pte2e9hXe04Hfuo9k9NIoKhumkhb\nMcZMAH4DTLTWljW4azpwpTEm1BiTAvQDFrZFjXWstSuttUnW2l7ev7ds3AkZtuODbQt8gNtIxRjT\nH3dSg534YNtKI+ojW4g/9Y/gd32k+sdW4C99pB/2j9DafaRt5bOvtNQP7kwy63BnZHmorevZo7ZT\ncMOgK4Bl3p9zcfPgZwHp3n/j27rWJmofQ/3ZsXp7v2TrgXfxnh2nrX+AYcBib/t+gBua9tm2BX4P\nrAFWAa/jzibkM20LvI07nqAa95/kTftqT9yQ/zPev7uVQKoP1LoeN3e87m/t+QbrP+StdS1wji+0\n7R73b6L+7Fi+2LYhwBve7+4S4ExfaVv9HPDzVB/Z8nX7fP/orc1v+kj1j61Sr0/2kf7UP+6nbVu1\njzTeJxYREREREREf5K/TI0VERERERI4KCm0iIiIiIiI+TKFNRERERETEhym0iYiIiIiI+DCFNhER\nERERER+m0CYiIiIiIuLDFNpERERERER8mEKbiIiIiIiID/t/zZP7q27t+UwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f75d4a37c50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train loss is: 1.6129793438911437\n",
      "Validation accuracy is: 0.6444000000000001\n",
      "Validation loss is: 0.8935291022062302\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# TRAIN!\n",
    "criterion = nn.NLLLoss().cuda()\n",
    "vbity = 250\n",
    "epoch = 2000\n",
    "\n",
    "runningloss = 0\n",
    "te_optimizer = None\n",
    "for it in range(20*epoch):\n",
    "    if it%10==0:\n",
    "        print(it)\n",
    "    \n",
    "    # Build batch\n",
    "    support, query, qtargs, stargs = batchmaker(way,trainshot,testshot,trainset,alterful=True)\n",
    "    \n",
    "    # Predict\n",
    "    embed.zero_grad()\n",
    "    embeds = embed(support)\n",
    "    qembeds = embed(query)\n",
    "    preds = model(embeds, stargs, qembeds, int((it/40000)**2)*80+20)\n",
    "\n",
    "    # Calculate Loss\n",
    "    loss = criterion(preds, qtargs)\n",
    "    runningloss += loss.data[0]\n",
    "    \n",
    "    # Backprop\n",
    "    if it%epoch == 0:\n",
    "        te_optimizer = optim.Adam(embed.parameters(),lr=.001/(2**(it//epoch)))\n",
    "    loss.backward()\n",
    "#     nn.utils.clip_grad_norm(model.parameters(), 1)\n",
    "    te_optimizer.step()\n",
    "    \n",
    "    # Report\n",
    "    if it%vbity == vbity-1:\n",
    "        display.clear_output(wait=True)\n",
    "\n",
    "        losstracker.append(runningloss/vbity)\n",
    "        evalloss, evalacc, _ = newevaluate(embed,model,criterion,evalway,trainshot,testshot,testset)\n",
    "        evallosstracker.append(evalloss)\n",
    "        evalacctracker.append(evalacc)\n",
    "        \n",
    "        pl.figure(1,figsize=(15,5))\n",
    "        pl.subplot(1,2,1)\n",
    "        pl.plot(losstracker)\n",
    "        pl.plot(evallosstracker)\n",
    "        pl.ylim((0,4))\n",
    "        pl.title(\"Loss: Training Blue, Validation Gold\")\n",
    "        pl.subplot(1,2,2)\n",
    "        pl.plot(evalacctracker[::-1])\n",
    "        pl.ylim((0,1))\n",
    "        pl.title(\"Validation Acc\")\n",
    "        pl.show()\n",
    "        \n",
    "        print(\"Train loss is: \"+str(runningloss/vbity)+\n",
    "              \"\\nValidation accuracy is: \"+str(evalacc)+\n",
    "              \"\\nValidation loss is: \"+str(evalloss)+\"\\n\")\n",
    "        runningloss = 0\n",
    "\n",
    "# META\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/dww78/anaconda3/lib/python3.6/site-packages/torch/serialization.py:147: UserWarning: Couldn't retrieve source code for container of type ENCODER. It won't be checked for correctness upon loading.\n",
      "  \"type \" + obj.__name__ + \". It won't be checked \"\n",
      "/home/dww78/anaconda3/lib/python3.6/site-packages/torch/serialization.py:147: UserWarning: Couldn't retrieve source code for container of type Block. It won't be checked for correctness upon loading.\n",
      "  \"type \" + obj.__name__ + \". It won't be checked \"\n",
      "/home/dww78/anaconda3/lib/python3.6/site-packages/torch/serialization.py:147: UserWarning: Couldn't retrieve source code for container of type smartPREDICTOR. It won't be checked for correctness upon loading.\n",
      "  \"type \" + obj.__name__ + \". It won't be checked \"\n"
     ]
    }
   ],
   "source": [
    "name = 'sklearn_lowshot'\n",
    "torch.save((embed,model,losstracker,evallosstracker,evalacctracker),'saved_models/'+name+'.pth')\n",
    "torch.save(embed.cpu().state_dict(),'saved_models/'+name+'_cpu.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7235450000000001"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(evalacctracker[-1000:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class PREDICTOR(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(PREDICTOR, self).__init__()\n",
    "        self.sm = nn.LogSoftmax()\n",
    "        \n",
    "    def forward(self, support, inp, way):\n",
    "        assert support.size(0)%way == 0,\"Error: classes are not the same size!\"\n",
    "        support = support.view(way,support.size(0)//way,-1)\n",
    "        centroids = torch.mean(support, 1).view(support.size(0),-1)\n",
    "#         temp1 = torch.stack([centroids]*inp.size(0))\n",
    "#         temp2 = torch.stack([inp]*centroids.size(0)).transpose(0,1)\n",
    "#         print(temp1.size(),temp2.size())\n",
    "        distmat = torch.sum((torch.stack([centroids]*inp.size(0))-torch.stack([inp]*centroids.size(0))\n",
    "                             .transpose(0,1))**2,2).squeeze()*-1\n",
    "        probs = self.sm(distmat)\n",
    "        return probs\n",
    "embed, _ = torch.load('saved_models/naive_2_noinfo.pth')\n",
    "embed = embed.cuda()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n6P8DsncXy0YbHmAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f83b89d9a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Build batch\n",
    "# embed, model, _,_,_ = torch.load('saved_models/smart_1.pth')\n",
    "support, query, qtargs, stargs = batchmaker(way,trainshot,testshot,valset,alterful=False)\n",
    "criterion = nn.NLLLoss().cuda()\n",
    "embed = embed.cuda()\n",
    "model = model.cuda()\n",
    "\n",
    "# Predict\n",
    "tracker = []\n",
    "embed.zero_grad()\n",
    "embeds = embed(support)\n",
    "model.initialize(way)\n",
    "tr_optimizer = optim.Adam(model.parameters(),lr=.01)\n",
    "for i in range(15):\n",
    "    model.zero_grad()\n",
    "    preds = model(embeds.detach())\n",
    "    loss = criterion(preds, stargs)\n",
    "    tracker.append(loss.data[0])\n",
    "    loss.backward()\n",
    "    tr_optimizer.step()\n",
    "# print(model.weights.data)\n",
    "# print(torch.var(model.weights.data.view(-1)),torch.max(model.weights.data),torch.min(model.weights.data))\n",
    "pl.figure()\n",
    "pl.plot(tracker)\n",
    "pl.ylim((0,4))\n",
    "pl.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Variable containing:\n",
       " 20.6023\n",
       "[torch.cuda.FloatTensor of size 1 (GPU 0)]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.mean(torch.norm(embeds,2,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# BASELINE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.91514739215374, 0.6385333333333335, 0.010876071111111113)\n"
     ]
    }
   ],
   "source": [
    "print(newevaluate(embed,model,criterion,evalway,trainshot,testshot,testset))\n",
    "# TESTSET"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.6446040686964989, 0.7650666666666668, 0.0088703288888888881)\n"
     ]
    }
   ],
   "source": [
    "print(newevaluate(embed,model,criterion,evalway,trainshot,testshot,trainset))\n",
    "# TRAINSET"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.7136507552862167, 0.7288000000000002, 0.012978559999999997)\n"
     ]
    }
   ],
   "source": [
    "print(newevaluate(embed,model,criterion,evalway,trainshot,testshot,testset))\n",
    "# TESTSET, 100 PATIENCE, NAIVE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.726845205873251, 0.7353333333333336, 0.0094271111111111101)\n"
     ]
    }
   ],
   "source": [
    "print(newevaluate(embed,model,criterion,evalway,trainshot,testshot,testset))\n",
    "# TESTSET, 100 PATIENCE, SMART"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# EXPERIMENTS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.9770350706577301, 0.6170666666666665, 0.010962062222222221)\n"
     ]
    }
   ],
   "source": [
    "print(newevaluate(embed,model,criterion,evalway,trainshot,testshot,testset))\n",
    "# TESTSET, 100 PATIENCE, SMART. LR .1, MOM 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.9578658026456833, 0.6214000000000002, 0.013506039999999999)\n"
     ]
    }
   ],
   "source": [
    "print(newevaluate(embed,model,criterion,evalway,trainshot,testshot,testset))\n",
    "# TESTSET, 100 PATIENCE, NAIVE. LR .1, MOM 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.9450288346409798, 0.6406666666666669, 0.013203111111111108)\n"
     ]
    }
   ],
   "source": [
    "print(newevaluate(embed,model,criterion,evalway,trainshot,testshot,testset))\n",
    "# TESTSET, 50 PATIENCE, SMART. LR .1, MOM 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0.9806817117333412, 0.6148000000000002, 0.011196071111111108)\n"
     ]
    }
   ],
   "source": [
    "print(newevaluate(embed,model,criterion,evalway,trainshot,testshot,testset))\n",
    "# TESTSET, 20 PATIENCE, SMART. LR 1, MOM .5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f184c4b0908>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from random import randint\n",
    "pl.figure()\n",
    "pl.imshow(testset[randint(0,len(testset)-1)][randint(0,len(testset[0])-1)])\n",
    "pl.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA20AAAE/CAYAAADVKysfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f79e26b8710>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train loss is: 1.4755005943775177\n",
      "Validation accuracy is: 0.6951111111111113\n",
      "Validation loss is: 0.7553736646970113\n",
      "\n",
      "188500\n",
      "188510\n",
      "188520\n",
      "188530\n",
      "188540\n",
      "188550\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-6-186fa9537132>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m     \u001b[0;31m# Build batch\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 14\u001b[0;31m     \u001b[0msupport\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mquery\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtargs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mbatchmaker\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mway\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtrainshot\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtestshot\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtrainset\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0malterful\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m     \u001b[0;31m# Predict\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-4-84462d890901>\u001b[0m in \u001b[0;36mbatchmaker\u001b[0;34m(way, trainshot, testshot, theset, alterful)\u001b[0m\n\u001b[1;32m     15\u001b[0m         li = [torch.cat([alter(trainset[cl][i]).view(1,3,84,84) for i in \n\u001b[1;32m     16\u001b[0m                          np.random.choice(600,trainshot+testshot)],dim=0).float()\n\u001b[0;32m---> 17\u001b[0;31m               for cl in classes]\n\u001b[0m\u001b[1;32m     18\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     19\u001b[0m         li = [torch.cat([standardize(trainset[cl][i]).view(1,3,84,84) for i in \n",
      "\u001b[0;32m<ipython-input-4-84462d890901>\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m(.0)\u001b[0m\n\u001b[1;32m     15\u001b[0m         li = [torch.cat([alter(trainset[cl][i]).view(1,3,84,84) for i in \n\u001b[1;32m     16\u001b[0m                          np.random.choice(600,trainshot+testshot)],dim=0).float()\n\u001b[0;32m---> 17\u001b[0;31m               for cl in classes]\n\u001b[0m\u001b[1;32m     18\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     19\u001b[0m         li = [torch.cat([standardize(trainset[cl][i]).view(1,3,84,84) for i in \n",
      "\u001b[0;32m<ipython-input-4-84462d890901>\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m(.0)\u001b[0m\n\u001b[1;32m     13\u001b[0m     \u001b[0mclasses\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mchoice\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtheset\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mway\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     14\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0malterful\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 15\u001b[0;31m         li = [torch.cat([alter(trainset[cl][i]).view(1,3,84,84) for i in \n\u001b[0m\u001b[1;32m     16\u001b[0m                          np.random.choice(600,trainshot+testshot)],dim=0).float()\n\u001b[1;32m     17\u001b[0m               for cl in classes]\n",
      "\u001b[0;32m/home/dww78/anaconda3/lib/python3.6/site-packages/torchvision-0.1.9-py3.6.egg/torchvision/transforms.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, img)\u001b[0m\n\u001b[1;32m     32\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__call__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     33\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mt\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtransforms\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 34\u001b[0;31m             \u001b[0mimg\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     35\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     36\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dww78/anaconda3/lib/python3.6/site-packages/torchvision-0.1.9-py3.6.egg/torchvision/transforms.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, img)\u001b[0m\n\u001b[1;32m     32\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__call__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     33\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mt\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtransforms\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 34\u001b[0;31m             \u001b[0mimg\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     35\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mimg\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     36\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dww78/anaconda3/lib/python3.6/site-packages/torchvision-0.1.9-py3.6.egg/torchvision/transforms.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, tensor)\u001b[0m\n\u001b[1;32m    154\u001b[0m         \u001b[0;31m# TODO: make efficient\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    155\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mt\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mm\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0ms\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtensor\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmean\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstd\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 156\u001b[0;31m             \u001b[0mt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msub_\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mm\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdiv_\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    157\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mtensor\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    158\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "# TRAIN!\n",
    "criterion = nn.NLLLoss().cuda()\n",
    "vbity = 100\n",
    "epoch = 2000\n",
    "\n",
    "losstracker,evallosstracker,evalacctracker = [],[],[]\n",
    "runningloss = 0\n",
    "optimizer = None\n",
    "for it in range(epoch*100):\n",
    "    if it%10==0:\n",
    "        print(it)\n",
    "    \n",
    "    # Build batch\n",
    "    support, query, qtargs, _ = batchmaker(way,trainshot,testshot,trainset,alterful=True)\n",
    "    \n",
    "    # Predict\n",
    "    embed.zero_grad()\n",
    "    model.zero_grad()\n",
    "    embeds = embed(support)\n",
    "    qembeds = embed(query)\n",
    "    preds = model(embeds,qembeds,way)\n",
    "\n",
    "    # Calculate Loss\n",
    "    loss = criterion(preds, targs)\n",
    "    runningloss += loss.data[0]\n",
    "    \n",
    "    # Backprop\n",
    "    if it%epoch == 0:\n",
    "        optimizer = optim.Adam(embed.parameters(),lr=.001/(2**(it//epoch)))\n",
    "    loss.backward()\n",
    "#     nn.utils.clip_grad_norm(model.parameters(), 1)\n",
    "    optimizer.step()\n",
    "    \n",
    "    # Report\n",
    "    if it%vbity == vbity-1:\n",
    "        display.clear_output(wait=True)\n",
    "\n",
    "        losstracker.append(runningloss/vbity)\n",
    "        evalloss, evalacc = evaluate(embed,model,criterion,evalway,trainshot,testshot)\n",
    "        evallosstracker.append(evalloss)\n",
    "        evalacctracker.append(evalacc)\n",
    "        \n",
    "        pl.figure(1,figsize=(15,5))\n",
    "        pl.subplot(1,2,1)\n",
    "        pl.plot(losstracker)\n",
    "        pl.plot(evallosstracker)\n",
    "        pl.ylim((0,4))\n",
    "        pl.title(\"Loss: Training Blue, Validation Gold\")\n",
    "        pl.subplot(1,2,2)\n",
    "        pl.plot(evalacctracker)\n",
    "        pl.ylim((0,1))\n",
    "        pl.title(\"Validation Acc\")\n",
    "        pl.show()\n",
    "        \n",
    "        print(\"Train loss is: \"+str(runningloss/vbity)+\n",
    "              \"\\nValidation accuracy is: \"+str(evalacc)+\n",
    "              \"\\nValidation loss is: \"+str(evalloss)+\"\\n\")\n",
    "        runningloss = 0\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "Paper model achieves (over last 1000):\n",
    "    0.71156711111111126\n",
    "after 188500 iters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Variable containing:\n",
       "-6.7075e-03 -5.0079e+00 -2.7911e+01\n",
       "-2.9057e+00 -5.6263e-02 -4.5704e+01\n",
       "-1.8272e+01 -4.3870e+01 -1.1600e-08\n",
       "[torch.FloatTensor of size 3x3]"
      ]
     },
     "execution_count": 213,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "support = supportmaker(nclass,shot)\n",
    "inp = inpmaker(nclass)\n",
    "embeds = embed(support)\n",
    "query = embed(inp)\n",
    "model(embeds,query)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}