Exercise-2_3-MNIST-classifier-solution.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "1399acaa",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pennylane as qml\n",
    "from pennylane import numpy as np\n",
    "\n",
    "from sklearn import datasets\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.preprocessing import StandardScaler, MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ae6fe725",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X = [-0.79923362 -0.26571862 -0.16474421 -0.17815209], Y = -1\n",
      "X = [-0.81630337 -0.2918343  -0.48757339 -0.44903275], Y = -1\n",
      "X = [-0.8857451  -0.29703903 -0.27973225  0.17301948], Y = -1\n",
      "X = [-0.76039807 -0.27000038 -0.26925903  0.32714174], Y = -1\n",
      "X = [-0.73871156 -0.11343435 -0.47219408 -0.49067011], Y = -1\n",
      "X = [ 0.78786325 -0.59573349 -0.4920569   0.05387855], Y =  1\n",
      "X = [ 0.39958822 -0.54294278  0.8919327  -0.61020082], Y =  1\n",
      "X = [ 0.59776664 -0.3451937  -0.95910006  0.02118022], Y =  1\n",
      "X = [ 0.21238989  0.90675747 -0.01334434 -0.16475638], Y =  1\n",
      "X = [ 0.5672546  -0.35245554 -0.42810247  0.62548925], Y =  1\n",
      "...\n"
     ]
    }
   ],
   "source": [
    "# Data preparation\n",
    "\n",
    "# Load digits dataset\n",
    "digits = datasets.load_digits(n_class=2)\n",
    "\n",
    "# Split dataset\n",
    "sample_train, sample_test, label_train, label_test = train_test_split(\n",
    "     digits.data, digits.target, test_size=0.2, random_state=22)\n",
    "\n",
    "# Reduce dimensions\n",
    "n_dim = 4\n",
    "pca = PCA(n_components=n_dim).fit(sample_train)\n",
    "sample_train = pca.transform(sample_train)\n",
    "sample_test = pca.transform(sample_test)\n",
    "\n",
    "# Normalise\n",
    "std_scale = StandardScaler().fit(sample_train)\n",
    "sample_train = std_scale.transform(sample_train)\n",
    "sample_test = std_scale.transform(sample_test)\n",
    "\n",
    "# Scale\n",
    "samples = np.append(sample_train, sample_test, axis=0)\n",
    "minmax_scale = MinMaxScaler((-1, 1)).fit(samples)\n",
    "sample_train = minmax_scale.transform(sample_train)\n",
    "sample_test = minmax_scale.transform(sample_test)\n",
    "\n",
    "# Select\n",
    "train_size = 100\n",
    "X = sample_train[:train_size]\n",
    "Y = label_train[:train_size]\n",
    "\n",
    "test_size = 20\n",
    "sample_test = sample_test[:test_size]\n",
    "label_test = label_test[:test_size]\n",
    "\n",
    "# shift label from {0, 1} to {-1, 1}\n",
    "Y = Y * 2 - np.ones(len(Y))\n",
    "label_test = label_test * 2 - np.ones(len(label_test))\n",
    "\n",
    "for i in range(10):\n",
    "    print(\"X = {}, Y = {: d}\".format(X[i], int(Y[i])))\n",
    "\n",
    "print(\"...\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c043ec69",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Initialize quantum device\n",
    "dev = qml.device(\"default.qubit\", wires=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "5c96f669",
   "metadata": {},
   "outputs": [],
   "source": [
    "def statepreparation(x):\n",
    "    for i in range(0,4):\n",
    "        qml.RX(x[i],wires=i)\n",
    "\n",
    "def layer(W):\n",
    "    # parametrized RZ RY RZ rotations\n",
    "    for i in range(0,3):\n",
    "        for j in range(0,4):\n",
    "            if i % 2 == 0:\n",
    "                qml.RZ(W[j, i], wires=j)\n",
    "            else:\n",
    "                qml.RY(W[j, i], wires=j)\n",
    "    \n",
    "    # circular entanglement\n",
    "    for i in range(0,3):\n",
    "        qml.CNOT(wires=[i,i+1])\n",
    "    qml.CNOT(wires=[3,0])\n",
    "\n",
    "@qml.qnode(dev, interface=\"autograd\")\n",
    "def circuit(weights, x):\n",
    "\n",
    "    statepreparation(x)\n",
    "\n",
    "    for W in weights:\n",
    "        try:\n",
    "            layer(W)\n",
    "        except:\n",
    "            import pdb\n",
    "            pdb.set_trace()\n",
    "\n",
    "    return qml.expval(qml.PauliZ(0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "75bb8ef2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(<Figure size 1800x500 with 1 Axes>, <Axes: >)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1800x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Create initial weights from normal distribution\n",
    "np.random.seed(0)\n",
    "num_qubits = 4\n",
    "num_layers = 2\n",
    "weights_init = 0.01 * np.random.randn(num_layers, num_qubits, 3, requires_grad=True)\n",
    "\n",
    "#print initialized circuit\n",
    "qml.draw_mpl(circuit,decimals=2, expansion_strategy='device')(weights_init,X[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b20e3edb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cost and accuracy functions\n",
    "\n",
    "def cost(weights, data, labels):\n",
    "    predictions = [circuit(weights, x) for x in data]\n",
    "    loss = 0\n",
    "    for l, p in zip(labels, predictions):\n",
    "        loss = loss + (l - p) ** 2\n",
    "\n",
    "    loss = loss / len(labels)\n",
    "    return loss\n",
    "\n",
    "def accuracy(weights, data, labels):\n",
    "    predictions = [np.sign(circuit(weights, x)) for x in data]\n",
    "    acc = 0\n",
    "    for l, p in zip(labels, predictions):\n",
    "        if abs(l - p) < 1e-5:\n",
    "            acc = acc + 1\n",
    "    acc = acc / len(labels)\n",
    "    return acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "7218f48d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cost:  Autograd ArrayBox with value [[[ 0.01764052  0.00400157  0.00978738]\n",
      "  [ 0.02240893  0.01867558 -0.00977278]\n",
      "  [ 0.00950088 -0.00151357 -0.00103219]\n",
      "  [ 0.00410599  0.00144044  0.01454274]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00333674  0.01494079 -0.00205158]\n",
      "  [ 0.00313068 -0.00854096 -0.0255299 ]\n",
      "  [ 0.00653619  0.00864436 -0.00742165]]]\n",
      "cost:  [[[ 0.01574464 -0.00288523  0.00978796]\n",
      "  [ 0.02080179  0.02359029 -0.00976531]\n",
      "  [ 0.00946168 -0.00141343 -0.00107218]\n",
      "  [ 0.0040149   0.00178199  0.01453288]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00334283  0.0203357  -0.00205158]\n",
      "  [ 0.00312572 -0.06934335 -0.0255299 ]\n",
      "  [ 0.00658722  0.01186837 -0.00742165]]]\n",
      "loss:  [[[ 0.01574464 -0.00288523  0.00978796]\n",
      "  [ 0.02080179  0.02359029 -0.00976531]\n",
      "  [ 0.00946168 -0.00141343 -0.00107218]\n",
      "  [ 0.0040149   0.00178199  0.01453288]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00334283  0.0203357  -0.00205158]\n",
      "  [ 0.00312572 -0.06934335 -0.0255299 ]\n",
      "  [ 0.00658722  0.01186837 -0.00742165]]]\n",
      "Iter:     1 | Cost: 1.5050352 | Accuracy: 0.4800000 \n",
      "cost:  Autograd ArrayBox with value [[[ 0.01574464 -0.00288523  0.00978796]\n",
      "  [ 0.02080179  0.02359029 -0.00976531]\n",
      "  [ 0.00946168 -0.00141343 -0.00107218]\n",
      "  [ 0.0040149   0.00178199  0.01453288]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00334283  0.0203357  -0.00205158]\n",
      "  [ 0.00312572 -0.06934335 -0.0255299 ]\n",
      "  [ 0.00658722  0.01186837 -0.00742165]]]\n",
      "cost:  [[[ 0.01711305 -0.01136768  0.00979685]\n",
      "  [ 0.01877552  0.03039172 -0.00975677]\n",
      "  [ 0.00938348 -0.00131877 -0.00115625]\n",
      "  [ 0.00384249  0.0024108   0.01447594]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00335687  0.02709196 -0.00205158]\n",
      "  [ 0.00309009 -0.1505477  -0.0255299 ]\n",
      "  [ 0.0066618   0.0162658  -0.00742165]]]\n",
      "loss:  [[[ 0.01711305 -0.01136768  0.00979685]\n",
      "  [ 0.01877552  0.03039172 -0.00975677]\n",
      "  [ 0.00938348 -0.00131877 -0.00115625]\n",
      "  [ 0.00384249  0.0024108   0.01447594]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00335687  0.02709196 -0.00205158]\n",
      "  [ 0.00309009 -0.1505477  -0.0255299 ]\n",
      "  [ 0.0066618   0.0162658  -0.00742165]]]\n",
      "Iter:     2 | Cost: 1.4853048 | Accuracy: 0.4800000 \n",
      "cost:  Autograd ArrayBox with value [[[ 0.01711305 -0.01136768  0.00979685]\n",
      "  [ 0.01877552  0.03039172 -0.00975677]\n",
      "  [ 0.00938348 -0.00131877 -0.00115625]\n",
      "  [ 0.00384249  0.0024108   0.01447594]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00335687  0.02709196 -0.00205158]\n",
      "  [ 0.00309009 -0.1505477  -0.0255299 ]\n",
      "  [ 0.0066618   0.0162658  -0.00742165]]]\n",
      "cost:  [[[ 0.02242712 -0.02342136  0.00983181]\n",
      "  [ 0.01620099  0.03965037 -0.00974939]\n",
      "  [ 0.00927089 -0.00127982 -0.00128031]\n",
      "  [ 0.00356289  0.00347079  0.01437669]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00339257  0.03536474 -0.00205158]\n",
      "  [ 0.00302947 -0.25791533 -0.0255299 ]\n",
      "  [ 0.00677821  0.02217817 -0.00742165]]]\n",
      "loss:  [[[ 0.02242712 -0.02342136  0.00983181]\n",
      "  [ 0.01620099  0.03965037 -0.00974939]\n",
      "  [ 0.00927089 -0.00127982 -0.00128031]\n",
      "  [ 0.00356289  0.00347079  0.01437669]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00339257  0.03536474 -0.00205158]\n",
      "  [ 0.00302947 -0.25791533 -0.0255299 ]\n",
      "  [ 0.00677821  0.02217817 -0.00742165]]]\n",
      "Iter:     3 | Cost: 1.4510240 | Accuracy: 0.4800000 \n",
      "cost:  Autograd ArrayBox with value [[[ 0.02242712 -0.02342136  0.00983181]\n",
      "  [ 0.01620099  0.03965037 -0.00974939]\n",
      "  [ 0.00927089 -0.00127982 -0.00128031]\n",
      "  [ 0.00356289  0.00347079  0.01437669]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00339257  0.03536474 -0.00205158]\n",
      "  [ 0.00302947 -0.25791533 -0.0255299 ]\n",
      "  [ 0.00677821  0.02217817 -0.00742165]]]\n",
      "cost:  [[[ 0.03307207 -0.04161609  0.00993666]\n",
      "  [ 0.01297145  0.0518452  -0.00974853]\n",
      "  [ 0.00915074 -0.0014125  -0.00141838]\n",
      "  [ 0.00315326  0.00512427  0.01426428]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00348412  0.04511187 -0.00205158]\n",
      "  [ 0.00297077 -0.39693598 -0.0255299 ]\n",
      "  [ 0.0069749   0.0299276  -0.00742165]]]\n",
      "loss:  [[[ 0.03307207 -0.04161609  0.00993666]\n",
      "  [ 0.01297145  0.0518452  -0.00974853]\n",
      "  [ 0.00915074 -0.0014125  -0.00141838]\n",
      "  [ 0.00315326  0.00512427  0.01426428]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00348412  0.04511187 -0.00205158]\n",
      "  [ 0.00297077 -0.39693598 -0.0255299 ]\n",
      "  [ 0.0069749   0.0299276  -0.00742165]]]\n",
      "Iter:     4 | Cost: 1.3942550 | Accuracy: 0.4800000 \n",
      "cost:  Autograd ArrayBox with value [[[ 0.03307207 -0.04161609  0.00993666]\n",
      "  [ 0.01297145  0.0518452  -0.00974853]\n",
      "  [ 0.00915074 -0.0014125  -0.00141838]\n",
      "  [ 0.00315326  0.00512427  0.01426428]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00348412  0.04511187 -0.00205158]\n",
      "  [ 0.00297077 -0.39693598 -0.0255299 ]\n",
      "  [ 0.0069749   0.0299276  -0.00742165]]]\n",
      "cost:  [[[ 0.0509384  -0.0687461   0.01020629]\n",
      "  [ 0.00910675  0.06692206 -0.00976398]\n",
      "  [ 0.00909643 -0.00195949 -0.00150025]\n",
      "  [ 0.00260889  0.00747608  0.01420692]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00370735  0.05589308 -0.00205158]\n",
      "  [ 0.00298392 -0.56982425 -0.0255299 ]\n",
      "  [ 0.00732928  0.03966401 -0.00742165]]]\n",
      "loss:  [[[ 0.0509384  -0.0687461   0.01020629]\n",
      "  [ 0.00910675  0.06692206 -0.00976398]\n",
      "  [ 0.00909643 -0.00195949 -0.00150025]\n",
      "  [ 0.00260889  0.00747608  0.01420692]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00370735  0.05589308 -0.00205158]\n",
      "  [ 0.00298392 -0.56982425 -0.0255299 ]\n",
      "  [ 0.00732928  0.03966401 -0.00742165]]]\n",
      "Iter:     5 | Cost: 1.3085429 | Accuracy: 0.4800000 \n",
      "cost:  Autograd ArrayBox with value [[[ 0.0509384  -0.0687461   0.01020629]\n",
      "  [ 0.00910675  0.06692206 -0.00976398]\n",
      "  [ 0.00909643 -0.00195949 -0.00150025]\n",
      "  [ 0.00260889  0.00747608  0.01420692]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00370735  0.05589308 -0.00205158]\n",
      "  [ 0.00298392 -0.56982425 -0.0255299 ]\n",
      "  [ 0.00732928  0.03966401 -0.00742165]]]\n",
      "cost:  [[[ 0.07746089 -0.10619377  0.01081118]\n",
      "  [ 0.00492688  0.08356603 -0.00980679]\n",
      "  [ 0.00925698 -0.00336644 -0.00138645]\n",
      "  [ 0.00196603  0.01041218  0.01430408]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00420395  0.06676076 -0.00205158]\n",
      "  [ 0.00320653 -0.76997856 -0.0255299 ]\n",
      "  [ 0.00797761  0.05113986 -0.00742165]]]\n",
      "loss:  [[[ 0.07746089 -0.10619377  0.01081118]\n",
      "  [ 0.00492688  0.08356603 -0.00980679]\n",
      "  [ 0.00925698 -0.00336644 -0.00138645]\n",
      "  [ 0.00196603  0.01041218  0.01430408]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00420395  0.06676076 -0.00205158]\n",
      "  [ 0.00320653 -0.76997856 -0.0255299 ]\n",
      "  [ 0.00797761  0.05113986 -0.00742165]]]\n",
      "Iter:     6 | Cost: 1.1981673 | Accuracy: 0.4700000 \n",
      "cost:  Autograd ArrayBox with value [[[ 0.07746089 -0.10619377  0.01081118]\n",
      "  [ 0.00492688  0.08356603 -0.00980679]\n",
      "  [ 0.00925698 -0.00336644 -0.00138645]\n",
      "  [ 0.00196603  0.01041218  0.01430408]]\n",
      "\n",
      " [[ 0.00761038  0.00121675  0.00443863]\n",
      "  [ 0.00420395  0.06676076 -0.00205158]\n",
      "  [ 0.00320653 -0.76997856 -0.0255299 ]\n",
      "  [ 0.00797761  0.05113986 -0.00742165]]]\n",
      "cost:  [[[ 1.11611978e-01 -1.51057771e-01  1.19790238e-02]\n",
      "  [ 1.12810803e-03  9.88900263e-02 -9.87608252e-03]\n",
      "  [ 9.84887694e-03 -6.27197395e-03 -8.79402621e-04]\n",
      "  [ 1.31046504e-03  1.34575050e-02  1.46174399e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 5.17517509e-03  7.66431465e-02 -2.05158264e-03]\n",
      "  [ 3.84072114e-03 -9.76990134e-01 -2.55298982e-02]\n",
      "  [ 9.10393197e-03  6.36173072e-02 -7.42165020e-03]]]\n",
      "loss:  [[[ 1.11611978e-01 -1.51057771e-01  1.19790238e-02]\n",
      "  [ 1.12810803e-03  9.88900263e-02 -9.87608252e-03]\n",
      "  [ 9.84887694e-03 -6.27197395e-03 -8.79402621e-04]\n",
      "  [ 1.31046504e-03  1.34575050e-02  1.46174399e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 5.17517509e-03  7.66431465e-02 -2.05158264e-03]\n",
      "  [ 3.84072114e-03 -9.76990134e-01 -2.55298982e-02]\n",
      "  [ 9.10393197e-03  6.36173072e-02 -7.42165020e-03]]]\n",
      "Iter:     7 | Cost: 1.0861280 | Accuracy: 0.4600000 \n",
      "cost:  Autograd ArrayBox with value [[[ 1.11611978e-01 -1.51057771e-01  1.19790238e-02]\n",
      "  [ 1.12810803e-03  9.88900263e-02 -9.87608252e-03]\n",
      "  [ 9.84887694e-03 -6.27197395e-03 -8.79402621e-04]\n",
      "  [ 1.31046504e-03  1.34575050e-02  1.46174399e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 5.17517509e-03  7.66431465e-02 -2.05158264e-03]\n",
      "  [ 3.84072114e-03 -9.76990134e-01 -2.55298982e-02]\n",
      "  [ 9.10393197e-03  6.36173072e-02 -7.42165020e-03]]]\n",
      "cost:  [[[ 1.48555628e-01 -1.95464646e-01  1.38975790e-02]\n",
      "  [-1.57928510e-03  1.09938477e-01 -9.94515181e-03]\n",
      "  [ 1.10483916e-02 -1.12340095e-02  1.77567885e-04]\n",
      "  [ 7.32654046e-04  1.59722509e-02  1.50754386e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 6.80544107e-03  8.52859891e-02 -2.05158264e-03]\n",
      "  [ 5.08196450e-03 -1.16172915e+00 -2.55298982e-02]\n",
      "  [ 1.08643394e-02  7.62788360e-02 -7.42165020e-03]]]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss:  [[[ 1.48555628e-01 -1.95464646e-01  1.38975790e-02]\n",
      "  [-1.57928510e-03  1.09938477e-01 -9.94515181e-03]\n",
      "  [ 1.10483916e-02 -1.12340095e-02  1.77567885e-04]\n",
      "  [ 7.32654046e-04  1.59722509e-02  1.50754386e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 6.80544107e-03  8.52859891e-02 -2.05158264e-03]\n",
      "  [ 5.08196450e-03 -1.16172915e+00 -2.55298982e-02]\n",
      "  [ 1.08643394e-02  7.62788360e-02 -7.42165020e-03]]]\n",
      "Iter:     8 | Cost: 1.0010855 | Accuracy: 0.4700000 \n",
      "cost:  Autograd ArrayBox with value [[[ 1.48555628e-01 -1.95464646e-01  1.38975790e-02]\n",
      "  [-1.57928510e-03  1.09938477e-01 -9.94515181e-03]\n",
      "  [ 1.10483916e-02 -1.12340095e-02  1.77567885e-04]\n",
      "  [ 7.32654046e-04  1.59722509e-02  1.50754386e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 6.80544107e-03  8.52859891e-02 -2.05158264e-03]\n",
      "  [ 5.08196450e-03 -1.16172915e+00 -2.55298982e-02]\n",
      "  [ 1.08643394e-02  7.62788360e-02 -7.42165020e-03]]]\n",
      "cost:  [[[ 1.82108786e-01 -2.31723980e-01  1.66339308e-02]\n",
      "  [-3.06199803e-03  1.16106260e-01 -9.97444726e-03]\n",
      "  [ 1.28625837e-02 -1.83628713e-02  1.79038315e-03]\n",
      "  [ 2.59990225e-04  1.76595265e-02  1.55115080e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 9.18111134e-03  9.35982952e-02 -2.05158264e-03]\n",
      "  [ 7.03079305e-03 -1.30399021e+00 -2.55298982e-02]\n",
      "  [ 1.33160267e-02  8.89909327e-02 -7.42165020e-03]]]\n",
      "loss:  [[[ 1.82108786e-01 -2.31723980e-01  1.66339308e-02]\n",
      "  [-3.06199803e-03  1.16106260e-01 -9.97444726e-03]\n",
      "  [ 1.28625837e-02 -1.83628713e-02  1.79038315e-03]\n",
      "  [ 2.59990225e-04  1.76595265e-02  1.55115080e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 9.18111134e-03  9.35982952e-02 -2.05158264e-03]\n",
      "  [ 7.03079305e-03 -1.30399021e+00 -2.55298982e-02]\n",
      "  [ 1.33160267e-02  8.89909327e-02 -7.42165020e-03]]]\n",
      "Iter:     9 | Cost: 0.9517600 | Accuracy: 0.5100000 \n",
      "cost:  Autograd ArrayBox with value [[[ 1.82108786e-01 -2.31723980e-01  1.66339308e-02]\n",
      "  [-3.06199803e-03  1.16106260e-01 -9.97444726e-03]\n",
      "  [ 1.28625837e-02 -1.83628713e-02  1.79038315e-03]\n",
      "  [ 2.59990225e-04  1.76595265e-02  1.55115080e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 9.18111134e-03  9.35982952e-02 -2.05158264e-03]\n",
      "  [ 7.03079305e-03 -1.30399021e+00 -2.55298982e-02]\n",
      "  [ 1.33160267e-02  8.89909327e-02 -7.42165020e-03]]]\n",
      "cost:  [[[ 2.08969055e-01 -2.57406733e-01  2.02049150e-02]\n",
      "  [-3.71374174e-03  1.18930326e-01 -9.93757656e-03]\n",
      "  [ 1.51754963e-02 -2.74141442e-02  3.86016042e-03]\n",
      "  [-1.37055748e-04  1.86724600e-02  1.58169769e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.23353360e-02  1.02801207e-01 -2.05158264e-03]\n",
      "  [ 9.71869120e-03 -1.40247942e+00 -2.55298982e-02]\n",
      "  [ 1.64714731e-02  1.02462712e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 2.08969055e-01 -2.57406733e-01  2.02049150e-02]\n",
      "  [-3.71374174e-03  1.18930326e-01 -9.93757656e-03]\n",
      "  [ 1.51754963e-02 -2.74141442e-02  3.86016042e-03]\n",
      "  [-1.37055748e-04  1.86724600e-02  1.58169769e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.23353360e-02  1.02801207e-01 -2.05158264e-03]\n",
      "  [ 9.71869120e-03 -1.40247942e+00 -2.55298982e-02]\n",
      "  [ 1.64714731e-02  1.02462712e-01 -7.42165020e-03]]]\n",
      "Iter:    10 | Cost: 0.9276087 | Accuracy: 0.5900000 \n",
      "cost:  Autograd ArrayBox with value [[[ 2.08969055e-01 -2.57406733e-01  2.02049150e-02]\n",
      "  [-3.71374174e-03  1.18930326e-01 -9.93757656e-03]\n",
      "  [ 1.51754963e-02 -2.74141442e-02  3.86016042e-03]\n",
      "  [-1.37055748e-04  1.86724600e-02  1.58169769e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.23353360e-02  1.02801207e-01 -2.05158264e-03]\n",
      "  [ 9.71869120e-03 -1.40247942e+00 -2.55298982e-02]\n",
      "  [ 1.64714731e-02  1.02462712e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 2.29364821e-01 -2.74118034e-01  2.47005544e-02]\n",
      "  [-3.95616259e-03  1.20079498e-01 -9.82126314e-03]\n",
      "  [ 1.78889479e-02 -3.81790214e-02  6.30288138e-03]\n",
      "  [-4.92998474e-04  1.92814394e-02  1.59821827e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.63582346e-02  1.13790838e-01 -2.05158264e-03]\n",
      "  [ 1.32108596e-02 -1.46686427e+00 -2.55298982e-02]\n",
      "  [ 2.04036309e-02  1.17717980e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 2.29364821e-01 -2.74118034e-01  2.47005544e-02]\n",
      "  [-3.95616259e-03  1.20079498e-01 -9.82126314e-03]\n",
      "  [ 1.78889479e-02 -3.81790214e-02  6.30288138e-03]\n",
      "  [-4.92998474e-04  1.92814394e-02  1.59821827e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.63582346e-02  1.13790838e-01 -2.05158264e-03]\n",
      "  [ 1.32108596e-02 -1.46686427e+00 -2.55298982e-02]\n",
      "  [ 2.04036309e-02  1.17717980e-01 -7.42165020e-03]]]\n",
      "Iter:    11 | Cost: 0.9161361 | Accuracy: 0.5800000 \n",
      "cost:  Autograd ArrayBox with value [[[ 2.29364821e-01 -2.74118034e-01  2.47005544e-02]\n",
      "  [-3.95616259e-03  1.20079498e-01 -9.82126314e-03]\n",
      "  [ 1.78889479e-02 -3.81790214e-02  6.30288138e-03]\n",
      "  [-4.92998474e-04  1.92814394e-02  1.59821827e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.63582346e-02  1.13790838e-01 -2.05158264e-03]\n",
      "  [ 1.32108596e-02 -1.46686427e+00 -2.55298982e-02]\n",
      "  [ 2.04036309e-02  1.17717980e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 2.45139349e-01 -2.84385931e-01  3.03359497e-02]\n",
      "  [-4.03271271e-03  1.20528447e-01 -9.61196666e-03]\n",
      "  [ 2.09745177e-02 -5.06481136e-02  9.08946134e-03]\n",
      "  [-8.27073501e-04  1.96750987e-02  1.60400617e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.14521249e-02  1.27252528e-01 -2.05158264e-03]\n",
      "  [ 1.76643855e-02 -1.50778725e+00 -2.55298982e-02]\n",
      "  [ 2.52894788e-02  1.35760043e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 2.45139349e-01 -2.84385931e-01  3.03359497e-02]\n",
      "  [-4.03271271e-03  1.20528447e-01 -9.61196666e-03]\n",
      "  [ 2.09745177e-02 -5.06481136e-02  9.08946134e-03]\n",
      "  [-8.27073501e-04  1.96750987e-02  1.60400617e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.14521249e-02  1.27252528e-01 -2.05158264e-03]\n",
      "  [ 1.76643855e-02 -1.50778725e+00 -2.55298982e-02]\n",
      "  [ 2.52894788e-02  1.35760043e-01 -7.42165020e-03]]]\n",
      "Iter:    12 | Cost: 0.9098021 | Accuracy: 0.6100000 \n",
      "cost:  Autograd ArrayBox with value [[[ 2.45139349e-01 -2.84385931e-01  3.03359497e-02]\n",
      "  [-4.03271271e-03  1.20528447e-01 -9.61196666e-03]\n",
      "  [ 2.09745177e-02 -5.06481136e-02  9.08946134e-03]\n",
      "  [-8.27073501e-04  1.96750987e-02  1.60400617e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.14521249e-02  1.27252528e-01 -2.05158264e-03]\n",
      "  [ 1.76643855e-02 -1.50778725e+00 -2.55298982e-02]\n",
      "  [ 2.52894788e-02  1.35760043e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 2.58363173e-01 -2.90278165e-01  3.74661962e-02]\n",
      "  [-4.05295335e-03  1.20728513e-01 -9.28807640e-03]\n",
      "  [ 2.44581432e-02 -6.49788296e-02  1.22305489e-02]\n",
      "  [-1.14724841e-03  1.99542725e-02  1.60256804e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.79505304e-02  1.43945351e-01 -2.05158264e-03]\n",
      "  [ 2.33470378e-02 -1.53325065e+00 -2.55298982e-02]\n",
      "  [ 3.14190091e-02  1.57608173e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 2.58363173e-01 -2.90278165e-01  3.74661962e-02]\n",
      "  [-4.05295335e-03  1.20728513e-01 -9.28807640e-03]\n",
      "  [ 2.44581432e-02 -6.49788296e-02  1.22305489e-02]\n",
      "  [-1.14724841e-03  1.99542725e-02  1.60256804e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.79505304e-02  1.43945351e-01 -2.05158264e-03]\n",
      "  [ 2.33470378e-02 -1.53325065e+00 -2.55298982e-02]\n",
      "  [ 3.14190091e-02  1.57608173e-01 -7.42165020e-03]]]\n",
      "Iter:    13 | Cost: 0.9048249 | Accuracy: 0.6500000 \n",
      "cost:  Autograd ArrayBox with value [[[ 2.58363173e-01 -2.90278165e-01  3.74661962e-02]\n",
      "  [-4.05295335e-03  1.20728513e-01 -9.28807640e-03]\n",
      "  [ 2.44581432e-02 -6.49788296e-02  1.22305489e-02]\n",
      "  [-1.14724841e-03  1.99542725e-02  1.60256804e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.79505304e-02  1.43945351e-01 -2.05158264e-03]\n",
      "  [ 2.33470378e-02 -1.53325065e+00 -2.55298982e-02]\n",
      "  [ 3.14190091e-02  1.57608173e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 2.70910476e-01 -2.93205618e-01  4.66147818e-02]\n",
      "  [-4.06163737e-03  1.20877751e-01 -8.81673215e-03]\n",
      "  [ 2.83983801e-02 -8.14367460e-02  1.57566274e-02]\n",
      "  [-1.45620254e-03  2.01681782e-02  1.59649721e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 3.63470878e-02  1.64920745e-01 -2.05158264e-03]\n",
      "  [ 3.06585478e-02 -1.54854084e+00 -2.55298982e-02]\n",
      "  [ 3.92130007e-02  1.84464145e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 2.70910476e-01 -2.93205618e-01  4.66147818e-02]\n",
      "  [-4.06163737e-03  1.20877751e-01 -8.81673215e-03]\n",
      "  [ 2.83983801e-02 -8.14367460e-02  1.57566274e-02]\n",
      "  [-1.45620254e-03  2.01681782e-02  1.59649721e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 3.63470878e-02  1.64920745e-01 -2.05158264e-03]\n",
      "  [ 3.06585478e-02 -1.54854084e+00 -2.55298982e-02]\n",
      "  [ 3.92130007e-02  1.84464145e-01 -7.42165020e-03]]]\n",
      "Iter:    14 | Cost: 0.8991018 | Accuracy: 0.6600000 \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cost:  Autograd ArrayBox with value [[[ 2.70910476e-01 -2.93205618e-01  4.66147818e-02]\n",
      "  [-4.06163737e-03  1.20877751e-01 -8.81673215e-03]\n",
      "  [ 2.83983801e-02 -8.14367460e-02  1.57566274e-02]\n",
      "  [-1.45620254e-03  2.01681782e-02  1.59649721e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 3.63470878e-02  1.64920745e-01 -2.05158264e-03]\n",
      "  [ 3.06585478e-02 -1.54854084e+00 -2.55298982e-02]\n",
      "  [ 3.92130007e-02  1.84464145e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 2.84488128e-01 -2.94047206e-01  5.85291738e-02]\n",
      "  [-4.07731836e-03  1.21073820e-01 -8.15193364e-03]\n",
      "  [ 3.28719831e-02 -1.00353038e-01  1.97031076e-02]\n",
      "  [-1.75444686e-03  2.03397364e-02  1.58747418e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 4.73490833e-02  1.91722609e-01 -2.05158264e-03]\n",
      "  [ 4.01677885e-02 -1.55694118e+00 -2.55298982e-02]\n",
      "  [ 4.92585864e-02  2.17909365e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 2.84488128e-01 -2.94047206e-01  5.85291738e-02]\n",
      "  [-4.07731836e-03  1.21073820e-01 -8.15193364e-03]\n",
      "  [ 3.28719831e-02 -1.00353038e-01  1.97031076e-02]\n",
      "  [-1.75444686e-03  2.03397364e-02  1.58747418e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 4.73490833e-02  1.91722609e-01 -2.05158264e-03]\n",
      "  [ 4.01677885e-02 -1.55694118e+00 -2.55298982e-02]\n",
      "  [ 4.92585864e-02  2.17909365e-01 -7.42165020e-03]]]\n",
      "Iter:    15 | Cost: 0.8908749 | Accuracy: 0.7000000 \n",
      "cost:  Autograd ArrayBox with value [[[ 2.84488128e-01 -2.94047206e-01  5.85291738e-02]\n",
      "  [-4.07731836e-03  1.21073820e-01 -8.15193364e-03]\n",
      "  [ 3.28719831e-02 -1.00353038e-01  1.97031076e-02]\n",
      "  [-1.75444686e-03  2.03397364e-02  1.58747418e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 4.73490833e-02  1.91722609e-01 -2.05158264e-03]\n",
      "  [ 4.01677885e-02 -1.55694118e+00 -2.55298982e-02]\n",
      "  [ 4.92585864e-02  2.17909365e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 3.00799628e-01 -2.93282839e-01  7.42598068e-02]\n",
      "  [-4.10954218e-03  1.21385888e-01 -7.23373380e-03]\n",
      "  [ 3.79608747e-02 -1.22079753e-01  2.40948422e-02]\n",
      "  [-2.04154338e-03  2.04786824e-02  1.57648696e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 6.19526223e-02  2.26630193e-01 -2.05158264e-03]\n",
      "  [ 5.26593025e-02 -1.56036183e+00 -2.55298982e-02]\n",
      "  [ 6.23528786e-02  2.60136392e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 3.00799628e-01 -2.93282839e-01  7.42598068e-02]\n",
      "  [-4.10954218e-03  1.21385888e-01 -7.23373380e-03]\n",
      "  [ 3.79608747e-02 -1.22079753e-01  2.40948422e-02]\n",
      "  [-2.04154338e-03  2.04786824e-02  1.57648696e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 6.19526223e-02  2.26630193e-01 -2.05158264e-03]\n",
      "  [ 5.26593025e-02 -1.56036183e+00 -2.55298982e-02]\n",
      "  [ 6.23528786e-02  2.60136392e-01 -7.42165020e-03]]]\n",
      "Iter:    16 | Cost: 0.8778051 | Accuracy: 0.7600000 \n",
      "cost:  Autograd ArrayBox with value [[[ 3.00799628e-01 -2.93282839e-01  7.42598068e-02]\n",
      "  [-4.10954218e-03  1.21385888e-01 -7.23373380e-03]\n",
      "  [ 3.79608747e-02 -1.22079753e-01  2.40948422e-02]\n",
      "  [-2.04154338e-03  2.04786824e-02  1.57648696e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 6.19526223e-02  2.26630193e-01 -2.05158264e-03]\n",
      "  [ 5.26593025e-02 -1.56036183e+00 -2.55298982e-02]\n",
      "  [ 6.23528786e-02  2.60136392e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 3.21703602e-01 -2.91059985e-01  9.52318664e-02]\n",
      "  [-4.16577270e-03  1.21890234e-01 -5.99141865e-03]\n",
      "  [ 4.37292913e-02 -1.46903880e-01  2.89213519e-02]\n",
      "  [-2.31644141e-03  2.05883240e-02  1.56402696e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 8.15091206e-02  2.72950545e-01 -2.05158264e-03]\n",
      "  [ 6.91528196e-02 -1.55976244e+00 -2.55298982e-02]\n",
      "  [ 7.95151619e-02  3.14210133e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 3.21703602e-01 -2.91059985e-01  9.52318664e-02]\n",
      "  [-4.16577270e-03  1.21890234e-01 -5.99141865e-03]\n",
      "  [ 4.37292913e-02 -1.46903880e-01  2.89213519e-02]\n",
      "  [-2.31644141e-03  2.05883240e-02  1.56402696e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 8.15091206e-02  2.72950545e-01 -2.05158264e-03]\n",
      "  [ 6.91528196e-02 -1.55976244e+00 -2.55298982e-02]\n",
      "  [ 7.95151619e-02  3.14210133e-01 -7.42165020e-03]]]\n",
      "Iter:    17 | Cost: 0.8560205 | Accuracy: 0.7800000 \n",
      "cost:  Autograd ArrayBox with value [[[ 3.21703602e-01 -2.91059985e-01  9.52318664e-02]\n",
      "  [-4.16577270e-03  1.21890234e-01 -5.99141865e-03]\n",
      "  [ 4.37292913e-02 -1.46903880e-01  2.89213519e-02]\n",
      "  [-2.31644141e-03  2.05883240e-02  1.56402696e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 8.15091206e-02  2.72950545e-01 -2.05158264e-03]\n",
      "  [ 6.91528196e-02 -1.55976244e+00 -2.55298982e-02]\n",
      "  [ 7.95151619e-02  3.14210133e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 3.49214312e-01 -2.87169118e-01  1.23188736e-01]\n",
      "  [-4.25289003e-03  1.22690118e-01 -4.35859748e-03]\n",
      "  [ 5.01729917e-02 -1.74852998e-01  3.40913942e-02]\n",
      "  [-2.57740721e-03  2.06704909e-02  1.55024425e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.07662112e-01  3.35217773e-01 -2.05158264e-03]\n",
      "  [ 9.07710997e-02 -1.55543437e+00 -2.55298982e-02]\n",
      "  [ 1.01838245e-01  3.84224575e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 3.49214312e-01 -2.87169118e-01  1.23188736e-01]\n",
      "  [-4.25289003e-03  1.22690118e-01 -4.35859748e-03]\n",
      "  [ 5.01729917e-02 -1.74852998e-01  3.40913942e-02]\n",
      "  [-2.57740721e-03  2.06704909e-02  1.55024425e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.07662112e-01  3.35217773e-01 -2.05158264e-03]\n",
      "  [ 9.07710997e-02 -1.55543437e+00 -2.55298982e-02]\n",
      "  [ 1.01838245e-01  3.84224575e-01 -7.42165020e-03]]]\n",
      "Iter:    18 | Cost: 0.8190731 | Accuracy: 0.8300000 \n",
      "cost:  Autograd ArrayBox with value [[[ 3.49214312e-01 -2.87169118e-01  1.23188736e-01]\n",
      "  [-4.25289003e-03  1.22690118e-01 -4.35859748e-03]\n",
      "  [ 5.01729917e-02 -1.74852998e-01  3.40913942e-02]\n",
      "  [-2.57740721e-03  2.06704909e-02  1.55024425e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.07662112e-01  3.35217773e-01 -2.05158264e-03]\n",
      "  [ 9.07710997e-02 -1.55543437e+00 -2.55298982e-02]\n",
      "  [ 1.01838245e-01  3.84224575e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 3.84966642e-01 -2.80902090e-01  1.59643831e-01]\n",
      "  [-4.37299310e-03  1.23917047e-01 -2.31627733e-03]\n",
      "  [ 5.71154011e-02 -2.05295833e-01  3.93572642e-02]\n",
      "  [-2.82167131e-03  2.07323853e-02  1.53514860e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.41789814e-01  4.18718325e-01 -2.05158264e-03]\n",
      "  [ 1.18113204e-01 -1.54729894e+00 -2.55298982e-02]\n",
      "  [ 1.29831567e-01  4.74813480e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 3.84966642e-01 -2.80902090e-01  1.59643831e-01]\n",
      "  [-4.37299310e-03  1.23917047e-01 -2.31627733e-03]\n",
      "  [ 5.71154011e-02 -2.05295833e-01  3.93572642e-02]\n",
      "  [-2.82167131e-03  2.07323853e-02  1.53514860e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.41789814e-01  4.18718325e-01 -2.05158264e-03]\n",
      "  [ 1.18113204e-01 -1.54729894e+00 -2.55298982e-02]\n",
      "  [ 1.29831567e-01  4.74813480e-01 -7.42165020e-03]]]\n",
      "Iter:    19 | Cost: 0.7581002 | Accuracy: 0.8200000 \n",
      "cost:  Autograd ArrayBox with value [[[ 3.84966642e-01 -2.80902090e-01  1.59643831e-01]\n",
      "  [-4.37299310e-03  1.23917047e-01 -2.31627733e-03]\n",
      "  [ 5.71154011e-02 -2.05295833e-01  3.93572642e-02]\n",
      "  [-2.82167131e-03  2.07323853e-02  1.53514860e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.41789814e-01  4.18718325e-01 -2.05158264e-03]\n",
      "  [ 1.18113204e-01 -1.54729894e+00 -2.55298982e-02]\n",
      "  [ 1.29831567e-01  4.74813480e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 4.28426420e-01 -2.70827187e-01  2.04139059e-01]\n",
      "  [-4.50996886e-03  1.25676024e-01  2.06250814e-05]\n",
      "  [ 6.40568946e-02 -2.36347076e-01  4.42344603e-02]\n",
      "  [-3.04452769e-03  2.07972230e-02  1.51906596e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.83242676e-01  5.26899089e-01 -2.05158264e-03]\n",
      "  [ 1.49538304e-01 -1.53551903e+00 -2.55298982e-02]\n",
      "  [ 1.61673527e-01  5.88668391e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 4.28426420e-01 -2.70827187e-01  2.04139059e-01]\n",
      "  [-4.50996886e-03  1.25676024e-01  2.06250814e-05]\n",
      "  [ 6.40568946e-02 -2.36347076e-01  4.42344603e-02]\n",
      "  [-3.04452769e-03  2.07972230e-02  1.51906596e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.83242676e-01  5.26899089e-01 -2.05158264e-03]\n",
      "  [ 1.49538304e-01 -1.53551903e+00 -2.55298982e-02]\n",
      "  [ 1.61673527e-01  5.88668391e-01 -7.42165020e-03]]]\n",
      "Iter:    20 | Cost: 0.6676409 | Accuracy: 0.8900000 \n",
      "cost:  Autograd ArrayBox with value [[[ 4.28426420e-01 -2.70827187e-01  2.04139059e-01]\n",
      "  [-4.50996886e-03  1.25676024e-01  2.06250814e-05]\n",
      "  [ 6.40568946e-02 -2.36347076e-01  4.42344603e-02]\n",
      "  [-3.04452769e-03  2.07972230e-02  1.51906596e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 1.83242676e-01  5.26899089e-01 -2.05158264e-03]\n",
      "  [ 1.49538304e-01 -1.53551903e+00 -2.55298982e-02]\n",
      "  [ 1.61673527e-01  5.88668391e-01 -7.42165020e-03]]]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cost:  [[[ 4.73975012e-01 -2.54856137e-01  2.51418646e-01]\n",
      "  [-4.60905479e-03  1.27866018e-01  2.32358013e-03]\n",
      "  [ 7.01604172e-02 -2.64716757e-01  4.80691511e-02]\n",
      "  [-3.23714911e-03  2.09088151e-02  1.50352035e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.26465353e-01  6.55500355e-01 -2.05158264e-03]\n",
      "  [ 1.78724620e-01 -1.52164713e+00 -2.55298982e-02]\n",
      "  [ 1.90805734e-01  7.21006102e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 4.73975012e-01 -2.54856137e-01  2.51418646e-01]\n",
      "  [-4.60905479e-03  1.27866018e-01  2.32358013e-03]\n",
      "  [ 7.01604172e-02 -2.64716757e-01  4.80691511e-02]\n",
      "  [-3.23714911e-03  2.09088151e-02  1.50352035e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.26465353e-01  6.55500355e-01 -2.05158264e-03]\n",
      "  [ 1.78724620e-01 -1.52164713e+00 -2.55298982e-02]\n",
      "  [ 1.90805734e-01  7.21006102e-01 -7.42165020e-03]]]\n",
      "Iter:    21 | Cost: 0.5589677 | Accuracy: 0.9100000 \n",
      "cost:  Autograd ArrayBox with value [[[ 4.73975012e-01 -2.54856137e-01  2.51418646e-01]\n",
      "  [-4.60905479e-03  1.27866018e-01  2.32358013e-03]\n",
      "  [ 7.01604172e-02 -2.64716757e-01  4.80691511e-02]\n",
      "  [-3.23714911e-03  2.09088151e-02  1.50352035e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.26465353e-01  6.55500355e-01 -2.05158264e-03]\n",
      "  [ 1.78724620e-01 -1.52164713e+00 -2.55298982e-02]\n",
      "  [ 1.90805734e-01  7.21006102e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 5.11704504e-01 -2.31336226e-01  2.92129362e-01]\n",
      "  [-4.57544462e-03  1.29942202e-01  4.18140818e-03]\n",
      "  [ 7.47496792e-02 -2.87390466e-01  5.04843205e-02]\n",
      "  [-3.38514333e-03  2.11039307e-02  1.49158718e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.61660195e-01  7.88948166e-01 -2.05158264e-03]\n",
      "  [ 1.96191280e-01 -1.50919668e+00 -2.55298982e-02]\n",
      "  [ 2.07620242e-01  8.56307390e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 5.11704504e-01 -2.31336226e-01  2.92129362e-01]\n",
      "  [-4.57544462e-03  1.29942202e-01  4.18140818e-03]\n",
      "  [ 7.47496792e-02 -2.87390466e-01  5.04843205e-02]\n",
      "  [-3.38514333e-03  2.11039307e-02  1.49158718e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.61660195e-01  7.88948166e-01 -2.05158264e-03]\n",
      "  [ 1.96191280e-01 -1.50919668e+00 -2.55298982e-02]\n",
      "  [ 2.07620242e-01  8.56307390e-01 -7.42165020e-03]]]\n",
      "Iter:    22 | Cost: 0.4592098 | Accuracy: 0.9500000 \n",
      "cost:  Autograd ArrayBox with value [[[ 5.11704504e-01 -2.31336226e-01  2.92129362e-01]\n",
      "  [-4.57544462e-03  1.29942202e-01  4.18140818e-03]\n",
      "  [ 7.47496792e-02 -2.87390466e-01  5.04843205e-02]\n",
      "  [-3.38514333e-03  2.11039307e-02  1.49158718e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.61660195e-01  7.88948166e-01 -2.05158264e-03]\n",
      "  [ 1.96191280e-01 -1.50919668e+00 -2.55298982e-02]\n",
      "  [ 2.07620242e-01  8.56307390e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 5.36567583e-01 -2.00302428e-01  3.21419676e-01]\n",
      "  [-4.31862885e-03  1.31079256e-01  5.45703865e-03]\n",
      "  [ 7.79335191e-02 -3.04011578e-01  5.18145930e-02]\n",
      "  [-3.47819995e-03  2.13779034e-02  1.48557700e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.83575166e-01  9.09930231e-01 -2.05158264e-03]\n",
      "  [ 1.98263217e-01 -1.50144863e+00 -2.55298982e-02]\n",
      "  [ 2.08507491e-01  9.77704894e-01 -7.42165020e-03]]]\n",
      "loss:  [[[ 5.36567583e-01 -2.00302428e-01  3.21419676e-01]\n",
      "  [-4.31862885e-03  1.31079256e-01  5.45703865e-03]\n",
      "  [ 7.79335191e-02 -3.04011578e-01  5.18145930e-02]\n",
      "  [-3.47819995e-03  2.13779034e-02  1.48557700e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.83575166e-01  9.09930231e-01 -2.05158264e-03]\n",
      "  [ 1.98263217e-01 -1.50144863e+00 -2.55298982e-02]\n",
      "  [ 2.08507491e-01  9.77704894e-01 -7.42165020e-03]]]\n",
      "Iter:    23 | Cost: 0.3841939 | Accuracy: 0.9600000 \n",
      "cost:  Autograd ArrayBox with value [[[ 5.36567583e-01 -2.00302428e-01  3.21419676e-01]\n",
      "  [-4.31862885e-03  1.31079256e-01  5.45703865e-03]\n",
      "  [ 7.79335191e-02 -3.04011578e-01  5.18145930e-02]\n",
      "  [-3.47819995e-03  2.13779034e-02  1.48557700e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.83575166e-01  9.09930231e-01 -2.05158264e-03]\n",
      "  [ 1.98263217e-01 -1.50144863e+00 -2.55298982e-02]\n",
      "  [ 2.08507491e-01  9.77704894e-01 -7.42165020e-03]]]\n",
      "cost:  [[[ 5.52457048e-01 -1.63109833e-01  3.42577265e-01]\n",
      "  [-3.81502200e-03  1.30837856e-01  6.30928085e-03]\n",
      "  [ 8.02511554e-02 -3.16120240e-01  5.26796148e-02]\n",
      "  [-3.52529342e-03  2.17056930e-02  1.48448753e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.95490187e-01  1.01128262e+00 -2.05158264e-03]\n",
      "  [ 1.89949085e-01 -1.49911192e+00 -2.55298982e-02]\n",
      "  [ 1.98620632e-01  1.07840871e+00 -7.42165020e-03]]]\n",
      "loss:  [[[ 5.52457048e-01 -1.63109833e-01  3.42577265e-01]\n",
      "  [-3.81502200e-03  1.30837856e-01  6.30928085e-03]\n",
      "  [ 8.02511554e-02 -3.16120240e-01  5.26796148e-02]\n",
      "  [-3.52529342e-03  2.17056930e-02  1.48448753e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.95490187e-01  1.01128262e+00 -2.05158264e-03]\n",
      "  [ 1.89949085e-01 -1.49911192e+00 -2.55298982e-02]\n",
      "  [ 1.98620632e-01  1.07840871e+00 -7.42165020e-03]]]\n",
      "Iter:    24 | Cost: 0.3315164 | Accuracy: 0.9700000 \n",
      "cost:  Autograd ArrayBox with value [[[ 5.52457048e-01 -1.63109833e-01  3.42577265e-01]\n",
      "  [-3.81502200e-03  1.30837856e-01  6.30928085e-03]\n",
      "  [ 8.02511554e-02 -3.16120240e-01  5.26796148e-02]\n",
      "  [-3.52529342e-03  2.17056930e-02  1.48448753e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 2.95490187e-01  1.01128262e+00 -2.05158264e-03]\n",
      "  [ 1.89949085e-01 -1.49911192e+00 -2.55298982e-02]\n",
      "  [ 1.98620632e-01  1.07840871e+00 -7.42165020e-03]]]\n",
      "cost:  [[[ 5.64709351e-01 -1.22085565e-01  3.59826718e-01]\n",
      "  [-3.11749334e-03  1.29376156e-01  6.90644104e-03]\n",
      "  [ 8.20788177e-02 -3.25194329e-01  5.34109843e-02]\n",
      "  [-3.54885541e-03  2.20562367e-02  1.48535964e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 3.02025543e-01  1.09415384e+00 -2.05158264e-03]\n",
      "  [ 1.76991359e-01 -1.50119404e+00 -2.55298982e-02]\n",
      "  [ 1.83898576e-01  1.15981349e+00 -7.42165020e-03]]]\n",
      "loss:  [[[ 5.64709351e-01 -1.22085565e-01  3.59826718e-01]\n",
      "  [-3.11749334e-03  1.29376156e-01  6.90644104e-03]\n",
      "  [ 8.20788177e-02 -3.25194329e-01  5.34109843e-02]\n",
      "  [-3.54885541e-03  2.20562367e-02  1.48535964e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 3.02025543e-01  1.09415384e+00 -2.05158264e-03]\n",
      "  [ 1.76991359e-01 -1.50119404e+00 -2.55298982e-02]\n",
      "  [ 1.83898576e-01  1.15981349e+00 -7.42165020e-03]]]\n",
      "Iter:    25 | Cost: 0.2942513 | Accuracy: 0.9800000 \n"
     ]
    }
   ],
   "source": [
    "# Train model\n",
    "\n",
    "weights = weights_init\n",
    "opt = qml.GradientDescentOptimizer(stepsize=0.4)\n",
    "\n",
    "for it in range(25):\n",
    "\n",
    "    weights, _, _ = opt.step(cost, weights, X, Y)\n",
    "\n",
    "    print(\n",
    "        \"Iter: {:5d} | Cost: {:0.7f} | Accuracy: {:0.7f} \".format(\n",
    "            it + 1, cost(weights, X, Y), accuracy(weights, X, Y)\n",
    "        )\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "9d67fc8e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cost:  [[[ 5.64709351e-01 -1.22085565e-01  3.59826718e-01]\n",
      "  [-3.11749334e-03  1.29376156e-01  6.90644104e-03]\n",
      "  [ 8.20788177e-02 -3.25194329e-01  5.34109843e-02]\n",
      "  [-3.54885541e-03  2.20562367e-02  1.48535964e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 3.02025543e-01  1.09415384e+00 -2.05158264e-03]\n",
      "  [ 1.76991359e-01 -1.50119404e+00 -2.55298982e-02]\n",
      "  [ 1.83898576e-01  1.15981349e+00 -7.42165020e-03]]]\n",
      "loss:  [[[ 5.64709351e-01 -1.22085565e-01  3.59826718e-01]\n",
      "  [-3.11749334e-03  1.29376156e-01  6.90644104e-03]\n",
      "  [ 8.20788177e-02 -3.25194329e-01  5.34109843e-02]\n",
      "  [-3.54885541e-03  2.20562367e-02  1.48535964e-02]]\n",
      "\n",
      " [[ 7.61037725e-03  1.21675016e-03  4.43863233e-03]\n",
      "  [ 3.02025543e-01  1.09415384e+00 -2.05158264e-03]\n",
      "  [ 1.76991359e-01 -1.50119404e+00 -2.55298982e-02]\n",
      "  [ 1.83898576e-01  1.15981349e+00 -7.42165020e-03]]]\n",
      "Evaluate test set:  Cost: 0.4331384 | Accuracy: 0.8500000 \n"
     ]
    }
   ],
   "source": [
    "# Evaluate model on test set; accuracy should be at least > 0.75\n",
    "\n",
    "print(\"Evaluate test set:  Cost: {:0.7f} | Accuracy: {:0.7f} \".format(\n",
    "            cost(weights, sample_test, label_test), accuracy(weights, sample_test, label_test)\n",
    "        ))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f7821a8e",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}