Keras Ch 6 Recurrent Temp Forecasting - Multi GPU.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#! wget https://s3.amazonaws.com/keras-datasets/jena_climate_2009_2016.csv.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#! unzip jena_climate_2009_2016.csv.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from util import print_curves"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Inspect Jena Weather Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "\n",
    "data_dir = './Downloads/'\n",
    "fname = os.path.join(data_dir, 'jena_climate_2009_2016.csv')\n",
    "df = pd.read_csv(fname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Date Time</th>\n",
       "      <th>p (mbar)</th>\n",
       "      <th>T (degC)</th>\n",
       "      <th>Tpot (K)</th>\n",
       "      <th>Tdew (degC)</th>\n",
       "      <th>rh (%)</th>\n",
       "      <th>VPmax (mbar)</th>\n",
       "      <th>VPact (mbar)</th>\n",
       "      <th>VPdef (mbar)</th>\n",
       "      <th>sh (g/kg)</th>\n",
       "      <th>H2OC (mmol/mol)</th>\n",
       "      <th>rho (g/m**3)</th>\n",
       "      <th>wv (m/s)</th>\n",
       "      <th>max. wv (m/s)</th>\n",
       "      <th>wd (deg)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>01.01.2009 00:10:00</td>\n",
       "      <td>996.52</td>\n",
       "      <td>-8.02</td>\n",
       "      <td>265.40</td>\n",
       "      <td>-8.90</td>\n",
       "      <td>93.3</td>\n",
       "      <td>3.33</td>\n",
       "      <td>3.11</td>\n",
       "      <td>0.22</td>\n",
       "      <td>1.94</td>\n",
       "      <td>3.12</td>\n",
       "      <td>1307.75</td>\n",
       "      <td>1.03</td>\n",
       "      <td>1.75</td>\n",
       "      <td>152.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>01.01.2009 00:20:00</td>\n",
       "      <td>996.57</td>\n",
       "      <td>-8.41</td>\n",
       "      <td>265.01</td>\n",
       "      <td>-9.28</td>\n",
       "      <td>93.4</td>\n",
       "      <td>3.23</td>\n",
       "      <td>3.02</td>\n",
       "      <td>0.21</td>\n",
       "      <td>1.89</td>\n",
       "      <td>3.03</td>\n",
       "      <td>1309.80</td>\n",
       "      <td>0.72</td>\n",
       "      <td>1.50</td>\n",
       "      <td>136.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>01.01.2009 00:30:00</td>\n",
       "      <td>996.53</td>\n",
       "      <td>-8.51</td>\n",
       "      <td>264.91</td>\n",
       "      <td>-9.31</td>\n",
       "      <td>93.9</td>\n",
       "      <td>3.21</td>\n",
       "      <td>3.01</td>\n",
       "      <td>0.20</td>\n",
       "      <td>1.88</td>\n",
       "      <td>3.02</td>\n",
       "      <td>1310.24</td>\n",
       "      <td>0.19</td>\n",
       "      <td>0.63</td>\n",
       "      <td>171.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>01.01.2009 00:40:00</td>\n",
       "      <td>996.51</td>\n",
       "      <td>-8.31</td>\n",
       "      <td>265.12</td>\n",
       "      <td>-9.07</td>\n",
       "      <td>94.2</td>\n",
       "      <td>3.26</td>\n",
       "      <td>3.07</td>\n",
       "      <td>0.19</td>\n",
       "      <td>1.92</td>\n",
       "      <td>3.08</td>\n",
       "      <td>1309.19</td>\n",
       "      <td>0.34</td>\n",
       "      <td>0.50</td>\n",
       "      <td>198.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>01.01.2009 00:50:00</td>\n",
       "      <td>996.51</td>\n",
       "      <td>-8.27</td>\n",
       "      <td>265.15</td>\n",
       "      <td>-9.04</td>\n",
       "      <td>94.1</td>\n",
       "      <td>3.27</td>\n",
       "      <td>3.08</td>\n",
       "      <td>0.19</td>\n",
       "      <td>1.92</td>\n",
       "      <td>3.09</td>\n",
       "      <td>1309.00</td>\n",
       "      <td>0.32</td>\n",
       "      <td>0.63</td>\n",
       "      <td>214.3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Date Time  p (mbar)  T (degC)  Tpot (K)  Tdew (degC)  rh (%)  \\\n",
       "0  01.01.2009 00:10:00    996.52     -8.02    265.40        -8.90    93.3   \n",
       "1  01.01.2009 00:20:00    996.57     -8.41    265.01        -9.28    93.4   \n",
       "2  01.01.2009 00:30:00    996.53     -8.51    264.91        -9.31    93.9   \n",
       "3  01.01.2009 00:40:00    996.51     -8.31    265.12        -9.07    94.2   \n",
       "4  01.01.2009 00:50:00    996.51     -8.27    265.15        -9.04    94.1   \n",
       "\n",
       "   VPmax (mbar)  VPact (mbar)  VPdef (mbar)  sh (g/kg)  H2OC (mmol/mol)  \\\n",
       "0          3.33          3.11          0.22       1.94             3.12   \n",
       "1          3.23          3.02          0.21       1.89             3.03   \n",
       "2          3.21          3.01          0.20       1.88             3.02   \n",
       "3          3.26          3.07          0.19       1.92             3.08   \n",
       "4          3.27          3.08          0.19       1.92             3.09   \n",
       "\n",
       "   rho (g/m**3)  wv (m/s)  max. wv (m/s)  wd (deg)  \n",
       "0       1307.75      1.03           1.75     152.3  \n",
       "1       1309.80      0.72           1.50     136.1  \n",
       "2       1310.24      0.19           0.63     171.6  \n",
       "3       1309.19      0.34           0.50     198.0  \n",
       "4       1309.00      0.32           0.63     214.3  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f85da2a6a10>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f864c1be8d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "df.plot('Date Time','T (degC)')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f85da16a490>]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f85da252510>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(1440),df['T (degC)'][:1440])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Preprocess Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import preprocessing\n",
    "\n",
    "#df = preprocessing.scale(df.drop('Date Time', axis=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_time = df\n",
    "df = df.drop('Date Time', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean = df.mean(0)\n",
    "std = df.std(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# standardize data\n",
    "df = df.sub(mean, axis=1).div(std, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "p (mbar)           1.0\n",
       "T (degC)           1.0\n",
       "Tpot (K)           1.0\n",
       "Tdew (degC)        1.0\n",
       "rh (%)             1.0\n",
       "VPmax (mbar)       1.0\n",
       "VPact (mbar)       1.0\n",
       "VPdef (mbar)       1.0\n",
       "sh (g/kg)          1.0\n",
       "H2OC (mmol/mol)    1.0\n",
       "rho (g/m**3)       1.0\n",
       "wv (m/s)           1.0\n",
       "max. wv (m/s)      1.0\n",
       "wd (deg)           1.0\n",
       "dtype: float64"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.std(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "p (mbar)          -3.526411e-13\n",
       "T (degC)           6.469275e-15\n",
       "Tpot (K)           4.183824e-13\n",
       "Tdew (degC)       -1.180013e-15\n",
       "rh (%)             2.505243e-14\n",
       "VPmax (mbar)      -1.337905e-13\n",
       "VPact (mbar)       7.613620e-15\n",
       "VPdef (mbar)      -5.937364e-14\n",
       "sh (g/kg)          4.250040e-14\n",
       "H2OC (mmol/mol)   -4.088514e-14\n",
       "rho (g/m**3)      -2.247232e-13\n",
       "wv (m/s)          -2.047628e-16\n",
       "max. wv (m/s)     -1.352686e-15\n",
       "wd (deg)          -2.302473e-14\n",
       "dtype: float64"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.mean(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create generator based on lookback, steps delay"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "lookback - many timesteps back the input data should go\n",
    "\n",
    "delay - how many timesteps in the future the target should be\n",
    "\n",
    "step - the period, in timesteps, at which you should sample data, in this case 6 for data every hour\n",
    "\n",
    "min-index and max-index - helpful for keeping validation and testing separate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "\n",
    "def generator(data, lookback, delay, min_index, max_index,\n",
    "              shuffle=False, batch_size=128, step=6,\n",
    "              reverse=False):\n",
    "    if max_index is None:\n",
    "        max_index = len(data) - delay - 1\n",
    "    i = min_index + lookback\n",
    "\n",
    "    # create offset so we don't see the same 1/step th of data in non-shuffled\n",
    "    # scenario\n",
    "    offset = 1\n",
    "\n",
    "    while True:\n",
    "        if shuffle:\n",
    "            rows = np.random.randint(\n",
    "                min_index + lookback, max_index, size=batch_size)\n",
    "        else:\n",
    "            if i + batch_size >= max_index:\n",
    "                # won't this reset the generator to the same place, esentially not giving us 1/6 of the data?\n",
    "                # might be somewhat trivial since we're taking hourly samples,\n",
    "                # so data only has changed by an hour and we're reading across\n",
    "                # 8 years total dataset\n",
    "                i = min_index + lookback + offset\n",
    "                offset += 1\n",
    "                if offset == step:\n",
    "                    offset = 0\n",
    "            rows = np.arange(i, min(i + batch_size, max_index))\n",
    "            i += len(rows)\n",
    "        samples = np.zeros((len(rows),\n",
    "                            lookback // step,\n",
    "                            data.shape[-1]))\n",
    "        targets = np.zeros((len(rows),))\n",
    "        for j, row in enumerate(rows):\n",
    "            indices = range(rows[j] - lookback, rows[j], step)\n",
    "            samples[j] = data[indices]\n",
    "            targets[j] = data[rows[j] + delay][1]\n",
    "        if reverse == True:\n",
    "            yield samples[:, ::-1, :], targets\n",
    "        else:\n",
    "            yield samples, targets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "lookback = 1440\n",
    "step = 6\n",
    "delay = 144\n",
    "batch_size = 128\n",
    "num_features = df.shape[1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create 3 generators that look at different segments of the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_gen = generator(\n",
    "    df.values,\n",
    "    lookback=lookback,\n",
    "    delay=delay,\n",
    "    min_index=0,\n",
    "    max_index=200000,\n",
    "    shuffle=True,\n",
    "    step=step,\n",
    "    batch_size=batch_size)\n",
    "\n",
    "val_gen = generator(\n",
    "    df.values,\n",
    "    lookback=lookback,\n",
    "    delay=delay,\n",
    "    min_index=200001,\n",
    "    max_index=300000,\n",
    "    step=step,\n",
    "    batch_size=batch_size)\n",
    "\n",
    "test_gen = generator(\n",
    "    df.values,\n",
    "    lookback=lookback,\n",
    "    delay=delay,\n",
    "    min_index=300001,\n",
    "    max_index=None,\n",
    "    step=step,\n",
    "    batch_size=batch_size)\n",
    "\n",
    "# how many train steps to see entire dataset\n",
    "train_steps = (200000 - lookback) // batch_size\n",
    "\n",
    "# how many val steps to see entire dataset\n",
    "val_steps = (300000 - 200001 - lookback) // batch_size\n",
    "\n",
    "# how many test steps to see entire dataset\n",
    "test_steps = (len(df) - 300001 - lookback) // batch_size\n",
    "\n",
    "# need to check if we're processing the cursor at all, i.e. are we only\n",
    "# seeing 1/6 of the data since we're sampling every 6 hours\n",
    "# Actually, we're shuffling, so that should take care of it, but only for\n",
    "# training, in theory you want to run through 6 times, offestting,\n",
    "# something like that"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Try Naive Method - Predict temp is same as 24 hours previous"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('naive mae: ', 0.30031224152324865)\n",
      "('Celcius Degress Error:', 2.5101540737249497)\n"
     ]
    }
   ],
   "source": [
    "def evaluate_naive_method():\n",
    "    batch_maes = []\n",
    "    # going through ALL steps was taking a long time, value is almost the same as printed\n",
    "    #for step in range(val_steps):\n",
    "    for step in range(500):\n",
    "        samples, targets = next(val_gen)\n",
    "        preds = samples[:,-1,1]\n",
    "        mae = np.abs(preds - targets)\n",
    "        batch_maes.append(mae)\n",
    "    return np.mean(batch_maes)\n",
    "\n",
    "naive_mae = evaluate_naive_method()\n",
    "print('naive mae: ', naive_mae)\n",
    "print('Celcius Degress Error:', naive_mae * std[0])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So we're off by 2.54 degrees Celsius"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Try Easier MLP Method - multi GPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten_5 (Flatten)          (None, 3360)              0         \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 32)                107552    \n",
      "_________________________________________________________________\n",
      "dense_10 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 107,585\n",
      "Trainable params: 107,585\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Flatten, Dense\n",
    "\n",
    "\n",
    "model = Sequential()\n",
    "model.add(Flatten(input_shape=(lookback // step, num_features)))\n",
    "model.add(Dense(32, activation= 'relu'))\n",
    "model.add(Dense(1))\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/3\n",
      "1551/1551 [==============================] - 13s 9ms/step - loss: 2.1068 - mean_absolute_error: 0.7237 - val_loss: 0.1939 - val_mean_absolute_error: 0.3477\n",
      "Epoch 2/3\n",
      "1551/1551 [==============================] - 14s 9ms/step - loss: 0.1344 - mean_absolute_error: 0.2866 - val_loss: 0.1732 - val_mean_absolute_error: 0.3280\n",
      "Epoch 3/3\n",
      "1551/1551 [==============================] - 14s 9ms/step - loss: 0.1027 - mean_absolute_error: 0.2519 - val_loss: 0.1975 - val_mean_absolute_error: 0.3541\n",
      "CPU times: user 1min 11s, sys: 7.82 s, total: 1min 19s\n",
      "Wall time: 41 s\n",
      "Min Val MAE: 0.328047437736\n",
      "Min Val Loss: 0.173184163891\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f85a00c0dd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f85a0166d90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model.compile(optimizer='rmsprop', loss='mse', metrics=['mae'])\n",
    "\n",
    "%time history = model.fit_generator(train_gen, epochs=3, steps_per_epoch=train_steps, validation_steps=val_steps, validation_data=val_gen)\n",
    "\n",
    "print_curves(history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten_4 (Flatten)          (None, 3360)              0         \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 32)                107552    \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 107,585\n",
      "Trainable params: 107,585\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from keras.utils import multi_gpu_model\n",
    "\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Flatten, Dense\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(Flatten(input_shape=(lookback // step, num_features)))\n",
    "    model.add(Dense(32, activation= 'relu'))\n",
    "    model.add(Dense(1))\n",
    "\n",
    "    model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/3\n",
      "1551/1551 [==============================] - 19s 13ms/step - loss: 2.0235 - mean_absolute_error: 0.6917 - val_loss: 0.2114 - val_mean_absolute_error: 0.3625\n",
      "Epoch 2/3\n",
      "1551/1551 [==============================] - 19s 12ms/step - loss: 0.1327 - mean_absolute_error: 0.2861 - val_loss: 0.1748 - val_mean_absolute_error: 0.3304\n",
      "Epoch 3/3\n",
      "1551/1551 [==============================] - 19s 12ms/step - loss: 0.1044 - mean_absolute_error: 0.2537 - val_loss: 0.1771 - val_mean_absolute_error: 0.3319\n",
      "CPU times: user 2min 23s, sys: 17.4 s, total: 2min 41s\n",
      "Wall time: 57.7 s\n",
      "Min Val MAE: 0.330429912088\n",
      "Min Val Loss: 0.174796134902\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f85ac15fd10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f85a01346d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer='rmsprop', loss='mse', metrics=['mae'])\n",
    "\n",
    "%time history = parallel_model.fit_generator(train_gen, epochs=3, steps_per_epoch=train_steps, validation_steps=val_steps, validation_data=val_gen)\n",
    "\n",
    "print_curves(history)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0,0.5,u'seconds 3 epochs')"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f85ac1d7410>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "p100_gpus = [1, 2, 4]\n",
    "times = [41, 48.1, 57.7]\n",
    "\n",
    "plt.plot(p100_gpus, times)\n",
    "plt.xlabel('num_gpus')\n",
    "plt.ylabel('seconds 3 epochs')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Basically the opposite of what we want. Either it's not splitting generator or too simple of a model to benefit from parallelization."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We're not really outperforming Naive. We have large hypothesis space, but it's failing to find simple solution to simple problem."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### First Recurrent Baseline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "gru_1 (GRU)                  (None, 32)                4512      \n",
      "_________________________________________________________________\n",
      "dense_11 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 4,545\n",
      "Trainable params: 4,545\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.layers import GRU\n",
    "\n",
    "model = Sequential()\n",
    "model.add(GRU(32, input_shape=(lookback // step, df.shape[1])))\n",
    "model.add(Dense(1))\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      " 44/500 [=>............................] - ETA: 1:50 - loss: 0.4543 - mean_absolute_error: 0.4543"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-40-9ff2a6d95f80>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m history = model.fit_generator(train_gen, epochs = 20, steps_per_epoch = 500,\n\u001b[0;32m----> 4\u001b[0;31m                              validation_data = val_gen, validation_steps = val_steps)\n\u001b[0m\u001b[1;32m      5\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0mprint_curves\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhistory\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/models.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   1225\u001b[0m                                         \u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1226\u001b[0m                                         \u001b[0mshuffle\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mshuffle\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1227\u001b[0;31m                                         initial_epoch=initial_epoch)\n\u001b[0m\u001b[1;32m   1228\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1229\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0minterfaces\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlegacy_generator_methods_support\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   2145\u001b[0m                     outs = self.train_on_batch(x, y,\n\u001b[1;32m   2146\u001b[0m                                                \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2147\u001b[0;31m                                                class_weight=class_weight)\n\u001b[0m\u001b[1;32m   2148\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2149\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mtrain_on_batch\u001b[0;34m(self, x, y, sample_weight, class_weight)\u001b[0m\n\u001b[1;32m   1837\u001b[0m             \u001b[0mins\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1838\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_train_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1839\u001b[0;31m         \u001b[0moutputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1840\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1841\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0moutputs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/backend/tensorflow_backend.pyc\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2355\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2356\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2357\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2358\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2359\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    887\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    888\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 889\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    890\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    891\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1118\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1119\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1120\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1121\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1122\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1315\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1316\u001b[0m       return self._do_call(_run_fn, self._session, feeds, fetches, targets,\n\u001b[0;32m-> 1317\u001b[0;31m                            options, run_metadata)\n\u001b[0m\u001b[1;32m   1318\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1319\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1321\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1322\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1323\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1324\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1325\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(session, feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1300\u001b[0m           return tf_session.TF_Run(session, options,\n\u001b[1;32m   1301\u001b[0m                                    \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1302\u001b[0;31m                                    status, run_metadata)\n\u001b[0m\u001b[1;32m   1303\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1304\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "model.compile(optimizer='rmsprop', loss='mae', metrics=['mae'])\n",
    "\n",
    "history = model.fit_generator(train_gen, epochs = 20, steps_per_epoch = 500,\n",
    "                             validation_data = val_gen, validation_steps = val_steps)\n",
    "\n",
    "print_curves(history)\n",
    "print(np.min(history.history['val_mean_absolute_error']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "gru_4 (GRU)                  (None, 32)                4512      \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 4,545\n",
      "Trainable params: 4,545\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.layers import GRU\n",
    "\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(GRU(32, input_shape=(lookback // step, df.shape[1])))\n",
    "    model.add(Dense(1))\n",
    "\n",
    "    model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      " 29/500 [>.............................] - ETA: 4:14 - loss: 0.4829 - mean_absolute_error: 0.4829"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-48-38b3c7598074>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m history = parallel_model.fit_generator(train_gen, epochs = 20, steps_per_epoch = 500,\n\u001b[0;32m----> 6\u001b[0;31m                              validation_data = val_gen, validation_steps = val_steps)\n\u001b[0m\u001b[1;32m      7\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0mprint_curves\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhistory\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   2145\u001b[0m                     outs = self.train_on_batch(x, y,\n\u001b[1;32m   2146\u001b[0m                                                \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2147\u001b[0;31m                                                class_weight=class_weight)\n\u001b[0m\u001b[1;32m   2148\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2149\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mtrain_on_batch\u001b[0;34m(self, x, y, sample_weight, class_weight)\u001b[0m\n\u001b[1;32m   1837\u001b[0m             \u001b[0mins\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1838\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_train_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1839\u001b[0;31m         \u001b[0moutputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1840\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1841\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0moutputs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/backend/tensorflow_backend.pyc\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2355\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2356\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2357\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2358\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2359\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    887\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    888\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 889\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    890\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    891\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1118\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1119\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1120\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1121\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1122\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1315\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1316\u001b[0m       return self._do_call(_run_fn, self._session, feeds, fetches, targets,\n\u001b[0;32m-> 1317\u001b[0;31m                            options, run_metadata)\n\u001b[0m\u001b[1;32m   1318\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1319\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1321\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1322\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1323\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1324\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1325\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(session, feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1300\u001b[0m           return tf_session.TF_Run(session, options,\n\u001b[1;32m   1301\u001b[0m                                    \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1302\u001b[0;31m                                    status, run_metadata)\n\u001b[0m\u001b[1;32m   1303\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1304\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer='rmsprop', loss='mae', metrics=['mae'])\n",
    "\n",
    "history = parallel_model.fit_generator(train_gen, epochs = 20, steps_per_epoch = 500,\n",
    "                             validation_data = val_gen, validation_steps = val_steps)\n",
    "\n",
    "print_curves(history)\n",
    "print(np.min(history.history['val_mean_absolute_error']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_val_mae = np.min(history.history['val_mean_absolute_error'])\n",
    "print(best_val_mae)\n",
    "print(best_val_mae * std[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Val MAE has improved quite a bit, but we're overfitting"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add Recurrent Dropout"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Need to add double the training steps more or less because of dropout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "500/500 [==============================] - 204s 407ms/step - loss: 0.3580 - mean_absolute_error: 0.3580 - val_loss: 0.2895 - val_mean_absolute_error: 0.2895\n",
      "Epoch 2/20\n",
      "500/500 [==============================] - 191s 382ms/step - loss: 0.3324 - mean_absolute_error: 0.3324 - val_loss: 0.2879 - val_mean_absolute_error: 0.2879\n",
      "Epoch 3/20\n",
      "500/500 [==============================] - 202s 404ms/step - loss: 0.3274 - mean_absolute_error: 0.3274 - val_loss: 0.2853 - val_mean_absolute_error: 0.2853\n",
      "Epoch 4/20\n",
      "500/500 [==============================] - 196s 391ms/step - loss: 0.3218 - mean_absolute_error: 0.3218 - val_loss: 0.2817 - val_mean_absolute_error: 0.2817\n",
      "Epoch 5/20\n",
      "499/500 [============================>.] - ETA: 0s - loss: 0.3178 - mean_absolute_error: 0.3178"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-54-959eb6496955>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     10\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m history = model.fit_generator(train_gen, epochs = 20, steps_per_epoch = 500,\n\u001b[0;32m---> 12\u001b[0;31m                              validation_data = val_gen, validation_steps = val_steps)\n\u001b[0m\u001b[1;32m     13\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     14\u001b[0m \u001b[0mprint_curves\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhistory\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/models.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   1225\u001b[0m                                         \u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1226\u001b[0m                                         \u001b[0mshuffle\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mshuffle\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1227\u001b[0;31m                                         initial_epoch=initial_epoch)\n\u001b[0m\u001b[1;32m   1228\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1229\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0minterfaces\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlegacy_generator_methods_support\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   2167\u001b[0m                                 \u001b[0mmax_queue_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmax_queue_size\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2168\u001b[0m                                 \u001b[0mworkers\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mworkers\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2169\u001b[0;31m                                 use_multiprocessing=use_multiprocessing)\n\u001b[0m\u001b[1;32m   2170\u001b[0m                         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2171\u001b[0m                             \u001b[0;31m# No need for try/except because\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mevaluate_generator\u001b[0;34m(self, generator, steps, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   2294\u001b[0m                                      \u001b[0;34m'or (x, y). Found: '\u001b[0m \u001b[0;34m+\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2295\u001b[0m                                      str(generator_output))\n\u001b[0;32m-> 2296\u001b[0;31m                 \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtest_on_batch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2297\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2298\u001b[0m                 \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mtest_on_batch\u001b[0;34m(self, x, y, sample_weight)\u001b[0m\n\u001b[1;32m   1879\u001b[0m             \u001b[0mins\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1880\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_test_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1881\u001b[0;31m         \u001b[0moutputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtest_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1882\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1883\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0moutputs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/backend/tensorflow_backend.pyc\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2355\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2356\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2357\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2358\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2359\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    887\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    888\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 889\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    890\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    891\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1118\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1119\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1120\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1121\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1122\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1315\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1316\u001b[0m       return self._do_call(_run_fn, self._session, feeds, fetches, targets,\n\u001b[0;32m-> 1317\u001b[0;31m                            options, run_metadata)\n\u001b[0m\u001b[1;32m   1318\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1319\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1321\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1322\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1323\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1324\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1325\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(session, feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1300\u001b[0m           return tf_session.TF_Run(session, options,\n\u001b[1;32m   1301\u001b[0m                                    \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1302\u001b[0;31m                                    status, run_metadata)\n\u001b[0m\u001b[1;32m   1303\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1304\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(GRU(32,\n",
    "             dropout=0.2,\n",
    "             recurrent_dropout=0.2,\n",
    "             # why can we do none?\n",
    "             input_shape=(None, df.shape[1])))\n",
    "model.add(Dense(1))\n",
    "\n",
    "model.compile(optimizer='rmsprop', loss='mae', metrics=['mae'])\n",
    "\n",
    "history = model.fit_generator(train_gen, epochs = 20, steps_per_epoch = 500,\n",
    "                             validation_data = val_gen, validation_steps = val_steps)\n",
    "\n",
    "print_curves(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Stack Recurrent Layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "371/500 [=====================>........] - ETA: 1:21 - loss: 0.3482 - mean_absolute_error: 0.3482"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-104-8c724e696edf>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     14\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m history = model.fit_generator(train_gen, epochs = 40, steps_per_epoch = 500,\n\u001b[0;32m---> 16\u001b[0;31m                              validation_data = val_gen, validation_steps = val_steps)\n\u001b[0m\u001b[1;32m     17\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m \u001b[0mprint_curves\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhistory\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/models.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   1225\u001b[0m                                         \u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1226\u001b[0m                                         \u001b[0mshuffle\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mshuffle\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1227\u001b[0;31m                                         initial_epoch=initial_epoch)\n\u001b[0m\u001b[1;32m   1228\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1229\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0minterfaces\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlegacy_generator_methods_support\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   2145\u001b[0m                     outs = self.train_on_batch(x, y,\n\u001b[1;32m   2146\u001b[0m                                                \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2147\u001b[0;31m                                                class_weight=class_weight)\n\u001b[0m\u001b[1;32m   2148\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2149\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mtrain_on_batch\u001b[0;34m(self, x, y, sample_weight, class_weight)\u001b[0m\n\u001b[1;32m   1837\u001b[0m             \u001b[0mins\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1838\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_train_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1839\u001b[0;31m         \u001b[0moutputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1840\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1841\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0moutputs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/backend/tensorflow_backend.pyc\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2355\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2356\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2357\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2358\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2359\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    887\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    888\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 889\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    890\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    891\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1118\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1119\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1120\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1121\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1122\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1315\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1316\u001b[0m       return self._do_call(_run_fn, self._session, feeds, fetches, targets,\n\u001b[0;32m-> 1317\u001b[0;31m                            options, run_metadata)\n\u001b[0m\u001b[1;32m   1318\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1319\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1321\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1322\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1323\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1324\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1325\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(session, feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1300\u001b[0m           return tf_session.TF_Run(session, options,\n\u001b[1;32m   1301\u001b[0m                                    \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1302\u001b[0;31m                                    status, run_metadata)\n\u001b[0m\u001b[1;32m   1303\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1304\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(GRU(32,\n",
    "             dropout=0.1,\n",
    "             recurrent_dropout=0.5,\n",
    "             return_sequences = True,\n",
    "             # why can we do none?\n",
    "             input_shape=(None, df.shape[1])))\n",
    "model.add(GRU(64, activation='relu',\n",
    "             dropout=0.1,\n",
    "             recurrent_dropout=0.5))\n",
    "model.add(Dense(1))\n",
    "\n",
    "model.compile(optimizer='rmsprop', loss='mae', metrics=['mae'])\n",
    "\n",
    "history = model.fit_generator(train_gen, epochs = 40, steps_per_epoch = 500,\n",
    "                             validation_data = val_gen, validation_steps = val_steps)\n",
    "\n",
    "print_curves(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have improved a little bit, but we're bottlenecking. We're no longer overfitting, so perhaps we need a larger network for larger hypothesis space."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bidirectional RNNs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_gen_reverse = generator(\n",
    "    df.values,\n",
    "    lookback=lookback,\n",
    "    delay=delay,\n",
    "    min_index=0,\n",
    "    max_index=200000,\n",
    "    shuffle=True,\n",
    "    step=step,\n",
    "    batch_size=batch_size,\n",
    "    reverse=True)\n",
    "\n",
    "val_gen_reverse = generator(\n",
    "    df.values,\n",
    "    lookback=lookback,\n",
    "    delay=delay,\n",
    "    min_index=200001,\n",
    "    max_index=300000,\n",
    "    step=step,\n",
    "    batch_size=batch_size,\n",
    "    reverse=True)\n",
    "\n",
    "test_gen_reverse = generator(\n",
    "    df.values,\n",
    "    lookback=lookback,\n",
    "    delay=delay,\n",
    "    min_index=300001,\n",
    "    max_index=None,\n",
    "    step=step,\n",
    "    batch_size=batch_size,\n",
    "    reverse=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "gru_15 (GRU)                 (None, 32)                4512      \n",
      "_________________________________________________________________\n",
      "dense_19 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 4,545\n",
      "Trainable params: 4,545\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/20\n",
      "  8/500 [..............................] - ETA: 3:33 - loss: 0.6307 - mean_absolute_error: 0.6307"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-127-9b4063ea96a9>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m history = model.fit_generator(train_gen_reverse, epochs = 20, steps_per_epoch = 500,\n\u001b[0;32m---> 10\u001b[0;31m                              validation_data = val_gen_reverse, validation_steps = val_steps)\n\u001b[0m\u001b[1;32m     11\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0mprint_curves\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhistory\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/models.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   1225\u001b[0m                                         \u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0muse_multiprocessing\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1226\u001b[0m                                         \u001b[0mshuffle\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mshuffle\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1227\u001b[0;31m                                         initial_epoch=initial_epoch)\n\u001b[0m\u001b[1;32m   1228\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1229\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0minterfaces\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlegacy_generator_methods_support\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/legacy/interfaces.pyc\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     85\u001b[0m                 warnings.warn('Update your `' + object_name +\n\u001b[1;32m     86\u001b[0m                               '` call to the Keras 2 API: ' + signature, stacklevel=2)\n\u001b[0;32m---> 87\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     88\u001b[0m         \u001b[0mwrapper\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_original_function\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     89\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mwrapper\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)\u001b[0m\n\u001b[1;32m   2145\u001b[0m                     outs = self.train_on_batch(x, y,\n\u001b[1;32m   2146\u001b[0m                                                \u001b[0msample_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2147\u001b[0;31m                                                class_weight=class_weight)\n\u001b[0m\u001b[1;32m   2148\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2149\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mtrain_on_batch\u001b[0;34m(self, x, y, sample_weight, class_weight)\u001b[0m\n\u001b[1;32m   1837\u001b[0m             \u001b[0mins\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0my\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0msample_weights\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1838\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_make_train_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1839\u001b[0;31m         \u001b[0moutputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1840\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1841\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0moutputs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/keras/backend/tensorflow_backend.pyc\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2355\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2356\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2357\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2358\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2359\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    887\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    888\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 889\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    890\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    891\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1118\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1119\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1120\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1121\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1122\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1315\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1316\u001b[0m       return self._do_call(_run_fn, self._session, feeds, fetches, targets,\n\u001b[0;32m-> 1317\u001b[0;31m                            options, run_metadata)\n\u001b[0m\u001b[1;32m   1318\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1319\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1321\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1322\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1323\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1324\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1325\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/home/dhodun/.local/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(session, feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1300\u001b[0m           return tf_session.TF_Run(session, options,\n\u001b[1;32m   1301\u001b[0m                                    \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1302\u001b[0;31m                                    status, run_metadata)\n\u001b[0m\u001b[1;32m   1303\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1304\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(GRU(32, input_shape=(lookback // step, df.shape[1])))\n",
    "model.add(Dense(1))\n",
    "\n",
    "model.summary()\n",
    "\n",
    "model.compile(optimizer='rmsprop', loss='mae', metrics=['mae'])\n",
    "\n",
    "history = model.fit_generator(train_gen_reverse, epochs = 20, steps_per_epoch = 500,\n",
    "                             validation_data = val_gen_reverse, validation_steps = val_steps)\n",
    "\n",
    "print_curves(history)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Performs much worse, as we might be able to expect, most of the predictive power comes from the data chronologically, i.e. temperature at time t depends on t-1, whereas temp at time t does not depend on t+1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's Try reverse chronological on IMDB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading data...\n"
     ]
    }
   ],
   "source": [
    "from keras.datasets import imdb\n",
    "from keras.preprocessing import sequence\n",
    "\n",
    "max_features = 10000\n",
    "maxlen = 500\n",
    "batch_size = 32\n",
    "\n",
    "print('Loading data...')\n",
    "(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=max_features)\n",
    "\n",
    "x_train = [x[::-1] for x in x_train]\n",
    "x_test = [x[::-1] for x in x_test]\n",
    "\n",
    "x_train = sequence.pad_sequences(x_train, maxlen=maxlen)\n",
    "x_test = sequence.pad_sequences(x_test, maxlen=maxlen)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_8 (Embedding)      (None, None, 128)         1280000   \n",
      "_________________________________________________________________\n",
      "lstm_8 (LSTM)                (None, 32)                20608     \n",
      "_________________________________________________________________\n",
      "dense_22 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,300,641\n",
      "Trainable params: 1,300,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:Tried to colocate training_20/RMSprop/gradients/replica_0_16/sequential_18/embedding_8/Gather_grad/Shape with an op embedding_8/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_20/RMSprop/gradients/replica_0_16/sequential_18/embedding_8/Gather_grad/ToInt32 with an op embedding_8/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_20/RMSprop/gradients/replica_1_16/sequential_18/embedding_8/Gather_grad/Shape with an op embedding_8/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_20/RMSprop/gradients/replica_1_16/sequential_18/embedding_8/Gather_grad/ToInt32 with an op embedding_8/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_20/RMSprop/gradients/replica_2_9/sequential_18/embedding_8/Gather_grad/Shape with an op embedding_8/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_20/RMSprop/gradients/replica_2_9/sequential_18/embedding_8/Gather_grad/ToInt32 with an op embedding_8/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_20/RMSprop/gradients/replica_3_9/sequential_18/embedding_8/Gather_grad/Shape with an op embedding_8/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_20/RMSprop/gradients/replica_3_9/sequential_18/embedding_8/Gather_grad/ToInt32 with an op embedding_8/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/10\n",
      "20000/20000 [==============================] - 13s 654us/step - loss: 0.6900 - acc: 0.5498 - val_loss: 0.6835 - val_acc: 0.6002\n",
      "Epoch 2/10\n",
      "20000/20000 [==============================] - 10s 499us/step - loss: 0.6663 - acc: 0.6591 - val_loss: 0.6444 - val_acc: 0.6398\n",
      "Epoch 3/10\n",
      "20000/20000 [==============================] - 10s 497us/step - loss: 0.5933 - acc: 0.7066 - val_loss: 0.5720 - val_acc: 0.7192\n",
      "Epoch 4/10\n",
      "20000/20000 [==============================] - 10s 501us/step - loss: 0.5318 - acc: 0.7539 - val_loss: 0.4917 - val_acc: 0.7810\n",
      "Epoch 5/10\n",
      "20000/20000 [==============================] - 10s 502us/step - loss: 0.4972 - acc: 0.7674 - val_loss: 0.4808 - val_acc: 0.7878\n",
      "Epoch 6/10\n",
      "20000/20000 [==============================] - 10s 507us/step - loss: 0.4324 - acc: 0.8150 - val_loss: 0.4456 - val_acc: 0.8056\n",
      "Epoch 7/10\n",
      "20000/20000 [==============================] - 10s 498us/step - loss: 0.4189 - acc: 0.8201 - val_loss: 0.4455 - val_acc: 0.8060\n",
      "Epoch 8/10\n",
      "20000/20000 [==============================] - 10s 503us/step - loss: 0.3375 - acc: 0.8686 - val_loss: 0.4043 - val_acc: 0.8336\n",
      "Epoch 9/10\n",
      "20000/20000 [==============================] - 10s 503us/step - loss: 0.3751 - acc: 0.8449 - val_loss: 0.3858 - val_acc: 0.8346\n",
      "Epoch 10/10\n",
      "20000/20000 [==============================] - 10s 501us/step - loss: 0.2703 - acc: 0.8996 - val_loss: 0.3811 - val_acc: 0.8470\n",
      "CPU times: user 4min 59s, sys: 30.4 s, total: 5min 29s\n",
      "Wall time: 1min 46s\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_9 (Embedding)      (None, None, 128)         1280000   \n",
      "_________________________________________________________________\n",
      "lstm_9 (LSTM)                (None, 32)                20608     \n",
      "_________________________________________________________________\n",
      "dense_23 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,300,641\n",
      "Trainable params: 1,300,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:Tried to colocate training_21/RMSprop/gradients/replica_0_17/sequential_19/embedding_9/Gather_grad/Shape with an op embedding_9/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_21/RMSprop/gradients/replica_0_17/sequential_19/embedding_9/Gather_grad/ToInt32 with an op embedding_9/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_21/RMSprop/gradients/replica_1_17/sequential_19/embedding_9/Gather_grad/Shape with an op embedding_9/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_21/RMSprop/gradients/replica_1_17/sequential_19/embedding_9/Gather_grad/ToInt32 with an op embedding_9/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_21/RMSprop/gradients/replica_2_10/sequential_19/embedding_9/Gather_grad/Shape with an op embedding_9/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_21/RMSprop/gradients/replica_2_10/sequential_19/embedding_9/Gather_grad/ToInt32 with an op embedding_9/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_21/RMSprop/gradients/replica_3_10/sequential_19/embedding_9/Gather_grad/Shape with an op embedding_9/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_21/RMSprop/gradients/replica_3_10/sequential_19/embedding_9/Gather_grad/ToInt32 with an op embedding_9/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/10\n",
      "20000/20000 [==============================] - 21s 1ms/step - loss: 0.6815 - acc: 0.5898 - val_loss: 0.6329 - val_acc: 0.6894\n",
      "Epoch 2/10\n",
      "20000/20000 [==============================] - 17s 865us/step - loss: 0.5990 - acc: 0.7030 - val_loss: 0.5337 - val_acc: 0.7552\n",
      "Epoch 3/10\n",
      "20000/20000 [==============================] - 17s 856us/step - loss: 0.4729 - acc: 0.7963 - val_loss: 0.5628 - val_acc: 0.7416\n",
      "Epoch 4/10\n",
      "20000/20000 [==============================] - 17s 843us/step - loss: 0.4530 - acc: 0.8035 - val_loss: 0.4552 - val_acc: 0.8050\n",
      "Epoch 5/10\n",
      "20000/20000 [==============================] - 16s 823us/step - loss: 0.3793 - acc: 0.8476 - val_loss: 0.4340 - val_acc: 0.8110\n",
      "Epoch 6/10\n",
      "20000/20000 [==============================] - 16s 815us/step - loss: 0.3342 - acc: 0.8745 - val_loss: 0.3928 - val_acc: 0.8386\n",
      "Epoch 7/10\n",
      "20000/20000 [==============================] - 16s 820us/step - loss: 0.3086 - acc: 0.8800 - val_loss: 0.3881 - val_acc: 0.8388\n",
      "Epoch 8/10\n",
      "20000/20000 [==============================] - 16s 821us/step - loss: 0.2847 - acc: 0.8933 - val_loss: 0.3929 - val_acc: 0.8444\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 9/10\n",
      "20000/20000 [==============================] - 16s 812us/step - loss: 0.2686 - acc: 0.8996 - val_loss: 0.3981 - val_acc: 0.8380\n",
      "Epoch 10/10\n",
      "20000/20000 [==============================] - 16s 815us/step - loss: 0.2464 - acc: 0.9114 - val_loss: 0.3922 - val_acc: 0.8336\n",
      "CPU times: user 8min 44s, sys: 51.2 s, total: 9min 36s\n",
      "Wall time: 2min 52s\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_10 (Embedding)     (None, None, 128)         1280000   \n",
      "_________________________________________________________________\n",
      "lstm_10 (LSTM)               (None, 32)                20608     \n",
      "_________________________________________________________________\n",
      "dense_24 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,300,641\n",
      "Trainable params: 1,300,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:Tried to colocate training_22/RMSprop/gradients/replica_0_18/sequential_20/embedding_10/Gather_grad/Shape with an op embedding_10/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_22/RMSprop/gradients/replica_0_18/sequential_20/embedding_10/Gather_grad/ToInt32 with an op embedding_10/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_22/RMSprop/gradients/replica_1_18/sequential_20/embedding_10/Gather_grad/Shape with an op embedding_10/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_22/RMSprop/gradients/replica_1_18/sequential_20/embedding_10/Gather_grad/ToInt32 with an op embedding_10/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_22/RMSprop/gradients/replica_2_11/sequential_20/embedding_10/Gather_grad/Shape with an op embedding_10/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_22/RMSprop/gradients/replica_2_11/sequential_20/embedding_10/Gather_grad/ToInt32 with an op embedding_10/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_22/RMSprop/gradients/replica_3_11/sequential_20/embedding_10/Gather_grad/Shape with an op embedding_10/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_22/RMSprop/gradients/replica_3_11/sequential_20/embedding_10/Gather_grad/ToInt32 with an op embedding_10/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/10\n",
      "20000/20000 [==============================] - 34s 2ms/step - loss: 0.6751 - acc: 0.6164 - val_loss: 0.5209 - val_acc: 0.7656\n",
      "Epoch 2/10\n",
      "20000/20000 [==============================] - 30s 1ms/step - loss: 0.4574 - acc: 0.8001 - val_loss: 0.5866 - val_acc: 0.6970\n",
      "Epoch 3/10\n",
      "20000/20000 [==============================] - 30s 1ms/step - loss: 0.3659 - acc: 0.8536 - val_loss: 0.3874 - val_acc: 0.8454\n",
      "Epoch 4/10\n",
      "20000/20000 [==============================] - 30s 2ms/step - loss: 0.3187 - acc: 0.8766 - val_loss: 0.3356 - val_acc: 0.8720\n",
      "Epoch 5/10\n",
      "20000/20000 [==============================] - 30s 1ms/step - loss: 0.2872 - acc: 0.8898 - val_loss: 0.3503 - val_acc: 0.8648\n",
      "Epoch 6/10\n",
      "20000/20000 [==============================] - 30s 1ms/step - loss: 0.2571 - acc: 0.9065 - val_loss: 0.3669 - val_acc: 0.8664\n",
      "Epoch 7/10\n",
      "20000/20000 [==============================] - 30s 2ms/step - loss: 0.2435 - acc: 0.9087 - val_loss: 0.3298 - val_acc: 0.8754\n",
      "Epoch 8/10\n",
      "20000/20000 [==============================] - 30s 1ms/step - loss: 0.2228 - acc: 0.9181 - val_loss: 0.3660 - val_acc: 0.8480\n",
      "Epoch 9/10\n",
      "20000/20000 [==============================] - 30s 1ms/step - loss: 0.2140 - acc: 0.9209 - val_loss: 0.4012 - val_acc: 0.8362\n",
      "Epoch 10/10\n",
      "20000/20000 [==============================] - 29s 1ms/step - loss: 0.1927 - acc: 0.9308 - val_loss: 0.3740 - val_acc: 0.8594\n",
      "CPU times: user 16min 7s, sys: 1min 32s, total: 17min 40s\n",
      "Wall time: 5min 4s\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_11 (Embedding)     (None, None, 128)         1280000   \n",
      "_________________________________________________________________\n",
      "lstm_11 (LSTM)               (None, 32)                20608     \n",
      "_________________________________________________________________\n",
      "dense_25 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,300,641\n",
      "Trainable params: 1,300,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:Tried to colocate training_23/RMSprop/gradients/replica_0_19/sequential_21/embedding_11/Gather_grad/Shape with an op embedding_11/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_23/RMSprop/gradients/replica_0_19/sequential_21/embedding_11/Gather_grad/ToInt32 with an op embedding_11/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_23/RMSprop/gradients/replica_1_19/sequential_21/embedding_11/Gather_grad/Shape with an op embedding_11/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_23/RMSprop/gradients/replica_1_19/sequential_21/embedding_11/Gather_grad/ToInt32 with an op embedding_11/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_23/RMSprop/gradients/replica_2_12/sequential_21/embedding_11/Gather_grad/Shape with an op embedding_11/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_23/RMSprop/gradients/replica_2_12/sequential_21/embedding_11/Gather_grad/ToInt32 with an op embedding_11/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_23/RMSprop/gradients/replica_3_12/sequential_21/embedding_11/Gather_grad/Shape with an op embedding_11/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_23/RMSprop/gradients/replica_3_12/sequential_21/embedding_11/Gather_grad/ToInt32 with an op embedding_11/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/10\n",
      "20000/20000 [==============================] - 61s 3ms/step - loss: 0.5866 - acc: 0.6971 - val_loss: 0.4460 - val_acc: 0.8206\n",
      "Epoch 2/10\n",
      "20000/20000 [==============================] - 56s 3ms/step - loss: 0.3908 - acc: 0.8472 - val_loss: 0.3779 - val_acc: 0.8454\n",
      "Epoch 3/10\n",
      "20000/20000 [==============================] - 56s 3ms/step - loss: 0.3135 - acc: 0.8808 - val_loss: 0.3393 - val_acc: 0.8690\n",
      "Epoch 4/10\n",
      "20000/20000 [==============================] - 56s 3ms/step - loss: 0.2831 - acc: 0.8954 - val_loss: 0.3384 - val_acc: 0.8676\n",
      "Epoch 5/10\n",
      "20000/20000 [==============================] - 56s 3ms/step - loss: 0.2338 - acc: 0.9141 - val_loss: 0.4282 - val_acc: 0.8434\n",
      "Epoch 6/10\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20000/20000 [==============================] - 56s 3ms/step - loss: 0.2190 - acc: 0.9203 - val_loss: 0.3993 - val_acc: 0.8430\n",
      "Epoch 7/10\n",
      "20000/20000 [==============================] - 55s 3ms/step - loss: 0.2035 - acc: 0.9266 - val_loss: 0.3798 - val_acc: 0.8548\n",
      "Epoch 8/10\n",
      "20000/20000 [==============================] - 55s 3ms/step - loss: 0.1926 - acc: 0.9312 - val_loss: 0.3759 - val_acc: 0.8756\n",
      "Epoch 9/10\n",
      "20000/20000 [==============================] - 54s 3ms/step - loss: 0.1692 - acc: 0.9394 - val_loss: 0.4174 - val_acc: 0.8738\n",
      "Epoch 10/10\n",
      "20000/20000 [==============================] - 55s 3ms/step - loss: 0.1540 - acc: 0.9462 - val_loss: 0.3856 - val_acc: 0.8540\n",
      "CPU times: user 30min 36s, sys: 2min 52s, total: 33min 29s\n",
      "Wall time: 9min 20s\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_12 (Embedding)     (None, None, 128)         1280000   \n",
      "_________________________________________________________________\n",
      "lstm_12 (LSTM)               (None, 32)                20608     \n",
      "_________________________________________________________________\n",
      "dense_26 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,300,641\n",
      "Trainable params: 1,300,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:Tried to colocate training_24/RMSprop/gradients/replica_0_20/sequential_22/embedding_12/Gather_grad/Shape with an op embedding_12/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_24/RMSprop/gradients/replica_0_20/sequential_22/embedding_12/Gather_grad/ToInt32 with an op embedding_12/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_24/RMSprop/gradients/replica_1_20/sequential_22/embedding_12/Gather_grad/Shape with an op embedding_12/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_24/RMSprop/gradients/replica_1_20/sequential_22/embedding_12/Gather_grad/ToInt32 with an op embedding_12/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_24/RMSprop/gradients/replica_2_13/sequential_22/embedding_12/Gather_grad/Shape with an op embedding_12/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_24/RMSprop/gradients/replica_2_13/sequential_22/embedding_12/Gather_grad/ToInt32 with an op embedding_12/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_24/RMSprop/gradients/replica_3_13/sequential_22/embedding_12/Gather_grad/Shape with an op embedding_12/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_24/RMSprop/gradients/replica_3_13/sequential_22/embedding_12/Gather_grad/ToInt32 with an op embedding_12/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/10\n",
      "20000/20000 [==============================] - 110s 6ms/step - loss: 0.5673 - acc: 0.7089 - val_loss: 0.4397 - val_acc: 0.8220\n",
      "Epoch 2/10\n",
      "20000/20000 [==============================] - 105s 5ms/step - loss: 0.3763 - acc: 0.8457 - val_loss: 0.3357 - val_acc: 0.8706\n",
      "Epoch 3/10\n",
      "20000/20000 [==============================] - 105s 5ms/step - loss: 0.2916 - acc: 0.8895 - val_loss: 0.4351 - val_acc: 0.8082\n",
      "Epoch 4/10\n",
      "20000/20000 [==============================] - 105s 5ms/step - loss: 0.2456 - acc: 0.9073 - val_loss: 0.3307 - val_acc: 0.8626\n",
      "Epoch 5/10\n",
      "20000/20000 [==============================] - 105s 5ms/step - loss: 0.2118 - acc: 0.9234 - val_loss: 0.3903 - val_acc: 0.8736\n",
      "Epoch 6/10\n",
      "20000/20000 [==============================] - 105s 5ms/step - loss: 0.1928 - acc: 0.9300 - val_loss: 0.6293 - val_acc: 0.8460\n",
      "Epoch 7/10\n",
      "20000/20000 [==============================] - 105s 5ms/step - loss: 0.1763 - acc: 0.9407 - val_loss: 0.3618 - val_acc: 0.8750\n",
      "Epoch 8/10\n",
      "20000/20000 [==============================] - 105s 5ms/step - loss: 0.1561 - acc: 0.9446 - val_loss: 0.4252 - val_acc: 0.8172\n",
      "Epoch 9/10\n",
      "20000/20000 [==============================] - 104s 5ms/step - loss: 0.1612 - acc: 0.9433 - val_loss: 0.5841 - val_acc: 0.8400\n",
      "Epoch 10/10\n",
      "20000/20000 [==============================] - 104s 5ms/step - loss: 0.1306 - acc: 0.9551 - val_loss: 0.5792 - val_acc: 0.7758\n",
      "CPU times: user 58min 36s, sys: 5min 27s, total: 1h 4min 4s\n",
      "Wall time: 17min 35s\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_13 (Embedding)     (None, None, 128)         1280000   \n",
      "_________________________________________________________________\n",
      "lstm_13 (LSTM)               (None, 32)                20608     \n",
      "_________________________________________________________________\n",
      "dense_27 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,300,641\n",
      "Trainable params: 1,300,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:Tried to colocate training_25/RMSprop/gradients/replica_0_21/sequential_23/embedding_13/Gather_grad/Shape with an op embedding_13/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_25/RMSprop/gradients/replica_0_21/sequential_23/embedding_13/Gather_grad/ToInt32 with an op embedding_13/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_25/RMSprop/gradients/replica_1_21/sequential_23/embedding_13/Gather_grad/Shape with an op embedding_13/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_25/RMSprop/gradients/replica_1_21/sequential_23/embedding_13/Gather_grad/ToInt32 with an op embedding_13/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_25/RMSprop/gradients/replica_2_14/sequential_23/embedding_13/Gather_grad/Shape with an op embedding_13/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_25/RMSprop/gradients/replica_2_14/sequential_23/embedding_13/Gather_grad/ToInt32 with an op embedding_13/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_25/RMSprop/gradients/replica_3_14/sequential_23/embedding_13/Gather_grad/Shape with an op embedding_13/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_25/RMSprop/gradients/replica_3_14/sequential_23/embedding_13/Gather_grad/ToInt32 with an op embedding_13/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/10\n",
      "20000/20000 [==============================] - 207s 10ms/step - loss: 0.5123 - acc: 0.7541 - val_loss: 0.3934 - val_acc: 0.8400\n",
      "Epoch 2/10\n",
      "20000/20000 [==============================] - 202s 10ms/step - loss: 0.3263 - acc: 0.8729 - val_loss: 0.4375 - val_acc: 0.8444\n",
      "Epoch 3/10\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20000/20000 [==============================] - 201s 10ms/step - loss: 0.2599 - acc: 0.9034 - val_loss: 0.3744 - val_acc: 0.8346\n",
      "Epoch 4/10\n",
      "20000/20000 [==============================] - 205s 10ms/step - loss: 0.2251 - acc: 0.9178 - val_loss: 0.3155 - val_acc: 0.8736\n",
      "Epoch 5/10\n",
      "20000/20000 [==============================] - 200s 10ms/step - loss: 0.1928 - acc: 0.9345 - val_loss: 0.3606 - val_acc: 0.8724\n",
      "Epoch 6/10\n",
      "20000/20000 [==============================] - 200s 10ms/step - loss: 0.1705 - acc: 0.9408 - val_loss: 0.3617 - val_acc: 0.8668\n",
      "Epoch 7/10\n",
      "20000/20000 [==============================] - 199s 10ms/step - loss: 0.1563 - acc: 0.9461 - val_loss: 0.4742 - val_acc: 0.8446\n",
      "Epoch 8/10\n",
      "20000/20000 [==============================] - 199s 10ms/step - loss: 0.1404 - acc: 0.9525 - val_loss: 0.5446 - val_acc: 0.8434\n",
      "Epoch 9/10\n",
      "20000/20000 [==============================] - 198s 10ms/step - loss: 0.1247 - acc: 0.9571 - val_loss: 0.5461 - val_acc: 0.8356\n",
      "Epoch 10/10\n",
      "20000/20000 [==============================] - 196s 10ms/step - loss: 0.1199 - acc: 0.9601 - val_loss: 0.3888 - val_acc: 0.8686\n",
      "CPU times: user 1h 52min, sys: 10min 38s, total: 2h 2min 38s\n",
      "Wall time: 33min 31s\n"
     ]
    }
   ],
   "source": [
    "from keras import layers\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(layers.Embedding(max_features, 128))\n",
    "    model.add(layers.LSTM(32))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "%time history_4096 = parallel_model.fit(x_train, y_train, epochs = 10, batch_size = 4096, validation_split=0.2)\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(layers.Embedding(max_features, 128))\n",
    "    model.add(layers.LSTM(32))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "%time history_2048 = parallel_model.fit(x_train, y_train, epochs = 10, batch_size = 2048, validation_split=0.2)\n",
    "\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(layers.Embedding(max_features, 128))\n",
    "    model.add(layers.LSTM(32))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "%time history_1028 = parallel_model.fit(x_train, y_train, epochs = 10, batch_size = 1028, validation_split=0.2)\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(layers.Embedding(max_features, 128))\n",
    "    model.add(layers.LSTM(32))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "%time history_512 = parallel_model.fit(x_train, y_train, epochs = 10, batch_size = 512, validation_split=0.2)\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(layers.Embedding(max_features, 128))\n",
    "    model.add(layers.LSTM(32))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "%time history_256 = parallel_model.fit(x_train, y_train, epochs = 10, batch_size = 256, validation_split=0.2)\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(layers.Embedding(max_features, 128))\n",
    "    model.add(layers.LSTM(32))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer='rmsprop', loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "%time history_128 = parallel_model.fit(x_train, y_train, epochs = 10, batch_size = 128, validation_split=0.2)\n",
    "\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_16 (Embedding)     (None, None, 128)         1280000   \n",
      "_________________________________________________________________\n",
      "lstm_16 (LSTM)               (None, 32)                20608     \n",
      "_________________________________________________________________\n",
      "dense_30 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,300,641\n",
      "Trainable params: 1,300,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:Tried to colocate training_28/RMSprop/gradients/replica_0_24/sequential_26/embedding_16/Gather_grad/Shape with an op embedding_16/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_28/RMSprop/gradients/replica_0_24/sequential_26/embedding_16/Gather_grad/ToInt32 with an op embedding_16/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_28/RMSprop/gradients/replica_1_24/sequential_26/embedding_16/Gather_grad/Shape with an op embedding_16/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_28/RMSprop/gradients/replica_1_24/sequential_26/embedding_16/Gather_grad/ToInt32 with an op embedding_16/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_28/RMSprop/gradients/replica_2_17/sequential_26/embedding_16/Gather_grad/Shape with an op embedding_16/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_28/RMSprop/gradients/replica_2_17/sequential_26/embedding_16/Gather_grad/ToInt32 with an op embedding_16/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_28/RMSprop/gradients/replica_3_17/sequential_26/embedding_16/Gather_grad/Shape with an op embedding_16/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_28/RMSprop/gradients/replica_3_17/sequential_26/embedding_16/Gather_grad/ToInt32 with an op embedding_16/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/30\n",
      "20000/20000 [==============================] - 16s 788us/step - loss: 0.6812 - acc: 0.5454 - val_loss: 0.6725 - val_acc: 0.5610\n",
      "Epoch 2/30\n",
      "20000/20000 [==============================] - 9s 460us/step - loss: 0.6234 - acc: 0.6546 - val_loss: 0.6292 - val_acc: 0.6470\n",
      "Epoch 3/30\n",
      "20000/20000 [==============================] - 9s 463us/step - loss: 0.5299 - acc: 0.7446 - val_loss: 0.7449 - val_acc: 0.6688\n",
      "Epoch 4/30\n",
      "20000/20000 [==============================] - 9s 458us/step - loss: 0.5552 - acc: 0.7339 - val_loss: 0.5584 - val_acc: 0.7104\n",
      "Epoch 5/30\n",
      "20000/20000 [==============================] - 9s 462us/step - loss: 0.4259 - acc: 0.8143 - val_loss: 0.5240 - val_acc: 0.7460\n",
      "Epoch 6/30\n",
      "20000/20000 [==============================] - 9s 467us/step - loss: 0.3980 - acc: 0.8335 - val_loss: 0.5412 - val_acc: 0.7480\n",
      "Epoch 7/30\n",
      "20000/20000 [==============================] - 9s 463us/step - loss: 0.3559 - acc: 0.8544 - val_loss: 0.4642 - val_acc: 0.7984\n",
      "Epoch 8/30\n",
      "20000/20000 [==============================] - 9s 469us/step - loss: 0.3486 - acc: 0.8564 - val_loss: 0.4803 - val_acc: 0.7772\n",
      "Epoch 9/30\n",
      "20000/20000 [==============================] - 9s 461us/step - loss: 0.2893 - acc: 0.8918 - val_loss: 0.4419 - val_acc: 0.8132\n",
      "Epoch 10/30\n",
      "20000/20000 [==============================] - 9s 457us/step - loss: 0.2552 - acc: 0.9068 - val_loss: 0.4639 - val_acc: 0.8012\n",
      "Epoch 11/30\n",
      "20000/20000 [==============================] - 9s 460us/step - loss: 0.3074 - acc: 0.8784 - val_loss: 0.4387 - val_acc: 0.8164\n",
      "Epoch 12/30\n",
      "20000/20000 [==============================] - 9s 458us/step - loss: 0.2395 - acc: 0.9129 - val_loss: 0.4787 - val_acc: 0.7980\n",
      "Epoch 13/30\n",
      "20000/20000 [==============================] - 9s 461us/step - loss: 0.2414 - acc: 0.9120 - val_loss: 0.4371 - val_acc: 0.8252\n",
      "Epoch 14/30\n",
      "20000/20000 [==============================] - 9s 464us/step - loss: 0.2298 - acc: 0.9159 - val_loss: 0.4374 - val_acc: 0.8148\n",
      "Epoch 15/30\n",
      "20000/20000 [==============================] - 9s 456us/step - loss: 0.1968 - acc: 0.9345 - val_loss: 0.4267 - val_acc: 0.8340\n",
      "Epoch 16/30\n",
      "20000/20000 [==============================] - 9s 463us/step - loss: 0.2144 - acc: 0.9204 - val_loss: 0.4462 - val_acc: 0.8116\n",
      "Epoch 17/30\n",
      "20000/20000 [==============================] - 9s 458us/step - loss: 0.1815 - acc: 0.9405 - val_loss: 0.4265 - val_acc: 0.8388\n",
      "Epoch 18/30\n",
      "20000/20000 [==============================] - 9s 462us/step - loss: 0.1766 - acc: 0.9399 - val_loss: 0.4917 - val_acc: 0.8056\n",
      "Epoch 19/30\n",
      "20000/20000 [==============================] - 9s 455us/step - loss: 0.1764 - acc: 0.9403 - val_loss: 0.4335 - val_acc: 0.8382\n",
      "Epoch 20/30\n",
      "20000/20000 [==============================] - 9s 467us/step - loss: 0.1512 - acc: 0.9518 - val_loss: 0.4374 - val_acc: 0.8404\n",
      "Epoch 21/30\n",
      "20000/20000 [==============================] - 9s 466us/step - loss: 0.1754 - acc: 0.9376 - val_loss: 0.4570 - val_acc: 0.8172\n",
      "Epoch 22/30\n",
      "20000/20000 [==============================] - 9s 458us/step - loss: 0.1438 - acc: 0.9570 - val_loss: 0.4431 - val_acc: 0.8418\n",
      "Epoch 23/30\n",
      "20000/20000 [==============================] - 9s 458us/step - loss: 0.1345 - acc: 0.9583 - val_loss: 0.4576 - val_acc: 0.8352\n",
      "Epoch 24/30\n",
      "20000/20000 [==============================] - 9s 460us/step - loss: 0.1591 - acc: 0.9465 - val_loss: 0.4527 - val_acc: 0.8346\n",
      "Epoch 25/30\n",
      "20000/20000 [==============================] - 9s 455us/step - loss: 0.1251 - acc: 0.9647 - val_loss: 0.4589 - val_acc: 0.8420\n",
      "Epoch 26/30\n",
      "20000/20000 [==============================] - 9s 457us/step - loss: 0.1241 - acc: 0.9636 - val_loss: 0.4876 - val_acc: 0.8280\n",
      "Epoch 27/30\n",
      "20000/20000 [==============================] - 9s 460us/step - loss: 0.1306 - acc: 0.9605 - val_loss: 0.4838 - val_acc: 0.8336\n",
      "Epoch 28/30\n",
      "20000/20000 [==============================] - 9s 461us/step - loss: 0.1143 - acc: 0.9673 - val_loss: 0.4723 - val_acc: 0.8406\n",
      "Epoch 29/30\n",
      "20000/20000 [==============================] - 9s 464us/step - loss: 0.1082 - acc: 0.9707 - val_loss: 0.4846 - val_acc: 0.8412\n",
      "Epoch 30/30\n",
      "20000/20000 [==============================] - 9s 456us/step - loss: 0.1063 - acc: 0.9705 - val_loss: 0.5116 - val_acc: 0.8264\n",
      "CPU times: user 13min 55s, sys: 1min 16s, total: 15min 12s\n",
      "Wall time: 4min 46s\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_17 (Embedding)     (None, None, 128)         1280000   \n",
      "_________________________________________________________________\n",
      "lstm_17 (LSTM)               (None, 32)                20608     \n",
      "_________________________________________________________________\n",
      "dense_31 (Dense)             (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 1,300,641\n",
      "Trainable params: 1,300,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "WARNING:tensorflow:Tried to colocate training_29/RMSprop/gradients/replica_0_25/sequential_27/embedding_17/Gather_grad/Shape with an op embedding_17/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_29/RMSprop/gradients/replica_0_25/sequential_27/embedding_17/Gather_grad/ToInt32 with an op embedding_17/embeddings/read that had a different device: /device:GPU:0 vs /device:CPU:0. Ignoring colocation property.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Tried to colocate training_29/RMSprop/gradients/replica_1_25/sequential_27/embedding_17/Gather_grad/Shape with an op embedding_17/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_29/RMSprop/gradients/replica_1_25/sequential_27/embedding_17/Gather_grad/ToInt32 with an op embedding_17/embeddings/read that had a different device: /device:GPU:1 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_29/RMSprop/gradients/replica_2_18/sequential_27/embedding_17/Gather_grad/Shape with an op embedding_17/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_29/RMSprop/gradients/replica_2_18/sequential_27/embedding_17/Gather_grad/ToInt32 with an op embedding_17/embeddings/read that had a different device: /device:GPU:2 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_29/RMSprop/gradients/replica_3_18/sequential_27/embedding_17/Gather_grad/Shape with an op embedding_17/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "WARNING:tensorflow:Tried to colocate training_29/RMSprop/gradients/replica_3_18/sequential_27/embedding_17/Gather_grad/ToInt32 with an op embedding_17/embeddings/read that had a different device: /device:GPU:3 vs /device:CPU:0. Ignoring colocation property.\n",
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/30\n",
      "20000/20000 [==============================] - 22s 1ms/step - loss: 0.7243 - acc: 0.5595 - val_loss: 0.6665 - val_acc: 0.6386\n",
      "Epoch 2/30\n",
      "20000/20000 [==============================] - 16s 784us/step - loss: 0.5837 - acc: 0.7103 - val_loss: 0.5808 - val_acc: 0.7118\n",
      "Epoch 3/30\n",
      "20000/20000 [==============================] - 16s 788us/step - loss: 0.4936 - acc: 0.7719 - val_loss: 0.4967 - val_acc: 0.7746\n",
      "Epoch 4/30\n",
      "20000/20000 [==============================] - 15s 771us/step - loss: 0.3996 - acc: 0.8364 - val_loss: 0.4627 - val_acc: 0.7982\n",
      "Epoch 5/30\n",
      "20000/20000 [==============================] - 16s 782us/step - loss: 0.3663 - acc: 0.8541 - val_loss: 0.4508 - val_acc: 0.8098\n",
      "Epoch 6/30\n",
      "20000/20000 [==============================] - 16s 780us/step - loss: 0.3281 - acc: 0.8724 - val_loss: 0.4326 - val_acc: 0.8246\n",
      "Epoch 7/30\n",
      "20000/20000 [==============================] - 16s 782us/step - loss: 0.3034 - acc: 0.8851 - val_loss: 0.4346 - val_acc: 0.8214\n",
      "Epoch 8/30\n",
      "20000/20000 [==============================] - 16s 779us/step - loss: 0.2764 - acc: 0.8975 - val_loss: 0.4510 - val_acc: 0.8118\n",
      "Epoch 9/30\n",
      "20000/20000 [==============================] - 16s 788us/step - loss: 0.2621 - acc: 0.9046 - val_loss: 0.4174 - val_acc: 0.8386\n",
      "Epoch 10/30\n",
      "20000/20000 [==============================] - 15s 767us/step - loss: 0.2363 - acc: 0.9172 - val_loss: 0.4199 - val_acc: 0.8368\n",
      "Epoch 11/30\n",
      "20000/20000 [==============================] - 15s 770us/step - loss: 0.2272 - acc: 0.9219 - val_loss: 0.4140 - val_acc: 0.8420\n",
      "Epoch 12/30\n",
      "20000/20000 [==============================] - 16s 781us/step - loss: 0.2080 - acc: 0.9286 - val_loss: 0.4145 - val_acc: 0.8450\n",
      "Epoch 13/30\n",
      "20000/20000 [==============================] - 16s 783us/step - loss: 0.2039 - acc: 0.9312 - val_loss: 0.4205 - val_acc: 0.8434\n",
      "Epoch 14/30\n",
      "20000/20000 [==============================] - 15s 773us/step - loss: 0.1893 - acc: 0.9364 - val_loss: 0.4169 - val_acc: 0.8484\n",
      "Epoch 15/30\n",
      "20000/20000 [==============================] - 20s 1ms/step - loss: 0.1808 - acc: 0.9410 - val_loss: 0.4267 - val_acc: 0.8390\n",
      "Epoch 16/30\n",
      "20000/20000 [==============================] - 16s 777us/step - loss: 0.1635 - acc: 0.9492 - val_loss: 0.4163 - val_acc: 0.8514\n",
      "Epoch 17/30\n",
      "20000/20000 [==============================] - 16s 786us/step - loss: 0.1660 - acc: 0.9466 - val_loss: 0.4170 - val_acc: 0.8508\n",
      "Epoch 18/30\n",
      "20000/20000 [==============================] - 15s 773us/step - loss: 0.1548 - acc: 0.9521 - val_loss: 0.4205 - val_acc: 0.8524\n",
      "Epoch 19/30\n",
      "20000/20000 [==============================] - 15s 767us/step - loss: 0.1451 - acc: 0.9570 - val_loss: 0.4415 - val_acc: 0.8386\n",
      "Epoch 20/30\n",
      "20000/20000 [==============================] - 15s 769us/step - loss: 0.1417 - acc: 0.9584 - val_loss: 0.4309 - val_acc: 0.8508\n",
      "Epoch 21/30\n",
      "20000/20000 [==============================] - 16s 782us/step - loss: 0.1325 - acc: 0.9620 - val_loss: 0.4347 - val_acc: 0.8504\n",
      "Epoch 22/30\n",
      "20000/20000 [==============================] - 15s 766us/step - loss: 0.1311 - acc: 0.9618 - val_loss: 0.4386 - val_acc: 0.8546\n",
      "Epoch 23/30\n",
      "20000/20000 [==============================] - 15s 767us/step - loss: 0.1199 - acc: 0.9664 - val_loss: 0.4546 - val_acc: 0.8472\n",
      "Epoch 24/30\n",
      "20000/20000 [==============================] - 15s 772us/step - loss: 0.1204 - acc: 0.9663 - val_loss: 0.4492 - val_acc: 0.8536\n",
      "Epoch 25/30\n",
      "20000/20000 [==============================] - 16s 778us/step - loss: 0.1149 - acc: 0.9679 - val_loss: 0.4743 - val_acc: 0.8442\n",
      "Epoch 26/30\n",
      "20000/20000 [==============================] - 15s 766us/step - loss: 0.1081 - acc: 0.9717 - val_loss: 0.4635 - val_acc: 0.8456\n",
      "Epoch 27/30\n",
      "20000/20000 [==============================] - 15s 762us/step - loss: 0.1026 - acc: 0.9733 - val_loss: 0.4631 - val_acc: 0.8508\n",
      "Epoch 28/30\n",
      "20000/20000 [==============================] - 15s 773us/step - loss: 0.1072 - acc: 0.9709 - val_loss: 0.4648 - val_acc: 0.8490\n",
      "Epoch 29/30\n",
      "20000/20000 [==============================] - 15s 770us/step - loss: 0.0946 - acc: 0.9760 - val_loss: 0.4720 - val_acc: 0.8506\n",
      "Epoch 30/30\n",
      "20000/20000 [==============================] - 16s 783us/step - loss: 0.0920 - acc: 0.9768 - val_loss: 0.4983 - val_acc: 0.8450\n",
      "CPU times: user 24min 29s, sys: 2min 15s, total: 26min 45s\n",
      "Wall time: 7min 58s\n"
     ]
    }
   ],
   "source": [
    "from keras.optimizers import RMSprop\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(layers.Embedding(max_features, 128))\n",
    "    model.add(layers.LSTM(32))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "\n",
    "\n",
    "parallel_model.compile(optimizer=RMSprop(lr=0.01, decay=0.1), loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "%time history_4096 = parallel_model.fit(x_train, y_train, epochs = 30, batch_size = 4096, validation_split=0.2)\n",
    "\n",
    "with tf.device('/cpu:0'):\n",
    "    model = Sequential()\n",
    "    model.add(layers.Embedding(max_features, 128))\n",
    "    model.add(layers.LSTM(32))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "    model.summary()\n",
    "    \n",
    "parallel_model = multi_gpu_model(model, gpus=4)\n",
    "\n",
    "parallel_model.compile(optimizer=RMSprop(lr=0.01, decay=0.1), loss='binary_crossentropy', metrics=['acc'])\n",
    "\n",
    "%time history_2048 = parallel_model.fit(x_train, y_train, epochs = 30, batch_size = 2048, validation_split=0.2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f7454e74090>"
      ]
     },
     "execution_count": 104,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7454e9f190>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7454e9f150>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "epochs = range(1, 10+1)\n",
    "epochs_30 = range(1, 30+1)\n",
    "plt.clf()\n",
    "\n",
    "plt.figure(figsize=(20,10)) \n",
    "plt.plot(epochs_30, history_4096.history['acc'], 'b', label='acc_4096')\n",
    "plt.plot(epochs_30, history_4096.history['val_acc'], 'b--', label='val_acc_4096')\n",
    "\n",
    "plt.plot(epochs_30, history_2048.history['acc'], 'r', label='acc_2048')\n",
    "plt.plot(epochs_30, history_2048.history['val_acc'], 'r--', label='val_acc_2048')\n",
    "\n",
    "plt.plot(epochs, history_1028.history['acc'], '-', color='orange', label='acc_1028')\n",
    "plt.plot(epochs, history_1028.history['val_acc'], '--', color='orange', label='val_acc_1028')\n",
    "\n",
    "plt.plot(epochs, history_512.history['acc'], 'g-', label='acc_512')\n",
    "plt.plot(epochs, history_512.history['val_acc'], 'g--', label='val_acc_512')\n",
    "\n",
    "plt.plot(epochs, history_256.history['acc'], 'y-', label='acc_256')\n",
    "plt.plot(epochs, history_256.history['val_acc'], 'y--', label='val_acc_256')\n",
    "\n",
    "plt.plot(epochs, history_128.history['acc'], '-', color='purple', label='acc_128')\n",
    "plt.plot(epochs, history_128.history['val_acc'], '--', color='purple', label='val_acc_128')\n",
    "\n",
    "\n",
    "\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('acc')\n",
    "plt.legend()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f7454d58c90>"
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7454e92c50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f7454eade90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "epochs = range(1, 10+1)\n",
    "epochs_30 = range(1, 30+1)\n",
    "plt.clf()\n",
    "\n",
    "plt.figure(figsize=(20,10)) \n",
    "plt.plot(epochs_30, history_4096.history['loss'], 'b', label='loss_4096')\n",
    "plt.plot(epochs_30, history_4096.history['val_loss'], 'b--', label='val_loss_4096')\n",
    "\n",
    "plt.plot(epochs_30, history_2048.history['loss'], 'r', label='loss_2048')\n",
    "plt.plot(epochs_30, history_2048.history['val_loss'], 'r--', label='val_loss_2048')\n",
    "\n",
    "plt.plot(epochs, history_1028.history['loss'], '-', color='orange', label='loss_1028')\n",
    "plt.plot(epochs, history_1028.history['val_loss'], '--', color='orange', label='val_loss_1028')\n",
    "\n",
    "plt.plot(epochs, history_512.history['loss'], 'g-', label='loss_512')\n",
    "plt.plot(epochs, history_512.history['val_loss'], 'g--', label='val_loss_512')\n",
    "\n",
    "plt.plot(epochs, history_256.history['loss'], 'y-', label='loss_256')\n",
    "plt.plot(epochs, history_256.history['val_loss'], 'y--', label='val_loss_256')\n",
    "\n",
    "plt.plot(epochs, history_128.history['loss'], '-', color='purple', label='loss_128')\n",
    "plt.plot(epochs, history_128.history['val_loss'], '--', color='purple', label='val_loss_128')\n",
    "\n",
    "\n",
    "\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('loss')\n",
    "plt.legend()"
   ]
  }
 ],
 "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": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}