Leaf-Classification.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "from skimage import data, io, filters, exposure\n",
    "from skimage.segmentation import slic, felzenszwalb\n",
    "from skimage.transform import resize\n",
    "from skimage.morphology import square, closing, binary_closing\n",
    "from matplotlib import pyplot as plt\n",
    "from skimage import color\n",
    "import numpy as np\n",
    "from scipy import ndimage as ndi\n",
    "from skimage.measure import moments_hu, moments, moments_central, moments_normalized"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_feature(name):\n",
    "    sample = io.imread(str(name))\n",
    "    sample = resize(sample, (148,224), mode='constant')\n",
    "\n",
    "    sample_hsv = color.convert_colorspace(sample,'RGB','HSV')\n",
    "    \n",
    "    # change green color to white, other color to black\n",
    "    for i in range(sample_hsv.shape[0]):\n",
    "        for j in range(sample_hsv.shape[1]):\n",
    "            if sample_hsv[i,j,0] > 0.2083 and sample_hsv[i,j,0] < 0.4167 and sample_hsv[i,j,2]>0.1  and sample_hsv[i,j,1]>0.1:\n",
    "                sample_hsv[i,j,1] = 0\n",
    "                sample_hsv[i,j,2] = 1\n",
    "            else:\n",
    "                sample_hsv[i,j,2] = 0\n",
    "\n",
    "    sample_rgb = color.convert_colorspace(sample_hsv,'HSV','RGB')\n",
    "\n",
    "    sample_grey = color.rgb2grey(sample_rgb)\n",
    "\n",
    "    # find center of the image\n",
    "    t = 0\n",
    "    res_i = 0\n",
    "    res_j = 0\n",
    "    for i in range(sample_grey.shape[0]-100):\n",
    "        for j in range(sample_grey.shape[1]-100):\n",
    "            if np.sum(sample_grey[i:i+100,j:j+100]) > t:\n",
    "                res_i = i\n",
    "                res_j = j\n",
    "                t = np.sum(sample_grey[i:i+100,j:j+100])\n",
    "\n",
    "    sample_cropped = sample_grey[res_i:res_i+100,res_j:res_j+100]\n",
    "\n",
    "    closed = binary_closing(sample_cropped, square(2))\n",
    "\n",
    "    closed = np.asarray(closed, dtype=np.double)\n",
    "\n",
    "    M = moments(closed)\n",
    "    cr = M[1, 0] / M[0, 0]\n",
    "    cc = M[0, 1] / M[0, 0]\n",
    "\n",
    "    mu = moments_central(closed, center=(cr,cc))\n",
    "    nu = moments_normalized(mu)\n",
    "    return moments_hu(nu)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature = np.empty((0,8), int)\n",
    "\n",
    "from os import listdir\n",
    "from os.path import isfile, join\n",
    "mypath = 'train_1'\n",
    "onlyfiles = [f for f in listdir(mypath) if isfile(join(mypath, f))]\n",
    "\n",
    "for o in onlyfiles:\n",
    "    feature = np.vstack((feature, np.hstack((get_feature(join(mypath,o)),0))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "mypath = 'train_2'\n",
    "onlyfiles = [f for f in listdir(mypath) if isfile(join(mypath, f))]\n",
    "\n",
    "for o in onlyfiles:\n",
    "    feature = np.vstack((feature,np.hstack((get_feature(join(mypath,o)),1))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(21, 8)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_test = np.empty((0,8), int)\n",
    "mypath = 'test'\n",
    "onlyfiles = [f for f in listdir(mypath) if isfile(join(mypath, f))]\n",
    "\n",
    "x = 0\n",
    "for o in onlyfiles:\n",
    "    x = int(o[8:10])\n",
    "    if x<67:\n",
    "        u=0\n",
    "    else:\n",
    "        u=1\n",
    "    feature_test = np.vstack((feature_test,np.hstack((get_feature(join(mypath,o)),u))))\n",
    "    x += 1\n",
    "feature_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/rian/.local/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:351: UserWarning: Got `batch_size` less than 1 or larger than sample size. It is going to be clipped\n",
      "  warnings.warn(\"Got `batch_size` less than 1 or larger than \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration 1, loss = 0.70239386\n",
      "Iteration 2, loss = 0.70062115\n",
      "Iteration 3, loss = 0.69908110\n",
      "Iteration 4, loss = 0.69774520\n",
      "Iteration 5, loss = 0.69658813\n",
      "Iteration 6, loss = 0.69558743\n",
      "Iteration 7, loss = 0.69472321\n",
      "Iteration 8, loss = 0.69397789\n",
      "Iteration 9, loss = 0.69333597\n",
      "Iteration 10, loss = 0.69278380\n",
      "Iteration 11, loss = 0.69230942\n",
      "Iteration 12, loss = 0.69190233\n",
      "Iteration 13, loss = 0.69155336\n",
      "Iteration 14, loss = 0.69125451\n",
      "Iteration 15, loss = 0.69099882\n",
      "Iteration 16, loss = 0.69078025\n",
      "Iteration 17, loss = 0.69059356\n",
      "Iteration 18, loss = 0.69043420\n",
      "Iteration 19, loss = 0.69029826\n",
      "Iteration 20, loss = 0.69018237\n",
      "Iteration 21, loss = 0.69008361\n",
      "Iteration 22, loss = 0.68999948\n",
      "Iteration 23, loss = 0.68992783\n",
      "Iteration 24, loss = 0.68986682\n",
      "Iteration 25, loss = 0.68981488\n",
      "Iteration 26, loss = 0.68977066\n",
      "Iteration 27, loss = 0.68973300\n",
      "Iteration 28, loss = 0.68970091\n",
      "Iteration 29, loss = 0.68967356\n",
      "Iteration 30, loss = 0.68965024\n",
      "Iteration 31, loss = 0.68963033\n",
      "Iteration 32, loss = 0.68961332\n",
      "Iteration 33, loss = 0.68959876\n",
      "Iteration 34, loss = 0.68958627\n",
      "Iteration 35, loss = 0.68957556\n",
      "Iteration 36, loss = 0.68956633\n",
      "Iteration 37, loss = 0.68955836\n",
      "Iteration 38, loss = 0.68955146\n",
      "Iteration 39, loss = 0.68954547\n",
      "Iteration 40, loss = 0.68954024\n",
      "Iteration 41, loss = 0.68953566\n",
      "Iteration 42, loss = 0.68953162\n",
      "Iteration 43, loss = 0.68952804\n",
      "Iteration 44, loss = 0.68952485\n",
      "Iteration 45, loss = 0.68952199\n",
      "Iteration 46, loss = 0.68951941\n",
      "Iteration 47, loss = 0.68951706\n",
      "Iteration 48, loss = 0.68951491\n",
      "Iteration 49, loss = 0.68951293\n",
      "Iteration 50, loss = 0.68951109\n",
      "Iteration 51, loss = 0.68950937\n",
      "Iteration 52, loss = 0.68950774\n",
      "Iteration 53, loss = 0.68950620\n",
      "Iteration 54, loss = 0.68950473\n",
      "Iteration 55, loss = 0.68950332\n",
      "Iteration 56, loss = 0.68950196\n",
      "Iteration 57, loss = 0.68950065\n",
      "Iteration 58, loss = 0.68949936\n",
      "Iteration 59, loss = 0.68949811\n",
      "Iteration 60, loss = 0.68949689\n",
      "Iteration 61, loss = 0.68949568\n",
      "Iteration 62, loss = 0.68949449\n",
      "Iteration 63, loss = 0.68949332\n",
      "Iteration 64, loss = 0.68949215\n",
      "Iteration 65, loss = 0.68949100\n",
      "Iteration 66, loss = 0.68948986\n",
      "Iteration 67, loss = 0.68948872\n",
      "Iteration 68, loss = 0.68948759\n",
      "Iteration 69, loss = 0.68948647\n",
      "Iteration 70, loss = 0.68948534\n",
      "Iteration 71, loss = 0.68948422\n",
      "Iteration 72, loss = 0.68948311\n",
      "Iteration 73, loss = 0.68948199\n",
      "Iteration 74, loss = 0.68948088\n",
      "Iteration 75, loss = 0.68947977\n",
      "Iteration 76, loss = 0.68947865\n",
      "Iteration 77, loss = 0.68947754\n",
      "Iteration 78, loss = 0.68947643\n",
      "Iteration 79, loss = 0.68947532\n",
      "Iteration 80, loss = 0.68947421\n",
      "Iteration 81, loss = 0.68947310\n",
      "Iteration 82, loss = 0.68947199\n",
      "Iteration 83, loss = 0.68947088\n",
      "Iteration 84, loss = 0.68946977\n",
      "Iteration 85, loss = 0.68946865\n",
      "Iteration 86, loss = 0.68946754\n",
      "Iteration 87, loss = 0.68946643\n",
      "Iteration 88, loss = 0.68946531\n",
      "Iteration 89, loss = 0.68946420\n",
      "Iteration 90, loss = 0.68946308\n",
      "Iteration 91, loss = 0.68946196\n",
      "Iteration 92, loss = 0.68946085\n",
      "Iteration 93, loss = 0.68945973\n",
      "Iteration 94, loss = 0.68945861\n",
      "Iteration 95, loss = 0.68945749\n",
      "Iteration 96, loss = 0.68945637\n",
      "Iteration 97, loss = 0.68945525\n",
      "Iteration 98, loss = 0.68945412\n",
      "Iteration 99, loss = 0.68945300\n",
      "Iteration 100, loss = 0.68945188\n",
      "Iteration 101, loss = 0.68945075\n",
      "Iteration 102, loss = 0.68944962\n",
      "Iteration 103, loss = 0.68944850\n",
      "Iteration 104, loss = 0.68944737\n",
      "Iteration 105, loss = 0.68944624\n",
      "Iteration 106, loss = 0.68944511\n",
      "Iteration 107, loss = 0.68944398\n",
      "Iteration 108, loss = 0.68944285\n",
      "Iteration 109, loss = 0.68944172\n",
      "Iteration 110, loss = 0.68944058\n",
      "Iteration 111, loss = 0.68943945\n",
      "Iteration 112, loss = 0.68943832\n",
      "Iteration 113, loss = 0.68943718\n",
      "Iteration 114, loss = 0.68943604\n",
      "Iteration 115, loss = 0.68943490\n",
      "Iteration 116, loss = 0.68943377\n",
      "Iteration 117, loss = 0.68943263\n",
      "Iteration 118, loss = 0.68943149\n",
      "Iteration 119, loss = 0.68943034\n",
      "Iteration 120, loss = 0.68942920\n",
      "Iteration 121, loss = 0.68942806\n",
      "Iteration 122, loss = 0.68942691\n",
      "Iteration 123, loss = 0.68942577\n",
      "Iteration 124, loss = 0.68942462\n",
      "Iteration 125, loss = 0.68942348\n",
      "Iteration 126, loss = 0.68942233\n",
      "Iteration 127, loss = 0.68942118\n",
      "Iteration 128, loss = 0.68942003\n",
      "Iteration 129, loss = 0.68941888\n",
      "Iteration 130, loss = 0.68941773\n",
      "Iteration 131, loss = 0.68941658\n",
      "Iteration 132, loss = 0.68941542\n",
      "Iteration 133, loss = 0.68941427\n",
      "Iteration 134, loss = 0.68941311\n",
      "Iteration 135, loss = 0.68941196\n",
      "Iteration 136, loss = 0.68941080\n",
      "Iteration 137, loss = 0.68940964\n",
      "Iteration 138, loss = 0.68940848\n",
      "Iteration 139, loss = 0.68940732\n",
      "Iteration 140, loss = 0.68940616\n",
      "Iteration 141, loss = 0.68940500\n",
      "Iteration 142, loss = 0.68940384\n",
      "Iteration 143, loss = 0.68940267\n",
      "Iteration 144, loss = 0.68940151\n",
      "Iteration 145, loss = 0.68940034\n",
      "Iteration 146, loss = 0.68939917\n",
      "Iteration 147, loss = 0.68939801\n",
      "Iteration 148, loss = 0.68939684\n",
      "Iteration 149, loss = 0.68939567\n",
      "Iteration 150, loss = 0.68939450\n",
      "Iteration 151, loss = 0.68939333\n",
      "Iteration 152, loss = 0.68939215\n",
      "Iteration 153, loss = 0.68939098\n",
      "Iteration 154, loss = 0.68938981\n",
      "Iteration 155, loss = 0.68938863\n",
      "Iteration 156, loss = 0.68938745\n",
      "Iteration 157, loss = 0.68938628\n",
      "Iteration 158, loss = 0.68938510\n",
      "Iteration 159, loss = 0.68938392\n",
      "Iteration 160, loss = 0.68938274\n",
      "Iteration 161, loss = 0.68938156\n",
      "Iteration 162, loss = 0.68938038\n",
      "Iteration 163, loss = 0.68937919\n",
      "Iteration 164, loss = 0.68937801\n",
      "Iteration 165, loss = 0.68937683\n",
      "Iteration 166, loss = 0.68937564\n",
      "Iteration 167, loss = 0.68937445\n",
      "Iteration 168, loss = 0.68937326\n",
      "Iteration 169, loss = 0.68937208\n",
      "Iteration 170, loss = 0.68937089\n",
      "Iteration 171, loss = 0.68936970\n",
      "Iteration 172, loss = 0.68936850\n",
      "Iteration 173, loss = 0.68936731\n",
      "Iteration 174, loss = 0.68936612\n",
      "Iteration 175, loss = 0.68936492\n",
      "Iteration 176, loss = 0.68936373\n",
      "Iteration 177, loss = 0.68936253\n",
      "Iteration 178, loss = 0.68936133\n",
      "Iteration 179, loss = 0.68936013\n",
      "Iteration 180, loss = 0.68935893\n",
      "Iteration 181, loss = 0.68935773\n",
      "Iteration 182, loss = 0.68935653\n",
      "Iteration 183, loss = 0.68935533\n",
      "Iteration 184, loss = 0.68935413\n",
      "Iteration 185, loss = 0.68935292\n",
      "Iteration 186, loss = 0.68935172\n",
      "Iteration 187, loss = 0.68935051\n",
      "Iteration 188, loss = 0.68934930\n",
      "Iteration 189, loss = 0.68934809\n",
      "Iteration 190, loss = 0.68934688\n",
      "Iteration 191, loss = 0.68934567\n",
      "Iteration 192, loss = 0.68934446\n",
      "Iteration 193, loss = 0.68934325\n",
      "Iteration 194, loss = 0.68934204\n",
      "Iteration 195, loss = 0.68934082\n",
      "Iteration 196, loss = 0.68933961\n",
      "Iteration 197, loss = 0.68933839\n",
      "Iteration 198, loss = 0.68933717\n",
      "Iteration 199, loss = 0.68933595\n",
      "Iteration 200, loss = 0.68933473\n",
      "Iteration 201, loss = 0.68933351\n",
      "Iteration 202, loss = 0.68933229\n",
      "Iteration 203, loss = 0.68933107\n",
      "Iteration 204, loss = 0.68932985\n",
      "Iteration 205, loss = 0.68932862\n",
      "Iteration 206, loss = 0.68932739\n",
      "Iteration 207, loss = 0.68932617\n",
      "Iteration 208, loss = 0.68932494\n",
      "Iteration 209, loss = 0.68932371\n",
      "Iteration 210, loss = 0.68932248\n",
      "Iteration 211, loss = 0.68932125\n",
      "Iteration 212, loss = 0.68932002\n",
      "Iteration 213, loss = 0.68931879\n",
      "Iteration 214, loss = 0.68931755\n",
      "Iteration 215, loss = 0.68931632\n",
      "Iteration 216, loss = 0.68931508\n",
      "Iteration 217, loss = 0.68931385\n",
      "Iteration 218, loss = 0.68931261\n",
      "Iteration 219, loss = 0.68931137\n",
      "Iteration 220, loss = 0.68931013\n",
      "Iteration 221, loss = 0.68930889\n",
      "Iteration 222, loss = 0.68930765\n",
      "Iteration 223, loss = 0.68930640\n",
      "Iteration 224, loss = 0.68930516\n",
      "Iteration 225, loss = 0.68930391\n",
      "Iteration 226, loss = 0.68930267\n",
      "Iteration 227, loss = 0.68930142\n",
      "Iteration 228, loss = 0.68930017\n",
      "Iteration 229, loss = 0.68929892\n",
      "Iteration 230, loss = 0.68929767\n",
      "Iteration 231, loss = 0.68929642\n",
      "Iteration 232, loss = 0.68929517\n",
      "Iteration 233, loss = 0.68929391\n",
      "Iteration 234, loss = 0.68929266\n",
      "Iteration 235, loss = 0.68929140\n",
      "Iteration 236, loss = 0.68929015\n",
      "Iteration 237, loss = 0.68928889\n",
      "Iteration 238, loss = 0.68928763\n",
      "Iteration 239, loss = 0.68928637\n",
      "Iteration 240, loss = 0.68928511\n",
      "Iteration 241, loss = 0.68928385\n",
      "Iteration 242, loss = 0.68928258\n",
      "Iteration 243, loss = 0.68928132\n",
      "Iteration 244, loss = 0.68928006\n",
      "Iteration 245, loss = 0.68927879\n",
      "Iteration 246, loss = 0.68927752\n",
      "Iteration 247, loss = 0.68927625\n",
      "Iteration 248, loss = 0.68927498\n",
      "Iteration 249, loss = 0.68927371\n",
      "Iteration 250, loss = 0.68927244\n",
      "Iteration 251, loss = 0.68927117\n",
      "Iteration 252, loss = 0.68926990\n",
      "Iteration 253, loss = 0.68926862\n",
      "Iteration 254, loss = 0.68926735\n",
      "Iteration 255, loss = 0.68926607\n",
      "Iteration 256, loss = 0.68926479\n",
      "Iteration 257, loss = 0.68926351\n",
      "Iteration 258, loss = 0.68926223\n",
      "Iteration 259, loss = 0.68926095\n",
      "Iteration 260, loss = 0.68925967\n",
      "Iteration 261, loss = 0.68925838\n",
      "Iteration 262, loss = 0.68925710\n",
      "Iteration 263, loss = 0.68925581\n",
      "Iteration 264, loss = 0.68925453\n",
      "Iteration 265, loss = 0.68925324\n",
      "Iteration 266, loss = 0.68925195\n",
      "Iteration 267, loss = 0.68925066\n",
      "Iteration 268, loss = 0.68924937\n",
      "Iteration 269, loss = 0.68924808\n",
      "Iteration 270, loss = 0.68924679\n",
      "Iteration 271, loss = 0.68924549\n",
      "Iteration 272, loss = 0.68924420\n",
      "Iteration 273, loss = 0.68924290\n",
      "Iteration 274, loss = 0.68924160\n",
      "Iteration 275, loss = 0.68924030\n",
      "Iteration 276, loss = 0.68923900\n",
      "Iteration 277, loss = 0.68923770\n",
      "Iteration 278, loss = 0.68923640\n",
      "Iteration 279, loss = 0.68923510\n",
      "Iteration 280, loss = 0.68923379\n",
      "Iteration 281, loss = 0.68923249\n",
      "Iteration 282, loss = 0.68923118\n",
      "Iteration 283, loss = 0.68922987\n",
      "Iteration 284, loss = 0.68922857\n",
      "Iteration 285, loss = 0.68922726\n",
      "Iteration 286, loss = 0.68922595\n",
      "Iteration 287, loss = 0.68922463\n",
      "Iteration 288, loss = 0.68922332\n",
      "Iteration 289, loss = 0.68922201\n",
      "Iteration 290, loss = 0.68922069\n",
      "Iteration 291, loss = 0.68921937\n",
      "Iteration 292, loss = 0.68921806\n",
      "Iteration 293, loss = 0.68921674\n",
      "Iteration 294, loss = 0.68921542\n",
      "Iteration 295, loss = 0.68921410\n",
      "Iteration 296, loss = 0.68921278\n",
      "Iteration 297, loss = 0.68921145\n",
      "Iteration 298, loss = 0.68921013\n",
      "Iteration 299, loss = 0.68920880\n",
      "Iteration 300, loss = 0.68920748\n",
      "Iteration 301, loss = 0.68920615\n",
      "Iteration 302, loss = 0.68920482\n",
      "Iteration 303, loss = 0.68920349\n",
      "Iteration 304, loss = 0.68920216\n",
      "Iteration 305, loss = 0.68920083\n",
      "Iteration 306, loss = 0.68919949\n",
      "Iteration 307, loss = 0.68919816\n",
      "Iteration 308, loss = 0.68919682\n",
      "Iteration 309, loss = 0.68919549\n",
      "Iteration 310, loss = 0.68919415\n",
      "Iteration 311, loss = 0.68919281\n",
      "Iteration 312, loss = 0.68919147\n",
      "Iteration 313, loss = 0.68919013\n",
      "Iteration 314, loss = 0.68918878\n",
      "Iteration 315, loss = 0.68918744\n",
      "Iteration 316, loss = 0.68918610\n",
      "Iteration 317, loss = 0.68918475\n",
      "Iteration 318, loss = 0.68918340\n",
      "Iteration 319, loss = 0.68918205\n",
      "Iteration 320, loss = 0.68918070\n",
      "Iteration 321, loss = 0.68917935\n",
      "Iteration 322, loss = 0.68917800\n",
      "Iteration 323, loss = 0.68917665\n",
      "Iteration 324, loss = 0.68917530\n",
      "Iteration 325, loss = 0.68917394\n",
      "Iteration 326, loss = 0.68917258\n",
      "Iteration 327, loss = 0.68917123\n",
      "Iteration 328, loss = 0.68916987\n",
      "Iteration 329, loss = 0.68916851\n",
      "Iteration 330, loss = 0.68916715\n",
      "Iteration 331, loss = 0.68916578\n",
      "Iteration 332, loss = 0.68916442\n",
      "Iteration 333, loss = 0.68916306\n",
      "Iteration 334, loss = 0.68916169\n",
      "Iteration 335, loss = 0.68916032\n",
      "Iteration 336, loss = 0.68915895\n",
      "Iteration 337, loss = 0.68915758\n",
      "Iteration 338, loss = 0.68915621\n",
      "Iteration 339, loss = 0.68915484\n",
      "Iteration 340, loss = 0.68915347\n",
      "Iteration 341, loss = 0.68915210\n",
      "Iteration 342, loss = 0.68915072\n",
      "Iteration 343, loss = 0.68914934\n",
      "Iteration 344, loss = 0.68914797\n",
      "Iteration 345, loss = 0.68914659\n",
      "Iteration 346, loss = 0.68914521\n",
      "Iteration 347, loss = 0.68914383\n",
      "Iteration 348, loss = 0.68914244\n",
      "Iteration 349, loss = 0.68914106\n",
      "Iteration 350, loss = 0.68913967\n",
      "Iteration 351, loss = 0.68913829\n",
      "Iteration 352, loss = 0.68913690\n",
      "Iteration 353, loss = 0.68913551\n",
      "Iteration 354, loss = 0.68913412\n",
      "Iteration 355, loss = 0.68913273\n",
      "Iteration 356, loss = 0.68913134\n",
      "Iteration 357, loss = 0.68912995\n",
      "Iteration 358, loss = 0.68912855\n",
      "Iteration 359, loss = 0.68912716\n",
      "Iteration 360, loss = 0.68912576\n",
      "Iteration 361, loss = 0.68912436\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration 362, loss = 0.68912296\n",
      "Iteration 363, loss = 0.68912156\n",
      "Iteration 364, loss = 0.68912016\n",
      "Iteration 365, loss = 0.68911876\n",
      "Iteration 366, loss = 0.68911735\n",
      "Iteration 367, loss = 0.68911595\n",
      "Iteration 368, loss = 0.68911454\n",
      "Iteration 369, loss = 0.68911313\n",
      "Iteration 370, loss = 0.68911172\n",
      "Iteration 371, loss = 0.68911031\n",
      "Iteration 372, loss = 0.68910890\n",
      "Iteration 373, loss = 0.68910749\n",
      "Iteration 374, loss = 0.68910607\n",
      "Iteration 375, loss = 0.68910466\n",
      "Iteration 376, loss = 0.68910324\n",
      "Iteration 377, loss = 0.68910182\n",
      "Iteration 378, loss = 0.68910040\n",
      "Iteration 379, loss = 0.68909898\n",
      "Iteration 380, loss = 0.68909756\n",
      "Iteration 381, loss = 0.68909614\n",
      "Iteration 382, loss = 0.68909472\n",
      "Iteration 383, loss = 0.68909329\n",
      "Iteration 384, loss = 0.68909186\n",
      "Iteration 385, loss = 0.68909044\n",
      "Iteration 386, loss = 0.68908901\n",
      "Iteration 387, loss = 0.68908758\n",
      "Iteration 388, loss = 0.68908615\n",
      "Iteration 389, loss = 0.68908471\n",
      "Iteration 390, loss = 0.68908328\n",
      "Iteration 391, loss = 0.68908184\n",
      "Iteration 392, loss = 0.68908041\n",
      "Iteration 393, loss = 0.68907897\n",
      "Iteration 394, loss = 0.68907753\n",
      "Iteration 395, loss = 0.68907609\n",
      "Iteration 396, loss = 0.68907465\n",
      "Iteration 397, loss = 0.68907320\n",
      "Iteration 398, loss = 0.68907176\n",
      "Iteration 399, loss = 0.68907031\n",
      "Iteration 400, loss = 0.68906887\n",
      "Iteration 401, loss = 0.68906742\n",
      "Iteration 402, loss = 0.68906597\n",
      "Iteration 403, loss = 0.68906452\n",
      "Iteration 404, loss = 0.68906307\n",
      "Iteration 405, loss = 0.68906162\n",
      "Iteration 406, loss = 0.68906016\n",
      "Iteration 407, loss = 0.68905871\n",
      "Iteration 408, loss = 0.68905725\n",
      "Iteration 409, loss = 0.68905579\n",
      "Iteration 410, loss = 0.68905433\n",
      "Iteration 411, loss = 0.68905287\n",
      "Iteration 412, loss = 0.68905141\n",
      "Iteration 413, loss = 0.68904994\n",
      "Iteration 414, loss = 0.68904848\n",
      "Iteration 415, loss = 0.68904701\n",
      "Iteration 416, loss = 0.68904555\n",
      "Iteration 417, loss = 0.68904408\n",
      "Iteration 418, loss = 0.68904261\n",
      "Iteration 419, loss = 0.68904114\n",
      "Iteration 420, loss = 0.68903966\n",
      "Iteration 421, loss = 0.68903819\n",
      "Iteration 422, loss = 0.68903671\n",
      "Iteration 423, loss = 0.68903524\n",
      "Iteration 424, loss = 0.68903376\n",
      "Iteration 425, loss = 0.68903228\n",
      "Iteration 426, loss = 0.68903080\n",
      "Iteration 427, loss = 0.68902932\n",
      "Iteration 428, loss = 0.68902784\n",
      "Iteration 429, loss = 0.68902635\n",
      "Iteration 430, loss = 0.68902487\n",
      "Iteration 431, loss = 0.68902338\n",
      "Iteration 432, loss = 0.68902189\n",
      "Iteration 433, loss = 0.68902040\n",
      "Iteration 434, loss = 0.68901891\n",
      "Iteration 435, loss = 0.68901742\n",
      "Iteration 436, loss = 0.68901593\n",
      "Iteration 437, loss = 0.68901443\n",
      "Iteration 438, loss = 0.68901293\n",
      "Iteration 439, loss = 0.68901144\n",
      "Iteration 440, loss = 0.68900994\n",
      "Iteration 441, loss = 0.68900844\n",
      "Iteration 442, loss = 0.68900694\n",
      "Iteration 443, loss = 0.68900543\n",
      "Iteration 444, loss = 0.68900393\n",
      "Iteration 445, loss = 0.68900242\n",
      "Iteration 446, loss = 0.68900092\n",
      "Iteration 447, loss = 0.68899941\n",
      "Iteration 448, loss = 0.68899790\n",
      "Iteration 449, loss = 0.68899639\n",
      "Iteration 450, loss = 0.68899487\n",
      "Iteration 451, loss = 0.68899336\n",
      "Iteration 452, loss = 0.68899185\n",
      "Iteration 453, loss = 0.68899033\n",
      "Iteration 454, loss = 0.68898881\n",
      "Iteration 455, loss = 0.68898729\n",
      "Iteration 456, loss = 0.68898577\n",
      "Iteration 457, loss = 0.68898425\n",
      "Iteration 458, loss = 0.68898273\n",
      "Iteration 459, loss = 0.68898120\n",
      "Iteration 460, loss = 0.68897968\n",
      "Iteration 461, loss = 0.68897815\n",
      "Iteration 462, loss = 0.68897662\n",
      "Iteration 463, loss = 0.68897509\n",
      "Iteration 464, loss = 0.68897356\n",
      "Iteration 465, loss = 0.68897203\n",
      "Iteration 466, loss = 0.68897049\n",
      "Iteration 467, loss = 0.68896896\n",
      "Iteration 468, loss = 0.68896742\n",
      "Iteration 469, loss = 0.68896588\n",
      "Iteration 470, loss = 0.68896434\n",
      "Iteration 471, loss = 0.68896280\n",
      "Iteration 472, loss = 0.68896126\n",
      "Iteration 473, loss = 0.68895971\n",
      "Iteration 474, loss = 0.68895817\n",
      "Iteration 475, loss = 0.68895662\n",
      "Iteration 476, loss = 0.68895507\n",
      "Iteration 477, loss = 0.68895353\n",
      "Iteration 478, loss = 0.68895197\n",
      "Iteration 479, loss = 0.68895042\n",
      "Iteration 480, loss = 0.68894887\n",
      "Iteration 481, loss = 0.68894731\n",
      "Iteration 482, loss = 0.68894576\n",
      "Iteration 483, loss = 0.68894420\n",
      "Iteration 484, loss = 0.68894264\n",
      "Iteration 485, loss = 0.68894108\n",
      "Iteration 486, loss = 0.68893952\n",
      "Iteration 487, loss = 0.68893795\n",
      "Iteration 488, loss = 0.68893639\n",
      "Iteration 489, loss = 0.68893482\n",
      "Iteration 490, loss = 0.68893326\n",
      "Iteration 491, loss = 0.68893169\n",
      "Iteration 492, loss = 0.68893012\n",
      "Iteration 493, loss = 0.68892854\n",
      "Iteration 494, loss = 0.68892697\n",
      "Iteration 495, loss = 0.68892540\n",
      "Iteration 496, loss = 0.68892382\n",
      "Iteration 497, loss = 0.68892224\n",
      "Iteration 498, loss = 0.68892066\n",
      "Iteration 499, loss = 0.68891908\n",
      "Iteration 500, loss = 0.68891750\n",
      "Iteration 501, loss = 0.68891592\n",
      "Iteration 502, loss = 0.68891433\n",
      "Iteration 503, loss = 0.68891275\n",
      "Iteration 504, loss = 0.68891116\n",
      "Iteration 505, loss = 0.68890957\n",
      "Iteration 506, loss = 0.68890798\n",
      "Iteration 507, loss = 0.68890639\n",
      "Iteration 508, loss = 0.68890479\n",
      "Iteration 509, loss = 0.68890320\n",
      "Iteration 510, loss = 0.68890160\n",
      "Iteration 511, loss = 0.68890000\n",
      "Iteration 512, loss = 0.68889840\n",
      "Iteration 513, loss = 0.68889680\n",
      "Iteration 514, loss = 0.68889520\n",
      "Iteration 515, loss = 0.68889360\n",
      "Iteration 516, loss = 0.68889199\n",
      "Iteration 517, loss = 0.68889039\n",
      "Iteration 518, loss = 0.68888878\n",
      "Iteration 519, loss = 0.68888717\n",
      "Iteration 520, loss = 0.68888556\n",
      "Iteration 521, loss = 0.68888394\n",
      "Iteration 522, loss = 0.68888233\n",
      "Iteration 523, loss = 0.68888072\n",
      "Iteration 524, loss = 0.68887910\n",
      "Iteration 525, loss = 0.68887748\n",
      "Iteration 526, loss = 0.68887586\n",
      "Iteration 527, loss = 0.68887424\n",
      "Iteration 528, loss = 0.68887262\n",
      "Iteration 529, loss = 0.68887099\n",
      "Iteration 530, loss = 0.68886937\n",
      "Iteration 531, loss = 0.68886774\n",
      "Iteration 532, loss = 0.68886611\n",
      "Iteration 533, loss = 0.68886448\n",
      "Iteration 534, loss = 0.68886285\n",
      "Iteration 535, loss = 0.68886121\n",
      "Iteration 536, loss = 0.68885958\n",
      "Iteration 537, loss = 0.68885794\n",
      "Iteration 538, loss = 0.68885631\n",
      "Iteration 539, loss = 0.68885467\n",
      "Iteration 540, loss = 0.68885303\n",
      "Iteration 541, loss = 0.68885138\n",
      "Iteration 542, loss = 0.68884974\n",
      "Iteration 543, loss = 0.68884809\n",
      "Iteration 544, loss = 0.68884645\n",
      "Iteration 545, loss = 0.68884480\n",
      "Iteration 546, loss = 0.68884315\n",
      "Iteration 547, loss = 0.68884150\n",
      "Iteration 548, loss = 0.68883984\n",
      "Iteration 549, loss = 0.68883819\n",
      "Iteration 550, loss = 0.68883653\n",
      "Iteration 551, loss = 0.68883488\n",
      "Iteration 552, loss = 0.68883322\n",
      "Iteration 553, loss = 0.68883156\n",
      "Iteration 554, loss = 0.68882990\n",
      "Iteration 555, loss = 0.68882823\n",
      "Iteration 556, loss = 0.68882657\n",
      "Iteration 557, loss = 0.68882490\n",
      "Iteration 558, loss = 0.68882323\n",
      "Iteration 559, loss = 0.68882156\n",
      "Iteration 560, loss = 0.68881989\n",
      "Iteration 561, loss = 0.68881822\n",
      "Iteration 562, loss = 0.68881654\n",
      "Iteration 563, loss = 0.68881487\n",
      "Iteration 564, loss = 0.68881319\n",
      "Iteration 565, loss = 0.68881151\n",
      "Iteration 566, loss = 0.68880983\n",
      "Iteration 567, loss = 0.68880815\n",
      "Iteration 568, loss = 0.68880647\n",
      "Iteration 569, loss = 0.68880478\n",
      "Iteration 570, loss = 0.68880309\n",
      "Iteration 571, loss = 0.68880141\n",
      "Iteration 572, loss = 0.68879972\n",
      "Iteration 573, loss = 0.68879802\n",
      "Iteration 574, loss = 0.68879633\n",
      "Iteration 575, loss = 0.68879464\n",
      "Iteration 576, loss = 0.68879294\n",
      "Iteration 577, loss = 0.68879124\n",
      "Iteration 578, loss = 0.68878954\n",
      "Iteration 579, loss = 0.68878784\n",
      "Iteration 580, loss = 0.68878614\n",
      "Iteration 581, loss = 0.68878444\n",
      "Iteration 582, loss = 0.68878273\n",
      "Iteration 583, loss = 0.68878102\n",
      "Iteration 584, loss = 0.68877931\n",
      "Iteration 585, loss = 0.68877760\n",
      "Iteration 586, loss = 0.68877589\n",
      "Iteration 587, loss = 0.68877418\n",
      "Iteration 588, loss = 0.68877246\n",
      "Iteration 589, loss = 0.68877075\n",
      "Iteration 590, loss = 0.68876903\n",
      "Iteration 591, loss = 0.68876731\n",
      "Iteration 592, loss = 0.68876559\n",
      "Iteration 593, loss = 0.68876386\n",
      "Iteration 594, loss = 0.68876214\n",
      "Iteration 595, loss = 0.68876041\n",
      "Iteration 596, loss = 0.68875868\n",
      "Iteration 597, loss = 0.68875695\n",
      "Iteration 598, loss = 0.68875522\n",
      "Iteration 599, loss = 0.68875349\n",
      "Iteration 600, loss = 0.68875176\n",
      "Iteration 601, loss = 0.68875002\n",
      "Iteration 602, loss = 0.68874828\n",
      "Iteration 603, loss = 0.68874654\n",
      "Iteration 604, loss = 0.68874480\n",
      "Iteration 605, loss = 0.68874306\n",
      "Iteration 606, loss = 0.68874131\n",
      "Iteration 607, loss = 0.68873957\n",
      "Iteration 608, loss = 0.68873782\n",
      "Iteration 609, loss = 0.68873607\n",
      "Iteration 610, loss = 0.68873432\n",
      "Iteration 611, loss = 0.68873257\n",
      "Iteration 612, loss = 0.68873081\n",
      "Iteration 613, loss = 0.68872906\n",
      "Iteration 614, loss = 0.68872730\n",
      "Iteration 615, loss = 0.68872554\n",
      "Iteration 616, loss = 0.68872378\n",
      "Iteration 617, loss = 0.68872202\n",
      "Iteration 618, loss = 0.68872026\n",
      "Iteration 619, loss = 0.68871849\n",
      "Iteration 620, loss = 0.68871672\n",
      "Iteration 621, loss = 0.68871495\n",
      "Iteration 622, loss = 0.68871318\n",
      "Iteration 623, loss = 0.68871141\n",
      "Iteration 624, loss = 0.68870964\n",
      "Iteration 625, loss = 0.68870786\n",
      "Iteration 626, loss = 0.68870608\n",
      "Iteration 627, loss = 0.68870430\n",
      "Iteration 628, loss = 0.68870252\n",
      "Iteration 629, loss = 0.68870074\n",
      "Iteration 630, loss = 0.68869896\n",
      "Iteration 631, loss = 0.68869717\n",
      "Iteration 632, loss = 0.68869538\n",
      "Iteration 633, loss = 0.68869360\n",
      "Iteration 634, loss = 0.68869180\n",
      "Iteration 635, loss = 0.68869001\n",
      "Iteration 636, loss = 0.68868822\n",
      "Iteration 637, loss = 0.68868642\n",
      "Iteration 638, loss = 0.68868463\n",
      "Iteration 639, loss = 0.68868283\n",
      "Iteration 640, loss = 0.68868102\n",
      "Iteration 641, loss = 0.68867922\n",
      "Iteration 642, loss = 0.68867742\n",
      "Iteration 643, loss = 0.68867561\n",
      "Iteration 644, loss = 0.68867380\n",
      "Iteration 645, loss = 0.68867199\n",
      "Iteration 646, loss = 0.68867018\n",
      "Iteration 647, loss = 0.68866837\n",
      "Iteration 648, loss = 0.68866656\n",
      "Iteration 649, loss = 0.68866474\n",
      "Iteration 650, loss = 0.68866292\n",
      "Iteration 651, loss = 0.68866110\n",
      "Iteration 652, loss = 0.68865928\n",
      "Iteration 653, loss = 0.68865746\n",
      "Iteration 654, loss = 0.68865563\n",
      "Iteration 655, loss = 0.68865381\n",
      "Iteration 656, loss = 0.68865198\n",
      "Iteration 657, loss = 0.68865015\n",
      "Iteration 658, loss = 0.68864832\n",
      "Iteration 659, loss = 0.68864648\n",
      "Iteration 660, loss = 0.68864465\n",
      "Iteration 661, loss = 0.68864281\n",
      "Iteration 662, loss = 0.68864097\n",
      "Iteration 663, loss = 0.68863913\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration 664, loss = 0.68863729\n",
      "Iteration 665, loss = 0.68863544\n",
      "Iteration 666, loss = 0.68863360\n",
      "Iteration 667, loss = 0.68863175\n",
      "Iteration 668, loss = 0.68862990\n",
      "Iteration 669, loss = 0.68862805\n",
      "Iteration 670, loss = 0.68862620\n",
      "Iteration 671, loss = 0.68862434\n",
      "Iteration 672, loss = 0.68862248\n",
      "Iteration 673, loss = 0.68862063\n",
      "Iteration 674, loss = 0.68861877\n",
      "Iteration 675, loss = 0.68861690\n",
      "Iteration 676, loss = 0.68861504\n",
      "Iteration 677, loss = 0.68861318\n",
      "Iteration 678, loss = 0.68861131\n",
      "Iteration 679, loss = 0.68860944\n",
      "Iteration 680, loss = 0.68860757\n",
      "Iteration 681, loss = 0.68860570\n",
      "Iteration 682, loss = 0.68860382\n",
      "Iteration 683, loss = 0.68860195\n",
      "Iteration 684, loss = 0.68860007\n",
      "Iteration 685, loss = 0.68859819\n",
      "Iteration 686, loss = 0.68859631\n",
      "Iteration 687, loss = 0.68859442\n",
      "Iteration 688, loss = 0.68859254\n",
      "Iteration 689, loss = 0.68859065\n",
      "Iteration 690, loss = 0.68858876\n",
      "Iteration 691, loss = 0.68858687\n",
      "Iteration 692, loss = 0.68858498\n",
      "Iteration 693, loss = 0.68858309\n",
      "Iteration 694, loss = 0.68858119\n",
      "Iteration 695, loss = 0.68857929\n",
      "Iteration 696, loss = 0.68857739\n",
      "Iteration 697, loss = 0.68857549\n",
      "Iteration 698, loss = 0.68857359\n",
      "Iteration 699, loss = 0.68857168\n",
      "Iteration 700, loss = 0.68856978\n",
      "Iteration 701, loss = 0.68856787\n",
      "Iteration 702, loss = 0.68856596\n",
      "Iteration 703, loss = 0.68856404\n",
      "Iteration 704, loss = 0.68856213\n",
      "Iteration 705, loss = 0.68856021\n",
      "Iteration 706, loss = 0.68855830\n",
      "Iteration 707, loss = 0.68855638\n",
      "Iteration 708, loss = 0.68855445\n",
      "Iteration 709, loss = 0.68855253\n",
      "Iteration 710, loss = 0.68855061\n",
      "Iteration 711, loss = 0.68854868\n",
      "Iteration 712, loss = 0.68854675\n",
      "Iteration 713, loss = 0.68854482\n",
      "Iteration 714, loss = 0.68854289\n",
      "Iteration 715, loss = 0.68854095\n",
      "Iteration 716, loss = 0.68853901\n",
      "Iteration 717, loss = 0.68853708\n",
      "Iteration 718, loss = 0.68853514\n",
      "Iteration 719, loss = 0.68853319\n",
      "Iteration 720, loss = 0.68853125\n",
      "Iteration 721, loss = 0.68852930\n",
      "Iteration 722, loss = 0.68852736\n",
      "Iteration 723, loss = 0.68852541\n",
      "Iteration 724, loss = 0.68852345\n",
      "Iteration 725, loss = 0.68852150\n",
      "Iteration 726, loss = 0.68851955\n",
      "Iteration 727, loss = 0.68851759\n",
      "Iteration 728, loss = 0.68851563\n",
      "Iteration 729, loss = 0.68851367\n",
      "Iteration 730, loss = 0.68851170\n",
      "Iteration 731, loss = 0.68850974\n",
      "Iteration 732, loss = 0.68850777\n",
      "Iteration 733, loss = 0.68850580\n",
      "Iteration 734, loss = 0.68850383\n",
      "Iteration 735, loss = 0.68850186\n",
      "Iteration 736, loss = 0.68849989\n",
      "Iteration 737, loss = 0.68849791\n",
      "Iteration 738, loss = 0.68849593\n",
      "Iteration 739, loss = 0.68849395\n",
      "Iteration 740, loss = 0.68849197\n",
      "Iteration 741, loss = 0.68848999\n",
      "Iteration 742, loss = 0.68848800\n",
      "Iteration 743, loss = 0.68848601\n",
      "Iteration 744, loss = 0.68848402\n",
      "Iteration 745, loss = 0.68848203\n",
      "Iteration 746, loss = 0.68848004\n",
      "Iteration 747, loss = 0.68847804\n",
      "Iteration 748, loss = 0.68847604\n",
      "Iteration 749, loss = 0.68847404\n",
      "Iteration 750, loss = 0.68847204\n",
      "Iteration 751, loss = 0.68847004\n",
      "Iteration 752, loss = 0.68846803\n",
      "Iteration 753, loss = 0.68846603\n",
      "Iteration 754, loss = 0.68846402\n",
      "Iteration 755, loss = 0.68846200\n",
      "Iteration 756, loss = 0.68845999\n",
      "Iteration 757, loss = 0.68845798\n",
      "Iteration 758, loss = 0.68845596\n",
      "Iteration 759, loss = 0.68845394\n",
      "Iteration 760, loss = 0.68845192\n",
      "Iteration 761, loss = 0.68844990\n",
      "Iteration 762, loss = 0.68844787\n",
      "Iteration 763, loss = 0.68844584\n",
      "Iteration 764, loss = 0.68844381\n",
      "Iteration 765, loss = 0.68844178\n",
      "Iteration 766, loss = 0.68843975\n",
      "Iteration 767, loss = 0.68843771\n",
      "Iteration 768, loss = 0.68843568\n",
      "Iteration 769, loss = 0.68843364\n",
      "Iteration 770, loss = 0.68843160\n",
      "Iteration 771, loss = 0.68842955\n",
      "Iteration 772, loss = 0.68842751\n",
      "Iteration 773, loss = 0.68842546\n",
      "Iteration 774, loss = 0.68842341\n",
      "Iteration 775, loss = 0.68842136\n",
      "Iteration 776, loss = 0.68841931\n",
      "Iteration 777, loss = 0.68841725\n",
      "Iteration 778, loss = 0.68841520\n",
      "Iteration 779, loss = 0.68841314\n",
      "Iteration 780, loss = 0.68841108\n",
      "Iteration 781, loss = 0.68840901\n",
      "Iteration 782, loss = 0.68840695\n",
      "Iteration 783, loss = 0.68840488\n",
      "Iteration 784, loss = 0.68840281\n",
      "Iteration 785, loss = 0.68840074\n",
      "Iteration 786, loss = 0.68839867\n",
      "Iteration 787, loss = 0.68839659\n",
      "Iteration 788, loss = 0.68839451\n",
      "Iteration 789, loss = 0.68839243\n",
      "Iteration 790, loss = 0.68839035\n",
      "Iteration 791, loss = 0.68838827\n",
      "Iteration 792, loss = 0.68838618\n",
      "Iteration 793, loss = 0.68838409\n",
      "Iteration 794, loss = 0.68838200\n",
      "Iteration 795, loss = 0.68837991\n",
      "Iteration 796, loss = 0.68837782\n",
      "Iteration 797, loss = 0.68837572\n",
      "Iteration 798, loss = 0.68837362\n",
      "Iteration 799, loss = 0.68837152\n",
      "Iteration 800, loss = 0.68836942\n",
      "Iteration 801, loss = 0.68836732\n",
      "Iteration 802, loss = 0.68836521\n",
      "Iteration 803, loss = 0.68836310\n",
      "Iteration 804, loss = 0.68836099\n",
      "Iteration 805, loss = 0.68835888\n",
      "Iteration 806, loss = 0.68835676\n",
      "Iteration 807, loss = 0.68835465\n",
      "Iteration 808, loss = 0.68835253\n",
      "Iteration 809, loss = 0.68835041\n",
      "Iteration 810, loss = 0.68834828\n",
      "Iteration 811, loss = 0.68834616\n",
      "Iteration 812, loss = 0.68834403\n",
      "Iteration 813, loss = 0.68834190\n",
      "Iteration 814, loss = 0.68833977\n",
      "Iteration 815, loss = 0.68833763\n",
      "Iteration 816, loss = 0.68833550\n",
      "Iteration 817, loss = 0.68833336\n",
      "Iteration 818, loss = 0.68833122\n",
      "Iteration 819, loss = 0.68832908\n",
      "Iteration 820, loss = 0.68832693\n",
      "Iteration 821, loss = 0.68832479\n",
      "Iteration 822, loss = 0.68832264\n",
      "Iteration 823, loss = 0.68832049\n",
      "Iteration 824, loss = 0.68831833\n",
      "Iteration 825, loss = 0.68831618\n",
      "Iteration 826, loss = 0.68831402\n",
      "Iteration 827, loss = 0.68831186\n",
      "Iteration 828, loss = 0.68830970\n",
      "Iteration 829, loss = 0.68830754\n",
      "Iteration 830, loss = 0.68830537\n",
      "Iteration 831, loss = 0.68830320\n",
      "Iteration 832, loss = 0.68830103\n",
      "Iteration 833, loss = 0.68829886\n",
      "Iteration 834, loss = 0.68829668\n",
      "Iteration 835, loss = 0.68829451\n",
      "Iteration 836, loss = 0.68829233\n",
      "Iteration 837, loss = 0.68829015\n",
      "Iteration 838, loss = 0.68828796\n",
      "Iteration 839, loss = 0.68828578\n",
      "Iteration 840, loss = 0.68828359\n",
      "Iteration 841, loss = 0.68828140\n",
      "Iteration 842, loss = 0.68827921\n",
      "Iteration 843, loss = 0.68827701\n",
      "Iteration 844, loss = 0.68827482\n",
      "Iteration 845, loss = 0.68827262\n",
      "Iteration 846, loss = 0.68827042\n",
      "Iteration 847, loss = 0.68826821\n",
      "Iteration 848, loss = 0.68826601\n",
      "Iteration 849, loss = 0.68826380\n",
      "Iteration 850, loss = 0.68826159\n",
      "Iteration 851, loss = 0.68825938\n",
      "Iteration 852, loss = 0.68825717\n",
      "Iteration 853, loss = 0.68825495\n",
      "Iteration 854, loss = 0.68825273\n",
      "Iteration 855, loss = 0.68825051\n",
      "Iteration 856, loss = 0.68824829\n",
      "Iteration 857, loss = 0.68824606\n",
      "Iteration 858, loss = 0.68824383\n",
      "Iteration 859, loss = 0.68824160\n",
      "Iteration 860, loss = 0.68823937\n",
      "Iteration 861, loss = 0.68823714\n",
      "Iteration 862, loss = 0.68823490\n",
      "Iteration 863, loss = 0.68823266\n",
      "Iteration 864, loss = 0.68823042\n",
      "Iteration 865, loss = 0.68822818\n",
      "Iteration 866, loss = 0.68822593\n",
      "Iteration 867, loss = 0.68822368\n",
      "Iteration 868, loss = 0.68822143\n",
      "Iteration 869, loss = 0.68821918\n",
      "Iteration 870, loss = 0.68821693\n",
      "Iteration 871, loss = 0.68821467\n",
      "Iteration 872, loss = 0.68821241\n",
      "Iteration 873, loss = 0.68821015\n",
      "Iteration 874, loss = 0.68820788\n",
      "Iteration 875, loss = 0.68820562\n",
      "Iteration 876, loss = 0.68820335\n",
      "Iteration 877, loss = 0.68820108\n",
      "Iteration 878, loss = 0.68819881\n",
      "Iteration 879, loss = 0.68819653\n",
      "Iteration 880, loss = 0.68819425\n",
      "Iteration 881, loss = 0.68819197\n",
      "Iteration 882, loss = 0.68818969\n",
      "Iteration 883, loss = 0.68818741\n",
      "Iteration 884, loss = 0.68818512\n",
      "Iteration 885, loss = 0.68818283\n",
      "Iteration 886, loss = 0.68818054\n",
      "Iteration 887, loss = 0.68817824\n",
      "Iteration 888, loss = 0.68817595\n",
      "Iteration 889, loss = 0.68817365\n",
      "Iteration 890, loss = 0.68817135\n",
      "Iteration 891, loss = 0.68816905\n",
      "Iteration 892, loss = 0.68816674\n",
      "Iteration 893, loss = 0.68816443\n",
      "Iteration 894, loss = 0.68816212\n",
      "Iteration 895, loss = 0.68815981\n",
      "Iteration 896, loss = 0.68815749\n",
      "Iteration 897, loss = 0.68815518\n",
      "Iteration 898, loss = 0.68815286\n",
      "Iteration 899, loss = 0.68815054\n",
      "Iteration 900, loss = 0.68814821\n",
      "Iteration 901, loss = 0.68814588\n",
      "Iteration 902, loss = 0.68814356\n",
      "Iteration 903, loss = 0.68814122\n",
      "Iteration 904, loss = 0.68813889\n",
      "Iteration 905, loss = 0.68813655\n",
      "Iteration 906, loss = 0.68813421\n",
      "Iteration 907, loss = 0.68813187\n",
      "Iteration 908, loss = 0.68812953\n",
      "Iteration 909, loss = 0.68812718\n",
      "Iteration 910, loss = 0.68812484\n",
      "Iteration 911, loss = 0.68812249\n",
      "Iteration 912, loss = 0.68812013\n",
      "Iteration 913, loss = 0.68811778\n",
      "Iteration 914, loss = 0.68811542\n",
      "Iteration 915, loss = 0.68811306\n",
      "Iteration 916, loss = 0.68811070\n",
      "Iteration 917, loss = 0.68810833\n",
      "Iteration 918, loss = 0.68810596\n",
      "Iteration 919, loss = 0.68810359\n",
      "Iteration 920, loss = 0.68810122\n",
      "Iteration 921, loss = 0.68809885\n",
      "Iteration 922, loss = 0.68809647\n",
      "Iteration 923, loss = 0.68809409\n",
      "Iteration 924, loss = 0.68809171\n",
      "Iteration 925, loss = 0.68808932\n",
      "Iteration 926, loss = 0.68808694\n",
      "Iteration 927, loss = 0.68808455\n",
      "Iteration 928, loss = 0.68808216\n",
      "Iteration 929, loss = 0.68807976\n",
      "Iteration 930, loss = 0.68807736\n",
      "Iteration 931, loss = 0.68807496\n",
      "Iteration 932, loss = 0.68807256\n",
      "Iteration 933, loss = 0.68807016\n",
      "Iteration 934, loss = 0.68806775\n",
      "Iteration 935, loss = 0.68806534\n",
      "Iteration 936, loss = 0.68806293\n",
      "Iteration 937, loss = 0.68806052\n",
      "Iteration 938, loss = 0.68805810\n",
      "Iteration 939, loss = 0.68805568\n",
      "Iteration 940, loss = 0.68805326\n",
      "Iteration 941, loss = 0.68805084\n",
      "Iteration 942, loss = 0.68804841\n",
      "Iteration 943, loss = 0.68804598\n",
      "Iteration 944, loss = 0.68804355\n",
      "Iteration 945, loss = 0.68804111\n",
      "Iteration 946, loss = 0.68803868\n",
      "Iteration 947, loss = 0.68803624\n",
      "Iteration 948, loss = 0.68803380\n",
      "Iteration 949, loss = 0.68803135\n",
      "Iteration 950, loss = 0.68802891\n",
      "Iteration 951, loss = 0.68802646\n",
      "Iteration 952, loss = 0.68802400\n",
      "Iteration 953, loss = 0.68802155\n",
      "Iteration 954, loss = 0.68801909\n",
      "Iteration 955, loss = 0.68801663\n",
      "Iteration 956, loss = 0.68801417\n",
      "Iteration 957, loss = 0.68801171\n",
      "Iteration 958, loss = 0.68800924\n",
      "Iteration 959, loss = 0.68800677\n",
      "Iteration 960, loss = 0.68800430\n",
      "Iteration 961, loss = 0.68800182\n",
      "Iteration 962, loss = 0.68799935\n",
      "Iteration 963, loss = 0.68799687\n",
      "Iteration 964, loss = 0.68799438\n",
      "Iteration 965, loss = 0.68799190\n",
      "Iteration 966, loss = 0.68798941\n",
      "Iteration 967, loss = 0.68798692\n",
      "Iteration 968, loss = 0.68798443\n",
      "Iteration 969, loss = 0.68798193\n",
      "Iteration 970, loss = 0.68797944\n",
      "Iteration 971, loss = 0.68797694\n",
      "Iteration 972, loss = 0.68797443\n",
      "Iteration 973, loss = 0.68797193\n",
      "Iteration 974, loss = 0.68796942\n",
      "Iteration 975, loss = 0.68796691\n",
      "Iteration 976, loss = 0.68796439\n",
      "Iteration 977, loss = 0.68796188\n",
      "Iteration 978, loss = 0.68795936\n",
      "Iteration 979, loss = 0.68795684\n",
      "Iteration 980, loss = 0.68795431\n",
      "Iteration 981, loss = 0.68795179\n",
      "Iteration 982, loss = 0.68794926\n",
      "Iteration 983, loss = 0.68794673\n",
      "Iteration 984, loss = 0.68794419\n",
      "Iteration 985, loss = 0.68794165\n",
      "Iteration 986, loss = 0.68793911\n",
      "Iteration 987, loss = 0.68793657\n",
      "Iteration 988, loss = 0.68793403\n",
      "Iteration 989, loss = 0.68793148\n",
      "Iteration 990, loss = 0.68792893\n",
      "Iteration 991, loss = 0.68792637\n",
      "Iteration 992, loss = 0.68792382\n",
      "Iteration 993, loss = 0.68792126\n",
      "Iteration 994, loss = 0.68791870\n",
      "Iteration 995, loss = 0.68791613\n",
      "Iteration 996, loss = 0.68791357\n",
      "Iteration 997, loss = 0.68791100\n",
      "Iteration 998, loss = 0.68790843\n",
      "Iteration 999, loss = 0.68790585\n",
      "Iteration 1000, loss = 0.68790327\n",
      "Iteration 1001, loss = 0.68790069\n",
      "Iteration 1002, loss = 0.68789811\n",
      "Iteration 1003, loss = 0.68789553\n",
      "Iteration 1004, loss = 0.68789294\n",
      "Iteration 1005, loss = 0.68789035\n",
      "Iteration 1006, loss = 0.68788775\n",
      "Iteration 1007, loss = 0.68788516\n",
      "Iteration 1008, loss = 0.68788256\n",
      "Iteration 1009, loss = 0.68787996\n",
      "Iteration 1010, loss = 0.68787735\n",
      "Iteration 1011, loss = 0.68787474\n",
      "Iteration 1012, loss = 0.68787213\n",
      "Iteration 1013, loss = 0.68786952\n",
      "Iteration 1014, loss = 0.68786691\n",
      "Iteration 1015, loss = 0.68786429\n",
      "Iteration 1016, loss = 0.68786167\n",
      "Iteration 1017, loss = 0.68785904\n",
      "Iteration 1018, loss = 0.68785642\n",
      "Iteration 1019, loss = 0.68785379\n",
      "Iteration 1020, loss = 0.68785116\n",
      "Iteration 1021, loss = 0.68784852\n",
      "Iteration 1022, loss = 0.68784588\n",
      "Iteration 1023, loss = 0.68784324\n",
      "Iteration 1024, loss = 0.68784060\n",
      "Iteration 1025, loss = 0.68783795\n",
      "Iteration 1026, loss = 0.68783530\n",
      "Iteration 1027, loss = 0.68783265\n",
      "Iteration 1028, loss = 0.68783000\n",
      "Iteration 1029, loss = 0.68782734\n",
      "Iteration 1030, loss = 0.68782468\n",
      "Iteration 1031, loss = 0.68782202\n",
      "Iteration 1032, loss = 0.68781935\n",
      "Iteration 1033, loss = 0.68781668\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration 1034, loss = 0.68781401\n",
      "Iteration 1035, loss = 0.68781134\n",
      "Iteration 1036, loss = 0.68780866\n",
      "Iteration 1037, loss = 0.68780598\n",
      "Iteration 1038, loss = 0.68780330\n",
      "Iteration 1039, loss = 0.68780061\n",
      "Iteration 1040, loss = 0.68779793\n",
      "Iteration 1041, loss = 0.68779523\n",
      "Iteration 1042, loss = 0.68779254\n",
      "Iteration 1043, loss = 0.68778984\n",
      "Iteration 1044, loss = 0.68778714\n",
      "Iteration 1045, loss = 0.68778444\n",
      "Iteration 1046, loss = 0.68778174\n",
      "Iteration 1047, loss = 0.68777903\n",
      "Iteration 1048, loss = 0.68777632\n",
      "Iteration 1049, loss = 0.68777360\n",
      "Iteration 1050, loss = 0.68777089\n",
      "Iteration 1051, loss = 0.68776817\n",
      "Iteration 1052, loss = 0.68776544\n",
      "Iteration 1053, loss = 0.68776272\n",
      "Iteration 1054, loss = 0.68775999\n",
      "Iteration 1055, loss = 0.68775726\n",
      "Iteration 1056, loss = 0.68775452\n",
      "Iteration 1057, loss = 0.68775179\n",
      "Iteration 1058, loss = 0.68774905\n",
      "Iteration 1059, loss = 0.68774630\n",
      "Iteration 1060, loss = 0.68774356\n",
      "Iteration 1061, loss = 0.68774081\n",
      "Iteration 1062, loss = 0.68773806\n",
      "Iteration 1063, loss = 0.68773530\n",
      "Iteration 1064, loss = 0.68773254\n",
      "Iteration 1065, loss = 0.68772978\n",
      "Iteration 1066, loss = 0.68772702\n",
      "Iteration 1067, loss = 0.68772425\n",
      "Iteration 1068, loss = 0.68772148\n",
      "Iteration 1069, loss = 0.68771871\n",
      "Iteration 1070, loss = 0.68771593\n",
      "Iteration 1071, loss = 0.68771316\n",
      "Iteration 1072, loss = 0.68771037\n",
      "Iteration 1073, loss = 0.68770759\n",
      "Iteration 1074, loss = 0.68770480\n",
      "Iteration 1075, loss = 0.68770201\n",
      "Iteration 1076, loss = 0.68769922\n",
      "Iteration 1077, loss = 0.68769642\n",
      "Iteration 1078, loss = 0.68769362\n",
      "Iteration 1079, loss = 0.68769082\n",
      "Iteration 1080, loss = 0.68768801\n",
      "Iteration 1081, loss = 0.68768521\n",
      "Iteration 1082, loss = 0.68768239\n",
      "Iteration 1083, loss = 0.68767958\n",
      "Iteration 1084, loss = 0.68767676\n",
      "Iteration 1085, loss = 0.68767394\n",
      "Iteration 1086, loss = 0.68767112\n",
      "Iteration 1087, loss = 0.68766829\n",
      "Iteration 1088, loss = 0.68766546\n",
      "Iteration 1089, loss = 0.68766263\n",
      "Iteration 1090, loss = 0.68765979\n",
      "Iteration 1091, loss = 0.68765696\n",
      "Iteration 1092, loss = 0.68765411\n",
      "Iteration 1093, loss = 0.68765127\n",
      "Iteration 1094, loss = 0.68764842\n",
      "Iteration 1095, loss = 0.68764557\n",
      "Iteration 1096, loss = 0.68764272\n",
      "Iteration 1097, loss = 0.68763986\n",
      "Iteration 1098, loss = 0.68763700\n",
      "Iteration 1099, loss = 0.68763414\n",
      "Iteration 1100, loss = 0.68763127\n",
      "Iteration 1101, loss = 0.68762840\n",
      "Iteration 1102, loss = 0.68762553\n",
      "Iteration 1103, loss = 0.68762265\n",
      "Iteration 1104, loss = 0.68761977\n",
      "Iteration 1105, loss = 0.68761689\n",
      "Iteration 1106, loss = 0.68761401\n",
      "Iteration 1107, loss = 0.68761112\n",
      "Iteration 1108, loss = 0.68760823\n",
      "Iteration 1109, loss = 0.68760533\n",
      "Iteration 1110, loss = 0.68760244\n",
      "Iteration 1111, loss = 0.68759954\n",
      "Iteration 1112, loss = 0.68759663\n",
      "Iteration 1113, loss = 0.68759372\n",
      "Iteration 1114, loss = 0.68759081\n",
      "Iteration 1115, loss = 0.68758790\n",
      "Iteration 1116, loss = 0.68758499\n",
      "Iteration 1117, loss = 0.68758207\n",
      "Iteration 1118, loss = 0.68757914\n",
      "Iteration 1119, loss = 0.68757622\n",
      "Iteration 1120, loss = 0.68757329\n",
      "Iteration 1121, loss = 0.68757036\n",
      "Iteration 1122, loss = 0.68756742\n",
      "Iteration 1123, loss = 0.68756448\n",
      "Iteration 1124, loss = 0.68756154\n",
      "Iteration 1125, loss = 0.68755860\n",
      "Iteration 1126, loss = 0.68755565\n",
      "Iteration 1127, loss = 0.68755270\n",
      "Iteration 1128, loss = 0.68754974\n",
      "Iteration 1129, loss = 0.68754678\n",
      "Iteration 1130, loss = 0.68754382\n",
      "Iteration 1131, loss = 0.68754086\n",
      "Iteration 1132, loss = 0.68753789\n",
      "Iteration 1133, loss = 0.68753492\n",
      "Iteration 1134, loss = 0.68753195\n",
      "Iteration 1135, loss = 0.68752897\n",
      "Iteration 1136, loss = 0.68752599\n",
      "Iteration 1137, loss = 0.68752301\n",
      "Iteration 1138, loss = 0.68752002\n",
      "Iteration 1139, loss = 0.68751703\n",
      "Iteration 1140, loss = 0.68751404\n",
      "Iteration 1141, loss = 0.68751104\n",
      "Iteration 1142, loss = 0.68750804\n",
      "Iteration 1143, loss = 0.68750504\n",
      "Iteration 1144, loss = 0.68750203\n",
      "Iteration 1145, loss = 0.68749902\n",
      "Iteration 1146, loss = 0.68749601\n",
      "Iteration 1147, loss = 0.68749299\n",
      "Iteration 1148, loss = 0.68748997\n",
      "Iteration 1149, loss = 0.68748695\n",
      "Iteration 1150, loss = 0.68748392\n",
      "Iteration 1151, loss = 0.68748089\n",
      "Iteration 1152, loss = 0.68747786\n",
      "Iteration 1153, loss = 0.68747482\n",
      "Iteration 1154, loss = 0.68747179\n",
      "Iteration 1155, loss = 0.68746874\n",
      "Iteration 1156, loss = 0.68746570\n",
      "Iteration 1157, loss = 0.68746265\n",
      "Iteration 1158, loss = 0.68745959\n",
      "Iteration 1159, loss = 0.68745654\n",
      "Iteration 1160, loss = 0.68745348\n",
      "Iteration 1161, loss = 0.68745042\n",
      "Iteration 1162, loss = 0.68744735\n",
      "Iteration 1163, loss = 0.68744428\n",
      "Iteration 1164, loss = 0.68744121\n",
      "Iteration 1165, loss = 0.68743813\n",
      "Iteration 1166, loss = 0.68743505\n",
      "Iteration 1167, loss = 0.68743197\n",
      "Iteration 1168, loss = 0.68742888\n",
      "Iteration 1169, loss = 0.68742579\n",
      "Iteration 1170, loss = 0.68742270\n",
      "Iteration 1171, loss = 0.68741960\n",
      "Iteration 1172, loss = 0.68741650\n",
      "Iteration 1173, loss = 0.68741340\n",
      "Iteration 1174, loss = 0.68741029\n",
      "Iteration 1175, loss = 0.68740718\n",
      "Iteration 1176, loss = 0.68740407\n",
      "Iteration 1177, loss = 0.68740095\n",
      "Iteration 1178, loss = 0.68739783\n",
      "Iteration 1179, loss = 0.68739471\n",
      "Iteration 1180, loss = 0.68739158\n",
      "Iteration 1181, loss = 0.68738845\n",
      "Iteration 1182, loss = 0.68738531\n",
      "Iteration 1183, loss = 0.68738218\n",
      "Iteration 1184, loss = 0.68737903\n",
      "Iteration 1185, loss = 0.68737589\n",
      "Iteration 1186, loss = 0.68737274\n",
      "Iteration 1187, loss = 0.68736959\n",
      "Iteration 1188, loss = 0.68736643\n",
      "Iteration 1189, loss = 0.68736328\n",
      "Iteration 1190, loss = 0.68736011\n",
      "Iteration 1191, loss = 0.68735695\n",
      "Iteration 1192, loss = 0.68735378\n",
      "Iteration 1193, loss = 0.68735061\n",
      "Iteration 1194, loss = 0.68734743\n",
      "Iteration 1195, loss = 0.68734425\n",
      "Iteration 1196, loss = 0.68734107\n",
      "Iteration 1197, loss = 0.68733788\n",
      "Iteration 1198, loss = 0.68733469\n",
      "Iteration 1199, loss = 0.68733150\n",
      "Iteration 1200, loss = 0.68732830\n",
      "Iteration 1201, loss = 0.68732510\n",
      "Iteration 1202, loss = 0.68732189\n",
      "Iteration 1203, loss = 0.68731869\n",
      "Iteration 1204, loss = 0.68731547\n",
      "Iteration 1205, loss = 0.68731226\n",
      "Iteration 1206, loss = 0.68730904\n",
      "Iteration 1207, loss = 0.68730582\n",
      "Iteration 1208, loss = 0.68730259\n",
      "Iteration 1209, loss = 0.68729936\n",
      "Iteration 1210, loss = 0.68729613\n",
      "Iteration 1211, loss = 0.68729289\n",
      "Iteration 1212, loss = 0.68728965\n",
      "Iteration 1213, loss = 0.68728641\n",
      "Iteration 1214, loss = 0.68728316\n",
      "Iteration 1215, loss = 0.68727991\n",
      "Iteration 1216, loss = 0.68727666\n",
      "Iteration 1217, loss = 0.68727340\n",
      "Iteration 1218, loss = 0.68727014\n",
      "Iteration 1219, loss = 0.68726687\n",
      "Iteration 1220, loss = 0.68726360\n",
      "Iteration 1221, loss = 0.68726033\n",
      "Iteration 1222, loss = 0.68725705\n",
      "Iteration 1223, loss = 0.68725377\n",
      "Iteration 1224, loss = 0.68725049\n",
      "Iteration 1225, loss = 0.68724720\n",
      "Iteration 1226, loss = 0.68724391\n",
      "Iteration 1227, loss = 0.68724061\n",
      "Iteration 1228, loss = 0.68723732\n",
      "Iteration 1229, loss = 0.68723401\n",
      "Iteration 1230, loss = 0.68723071\n",
      "Iteration 1231, loss = 0.68722740\n",
      "Iteration 1232, loss = 0.68722408\n",
      "Iteration 1233, loss = 0.68722077\n",
      "Iteration 1234, loss = 0.68721745\n",
      "Iteration 1235, loss = 0.68721412\n",
      "Iteration 1236, loss = 0.68721079\n",
      "Iteration 1237, loss = 0.68720746\n",
      "Iteration 1238, loss = 0.68720413\n",
      "Iteration 1239, loss = 0.68720079\n",
      "Iteration 1240, loss = 0.68719744\n",
      "Iteration 1241, loss = 0.68719410\n",
      "Iteration 1242, loss = 0.68719075\n",
      "Iteration 1243, loss = 0.68718739\n",
      "Iteration 1244, loss = 0.68718403\n",
      "Iteration 1245, loss = 0.68718067\n",
      "Iteration 1246, loss = 0.68717731\n",
      "Iteration 1247, loss = 0.68717394\n",
      "Iteration 1248, loss = 0.68717056\n",
      "Iteration 1249, loss = 0.68716719\n",
      "Iteration 1250, loss = 0.68716381\n",
      "Iteration 1251, loss = 0.68716042\n",
      "Iteration 1252, loss = 0.68715703\n",
      "Iteration 1253, loss = 0.68715364\n",
      "Iteration 1254, loss = 0.68715024\n",
      "Iteration 1255, loss = 0.68714684\n",
      "Iteration 1256, loss = 0.68714344\n",
      "Iteration 1257, loss = 0.68714003\n",
      "Iteration 1258, loss = 0.68713662\n",
      "Iteration 1259, loss = 0.68713321\n",
      "Iteration 1260, loss = 0.68712979\n",
      "Iteration 1261, loss = 0.68712636\n",
      "Iteration 1262, loss = 0.68712294\n",
      "Iteration 1263, loss = 0.68711951\n",
      "Iteration 1264, loss = 0.68711607\n",
      "Iteration 1265, loss = 0.68711263\n",
      "Iteration 1266, loss = 0.68710919\n",
      "Iteration 1267, loss = 0.68710575\n",
      "Iteration 1268, loss = 0.68710230\n",
      "Iteration 1269, loss = 0.68709884\n",
      "Iteration 1270, loss = 0.68709538\n",
      "Iteration 1271, loss = 0.68709192\n",
      "Iteration 1272, loss = 0.68708846\n",
      "Iteration 1273, loss = 0.68708499\n",
      "Iteration 1274, loss = 0.68708151\n",
      "Iteration 1275, loss = 0.68707803\n",
      "Iteration 1276, loss = 0.68707455\n",
      "Iteration 1277, loss = 0.68707107\n",
      "Iteration 1278, loss = 0.68706758\n",
      "Iteration 1279, loss = 0.68706409\n",
      "Iteration 1280, loss = 0.68706059\n",
      "Iteration 1281, loss = 0.68705709\n",
      "Iteration 1282, loss = 0.68705358\n",
      "Iteration 1283, loss = 0.68705007\n",
      "Iteration 1284, loss = 0.68704656\n",
      "Iteration 1285, loss = 0.68704304\n",
      "Iteration 1286, loss = 0.68703952\n",
      "Iteration 1287, loss = 0.68703600\n",
      "Iteration 1288, loss = 0.68703247\n",
      "Iteration 1289, loss = 0.68702893\n",
      "Iteration 1290, loss = 0.68702540\n",
      "Iteration 1291, loss = 0.68702186\n",
      "Iteration 1292, loss = 0.68701831\n",
      "Iteration 1293, loss = 0.68701476\n",
      "Iteration 1294, loss = 0.68701121\n",
      "Iteration 1295, loss = 0.68700765\n",
      "Iteration 1296, loss = 0.68700409\n",
      "Iteration 1297, loss = 0.68700052\n",
      "Iteration 1298, loss = 0.68699695\n",
      "Iteration 1299, loss = 0.68699338\n",
      "Iteration 1300, loss = 0.68698980\n",
      "Iteration 1301, loss = 0.68698622\n",
      "Iteration 1302, loss = 0.68698264\n",
      "Iteration 1303, loss = 0.68697905\n",
      "Iteration 1304, loss = 0.68697545\n",
      "Iteration 1305, loss = 0.68697185\n",
      "Iteration 1306, loss = 0.68696825\n",
      "Iteration 1307, loss = 0.68696465\n",
      "Iteration 1308, loss = 0.68696103\n",
      "Iteration 1309, loss = 0.68695742\n",
      "Iteration 1310, loss = 0.68695380\n",
      "Iteration 1311, loss = 0.68695018\n",
      "Iteration 1312, loss = 0.68694655\n",
      "Iteration 1313, loss = 0.68694292\n",
      "Iteration 1314, loss = 0.68693929\n",
      "Iteration 1315, loss = 0.68693565\n",
      "Iteration 1316, loss = 0.68693200\n",
      "Iteration 1317, loss = 0.68692835\n",
      "Iteration 1318, loss = 0.68692470\n",
      "Iteration 1319, loss = 0.68692105\n",
      "Iteration 1320, loss = 0.68691739\n",
      "Iteration 1321, loss = 0.68691372\n",
      "Iteration 1322, loss = 0.68691005\n",
      "Iteration 1323, loss = 0.68690638\n",
      "Iteration 1324, loss = 0.68690270\n",
      "Iteration 1325, loss = 0.68689902\n",
      "Iteration 1326, loss = 0.68689534\n",
      "Iteration 1327, loss = 0.68689165\n",
      "Iteration 1328, loss = 0.68688795\n",
      "Iteration 1329, loss = 0.68688425\n",
      "Iteration 1330, loss = 0.68688055\n",
      "Iteration 1331, loss = 0.68687684\n",
      "Iteration 1332, loss = 0.68687313\n",
      "Iteration 1333, loss = 0.68686942\n",
      "Iteration 1334, loss = 0.68686570\n",
      "Iteration 1335, loss = 0.68686197\n",
      "Iteration 1336, loss = 0.68685825\n",
      "Iteration 1337, loss = 0.68685451\n",
      "Iteration 1338, loss = 0.68685078\n",
      "Iteration 1339, loss = 0.68684704\n",
      "Iteration 1340, loss = 0.68684329\n",
      "Iteration 1341, loss = 0.68683954\n",
      "Iteration 1342, loss = 0.68683579\n",
      "Iteration 1343, loss = 0.68683203\n",
      "Iteration 1344, loss = 0.68682827\n",
      "Iteration 1345, loss = 0.68682450\n",
      "Iteration 1346, loss = 0.68682073\n",
      "Iteration 1347, loss = 0.68681695\n",
      "Iteration 1348, loss = 0.68681317\n",
      "Iteration 1349, loss = 0.68680939\n",
      "Iteration 1350, loss = 0.68680560\n",
      "Iteration 1351, loss = 0.68680180\n",
      "Iteration 1352, loss = 0.68679801\n",
      "Iteration 1353, loss = 0.68679420\n",
      "Iteration 1354, loss = 0.68679040\n",
      "Iteration 1355, loss = 0.68678659\n",
      "Iteration 1356, loss = 0.68678277\n",
      "Iteration 1357, loss = 0.68677895\n",
      "Iteration 1358, loss = 0.68677513\n",
      "Iteration 1359, loss = 0.68677130\n",
      "Iteration 1360, loss = 0.68676747\n",
      "Iteration 1361, loss = 0.68676363\n",
      "Iteration 1362, loss = 0.68675979\n",
      "Iteration 1363, loss = 0.68675594\n",
      "Iteration 1364, loss = 0.68675209\n",
      "Iteration 1365, loss = 0.68674824\n",
      "Iteration 1366, loss = 0.68674438\n",
      "Iteration 1367, loss = 0.68674051\n",
      "Iteration 1368, loss = 0.68673664\n",
      "Iteration 1369, loss = 0.68673277\n",
      "Iteration 1370, loss = 0.68672889\n",
      "Iteration 1371, loss = 0.68672501\n",
      "Iteration 1372, loss = 0.68672112\n",
      "Iteration 1373, loss = 0.68671723\n",
      "Iteration 1374, loss = 0.68671334\n",
      "Iteration 1375, loss = 0.68670944\n",
      "Iteration 1376, loss = 0.68670553\n",
      "Iteration 1377, loss = 0.68670162\n",
      "Iteration 1378, loss = 0.68669771\n",
      "Iteration 1379, loss = 0.68669379\n",
      "Iteration 1380, loss = 0.68668987\n",
      "Iteration 1381, loss = 0.68668594\n",
      "Iteration 1382, loss = 0.68668201\n",
      "Iteration 1383, loss = 0.68667807\n",
      "Iteration 1384, loss = 0.68667413\n",
      "Iteration 1385, loss = 0.68667018\n",
      "Iteration 1386, loss = 0.68666623\n",
      "Iteration 1387, loss = 0.68666228\n",
      "Iteration 1388, loss = 0.68665832\n",
      "Iteration 1389, loss = 0.68665435\n",
      "Iteration 1390, loss = 0.68665038\n",
      "Iteration 1391, loss = 0.68664641\n",
      "Iteration 1392, loss = 0.68664243\n",
      "Iteration 1393, loss = 0.68663845\n",
      "Iteration 1394, loss = 0.68663446\n",
      "Iteration 1395, loss = 0.68663047\n",
      "Iteration 1396, loss = 0.68662647\n",
      "Iteration 1397, loss = 0.68662247\n",
      "Iteration 1398, loss = 0.68661846\n",
      "Iteration 1399, loss = 0.68661445\n",
      "Iteration 1400, loss = 0.68661044\n",
      "Iteration 1401, loss = 0.68660642\n",
      "Iteration 1402, loss = 0.68660239\n",
      "Iteration 1403, loss = 0.68659836\n",
      "Iteration 1404, loss = 0.68659433\n",
      "Iteration 1405, loss = 0.68659029\n",
      "Iteration 1406, loss = 0.68658624\n",
      "Iteration 1407, loss = 0.68658220\n",
      "Iteration 1408, loss = 0.68657814\n",
      "Iteration 1409, loss = 0.68657408\n",
      "Iteration 1410, loss = 0.68657002\n",
      "Iteration 1411, loss = 0.68656595\n",
      "Iteration 1412, loss = 0.68656188\n",
      "Iteration 1413, loss = 0.68655780\n",
      "Iteration 1414, loss = 0.68655372\n",
      "Iteration 1415, loss = 0.68654963\n",
      "Iteration 1416, loss = 0.68654554\n",
      "Iteration 1417, loss = 0.68654145\n",
      "Iteration 1418, loss = 0.68653734\n",
      "Iteration 1419, loss = 0.68653324\n",
      "Iteration 1420, loss = 0.68652913\n",
      "Iteration 1421, loss = 0.68652501\n",
      "Iteration 1422, loss = 0.68652089\n",
      "Iteration 1423, loss = 0.68651676\n",
      "Iteration 1424, loss = 0.68651263\n",
      "Iteration 1425, loss = 0.68650850\n",
      "Iteration 1426, loss = 0.68650436\n",
      "Iteration 1427, loss = 0.68650021\n",
      "Iteration 1428, loss = 0.68649606\n",
      "Iteration 1429, loss = 0.68649191\n",
      "Iteration 1430, loss = 0.68648775\n",
      "Iteration 1431, loss = 0.68648358\n",
      "Iteration 1432, loss = 0.68647941\n",
      "Iteration 1433, loss = 0.68647524\n",
      "Iteration 1434, loss = 0.68647106\n",
      "Iteration 1435, loss = 0.68646687\n",
      "Iteration 1436, loss = 0.68646268\n",
      "Iteration 1437, loss = 0.68645849\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration 1438, loss = 0.68645429\n",
      "Iteration 1439, loss = 0.68645009\n",
      "Iteration 1440, loss = 0.68644588\n",
      "Iteration 1441, loss = 0.68644166\n",
      "Iteration 1442, loss = 0.68643744\n",
      "Iteration 1443, loss = 0.68643322\n",
      "Iteration 1444, loss = 0.68642899\n",
      "Iteration 1445, loss = 0.68642476\n",
      "Iteration 1446, loss = 0.68642052\n",
      "Iteration 1447, loss = 0.68641627\n",
      "Iteration 1448, loss = 0.68641202\n",
      "Iteration 1449, loss = 0.68640777\n",
      "Iteration 1450, loss = 0.68640351\n",
      "Iteration 1451, loss = 0.68639924\n",
      "Iteration 1452, loss = 0.68639497\n",
      "Iteration 1453, loss = 0.68639070\n",
      "Iteration 1454, loss = 0.68638642\n",
      "Iteration 1455, loss = 0.68638213\n",
      "Iteration 1456, loss = 0.68637784\n",
      "Iteration 1457, loss = 0.68637355\n",
      "Iteration 1458, loss = 0.68636925\n",
      "Iteration 1459, loss = 0.68636494\n",
      "Iteration 1460, loss = 0.68636063\n",
      "Iteration 1461, loss = 0.68635632\n",
      "Iteration 1462, loss = 0.68635200\n",
      "Iteration 1463, loss = 0.68634767\n",
      "Iteration 1464, loss = 0.68634334\n",
      "Iteration 1465, loss = 0.68633900\n",
      "Iteration 1466, loss = 0.68633466\n",
      "Iteration 1467, loss = 0.68633032\n",
      "Iteration 1468, loss = 0.68632596\n",
      "Iteration 1469, loss = 0.68632161\n",
      "Iteration 1470, loss = 0.68631724\n",
      "Iteration 1471, loss = 0.68631288\n",
      "Iteration 1472, loss = 0.68630850\n",
      "Iteration 1473, loss = 0.68630413\n",
      "Iteration 1474, loss = 0.68629974\n",
      "Iteration 1475, loss = 0.68629536\n",
      "Iteration 1476, loss = 0.68629096\n",
      "Iteration 1477, loss = 0.68628656\n",
      "Iteration 1478, loss = 0.68628216\n",
      "Iteration 1479, loss = 0.68627775\n",
      "Iteration 1480, loss = 0.68627333\n",
      "Iteration 1481, loss = 0.68626891\n",
      "Iteration 1482, loss = 0.68626449\n",
      "Iteration 1483, loss = 0.68626006\n",
      "Iteration 1484, loss = 0.68625562\n",
      "Iteration 1485, loss = 0.68625118\n",
      "Iteration 1486, loss = 0.68624674\n",
      "Iteration 1487, loss = 0.68624228\n",
      "Iteration 1488, loss = 0.68623783\n",
      "Iteration 1489, loss = 0.68623336\n",
      "Iteration 1490, loss = 0.68622890\n",
      "Iteration 1491, loss = 0.68622442\n",
      "Iteration 1492, loss = 0.68621995\n",
      "Iteration 1493, loss = 0.68621546\n",
      "Iteration 1494, loss = 0.68621097\n",
      "Iteration 1495, loss = 0.68620648\n",
      "Iteration 1496, loss = 0.68620198\n",
      "Iteration 1497, loss = 0.68619747\n",
      "Iteration 1498, loss = 0.68619296\n",
      "Iteration 1499, loss = 0.68618844\n",
      "Iteration 1500, loss = 0.68618392\n",
      "0.47619047619047616\n",
      "[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/rian/.local/lib/python3.6/site-packages/sklearn/neural_network/multilayer_perceptron.py:562: ConvergenceWarning: Stochastic Optimizer: Maximum iterations (1500) reached and the optimization hasn't converged yet.\n",
      "  % self.max_iter, ConvergenceWarning)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.neural_network import MLPClassifier\n",
    "# 15 0.6 50\n",
    "# 9-3, 200, 0.1, 1500\n",
    "mlp = MLPClassifier(hidden_layer_sizes=(9,3),\n",
    "                    solver='sgd',\n",
    "                    batch_size=200,\n",
    "                    activation='relu',\n",
    "                    learning_rate='adaptive',\n",
    "                    learning_rate_init=0.1,\n",
    "                    max_iter=1500,\n",
    "                    momentum=0,\n",
    "                    tol=0.0,\n",
    "                    alpha=0,\n",
    "                    verbose=True,\n",
    "                    random_state=123)\n",
    "train_x = feature[:,:-1]\n",
    "train_y = feature[:,-1]\n",
    "test_x = feature_test[:,:-1]\n",
    "test_y = feature_test[:,-1]\n",
    "test_x.shape, test_y.shape\n",
    "\n",
    "mlp.fit(train_x, train_y)\n",
    "print(mlp.score(test_x,test_y))\n",
    "print(mlp.predict(test_x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}