polyaxon_lstm_sin.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "from tensorflow.contrib import learn\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "from data_processing import generate_data\n",
    "from polyaxon_experiement import create_experiment\n",
    "from polyaxon.processing.input_fn import numpy_input_fn\n",
    "\n",
    "tf.logging.set_verbosity(tf.logging.INFO)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Using config: {'_save_checkpoints_steps': 100, '_task_type': None, '_task_id': 0, '_num_ps_replicas': 0, '_tf_random_seed': 1, '_tf_config': gpu_options {\n",
      "  per_process_gpu_memory_fraction: 1.0\n",
      "}\n",
      ", '_keep_checkpoint_max': 5, '_environment': 'local', '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x112ec98d0>, '_save_summary_steps': 100, '_is_chief': True, '_session_config': None, '_model_dir': None, '_save_checkpoints_secs': None, '_master': '', '_keep_checkpoint_every_n_hours': 4, '_num_worker_replicas': 0, '_evaluation_master': ''}\n"
     ]
    }
   ],
   "source": [
    "TIMESTEPS = 7\n",
    "X, y = generate_data(np.sin, np.linspace(0, 100, 10000, dtype=np.float32), TIMESTEPS, seperate=False)\n",
    "xp = create_experiment('/tmp/polyaxon_logs/sin', X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Saving checkpoints for 1 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 0.16403, step = 1\n",
      "INFO:tensorflow:Saving checkpoints for 100 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.00252027.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-15:59:14\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-100\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-15:59:15\n",
      "INFO:tensorflow:Saving dict for global step 100: global_step = 100, loss = 0.00379716, streaming_mean_absolute_error = 0.0556576, streaming_root_mean_squared_error = 0.0616211\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-100\n",
      "INFO:tensorflow:Saving checkpoints for 101 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 0.00168126, step = 101\n",
      "INFO:tensorflow:Saving checkpoints for 200 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.00895816.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-15:59:25\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-200\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-15:59:26\n",
      "INFO:tensorflow:Saving dict for global step 200: global_step = 200, loss = 0.000530922, streaming_mean_absolute_error = 0.0212256, streaming_root_mean_squared_error = 0.0230417\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-200\n",
      "INFO:tensorflow:Saving checkpoints for 201 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 0.000283245, step = 201\n",
      "INFO:tensorflow:Saving checkpoints for 300 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.00656085.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-15:59:36\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-300\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-15:59:37\n",
      "INFO:tensorflow:Saving dict for global step 300: global_step = 300, loss = 0.000288103, streaming_mean_absolute_error = 0.0152408, streaming_root_mean_squared_error = 0.0169736\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-300\n",
      "INFO:tensorflow:Saving checkpoints for 301 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 7.94632e-05, step = 301\n",
      "INFO:tensorflow:Saving checkpoints for 400 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.000828735.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-15:59:47\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-400\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-15:59:48\n",
      "INFO:tensorflow:Saving dict for global step 400: global_step = 400, loss = 0.00041661, streaming_mean_absolute_error = 0.0177949, streaming_root_mean_squared_error = 0.020411\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-400\n",
      "INFO:tensorflow:Saving checkpoints for 401 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 0.000106203, step = 401\n",
      "INFO:tensorflow:Saving checkpoints for 500 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.0039412.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-15:59:58\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-500\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Finished evaluation at 2017-06-15-15:59:58\n",
      "INFO:tensorflow:Saving dict for global step 500: global_step = 500, loss = 8.96396e-05, streaming_mean_absolute_error = 0.00771438, streaming_root_mean_squared_error = 0.00946782\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-500\n",
      "INFO:tensorflow:Saving checkpoints for 501 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 1.15566e-05, step = 501\n",
      "INFO:tensorflow:Saving checkpoints for 600 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.00338809.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-16:00:08\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-600\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-16:00:08\n",
      "INFO:tensorflow:Saving dict for global step 600: global_step = 600, loss = 0.000660179, streaming_mean_absolute_error = 0.0215291, streaming_root_mean_squared_error = 0.0256939\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-600\n",
      "INFO:tensorflow:Saving checkpoints for 601 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 4.22406e-05, step = 601\n",
      "INFO:tensorflow:Saving checkpoints for 700 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.000397511.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-16:00:18\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-700\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-16:00:18\n",
      "INFO:tensorflow:Saving dict for global step 700: global_step = 700, loss = 0.000111751, streaming_mean_absolute_error = 0.0100259, streaming_root_mean_squared_error = 0.0105712\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-700\n",
      "INFO:tensorflow:Saving checkpoints for 701 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 3.31209e-05, step = 701\n",
      "INFO:tensorflow:Saving checkpoints for 800 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.000232355.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-16:00:28\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-800\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-16:00:28\n",
      "INFO:tensorflow:Saving dict for global step 800: global_step = 800, loss = 0.000126429, streaming_mean_absolute_error = 0.0109421, streaming_root_mean_squared_error = 0.011244\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-800\n",
      "INFO:tensorflow:Saving checkpoints for 801 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 6.64472e-05, step = 801\n",
      "INFO:tensorflow:Saving checkpoints for 900 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.00168738.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-16:00:38\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-900\n",
      "INFO:tensorflow:Evaluation [1/10]\n",
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-16:00:38\n",
      "INFO:tensorflow:Saving dict for global step 900: global_step = 900, loss = 0.000607481, streaming_mean_absolute_error = 0.0241623, streaming_root_mean_squared_error = 0.0246471\n",
      "INFO:tensorflow:Training model for 100 steps\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Create CheckpointSaverHook.\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-900\n",
      "INFO:tensorflow:Saving checkpoints for 901 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:loss = 0.000355484, step = 901\n",
      "INFO:tensorflow:Saving checkpoints for 1000 into /tmp/polyaxon_logs/sin/model.ckpt.\n",
      "INFO:tensorflow:Loss for final step: 0.000158104.\n",
      "INFO:tensorflow:Evaluating model now.\n",
      "WARNING:tensorflow:enqueue_data was called with shuffle=False and num_threads > 1. This will create multiple threads, all reading the array/dataframe in order. If you want examples read in order, use one thread; if you want multiple threads, enable shuffling.\n",
      "INFO:tensorflow:Starting evaluation at 2017-06-15-16:00:48\n",
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-1000\n",
      "INFO:tensorflow:Evaluation [1/10]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Evaluation [2/10]\n",
      "INFO:tensorflow:Evaluation [3/10]\n",
      "INFO:tensorflow:Evaluation [4/10]\n",
      "INFO:tensorflow:Evaluation [5/10]\n",
      "INFO:tensorflow:Evaluation [6/10]\n",
      "INFO:tensorflow:Evaluation [7/10]\n",
      "INFO:tensorflow:Evaluation [8/10]\n",
      "INFO:tensorflow:Evaluation [9/10]\n",
      "INFO:tensorflow:Evaluation [10/10]\n",
      "INFO:tensorflow:Finished evaluation at 2017-06-15-16:00:48\n",
      "INFO:tensorflow:Saving dict for global step 1000: global_step = 1000, loss = 0.000126559, streaming_mean_absolute_error = 0.0104135, streaming_root_mean_squared_error = 0.0112498\n",
      "INFO:tensorflow:Stop training model as max steps reached\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'global_step': 1000,\n",
       "  'loss': 0.00012655852,\n",
       "  'streaming_mean_absolute_error': 0.010413513,\n",
       "  'streaming_root_mean_squared_error': 0.011249823},\n",
       " [])"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xp.continuous_train_and_evaluate()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from /tmp/polyaxon_logs/sin/model.ckpt-1000\n",
      "MSE: 0.001129\n"
     ]
    }
   ],
   "source": [
    "predicted = [i['results'] for i in xp.estimator.predict(numpy_input_fn({'x': X['test']}, shuffle=False))]\n",
    "rmse = np.sqrt(((predicted - y['test']) ** 2).mean(axis=0))\n",
    "score = mean_squared_error(predicted, y['test'])\n",
    "print (\"MSE: %f\" % score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x116757d30>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEACAYAAAC3adEgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\noGmapj1EFwZN0zTtIbowaJqmaQ/5fzBO8gn53xmpAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x116773a90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_predicted, = plt.plot(predicted, label='predicted')\n",
    "plot_test, = plt.plot(y['test'], label='test')\n",
    "plt.legend(handles=[plot_predicted, plot_test])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    ""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    ""
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3.0
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}