proto_L2.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",
    "%matplotlib inline\n",
    "import pylab as pl\n",
    "from IPython import display\n",
    "import time"
   ]
  },
  {
   "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",
    "#         self.final = nn.Conv2d(64,64, kernel_size=1, padding=0)\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",
    "#         out = self.final(out)\n",
    "        return out.view(out.size(0),-1)\n",
    "    \n",
    "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"
   ]
  },
  {
   "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": 7,
   "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 evaluate(embed,model,criterion,way,trainshot,testshot,reps,theset):\n",
    "    losstracker = 0\n",
    "    acctracker = 0\n",
    "    acclist = []\n",
    "    for i in range(reps):\n",
    "        s,q,qt,st = batchmaker(way,trainshot,testshot,theset)\n",
    "        embeds = embed(s)\n",
    "        qembeds = embed(q)\n",
    "        preds = model(embeds,qembeds,way)\n",
    "        \n",
    "        # Val loss\n",
    "        loss = criterion(preds,qt)\n",
    "        losstracker += loss.data[0]\n",
    "        \n",
    "        # Val accuracy\n",
    "        _,bins = torch.max(preds,1)\n",
    "        acc = torch.sum(torch.eq(bins,qt)).data[0]/testshot/way\n",
    "        acctracker += acc\n",
    "        acclist.append(acc)\n",
    "    return losstracker/reps,acctracker/reps,np.var(acclist)\n",
    "\n",
    "way = 20\n",
    "evalway = 5\n",
    "trainshot = 30\n",
    "testshot = 5\n",
    "reps = 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "torch.cuda.set_device(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "113088  parameters in neural net.\n"
     ]
    }
   ],
   "source": [
    "embed = ENCODER().cuda()\n",
    "model = PREDICTOR().cuda()\n",
    "losstracker,l2losstracker,evallosstracker,evalacctracker = [],[],[],[]\n",
    "optimizer = None\n",
    "nweights = sum([i.numel() for i in list(embed.parameters())])\n",
    "print(nweights,\" parameters in neural net.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA20AAAE/CAYAAADVKysfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/X8AMzQoi/LdggAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f68b1845438>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train loss is: 1.190390727519989\n",
      "L2 norm is: 5.894390897750855\n",
      "Validation accuracy is: 0.7300000000000003\n",
      "Validation loss is: 0.636213553249836\n",
      "\n",
      "5860.016466856003\n"
     ]
    }
   ],
   "source": [
    "# TRAIN!\n",
    "criterion = nn.NLLLoss().cuda()\n",
    "vbity = 200\n",
    "epoch = 2000\n",
    "start = time.time()\n",
    "\n",
    "runningloss, runningl2loss = 0, 0\n",
    "for it in range(10*epoch):\n",
    "    if it%10==0:\n",
    "        print(it)\n",
    "    \n",
    "    # Build batch\n",
    "    support, query, targs, _ = 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",
    "    l2loss = torch.mean(torch.norm(qembeds, dim=1))\n",
    "    totalloss = loss+l2loss/100\n",
    "    runningloss += loss.data[0]\n",
    "    runningl2loss += l2loss.data[0]\n",
    "    \n",
    "    # Backprop\n",
    "    if it%epoch == 0:\n",
    "        optimizer = optim.Adam(embed.parameters(),lr=.001/(2**(it//epoch)))\n",
    "    totalloss.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",
    "        l2losstracker.append(runningl2loss/vbity/10)\n",
    "        embed = embed.eval()\n",
    "        evalloss, evalacc, _ = evaluate(embed,model,criterion,evalway,trainshot,testshot,reps,testset)\n",
    "        embed = embed.train()\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.plot(l2losstracker)\n",
    "        pl.ylim((0,3))\n",
    "        pl.title(\"Loss: Training Blue, Validation Gold\")\n",
    "        pl.subplot(1,2,2)\n",
    "        pl.plot(evalacctracker[::-1])\n",
    "        pl.ylim((0.3,.8))\n",
    "        pl.title(\"Validation Acc\")\n",
    "        pl.show()\n",
    "        \n",
    "        print(\"Train loss is: \"+str(runningloss/vbity)+\n",
    "              \"\\nL2 norm is: \"+str(runningl2loss/vbity)+\n",
    "              \"\\nValidation accuracy is: \"+str(evalacc)+\n",
    "              \"\\nValidation loss is: \"+str(evalloss)+\"\\n\")\n",
    "        runningloss = 0\n",
    "        runningl2loss = 0\n",
    "        print(time.time()-start)\n",
    "\n",
    "# MORE RECENT TEST\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1.0421067474782466, 0.6425333333333333, 0.010061137777777777)\n"
     ]
    }
   ],
   "source": [
    "criterion = nn.NLLLoss().cuda()\n",
    "embed = embed.cuda()\n",
    "embed = embed.eval()\n",
    "print(evaluate(embed,model,criterion,evalway,5,testshot*3,200,testset))\n",
    "embed = embed.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.01318637251104336"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "1.96*np.sqrt(.0090525/200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.65521600000000002"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(evalacctracker[-50:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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 PREDICTOR. It won't be checked for correctness upon loading.\n",
      "  \"type \" + obj.__name__ + \". It won't be checked \"\n"
     ]
    }
   ],
   "source": [
    "name = 'L2_highshot'\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": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "embed,model,losstracker,evallosstracker,evalacctracker = torch.load('saved_models/naive_4.pth')"
   ]
  },
  {
   "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": 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
}