readTFR.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "sess = tf.InteractiveSession()\n",
    "reader = tf.TFRecordReader()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "# filename_queue = tf.train.string_input_producer(['./datasets/square_room_100steps_2.2m_1000000/0000-of-0099.tfrecord'])\n",
    "filename_queue = tf.train.string_input_producer(['./example1.tfrecord'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "target_pos: [[ 0.16789106  0.0397125 ]\n",
      " [ 0.17730418  0.06788748]\n",
      " [ 0.18118468  0.08794076]\n",
      " [ 0.17628641  0.10716452]\n",
      " [ 0.16687028  0.13080056]\n",
      " [ 0.15133652  0.1492285 ]\n",
      " [ 0.13064611  0.16751076]\n",
      " [ 0.1099921   0.1797836 ]\n",
      " [ 0.08864851  0.1918129 ]\n",
      " [ 0.06653202  0.20033191]\n",
      " [ 0.0509055   0.19938189]\n",
      " [ 0.03569192  0.18910807]\n",
      " [ 0.01652917  0.17823593]\n",
      " [-0.00150753  0.17540517]\n",
      " [-0.01799637  0.17148489]\n",
      " [-0.04065638  0.16649796]\n",
      " [-0.06598745  0.1648785 ]\n",
      " [-0.08325773  0.16980955]\n",
      " [-0.10907358  0.17440842]\n",
      " [-0.13036852  0.17724015]\n",
      " [-0.15194525  0.19276001]\n",
      " [-0.17616162  0.20708695]\n",
      " [-0.19382176  0.21443062]\n",
      " [-0.21773294  0.2244129 ]\n",
      " [-0.23798586  0.23841207]\n",
      " [-0.25026584  0.25296351]\n",
      " [-0.26049852  0.26624751]\n",
      " [-0.26850599  0.2823551 ]\n",
      " [-0.28137031  0.30331463]\n",
      " [-0.28908867  0.32190934]\n",
      " [-0.29064542  0.34645382]\n",
      " [-0.29314369  0.36233377]\n",
      " [-0.28895232  0.39396751]\n",
      " [-0.29191801  0.41702205]\n",
      " [-0.29882398  0.44097596]\n",
      " [-0.31133732  0.46339181]\n",
      " [-0.32495701  0.48391005]\n",
      " [-0.32901078  0.50745565]\n",
      " [-0.33408189  0.52826113]\n",
      " [-0.33768451  0.5431242 ]\n",
      " [-0.34235203  0.56101984]\n",
      " [-0.34821975  0.58217788]\n",
      " [-0.34941772  0.60317385]\n",
      " [-0.36339995  0.6201039 ]\n",
      " [-0.37944818  0.63971466]\n",
      " [-0.39289081  0.6602965 ]\n",
      " [-0.39968723  0.67930627]\n",
      " [-0.40741155  0.69899017]\n",
      " [-0.42695448  0.71690172]\n",
      " [-0.43903944  0.73375028]\n",
      " [-0.44582114  0.75761586]\n",
      " [-0.43382218  0.77328378]\n",
      " [-0.41702819  0.77566481]\n",
      " [-0.39824608  0.7733956 ]\n",
      " [-0.37649682  0.7712456 ]\n",
      " [-0.35222742  0.7652548 ]\n",
      " [-0.32803336  0.7594493 ]\n",
      " [-0.30504048  0.76138407]\n",
      " [-0.2814675   0.76611149]\n",
      " [-0.25777757  0.76327276]\n",
      " [-0.23844025  0.76226276]\n",
      " [-0.20823996  0.7562567 ]\n",
      " [-0.18948869  0.75631583]\n",
      " [-0.16775356  0.75577611]\n",
      " [-0.14501289  0.75269526]\n",
      " [-0.11908989  0.74127799]\n",
      " [-0.09699328  0.72744167]\n",
      " [-0.07790814  0.70770222]\n",
      " [-0.07176384  0.69278574]\n",
      " [-0.06865078  0.67827791]\n",
      " [-0.0631876   0.65552849]\n",
      " [-0.06295022  0.63896692]\n",
      " [-0.06544045  0.61367869]\n",
      " [-0.0740825   0.5967713 ]\n",
      " [-0.08132631  0.57766372]\n",
      " [-0.08851404  0.56200254]\n",
      " [-0.08941882  0.53839648]\n",
      " [-0.0772913   0.52415156]\n",
      " [-0.06312168  0.51314741]\n",
      " [-0.03864142  0.50036561]\n",
      " [-0.01470841  0.49421132]\n",
      " [ 0.01066761  0.49415088]\n",
      " [ 0.03530373  0.49705485]\n",
      " [ 0.06069348  0.49202728]\n",
      " [ 0.0817477   0.48798373]\n",
      " [ 0.09534949  0.4775573 ]\n",
      " [ 0.10434284  0.45946115]\n",
      " [ 0.10992885  0.44356224]\n",
      " [ 0.12267777  0.4207772 ]\n",
      " [ 0.13318513  0.40803593]\n",
      " [ 0.14917843  0.39411393]\n",
      " [ 0.17038041  0.38610259]\n",
      " [ 0.1884826   0.39100966]\n",
      " [ 0.21025339  0.39652839]\n",
      " [ 0.23148842  0.40457204]\n",
      " [ 0.24064173  0.41526338]\n",
      " [ 0.25219876  0.43181238]\n",
      " [ 0.26813519  0.43896854]\n",
      " [ 0.29567817  0.44373178]\n",
      " [ 0.30593115  0.44724825]]\n",
      "target_hd: [[ 0.73634267]\n",
      " [ 0.66448343]\n",
      " [ 0.77527589]\n",
      " [ 1.02091265]\n",
      " [ 1.15677929]\n",
      " [ 0.78456169]\n",
      " [ 1.2135061 ]\n",
      " [ 0.66780931]\n",
      " [ 0.53124517]\n",
      " [ 1.02985346]\n",
      " [ 1.13827884]\n",
      " [ 0.56416434]\n",
      " [ 1.13481784]\n",
      " [ 1.13437033]\n",
      " [ 1.00798261]\n",
      " [ 1.23360908]\n",
      " [ 0.94903922]\n",
      " [ 1.26141036]\n",
      " [ 0.5437507 ]\n",
      " [ 0.87889707]\n",
      " [ 0.95747477]\n",
      " [ 0.67802221]\n",
      " [ 0.70092118]\n",
      " [ 0.64832664]\n",
      " [ 0.69200474]\n",
      " [ 0.68390495]\n",
      " [ 1.1300019 ]\n",
      " [ 1.43376076]\n",
      " [ 1.62127292]\n",
      " [ 1.50881326]\n",
      " [ 1.90815938]\n",
      " [ 1.91646743]\n",
      " [ 2.07147622]\n",
      " [ 2.79220986]\n",
      " [ 2.97339439]\n",
      " [-3.13111782]\n",
      " [ 2.90042257]\n",
      " [ 3.0756247 ]\n",
      " [-2.8776927 ]\n",
      " [ 2.94202304]\n",
      " [ 2.9974544 ]\n",
      " [-2.95047212]\n",
      " [-2.878088  ]\n",
      " [ 2.95315599]\n",
      " [ 2.64274836]\n",
      " [ 2.14728189]\n",
      " [ 2.10884261]\n",
      " [ 2.25873756]\n",
      " [ 2.65895867]\n",
      " [ 2.59026384]\n",
      " [ 2.37492013]\n",
      " [ 2.51142073]\n",
      " [ 2.68748569]\n",
      " [ 2.51486921]\n",
      " [ 2.26631141]\n",
      " [ 2.28934264]\n",
      " [ 2.43358207]\n",
      " [ 2.73836398]\n",
      " [ 2.58699799]\n",
      " [ 2.31739068]\n",
      " [ 2.0762105 ]\n",
      " [ 1.86384022]\n",
      " [ 1.32219136]\n",
      " [ 1.59439552]\n",
      " [ 2.01543689]\n",
      " [ 2.10335064]\n",
      " [ 1.61920702]\n",
      " [ 1.77373898]\n",
      " [ 1.34915018]\n",
      " [ 0.37883732]\n",
      " [-0.12575568]\n",
      " [-0.00352147]\n",
      " [ 0.46106958]\n",
      " [ 0.80886751]\n",
      " [ 1.22731638]\n",
      " [ 1.76738417]\n",
      " [ 1.5530684 ]\n",
      " [ 1.58799267]\n",
      " [ 2.08755302]\n",
      " [ 2.27372813]\n",
      " [ 2.53335071]\n",
      " [-3.08567905]\n",
      " [-2.9487834 ]\n",
      " [-2.75709844]\n",
      " [-2.50694656]\n",
      " [-2.10253978]\n",
      " [-2.36092186]\n",
      " [-2.14016342]\n",
      " [-2.78596759]\n",
      " [-2.70180917]\n",
      " [-2.01605868]\n",
      " [-1.49226606]\n",
      " [-1.74775684]\n",
      " [-2.21310997]\n",
      " [-2.26609159]\n",
      " [-1.80427516]\n",
      " [-1.20237851]\n",
      " [-1.53935921]\n",
      " [-1.63808382]\n",
      " [-2.18249321]]\n",
      "ego_vel: [[ 0.22561221  0.75478506 -0.65597218]\n",
      " [ 0.14694074  0.89227808  0.45148623]\n",
      " [ 0.1106957  -0.39912137  0.91689807]\n",
      " [ 0.00444349 -0.66544378  0.74644798]\n",
      " [ 0.00250759 -0.96204859 -0.27287808]\n",
      " [ 0.00250003  0.97177613  0.23590493]\n",
      " [ 0.0025     -0.84159917  0.54010266]\n",
      " [ 0.0025     -0.6746037  -0.7381801 ]\n",
      " [ 0.06948025 -0.90284044 -0.42997569]\n",
      " [ 0.18522677  0.98727757 -0.15900613]\n",
      " [ 0.11738774 -0.47757101  0.87859315]\n",
      " [ 0.15581043  0.99930215  0.03735331]\n",
      " [ 0.13916002 -0.34634149 -0.9381085 ]\n",
      " [ 0.15084584 -0.98530638 -0.1707961 ]\n",
      " [ 0.18935546  0.58143538 -0.81359261]\n",
      " [ 0.09871248 -0.48014092  0.87719136]\n",
      " [ 0.14818396 -0.44695625  0.89455581]\n",
      " [ 0.18594703  0.39950046  0.91673297]\n",
      " [ 0.17088607  0.97972953  0.20032494]\n",
      " [ 0.18265887  0.02400011  0.99971193]\n",
      " [ 0.16458672  0.38841256 -0.9214856 ]\n",
      " [ 0.15000963 -0.92914981 -0.36970338]\n",
      " [ 0.17886183 -0.56241471  0.8268553 ]\n",
      " [ 0.1671187   0.74405634  0.66811687]\n",
      " [ 0.19360736  0.99559814 -0.09372516]\n",
      " [ 0.14706317 -0.57444775 -0.81854129]\n",
      " [ 0.18036179  0.59938109  0.8004638 ]\n",
      " [ 0.18941639  0.06032866  0.99817854]\n",
      " [ 0.14254919 -0.08638027  0.99626225]\n",
      " [ 0.14247712  0.87648052  0.4814373 ]\n",
      " [ 0.17221752 -0.93536639  0.35368025]\n",
      " [ 0.17932069  0.47099745 -0.88213456]\n",
      " [ 0.2026867  -0.2090677  -0.97790116]\n",
      " [ 0.12609276 -0.1101962   0.99390984]\n",
      " [ 0.17138405  0.96556115  0.26017618]\n",
      " [ 0.13020997  0.99873477  0.05028732]\n",
      " [ 0.1412261  -0.86267781 -0.50575393]\n",
      " [ 0.18413493 -0.13359094  0.99103653]\n",
      " [ 0.19115157 -0.27877492  0.96035647]\n",
      " [ 0.17206489 -0.00936235  0.99995619]\n",
      " [ 0.20804456 -0.26937783  0.96303457]\n",
      " [ 0.18343997  0.2267186   0.97396028]\n",
      " [ 0.12743111 -0.94472897  0.3278524 ]\n",
      " [ 0.11895178 -0.1292178  -0.99161625]\n",
      " [ 0.12218695  0.19173712 -0.98144633]\n",
      " [ 0.15905876  0.61962539  0.78489769]\n",
      " [ 0.13428903 -0.08837649  0.99608713]\n",
      " [ 0.11854813 -0.88545537  0.46472439]\n",
      " [ 0.14313602 -0.27459314 -0.96156049]\n",
      " [ 0.1991322   0.9563697   0.29215914]\n",
      " [ 0.13004334  0.58766121 -0.80910712]\n",
      " [ 0.18471584 -0.0682756  -0.99766648]\n",
      " [ 0.14917471 -0.13037564  0.99146467]\n",
      " [ 0.12616149  0.84167272 -0.53998798]\n",
      " [ 0.20141742  0.12079083  0.99267799]\n",
      " [ 0.14690074  0.87007856 -0.49291307]\n",
      " [ 0.15758102 -0.04052253  0.99917865]\n",
      " [ 0.12190241 -0.16439471  0.98639464]\n",
      " [ 0.14114717  0.26732451  0.96360654]\n",
      " [ 0.16518693  0.89868045 -0.438604  ]\n",
      " [ 0.18499425 -0.78325415 -0.62170166]\n",
      " [ 0.1868231  -0.16202733  0.98678625]\n",
      " [ 0.20462577 -0.86366546  0.50406545]\n",
      " [ 0.15256882 -0.99846983  0.05529951]\n",
      " [ 0.17208576 -0.97633296 -0.21627288]\n",
      " [ 0.13762394 -0.51963538  0.85438812]\n",
      " [ 0.19098809 -0.0369237  -0.99931806]\n",
      " [ 0.09605212 -0.94267857 -0.33370209]\n",
      " [ 0.13218114  0.54060233 -0.84127825]\n",
      " [ 0.18259338  0.84530032  0.53429145]\n",
      " [ 0.15089616 -0.82509083  0.56500012]\n",
      " [ 0.2108694  -0.99128872  0.13170697]\n",
      " [ 0.16788182  0.93442273  0.35616583]\n",
      " [ 0.12389984  0.81516385 -0.57923037]\n",
      " [ 0.14548983  0.22132674 -0.9751997 ]\n",
      " [ 0.14553894 -0.73321778  0.67999387]\n",
      " [ 0.17968073  0.56262517 -0.82671213]\n",
      " [ 0.19997676  0.9724645   0.23305097]\n",
      " [ 0.16726585 -0.7029056   0.71128315]\n",
      " [ 0.12137307 -0.99945128  0.03312291]\n",
      " [ 0.14411265 -0.40678793  0.9135226 ]\n",
      " [ 0.11135659 -0.64327103 -0.76563853]\n",
      " [ 0.19674449 -0.9999615   0.00877776]\n",
      " [ 0.14086215 -0.45952991  0.88816231]\n",
      " [ 0.18361235  0.73458952  0.6785118 ]\n",
      " [ 0.16736136 -0.22458309 -0.97445494]\n",
      " [ 0.11014    -0.94106841  0.33821633]\n",
      " [ 0.22119345 -0.92505449 -0.37983435]\n",
      " [ 0.16141056  0.94191951  0.33583868]\n",
      " [ 0.16459626 -0.72618794 -0.68749624]\n",
      " [ 0.18067999 -0.16813111  0.98576462]\n",
      " [ 0.15870436  0.06256977 -0.99804062]\n",
      " [ 0.14059898  0.51881635  0.85488576]\n",
      " [ 0.19989385 -0.23448482 -0.97211981]\n",
      " [ 0.16359557 -0.9783473   0.20696984]\n",
      " [ 0.1601924  -0.91784239  0.396945  ]\n",
      " [ 0.14356463  0.9140479   0.40560624]\n",
      " [ 0.15788403  0.82337493 -0.56749785]\n",
      " [ 0.20549591  0.39508361 -0.91864514]\n",
      " [ 0.15811889 -0.79993588  0.6000855 ]]\n",
      "init_pos: [-0.27034846 -0.26003239]\n",
      "init_hd: [ 1.41786706]\n"
     ]
    }
   ],
   "source": [
    "_, serialized_example = reader.read(filename_queue)\n",
    "\n",
    "feature_map = {\n",
    "        'init_pos':\n",
    "            tf.FixedLenFeature(shape=[2], dtype=tf.float32),\n",
    "        'init_hd':\n",
    "            tf.FixedLenFeature(shape=[1], dtype=tf.float32),\n",
    "        'ego_vel':\n",
    "            tf.FixedLenFeature(\n",
    "                shape=[100, 3],\n",
    "                dtype=tf.float32),\n",
    "        'target_pos':\n",
    "            tf.FixedLenFeature(\n",
    "                shape=[100, 2],\n",
    "                dtype=tf.float32),\n",
    "        'target_hd':\n",
    "            tf.FixedLenFeature(\n",
    "                shape=[100, 1],\n",
    "                dtype=tf.float32),\n",
    "    }\n",
    "\n",
    "read_data = tf.parse_single_example(serialized=serialized_example,\n",
    "                                    features=feature_map)\n",
    "\n",
    "tf.train.start_queue_runners(sess)\n",
    "\n",
    "# Print features\n",
    "for name, tensor in read_data.items():\n",
    "    print('{}: {}'.format(name, tensor.eval()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-0.17160062, -0.16244636, -0.15821284, -0.1641849 , -0.16471101,\n",
       "       -0.17062843, -0.17133029, -0.17155679, -0.17184208, -0.17876536,\n",
       "       -0.19547674, -0.21283603, -0.2256929 , -0.24282892, -0.25904962,\n",
       "       -0.28215125, -0.30966181, -0.33541924, -0.35333151, -0.37279519,\n",
       "       -0.39492521, -0.41110572, -0.43669641, -0.4525272 , -0.47079983,\n",
       "       -0.47213516, -0.48205933, -0.50336748, -0.52002811, -0.54052699,\n",
       "       -0.56658524, -0.58640081, -0.60685253, -0.62643248, -0.64322537,\n",
       "       -0.6581763 , -0.65652871, -0.64224541, -0.62879515, -0.6112954 ,\n",
       "       -0.59258038, -0.57409555, -0.55598348, -0.53933108, -0.51950413,\n",
       "       -0.50969833, -0.50135624, -0.50126737, -0.49905571, -0.49087292,\n",
       "       -0.48253179, -0.46288931, -0.4497551 , -0.43154129, -0.40728655,\n",
       "       -0.38462624, -0.36090463, -0.34236431, -0.31847057, -0.3046087 ,\n",
       "       -0.29466343, -0.28204045, -0.26834753, -0.25575718, -0.24083748,\n",
       "       -0.22793901, -0.21558861, -0.18750502, -0.15820356, -0.13732593,\n",
       "       -0.10678011, -0.07779846, -0.0490781 , -0.02684332, -0.00240542,\n",
       "        0.02219031,  0.04681126,  0.07350826,  0.08986647,  0.11382597,\n",
       "        0.129934  ,  0.15197898,  0.17092368,  0.18595111,  0.20839725,\n",
       "        0.24201508,  0.26227212,  0.28605199,  0.3053439 ,  0.31204039,\n",
       "        0.31623253,  0.32842413,  0.35002887,  0.37635064,  0.39033496,\n",
       "        0.41198239,  0.42847842,  0.442779  ,  0.44926083,  0.45204741], dtype=float32)"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "read_data['target_pos'][:,0].eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for _ in range(10):\n",
    "    read_data = tf.parse_single_example(serialized=serialized_example,\n",
    "                                        features=feature_map)\n",
    "\n",
    "    tf.train.start_queue_runners(sess)\n",
    "    plt.plot(read_data['target_pos'][:,0].eval(), read_data['target_pos'][:,1].eval())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.9987786546592783"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "target_pos = read_data['target_pos'].eval()\n",
    "path_length = 0\n",
    "for i in range(99):\n",
    "    path_length += np.sqrt(np.abs(target_pos[:,0][i]-target_pos[:,0][i+1])**2 + np.abs(target_pos[:,1][i]-target_pos[:,1][i+1])**2)\n",
    "path_length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.1215554925221793"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "path_length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "_, raw_data = reader.read_up_to(filename_queue, num_records=64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "example = tf.parse_example(serialized=raw_data, features=feature_map)\n",
    "batch = [\n",
    "    example['init_pos'], example['init_hd'],\n",
    "    example['ego_vel'][:, :100, :],\n",
    "    example['target_pos'][:, :100, :],\n",
    "    example['target_hd'][:, :100, :]\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "nest = tf.contrib.framework.nest\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# tf.train.start_queue_runners(sess)\n",
    "\n",
    "dtypes = nest.map_structure(lambda x: x.dtype, batch)\n",
    "shapes = nest.map_structure(lambda x: x.shape[1:], batch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "_queue = tf.RandomShuffleQueue(\n",
    "          capacity=256,\n",
    "          min_after_dequeue=128,\n",
    "          dtypes=dtypes,\n",
    "          shapes=shapes,\n",
    "          seed=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "in_pos, in_hd, ego_vel, target_pos, target_hd = _queue.dequeue()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "import dataset_reader \n",
    "import tensorflow as tf\n",
    "\n",
    "# Task config\n",
    "tf.flags.DEFINE_string('task_dataset_info', 'square_room',\n",
    "                       'Name of the room in which the experiment is performed.')\n",
    "tf.flags.DEFINE_string('task_root',\n",
    "                       None,\n",
    "                       'Dataset path.')\n",
    "tf.flags.DEFINE_float('task_env_size', 2.2,\n",
    "                      'Environment size (meters).')\n",
    "tf.flags.DEFINE_list('task_n_pc', [256],\n",
    "                     'Number of target place cells.')\n",
    "tf.flags.DEFINE_list('task_pc_scale', [0.01],\n",
    "                     'Place cell standard deviation parameter (meters).')\n",
    "tf.flags.DEFINE_list('task_n_hdc', [12],\n",
    "                     'Number of target head direction cells.')\n",
    "tf.flags.DEFINE_list('task_hdc_concentration', [20.],\n",
    "                     'Head direction concentration parameter.')\n",
    "tf.flags.DEFINE_integer('task_neurons_seed', 8341,\n",
    "                        'Seeds.')\n",
    "tf.flags.DEFINE_string('task_targets_type', 'softmax',\n",
    "                       'Type of target, soft or hard.')\n",
    "tf.flags.DEFINE_string('task_lstm_init_type', 'softmax',\n",
    "                       'Type of LSTM initialisation, soft or hard.')\n",
    "tf.flags.DEFINE_bool('task_velocity_inputs', True,\n",
    "                     'Input velocity.')\n",
    "tf.flags.DEFINE_list('task_velocity_noise', [0.0, 0.0, 0.0],\n",
    "                     'Add noise to velocity.')\n",
    "\n",
    "# Model config\n",
    "tf.flags.DEFINE_integer('model_nh_lstm', 128, 'Number of hidden units in LSTM.')\n",
    "tf.flags.DEFINE_integer('model_nh_bottleneck', 256,\n",
    "                        'Number of hidden units in linear bottleneck.')\n",
    "tf.flags.DEFINE_list('model_dropout_rates', [0.5],\n",
    "                     'List of floats with dropout rates.')\n",
    "tf.flags.DEFINE_float('model_weight_decay', 1e-5,\n",
    "                      'Weight decay regularisation')\n",
    "tf.flags.DEFINE_bool('model_bottleneck_has_bias', False,\n",
    "                     'Whether to include a bias in linear bottleneck')\n",
    "tf.flags.DEFINE_float('model_init_weight_disp', 0.0,\n",
    "                      'Initial weight displacement.')\n",
    "\n",
    "# Training config\n",
    "tf.flags.DEFINE_integer('training_epochs', 1000, 'Number of training epochs.')\n",
    "tf.flags.DEFINE_integer('training_steps_per_epoch', 1000,\n",
    "                        'Number of optimization steps per epoch.')\n",
    "tf.flags.DEFINE_integer('training_minibatch_size', 100,\n",
    "                        'Size of the training minibatch.')\n",
    "tf.flags.DEFINE_integer('training_evaluation_minibatch_size', 4000,\n",
    "                        'Size of the minibatch during evaluation.')\n",
    "tf.flags.DEFINE_string('training_clipping_function', 'utils.clip_all_gradients',\n",
    "                       'Function for gradient clipping.')\n",
    "tf.flags.DEFINE_float('training_clipping', 1e-5,\n",
    "                      'The absolute value to clip by.')\n",
    "\n",
    "tf.flags.DEFINE_string('training_optimizer_class', 'tf.train.RMSPropOptimizer',\n",
    "                       'The optimizer used for training.')\n",
    "tf.flags.DEFINE_string('training_optimizer_options',\n",
    "                       '{\"learning_rate\": 1e-5, \"momentum\": 0.9}',\n",
    "                       'Defines a dict with opts passed to the optimizer.')\n",
    "\n",
    "# Store\n",
    "tf.flags.DEFINE_string('saver_results_directory',\n",
    "                       None,\n",
    "                       'Path to directory for saving results.')\n",
    "tf.flags.DEFINE_integer('saver_eval_time', 20,\n",
    "                        'Frequency at which results are saved.')\n",
    "\n",
    "# Require flags\n",
    "tf.flags.mark_flag_as_required('task_root')\n",
    "tf.flags.mark_flag_as_required('saver_results_directory')\n",
    "FLAGS = tf.flags.FLAGS\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_reader = dataset_reader.DataReader(\n",
    "      'square_room', root='./datasets', num_threads=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_traj = data_reader.read(batch_size=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(<tf.Tensor 'random_shuffle_queue_2_DequeueMany:0' shape=(10, 2) dtype=float32>,\n",
       " <tf.Tensor 'random_shuffle_queue_2_DequeueMany:1' shape=(10, 1) dtype=float32>,\n",
       " <tf.Tensor 'random_shuffle_queue_2_DequeueMany:2' shape=(10, 100, 3) dtype=float32>,\n",
       " <tf.Tensor 'random_shuffle_queue_2_DequeueMany:3' shape=(10, 100, 2) dtype=float32>,\n",
       " <tf.Tensor 'random_shuffle_queue_2_DequeueMany:4' shape=(10, 100, 1) dtype=float32>)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_traj"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "init_pos, init_hd, ego_vel, target_pos, target_hd = train_traj"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    label_numpy = init_pos.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "ename": "AttributeError",
     "evalue": "'Tensor' object has no attribute 'numpy'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0mTraceback (most recent call last)",
      "\u001b[0;32m<ipython-input-31-0ac4c422d313>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0minit_pos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0m: 'Tensor' object has no attribute 'numpy'"
     ]
    }
   ],
   "source": [
    "init_pos.numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    " print(type(tf.Session().run(init_pos)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}