Skip to content

Instantly share code, notes, and snippets.

@EMCP
Forked from martinwicke/automobile.ipynb
Created August 2, 2017 11:16
Show Gist options
  • Save EMCP/3b8f6f0f56a5f507a12d351789a5bb19 to your computer and use it in GitHub Desktop.
Save EMCP/3b8f6f0f56a5f507a12d351789a5bb19 to your computer and use it in GitHub Desktop.
Estimator demo using Automobile dataset
Display the source blob
Display the rendered blob
Raw
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"from __future__ import print_function\n",
"from __future__ import division\n",
"from __future__ import absolute_import"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# First thing to do: Download https://archive.ics.uci.edu/ml/machine-learning-databases/autos/imports-85.data"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# We're using pandas to read the CSV file. This is easy for small datasets, but for large and complex datasets,\n",
"# tensorflow parsing and processing functions are more powerful.\n",
"import pandas as pd\n",
"import numpy as np"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# The CSV file does not have a header, so we have to fill in column names.\n",
"names = [\n",
" 'symboling', \n",
" 'normalized-losses', \n",
" 'make', \n",
" 'fuel-type', \n",
" 'aspiration',\n",
" 'num-of-doors',\n",
" 'body-style',\n",
" 'drive-wheels',\n",
" 'engine-location',\n",
" 'wheel-base',\n",
" 'length',\n",
" 'width',\n",
" 'height',\n",
" 'curb-weight',\n",
" 'engine-type',\n",
" 'num-of-cylinders',\n",
" 'engine-size',\n",
" 'fuel-system',\n",
" 'bore',\n",
" 'stroke',\n",
" 'compression-ratio',\n",
" 'horsepower',\n",
" 'peak-rpm',\n",
" 'city-mpg',\n",
" 'highway-mpg',\n",
" 'price',\n",
"]\n",
"\n",
"# We also have to specify dtypes.\n",
"dtypes = {\n",
" 'symboling': np.int32, \n",
" 'normalized-losses': np.float32, \n",
" 'make': str, \n",
" 'fuel-type': str, \n",
" 'aspiration': str,\n",
" 'num-of-doors': str,\n",
" 'body-style': str,\n",
" 'drive-wheels': str,\n",
" 'engine-location': str,\n",
" 'wheel-base': np.float32,\n",
" 'length': np.float32,\n",
" 'width': np.float32,\n",
" 'height': np.float32,\n",
" 'curb-weight': np.float32,\n",
" 'engine-type': str,\n",
" 'num-of-cylinders': str,\n",
" 'engine-size': np.float32,\n",
" 'fuel-system': str,\n",
" 'bore': np.float32,\n",
" 'stroke': np.float32,\n",
" 'compression-ratio': np.float32,\n",
" 'horsepower': np.float32,\n",
" 'peak-rpm': np.float32,\n",
" 'city-mpg': np.float32,\n",
" 'highway-mpg': np.float32,\n",
" 'price': np.float32, \n",
"}"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# Read the file.\n",
"df = pd.read_csv('imports-85.data', names=names, dtype=dtypes, na_values='?')"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# Some rows don't have price data, we can't use those.\n",
"df = df.dropna(axis='rows', how='any', subset=['price'])"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# Fill missing values in continuous columns with zeros instead of NaN.\n",
"float_columns = [k for k,v in dtypes.items() if v == np.float32]\n",
"df[float_columns] = df[float_columns].fillna(value=0., axis='columns')\n",
"# Fill missing values in continuous columns with '' instead of NaN (NaN mixed with strings is very bad for us).\n",
"string_columns = [k for k,v in dtypes.items() if v == str]\n",
"df[string_columns] = df[string_columns].fillna(value='', axis='columns')"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# Split the data into a training set and an eval set.\n",
"training_data = df[:160]\n",
"eval_data = df[160:]\n",
"\n",
"# Separate input features from labels\n",
"training_label = training_data.pop('price')\n",
"eval_label = eval_data.pop('price')"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"please make sure that version >= 1.2:\n",
"1.2.0-rc0\n"
]
}
],
"source": [
"# Now we can start using some TensorFlow.\n",
"import tensorflow as tf\n",
"print('please make sure that version >= 1.2:')\n",
"print(tf.__version__)"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# Make input function for training: \n",
"# num_epochs=None -> will cycle through input data forever\n",
"# shuffle=True -> randomize order of input data\n",
"training_input_fn = tf.estimator.inputs.pandas_input_fn(x=training_data, y=training_label, batch_size=64, shuffle=True, num_epochs=None)\n",
"\n",
"# Make input function for evaluation:\n",
"# shuffle=False -> do not randomize input data\n",
"eval_input_fn = tf.estimator.inputs.pandas_input_fn(x=eval_data, y=eval_label, batch_size=64, shuffle=False)"
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"# Describe how the model should interpret the inputs. The names of the feature columns have to match the names\n",
"# of the series in the dataframe.\n",
"\n",
"symboling = tf.feature_column.numeric_column('symboling')\n",
"normalized_losses = tf.feature_column.numeric_column('normalized-losses')\n",
"make = tf.feature_column.categorical_column_with_hash_bucket('make', 50)\n",
"fuel_type = tf.feature_column.categorical_column_with_vocabulary_list('fuel-type', vocabulary_list=['diesel', 'gas'])\n",
"aspiration = tf.feature_column.categorical_column_with_vocabulary_list('aspiration', vocabulary_list=['std', 'turbo'])\n",
"num_of_doors = tf.feature_column.categorical_column_with_vocabulary_list('num-of-doors', vocabulary_list=['two', 'four'])\n",
"body_style = tf.feature_column.categorical_column_with_vocabulary_list('body-style', vocabulary_list=['hardtop', 'wagon', 'sedan', 'hatchback', 'convertible'])\n",
"drive_wheels = tf.feature_column.categorical_column_with_vocabulary_list('drive-wheels', vocabulary_list=['4wd', 'rwd', 'fwd'])\n",
"engine_location = tf.feature_column.categorical_column_with_vocabulary_list('engine-location', vocabulary_list=['front', 'rear'])\n",
"wheel_base = tf.feature_column.numeric_column('wheel-base')\n",
"length = tf.feature_column.numeric_column('length')\n",
"width = tf.feature_column.numeric_column('width')\n",
"height = tf.feature_column.numeric_column('height')\n",
"curb_weight = tf.feature_column.numeric_column('curb-weight')\n",
"engine_type = tf.feature_column.categorical_column_with_vocabulary_list('engine-type', ['dohc', 'dohcv', 'l', 'ohc', 'ohcf', 'ohcv', 'rotor'])\n",
"num_of_cylinders = tf.feature_column.categorical_column_with_vocabulary_list('num-of-cylinders', ['eight', 'five', 'four', 'six', 'three', 'twelve', 'two'])\n",
"engine_size = tf.feature_column.numeric_column('engine-size')\n",
"fuel_system = tf.feature_column.categorical_column_with_vocabulary_list('fuel-system', ['1bbl', '2bbl', '4bbl', 'idi', 'mfi', 'mpfi', 'spdi', 'spfi'])\n",
"bore = tf.feature_column.numeric_column('bore')\n",
"stroke = tf.feature_column.numeric_column('stroke')\n",
"compression_ratio = tf.feature_column.numeric_column('compression-ratio')\n",
"horsepower = tf.feature_column.numeric_column('horsepower')\n",
"peak_rpm = tf.feature_column.numeric_column('peak-rpm')\n",
"city_mpg = tf.feature_column.numeric_column('city-mpg')\n",
"highway_mpg = tf.feature_column.numeric_column('highway-mpg')"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"linear_features = [symboling, normalized_losses, make, fuel_type, aspiration, num_of_doors,\n",
" body_style, drive_wheels, engine_location, wheel_base, length, width,\n",
" height, curb_weight, engine_type, num_of_cylinders, engine_size, fuel_system,\n",
" bore, stroke, compression_ratio, horsepower, peak_rpm, city_mpg, highway_mpg]"
]
},
{
"cell_type": "code",
"execution_count": 21,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"INFO:tensorflow:Using default config.\n",
"WARNING:tensorflow:Using temporary folder as model directory: /tmp/tmp6ZlFc0\n",
"INFO:tensorflow:Using config: {'_save_checkpoints_secs': 600, '_num_ps_replicas': 0, '_keep_checkpoint_max': 5, '_task_type': None, '_is_chief': True, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fc2d61bac10>, '_model_dir': '/tmp/tmp6ZlFc0', '_save_checkpoints_steps': None, '_keep_checkpoint_every_n_hours': 10000, '_session_config': None, '_tf_random_seed': None, '_environment': 'local', '_num_worker_replicas': 0, '_task_id': 0, '_save_summary_steps': 100, '_tf_config': gpu_options {\n",
" per_process_gpu_memory_fraction: 1\n",
"}\n",
", '_evaluation_master': '', '_master': ''}\n"
]
}
],
"source": [
"regressor = tf.contrib.learn.LinearRegressor(feature_columns=linear_features)"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"WARNING:tensorflow:From /usr/local/tensorflow/local/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py:625: scalar_summary (from tensorflow.python.ops.logging_ops) is deprecated and will be removed after 2016-11-30.\n",
"Instructions for updating:\n",
"Please switch to tf.summary.scalar. Note that tf.summary.scalar uses the node name instead of the tag. This means that TensorFlow will automatically de-duplicate summary names based on the scope they are created in. Also, passing a tensor or list of tags to a scalar summary op is no longer supported.\n",
"INFO:tensorflow:Create CheckpointSaverHook.\n",
"INFO:tensorflow:Saving checkpoints for 1 into /tmp/tmpNU0eqz/model.ckpt.\n",
"INFO:tensorflow:loss = 2.413e+08, step = 1\n",
"INFO:tensorflow:global_step/sec: 176.967\n",
"INFO:tensorflow:loss = 8.0506e+07, step = 101 (0.570 sec)\n",
"INFO:tensorflow:global_step/sec: 199.97\n",
"INFO:tensorflow:loss = 6.6745e+07, step = 201 (0.496 sec)\n",
"INFO:tensorflow:global_step/sec: 216.747\n",
"INFO:tensorflow:loss = 6.00864e+07, step = 301 (0.465 sec)\n",
"INFO:tensorflow:global_step/sec: 191.496\n",
"INFO:tensorflow:loss = 3.47049e+07, step = 401 (0.522 sec)\n",
"INFO:tensorflow:global_step/sec: 208.642\n",
"INFO:tensorflow:loss = 3.39389e+07, step = 501 (0.478 sec)\n",
"INFO:tensorflow:global_step/sec: 202.928\n",
"INFO:tensorflow:loss = 4.3715e+07, step = 601 (0.494 sec)\n",
"INFO:tensorflow:global_step/sec: 195.74\n",
"INFO:tensorflow:loss = 4.14391e+07, step = 701 (0.510 sec)\n",
"INFO:tensorflow:global_step/sec: 207.903\n",
"INFO:tensorflow:loss = 3.20546e+07, step = 801 (0.481 sec)\n",
"INFO:tensorflow:global_step/sec: 198.331\n",
"INFO:tensorflow:loss = 3.13368e+07, step = 901 (0.505 sec)\n",
"INFO:tensorflow:global_step/sec: 207.86\n",
"INFO:tensorflow:loss = 3.19198e+07, step = 1001 (0.482 sec)\n",
"INFO:tensorflow:global_step/sec: 200.102\n",
"INFO:tensorflow:loss = 3.91971e+07, step = 1101 (0.499 sec)\n",
"INFO:tensorflow:global_step/sec: 194.879\n",
"INFO:tensorflow:loss = 3.59553e+07, step = 1201 (0.513 sec)\n",
"INFO:tensorflow:global_step/sec: 195.484\n",
"INFO:tensorflow:loss = 3.05773e+07, step = 1301 (0.517 sec)\n",
"INFO:tensorflow:global_step/sec: 190.228\n",
"INFO:tensorflow:loss = 3.37149e+07, step = 1401 (0.517 sec)\n",
"INFO:tensorflow:global_step/sec: 200.541\n",
"INFO:tensorflow:loss = 1.99288e+07, step = 1501 (0.504 sec)\n",
"INFO:tensorflow:global_step/sec: 198.025\n",
"INFO:tensorflow:loss = 2.11798e+07, step = 1601 (0.505 sec)\n",
"INFO:tensorflow:global_step/sec: 197.413\n",
"INFO:tensorflow:loss = 2.43542e+07, step = 1701 (0.502 sec)\n",
"INFO:tensorflow:global_step/sec: 204.236\n",
"INFO:tensorflow:loss = 3.86619e+07, step = 1801 (0.493 sec)\n",
"INFO:tensorflow:global_step/sec: 198.07\n",
"INFO:tensorflow:loss = 1.96792e+07, step = 1901 (0.504 sec)\n",
"INFO:tensorflow:global_step/sec: 217.409\n",
"INFO:tensorflow:loss = 2.44097e+07, step = 2001 (0.461 sec)\n",
"INFO:tensorflow:global_step/sec: 210.291\n",
"INFO:tensorflow:loss = 1.76028e+07, step = 2101 (0.476 sec)\n",
"INFO:tensorflow:global_step/sec: 199.749\n",
"INFO:tensorflow:loss = 3.83393e+07, step = 2201 (0.500 sec)\n",
"INFO:tensorflow:global_step/sec: 214.489\n",
"INFO:tensorflow:loss = 1.82325e+07, step = 2301 (0.466 sec)\n",
"INFO:tensorflow:global_step/sec: 202.679\n",
"INFO:tensorflow:loss = 3.6999e+07, step = 2401 (0.491 sec)\n",
"INFO:tensorflow:global_step/sec: 212.171\n",
"INFO:tensorflow:loss = 3.70806e+07, step = 2501 (0.475 sec)\n",
"INFO:tensorflow:global_step/sec: 197.062\n",
"INFO:tensorflow:loss = 3.35983e+07, step = 2601 (0.507 sec)\n",
"INFO:tensorflow:global_step/sec: 204.514\n",
"INFO:tensorflow:loss = 3.34915e+07, step = 2701 (0.489 sec)\n",
"INFO:tensorflow:global_step/sec: 204.905\n",
"INFO:tensorflow:loss = 1.7547e+07, step = 2801 (0.488 sec)\n",
"INFO:tensorflow:global_step/sec: 204.146\n",
"INFO:tensorflow:loss = 2.43095e+07, step = 2901 (0.489 sec)\n",
"INFO:tensorflow:global_step/sec: 196.715\n",
"INFO:tensorflow:loss = 1.35012e+07, step = 3001 (0.508 sec)\n",
"INFO:tensorflow:global_step/sec: 198.609\n",
"INFO:tensorflow:loss = 3.57368e+07, step = 3101 (0.500 sec)\n",
"INFO:tensorflow:global_step/sec: 199.429\n",
"INFO:tensorflow:loss = 1.60761e+07, step = 3201 (0.505 sec)\n",
"INFO:tensorflow:global_step/sec: 198.273\n",
"INFO:tensorflow:loss = 2.35629e+07, step = 3301 (0.504 sec)\n",
"INFO:tensorflow:global_step/sec: 195.878\n",
"INFO:tensorflow:loss = 2.66036e+07, step = 3401 (0.511 sec)\n",
"INFO:tensorflow:global_step/sec: 212.195\n",
"INFO:tensorflow:loss = 2.08395e+07, step = 3501 (0.471 sec)\n",
"INFO:tensorflow:global_step/sec: 193.941\n",
"INFO:tensorflow:loss = 2.37839e+07, step = 3601 (0.517 sec)\n",
"INFO:tensorflow:global_step/sec: 204.155\n",
"INFO:tensorflow:loss = 2.1417e+07, step = 3701 (0.490 sec)\n",
"INFO:tensorflow:global_step/sec: 199.325\n",
"INFO:tensorflow:loss = 2.40825e+07, step = 3801 (0.501 sec)\n",
"INFO:tensorflow:global_step/sec: 201.735\n",
"INFO:tensorflow:loss = 2.53497e+07, step = 3901 (0.495 sec)\n",
"INFO:tensorflow:global_step/sec: 195.295\n",
"INFO:tensorflow:loss = 2.46902e+07, step = 4001 (0.512 sec)\n",
"INFO:tensorflow:global_step/sec: 204.348\n",
"INFO:tensorflow:loss = 1.41864e+07, step = 4101 (0.491 sec)\n",
"INFO:tensorflow:global_step/sec: 192.389\n",
"INFO:tensorflow:loss = 1.93114e+07, step = 4201 (0.520 sec)\n",
"INFO:tensorflow:global_step/sec: 201.688\n",
"INFO:tensorflow:loss = 2.07875e+07, step = 4301 (0.494 sec)\n",
"INFO:tensorflow:global_step/sec: 203.185\n",
"INFO:tensorflow:loss = 1.92451e+07, step = 4401 (0.492 sec)\n",
"INFO:tensorflow:global_step/sec: 196.942\n",
"INFO:tensorflow:loss = 2.25099e+07, step = 4501 (0.509 sec)\n",
"INFO:tensorflow:global_step/sec: 200.913\n",
"INFO:tensorflow:loss = 3.70702e+07, step = 4601 (0.497 sec)\n",
"INFO:tensorflow:global_step/sec: 201.091\n",
"INFO:tensorflow:loss = 2.03945e+07, step = 4701 (0.497 sec)\n",
"INFO:tensorflow:global_step/sec: 222.047\n",
"INFO:tensorflow:loss = 1.86867e+07, step = 4801 (0.449 sec)\n",
"INFO:tensorflow:global_step/sec: 208.696\n",
"INFO:tensorflow:loss = 3.36875e+07, step = 4901 (0.480 sec)\n",
"INFO:tensorflow:global_step/sec: 207.015\n",
"INFO:tensorflow:loss = 2.07768e+07, step = 5001 (0.481 sec)\n",
"INFO:tensorflow:global_step/sec: 196.225\n",
"INFO:tensorflow:loss = 2.37784e+07, step = 5101 (0.509 sec)\n",
"INFO:tensorflow:global_step/sec: 218.157\n",
"INFO:tensorflow:loss = 2.1156e+07, step = 5201 (0.462 sec)\n",
"INFO:tensorflow:global_step/sec: 211.054\n",
"INFO:tensorflow:loss = 2.725e+07, step = 5301 (0.478 sec)\n",
"INFO:tensorflow:global_step/sec: 209.87\n",
"INFO:tensorflow:loss = 2.8351e+07, step = 5401 (0.473 sec)\n",
"INFO:tensorflow:global_step/sec: 221.093\n",
"INFO:tensorflow:loss = 1.76603e+07, step = 5501 (0.447 sec)\n",
"INFO:tensorflow:global_step/sec: 214.878\n",
"INFO:tensorflow:loss = 2.43624e+07, step = 5601 (0.470 sec)\n",
"INFO:tensorflow:global_step/sec: 210.194\n",
"INFO:tensorflow:loss = 1.40775e+07, step = 5701 (0.475 sec)\n",
"INFO:tensorflow:global_step/sec: 229.605\n",
"INFO:tensorflow:loss = 1.59308e+07, step = 5801 (0.433 sec)\n",
"INFO:tensorflow:global_step/sec: 234.067\n",
"INFO:tensorflow:loss = 2.976e+07, step = 5901 (0.432 sec)\n",
"INFO:tensorflow:global_step/sec: 214.035\n",
"INFO:tensorflow:loss = 2.19687e+07, step = 6001 (0.464 sec)\n",
"INFO:tensorflow:global_step/sec: 223.271\n",
"INFO:tensorflow:loss = 2.21959e+07, step = 6101 (0.453 sec)\n",
"INFO:tensorflow:global_step/sec: 207.604\n",
"INFO:tensorflow:loss = 1.83792e+07, step = 6201 (0.478 sec)\n",
"INFO:tensorflow:global_step/sec: 208.606\n",
"INFO:tensorflow:loss = 1.93478e+07, step = 6301 (0.480 sec)\n",
"INFO:tensorflow:global_step/sec: 197.773\n",
"INFO:tensorflow:loss = 2.01267e+07, step = 6401 (0.506 sec)\n",
"INFO:tensorflow:global_step/sec: 209.845\n",
"INFO:tensorflow:loss = 1.29307e+07, step = 6501 (0.479 sec)\n",
"INFO:tensorflow:global_step/sec: 194.682\n",
"INFO:tensorflow:loss = 2.8945e+07, step = 6601 (0.511 sec)\n",
"INFO:tensorflow:global_step/sec: 204.713\n",
"INFO:tensorflow:loss = 2.75792e+07, step = 6701 (0.487 sec)\n",
"INFO:tensorflow:global_step/sec: 196.828\n",
"INFO:tensorflow:loss = 1.97477e+07, step = 6801 (0.505 sec)\n",
"INFO:tensorflow:global_step/sec: 205.33\n",
"INFO:tensorflow:loss = 1.99781e+07, step = 6901 (0.491 sec)\n",
"INFO:tensorflow:global_step/sec: 186.738\n",
"INFO:tensorflow:loss = 2.87054e+07, step = 7001 (0.534 sec)\n",
"INFO:tensorflow:global_step/sec: 218.389\n",
"INFO:tensorflow:loss = 1.15199e+07, step = 7101 (0.456 sec)\n",
"INFO:tensorflow:global_step/sec: 206.017\n",
"INFO:tensorflow:loss = 2.23859e+07, step = 7201 (0.488 sec)\n",
"INFO:tensorflow:global_step/sec: 210.881\n",
"INFO:tensorflow:loss = 2.92122e+07, step = 7301 (0.475 sec)\n",
"INFO:tensorflow:global_step/sec: 203.424\n",
"INFO:tensorflow:loss = 2.21403e+07, step = 7401 (0.491 sec)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"INFO:tensorflow:global_step/sec: 191.929\n",
"INFO:tensorflow:loss = 2.00178e+07, step = 7501 (0.518 sec)\n",
"INFO:tensorflow:global_step/sec: 203.697\n",
"INFO:tensorflow:loss = 1.53318e+07, step = 7601 (0.495 sec)\n",
"INFO:tensorflow:global_step/sec: 202.307\n",
"INFO:tensorflow:loss = 2.0854e+07, step = 7701 (0.495 sec)\n",
"INFO:tensorflow:global_step/sec: 203.504\n",
"INFO:tensorflow:loss = 1.40464e+07, step = 7801 (0.490 sec)\n",
"INFO:tensorflow:global_step/sec: 194.591\n",
"INFO:tensorflow:loss = 1.54422e+07, step = 7901 (0.513 sec)\n",
"INFO:tensorflow:global_step/sec: 199.267\n",
"INFO:tensorflow:loss = 2.12707e+07, step = 8001 (0.504 sec)\n",
"INFO:tensorflow:global_step/sec: 195.141\n",
"INFO:tensorflow:loss = 2.22916e+07, step = 8101 (0.509 sec)\n",
"INFO:tensorflow:global_step/sec: 194.245\n",
"INFO:tensorflow:loss = 1.87815e+07, step = 8201 (0.518 sec)\n",
"INFO:tensorflow:global_step/sec: 205.331\n",
"INFO:tensorflow:loss = 1.70511e+07, step = 8301 (0.484 sec)\n",
"INFO:tensorflow:global_step/sec: 193.251\n",
"INFO:tensorflow:loss = 1.845e+07, step = 8401 (0.518 sec)\n",
"INFO:tensorflow:global_step/sec: 194.904\n",
"INFO:tensorflow:loss = 1.34779e+07, step = 8501 (0.515 sec)\n",
"INFO:tensorflow:global_step/sec: 202.893\n",
"INFO:tensorflow:loss = 2.24741e+07, step = 8601 (0.490 sec)\n",
"INFO:tensorflow:global_step/sec: 198.084\n",
"INFO:tensorflow:loss = 1.51334e+07, step = 8701 (0.505 sec)\n",
"INFO:tensorflow:global_step/sec: 195.539\n",
"INFO:tensorflow:loss = 1.41038e+07, step = 8801 (0.515 sec)\n",
"INFO:tensorflow:global_step/sec: 196.798\n",
"INFO:tensorflow:loss = 1.2088e+07, step = 8901 (0.508 sec)\n",
"INFO:tensorflow:global_step/sec: 201.058\n",
"INFO:tensorflow:loss = 2.5259e+07, step = 9001 (0.499 sec)\n",
"INFO:tensorflow:global_step/sec: 206.198\n",
"INFO:tensorflow:loss = 1.56701e+07, step = 9101 (0.484 sec)\n",
"INFO:tensorflow:global_step/sec: 205.651\n",
"INFO:tensorflow:loss = 1.74854e+07, step = 9201 (0.483 sec)\n",
"INFO:tensorflow:global_step/sec: 214.525\n",
"INFO:tensorflow:loss = 1.77746e+07, step = 9301 (0.466 sec)\n",
"INFO:tensorflow:global_step/sec: 199.056\n",
"INFO:tensorflow:loss = 2.04472e+07, step = 9401 (0.503 sec)\n",
"INFO:tensorflow:global_step/sec: 191.095\n",
"INFO:tensorflow:loss = 3.08347e+07, step = 9501 (0.526 sec)\n",
"INFO:tensorflow:global_step/sec: 197.268\n",
"INFO:tensorflow:loss = 1.21763e+07, step = 9601 (0.507 sec)\n",
"INFO:tensorflow:global_step/sec: 209.745\n",
"INFO:tensorflow:loss = 2.25766e+07, step = 9701 (0.476 sec)\n",
"INFO:tensorflow:global_step/sec: 211.345\n",
"INFO:tensorflow:loss = 1.2244e+07, step = 9801 (0.475 sec)\n",
"INFO:tensorflow:global_step/sec: 200.21\n",
"INFO:tensorflow:loss = 1.46577e+07, step = 9901 (0.500 sec)\n",
"INFO:tensorflow:Saving checkpoints for 10000 into /tmp/tmpNU0eqz/model.ckpt.\n",
"INFO:tensorflow:Loss for final step: 1.43916e+07.\n"
]
},
{
"data": {
"text/plain": [
"LinearRegressor(params={'gradient_clip_norm': None, 'head': <tensorflow.contrib.learn.python.learn.estimators.head._RegressionHead object at 0x7f34ca5ac190>, 'joint_weights': False, 'optimizer': None, 'feature_columns': [_NumericColumn(key='symboling', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='normalized-losses', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _HashedCategoricalColumn(key='make', hash_bucket_size=50, dtype=tf.string), _VocabularyListCategoricalColumn(key='fuel-type', vocabulary_list=('diesel', 'gas'), dtype=tf.string, default_value=-1), _VocabularyListCategoricalColumn(key='aspiration', vocabulary_list=('std', 'turbo'), dtype=tf.string, default_value=-1), _VocabularyListCategoricalColumn(key='num-of-doors', vocabulary_list=('two', 'four'), dtype=tf.string, default_value=-1), _VocabularyListCategoricalColumn(key='body-style', vocabulary_list=('hardtop', 'wagon', 'sedan', 'hatchback', 'convertible'), dtype=tf.string, default_value=-1), _VocabularyListCategoricalColumn(key='drive-wheels', vocabulary_list=('4wd', 'rwd', 'fwd'), dtype=tf.string, default_value=-1), _VocabularyListCategoricalColumn(key='engine-location', vocabulary_list=('front', 'rear'), dtype=tf.string, default_value=-1), _NumericColumn(key='wheel-base', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='length', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='width', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='height', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='curb-weight', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _VocabularyListCategoricalColumn(key='engine-type', vocabulary_list=('dohc', 'dohcv', 'l', 'ohc', 'ohcf', 'ohcv', 'rotor'), dtype=tf.string, default_value=-1), _VocabularyListCategoricalColumn(key='num-of-cylinders', vocabulary_list=('eight', 'five', 'four', 'six', 'three', 'twelve', 'two'), dtype=tf.string, default_value=-1), _NumericColumn(key='engine-size', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _VocabularyListCategoricalColumn(key='fuel-system', vocabulary_list=('1bbl', '2bbl', '4bbl', 'idi', 'mfi', 'mpfi', 'spdi', 'spfi'), dtype=tf.string, default_value=-1), _NumericColumn(key='bore', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='stroke', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='compression-ratio', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='horsepower', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='peak-rpm', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='city-mpg', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='highway-mpg', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None)]})"
]
},
"execution_count": 13,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"regressor.fit(input_fn=training_input_fn, steps=10000)"
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"WARNING:tensorflow:From /usr/local/tensorflow/local/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py:625: scalar_summary (from tensorflow.python.ops.logging_ops) is deprecated and will be removed after 2016-11-30.\n",
"Instructions for updating:\n",
"Please switch to tf.summary.scalar. Note that tf.summary.scalar uses the node name instead of the tag. This means that TensorFlow will automatically de-duplicate summary names based on the scope they are created in. Also, passing a tensor or list of tags to a scalar summary op is no longer supported.\n",
"INFO:tensorflow:Starting evaluation at 2017-05-17-21:48:53\n",
"INFO:tensorflow:Restoring parameters from /tmp/tmpNU0eqz/model.ckpt-10000\n",
"INFO:tensorflow:Finished evaluation at 2017-05-17-21:48:54\n",
"INFO:tensorflow:Saving dict for global step 10000: global_step = 10000, loss = 7.96542e+06\n"
]
},
{
"data": {
"text/plain": [
"{'global_step': 10000, 'loss': 7965423.5}"
]
},
"execution_count": 14,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"regressor.evaluate(input_fn=eval_input_fn)"
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"dnn_features = [\n",
" #numerical features\n",
" symboling, normalized_losses, wheel_base, length, width, height, curb_weight, engine_size,\n",
" bore, stroke, compression_ratio, horsepower, peak_rpm, city_mpg, highway_mpg, \n",
" # densify categorical features:\n",
" tf.feature_column.indicator_column(make),\n",
" tf.feature_column.indicator_column(fuel_type),\n",
" tf.feature_column.indicator_column(aspiration),\n",
" tf.feature_column.indicator_column(num_of_doors),\n",
" tf.feature_column.indicator_column(body_style),\n",
" tf.feature_column.indicator_column(drive_wheels), \n",
" tf.feature_column.indicator_column(engine_location),\n",
" tf.feature_column.indicator_column(engine_type),\n",
" tf.feature_column.indicator_column(num_of_cylinders),\n",
" tf.feature_column.indicator_column(fuel_system),\n",
"]"
]
},
{
"cell_type": "code",
"execution_count": 16,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"INFO:tensorflow:Using default config.\n",
"WARNING:tensorflow:Using temporary folder as model directory: /tmp/tmpnqPn3Q\n",
"INFO:tensorflow:Using config: {'_save_checkpoints_secs': 600, '_num_ps_replicas': 0, '_keep_checkpoint_max': 5, '_task_type': None, '_is_chief': True, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fc2d60a9bd0>, '_model_dir': '/tmp/tmpnqPn3Q', '_save_checkpoints_steps': None, '_keep_checkpoint_every_n_hours': 10000, '_session_config': None, '_tf_random_seed': None, '_environment': 'local', '_num_worker_replicas': 0, '_task_id': 0, '_save_summary_steps': 100, '_tf_config': gpu_options {\n",
" per_process_gpu_memory_fraction: 1\n",
"}\n",
", '_evaluation_master': '', '_master': ''}\n"
]
}
],
"source": [
"dnnregressor = tf.contrib.learn.DNNRegressor(feature_columns=dnn_features, hidden_units=[50, 30, 10])"
]
},
{
"cell_type": "code",
"execution_count": 17,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"WARNING:tensorflow:From /usr/local/tensorflow/local/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py:625: scalar_summary (from tensorflow.python.ops.logging_ops) is deprecated and will be removed after 2016-11-30.\n",
"Instructions for updating:\n",
"Please switch to tf.summary.scalar. Note that tf.summary.scalar uses the node name instead of the tag. This means that TensorFlow will automatically de-duplicate summary names based on the scope they are created in. Also, passing a tensor or list of tags to a scalar summary op is no longer supported.\n",
"INFO:tensorflow:Create CheckpointSaverHook.\n",
"INFO:tensorflow:Saving checkpoints for 1 into /tmp/tmpnqPn3Q/model.ckpt.\n",
"INFO:tensorflow:loss = 3.22914e+08, step = 1\n",
"INFO:tensorflow:global_step/sec: 219.187\n",
"INFO:tensorflow:loss = 3.491e+07, step = 101 (0.464 sec)\n",
"INFO:tensorflow:global_step/sec: 205.938\n",
"INFO:tensorflow:loss = 1.59505e+07, step = 201 (0.482 sec)\n",
"INFO:tensorflow:global_step/sec: 205.809\n",
"INFO:tensorflow:loss = 1.67622e+07, step = 301 (0.488 sec)\n",
"INFO:tensorflow:global_step/sec: 197.773\n",
"INFO:tensorflow:loss = 1.92105e+07, step = 401 (0.507 sec)\n",
"INFO:tensorflow:global_step/sec: 195.636\n",
"INFO:tensorflow:loss = 1.33924e+07, step = 501 (0.514 sec)\n",
"INFO:tensorflow:global_step/sec: 196.098\n",
"INFO:tensorflow:loss = 2.0106e+07, step = 601 (0.507 sec)\n",
"INFO:tensorflow:global_step/sec: 200.687\n",
"INFO:tensorflow:loss = 1.12352e+07, step = 701 (0.498 sec)\n",
"INFO:tensorflow:global_step/sec: 204.846\n",
"INFO:tensorflow:loss = 1.22986e+07, step = 801 (0.488 sec)\n",
"INFO:tensorflow:global_step/sec: 213.601\n",
"INFO:tensorflow:loss = 1.2512e+07, step = 901 (0.472 sec)\n",
"INFO:tensorflow:global_step/sec: 211.612\n",
"INFO:tensorflow:loss = 1.48472e+07, step = 1001 (0.464 sec)\n",
"INFO:tensorflow:global_step/sec: 220.501\n",
"INFO:tensorflow:loss = 8.40844e+06, step = 1101 (0.461 sec)\n",
"INFO:tensorflow:global_step/sec: 211.484\n",
"INFO:tensorflow:loss = 8.44797e+06, step = 1201 (0.472 sec)\n",
"INFO:tensorflow:global_step/sec: 212.557\n",
"INFO:tensorflow:loss = 7.00385e+06, step = 1301 (0.467 sec)\n",
"INFO:tensorflow:global_step/sec: 230.789\n",
"INFO:tensorflow:loss = 9.14419e+06, step = 1401 (0.429 sec)\n",
"INFO:tensorflow:global_step/sec: 205.606\n",
"INFO:tensorflow:loss = 9.5279e+06, step = 1501 (0.489 sec)\n",
"INFO:tensorflow:global_step/sec: 206.976\n",
"INFO:tensorflow:loss = 6.6255e+06, step = 1601 (0.488 sec)\n",
"INFO:tensorflow:global_step/sec: 197.825\n",
"INFO:tensorflow:loss = 6.00631e+06, step = 1701 (0.502 sec)\n",
"INFO:tensorflow:global_step/sec: 195.028\n",
"INFO:tensorflow:loss = 7.70543e+06, step = 1801 (0.509 sec)\n",
"INFO:tensorflow:global_step/sec: 219.272\n",
"INFO:tensorflow:loss = 9.4826e+06, step = 1901 (0.464 sec)\n",
"INFO:tensorflow:global_step/sec: 190.787\n",
"INFO:tensorflow:loss = 9.36445e+06, step = 2001 (0.523 sec)\n",
"INFO:tensorflow:global_step/sec: 193.815\n",
"INFO:tensorflow:loss = 1.04711e+07, step = 2101 (0.516 sec)\n",
"INFO:tensorflow:global_step/sec: 209.743\n",
"INFO:tensorflow:loss = 7.58201e+06, step = 2201 (0.477 sec)\n",
"INFO:tensorflow:global_step/sec: 204.692\n",
"INFO:tensorflow:loss = 8.64958e+06, step = 2301 (0.491 sec)\n",
"INFO:tensorflow:global_step/sec: 203.376\n",
"INFO:tensorflow:loss = 6.88543e+06, step = 2401 (0.492 sec)\n",
"INFO:tensorflow:global_step/sec: 202.242\n",
"INFO:tensorflow:loss = 4.69487e+06, step = 2501 (0.494 sec)\n",
"INFO:tensorflow:global_step/sec: 240.692\n",
"INFO:tensorflow:loss = 5.28382e+06, step = 2601 (0.416 sec)\n",
"INFO:tensorflow:global_step/sec: 209.215\n",
"INFO:tensorflow:loss = 4.16122e+06, step = 2701 (0.480 sec)\n",
"INFO:tensorflow:global_step/sec: 198.782\n",
"INFO:tensorflow:loss = 7.92784e+06, step = 2801 (0.497 sec)\n",
"INFO:tensorflow:global_step/sec: 192.573\n",
"INFO:tensorflow:loss = 7.9074e+06, step = 2901 (0.522 sec)\n",
"INFO:tensorflow:global_step/sec: 212.379\n",
"INFO:tensorflow:loss = 6.87462e+06, step = 3001 (0.470 sec)\n",
"INFO:tensorflow:global_step/sec: 205.104\n",
"INFO:tensorflow:loss = 6.78068e+06, step = 3101 (0.488 sec)\n",
"INFO:tensorflow:global_step/sec: 215.02\n",
"INFO:tensorflow:loss = 6.94585e+06, step = 3201 (0.463 sec)\n",
"INFO:tensorflow:global_step/sec: 197.712\n",
"INFO:tensorflow:loss = 1.02954e+07, step = 3301 (0.507 sec)\n",
"INFO:tensorflow:global_step/sec: 197.414\n",
"INFO:tensorflow:loss = 5.35531e+06, step = 3401 (0.505 sec)\n",
"INFO:tensorflow:global_step/sec: 204.421\n",
"INFO:tensorflow:loss = 4.74769e+06, step = 3501 (0.490 sec)\n",
"INFO:tensorflow:global_step/sec: 202.362\n",
"INFO:tensorflow:loss = 3.56324e+06, step = 3601 (0.495 sec)\n",
"INFO:tensorflow:global_step/sec: 208.304\n",
"INFO:tensorflow:loss = 7.61136e+06, step = 3701 (0.475 sec)\n",
"INFO:tensorflow:global_step/sec: 222.428\n",
"INFO:tensorflow:loss = 6.08336e+06, step = 3801 (0.447 sec)\n",
"INFO:tensorflow:global_step/sec: 209.433\n",
"INFO:tensorflow:loss = 3.95651e+06, step = 3901 (0.485 sec)\n",
"INFO:tensorflow:global_step/sec: 187.938\n",
"INFO:tensorflow:loss = 5.55149e+06, step = 4001 (0.533 sec)\n",
"INFO:tensorflow:global_step/sec: 205.872\n",
"INFO:tensorflow:loss = 6.80143e+06, step = 4101 (0.486 sec)\n",
"INFO:tensorflow:global_step/sec: 183.963\n",
"INFO:tensorflow:loss = 7.18965e+06, step = 4201 (0.542 sec)\n",
"INFO:tensorflow:global_step/sec: 193.573\n",
"INFO:tensorflow:loss = 4.6761e+06, step = 4301 (0.517 sec)\n",
"INFO:tensorflow:global_step/sec: 200.71\n",
"INFO:tensorflow:loss = 2.98158e+06, step = 4401 (0.500 sec)\n",
"INFO:tensorflow:global_step/sec: 207.15\n",
"INFO:tensorflow:loss = 6.64037e+06, step = 4501 (0.484 sec)\n",
"INFO:tensorflow:global_step/sec: 199.069\n",
"INFO:tensorflow:loss = 4.07609e+06, step = 4601 (0.499 sec)\n",
"INFO:tensorflow:global_step/sec: 197.118\n",
"INFO:tensorflow:loss = 2.61921e+06, step = 4701 (0.504 sec)\n",
"INFO:tensorflow:global_step/sec: 192.8\n",
"INFO:tensorflow:loss = 5.55107e+06, step = 4801 (0.516 sec)\n",
"INFO:tensorflow:global_step/sec: 189.872\n",
"INFO:tensorflow:loss = 3.51778e+06, step = 4901 (0.535 sec)\n",
"INFO:tensorflow:global_step/sec: 211.093\n",
"INFO:tensorflow:loss = 4.04648e+06, step = 5001 (0.472 sec)\n",
"INFO:tensorflow:global_step/sec: 184.25\n",
"INFO:tensorflow:loss = 3.56233e+06, step = 5101 (0.541 sec)\n",
"INFO:tensorflow:global_step/sec: 210.133\n",
"INFO:tensorflow:loss = 3.7764e+06, step = 5201 (0.482 sec)\n",
"INFO:tensorflow:global_step/sec: 198.005\n",
"INFO:tensorflow:loss = 2.44377e+06, step = 5301 (0.495 sec)\n",
"INFO:tensorflow:global_step/sec: 198.66\n",
"INFO:tensorflow:loss = 6.5769e+06, step = 5401 (0.511 sec)\n",
"INFO:tensorflow:global_step/sec: 190.708\n",
"INFO:tensorflow:loss = 3.70076e+06, step = 5501 (0.524 sec)\n",
"INFO:tensorflow:global_step/sec: 204.201\n",
"INFO:tensorflow:loss = 6.93209e+06, step = 5601 (0.483 sec)\n",
"INFO:tensorflow:global_step/sec: 202.983\n",
"INFO:tensorflow:loss = 3.69568e+06, step = 5701 (0.496 sec)\n",
"INFO:tensorflow:global_step/sec: 206.596\n",
"INFO:tensorflow:loss = 2.59145e+06, step = 5801 (0.487 sec)\n",
"INFO:tensorflow:global_step/sec: 211.526\n",
"INFO:tensorflow:loss = 4.70694e+06, step = 5901 (0.470 sec)\n",
"INFO:tensorflow:global_step/sec: 213.76\n",
"INFO:tensorflow:loss = 2.60545e+06, step = 6001 (0.466 sec)\n",
"INFO:tensorflow:global_step/sec: 220.998\n",
"INFO:tensorflow:loss = 5.55105e+06, step = 6101 (0.453 sec)\n",
"INFO:tensorflow:global_step/sec: 200.507\n",
"INFO:tensorflow:loss = 3.21904e+06, step = 6201 (0.496 sec)\n",
"INFO:tensorflow:global_step/sec: 222.679\n",
"INFO:tensorflow:loss = 4.14268e+06, step = 6301 (0.451 sec)\n",
"INFO:tensorflow:global_step/sec: 229.058\n",
"INFO:tensorflow:loss = 4.81932e+06, step = 6401 (0.436 sec)\n",
"INFO:tensorflow:global_step/sec: 208.461\n",
"INFO:tensorflow:loss = 3.18166e+06, step = 6501 (0.483 sec)\n",
"INFO:tensorflow:global_step/sec: 218.974\n",
"INFO:tensorflow:loss = 3.5313e+06, step = 6601 (0.456 sec)\n",
"INFO:tensorflow:global_step/sec: 207.385\n",
"INFO:tensorflow:loss = 2.94534e+06, step = 6701 (0.482 sec)\n",
"INFO:tensorflow:global_step/sec: 213.043\n",
"INFO:tensorflow:loss = 2.06027e+06, step = 6801 (0.471 sec)\n",
"INFO:tensorflow:global_step/sec: 198.342\n",
"INFO:tensorflow:loss = 3.38289e+06, step = 6901 (0.498 sec)\n",
"INFO:tensorflow:global_step/sec: 198.456\n",
"INFO:tensorflow:loss = 4.20299e+06, step = 7001 (0.505 sec)\n",
"INFO:tensorflow:global_step/sec: 200.212\n",
"INFO:tensorflow:loss = 3.04473e+06, step = 7101 (0.507 sec)\n",
"INFO:tensorflow:global_step/sec: 198.17\n",
"INFO:tensorflow:loss = 3.72165e+06, step = 7201 (0.496 sec)\n",
"INFO:tensorflow:global_step/sec: 212.793\n",
"INFO:tensorflow:loss = 4.78442e+06, step = 7301 (0.470 sec)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"INFO:tensorflow:global_step/sec: 198.116\n",
"INFO:tensorflow:loss = 4.21432e+06, step = 7401 (0.510 sec)\n",
"INFO:tensorflow:global_step/sec: 202.815\n",
"INFO:tensorflow:loss = 3.10884e+06, step = 7501 (0.493 sec)\n",
"INFO:tensorflow:global_step/sec: 199.3\n",
"INFO:tensorflow:loss = 2.30774e+06, step = 7601 (0.498 sec)\n",
"INFO:tensorflow:global_step/sec: 227.64\n",
"INFO:tensorflow:loss = 3.16538e+06, step = 7701 (0.445 sec)\n",
"INFO:tensorflow:global_step/sec: 216.803\n",
"INFO:tensorflow:loss = 2.16325e+06, step = 7801 (0.456 sec)\n",
"INFO:tensorflow:global_step/sec: 222.97\n",
"INFO:tensorflow:loss = 4.19254e+06, step = 7901 (0.446 sec)\n",
"INFO:tensorflow:global_step/sec: 222.506\n",
"INFO:tensorflow:loss = 3.27005e+06, step = 8001 (0.454 sec)\n",
"INFO:tensorflow:global_step/sec: 220.794\n",
"INFO:tensorflow:loss = 3.39485e+06, step = 8101 (0.451 sec)\n",
"INFO:tensorflow:global_step/sec: 204.882\n",
"INFO:tensorflow:loss = 2.88965e+06, step = 8201 (0.493 sec)\n",
"INFO:tensorflow:global_step/sec: 210.89\n",
"INFO:tensorflow:loss = 4.03541e+06, step = 8301 (0.473 sec)\n",
"INFO:tensorflow:global_step/sec: 198.464\n",
"INFO:tensorflow:loss = 1.8358e+06, step = 8401 (0.508 sec)\n",
"INFO:tensorflow:global_step/sec: 198.038\n",
"INFO:tensorflow:loss = 2.85165e+06, step = 8501 (0.500 sec)\n",
"INFO:tensorflow:global_step/sec: 198.361\n",
"INFO:tensorflow:loss = 1.68505e+06, step = 8601 (0.504 sec)\n",
"INFO:tensorflow:global_step/sec: 213.096\n",
"INFO:tensorflow:loss = 3.16152e+06, step = 8701 (0.471 sec)\n",
"INFO:tensorflow:global_step/sec: 209.367\n",
"INFO:tensorflow:loss = 2.23255e+06, step = 8801 (0.475 sec)\n",
"INFO:tensorflow:global_step/sec: 199.187\n",
"INFO:tensorflow:loss = 2.44422e+06, step = 8901 (0.503 sec)\n",
"INFO:tensorflow:global_step/sec: 198.963\n",
"INFO:tensorflow:loss = 2.98126e+06, step = 9001 (0.504 sec)\n",
"INFO:tensorflow:global_step/sec: 192.586\n",
"INFO:tensorflow:loss = 3.03181e+06, step = 9101 (0.524 sec)\n",
"INFO:tensorflow:global_step/sec: 201.384\n",
"INFO:tensorflow:loss = 2.13724e+06, step = 9201 (0.488 sec)\n",
"INFO:tensorflow:global_step/sec: 199.785\n",
"INFO:tensorflow:loss = 2.05773e+06, step = 9301 (0.503 sec)\n",
"INFO:tensorflow:global_step/sec: 212.812\n",
"INFO:tensorflow:loss = 2.25637e+06, step = 9401 (0.472 sec)\n",
"INFO:tensorflow:global_step/sec: 197.215\n",
"INFO:tensorflow:loss = 2.08347e+06, step = 9501 (0.504 sec)\n",
"INFO:tensorflow:global_step/sec: 205.228\n",
"INFO:tensorflow:loss = 1.82127e+06, step = 9601 (0.488 sec)\n",
"INFO:tensorflow:global_step/sec: 205.092\n",
"INFO:tensorflow:loss = 3.25608e+06, step = 9701 (0.490 sec)\n",
"INFO:tensorflow:global_step/sec: 209.62\n",
"INFO:tensorflow:loss = 2.75633e+06, step = 9801 (0.472 sec)\n",
"INFO:tensorflow:global_step/sec: 203.143\n",
"INFO:tensorflow:loss = 2.25092e+06, step = 9901 (0.497 sec)\n",
"INFO:tensorflow:Saving checkpoints for 10000 into /tmp/tmpnqPn3Q/model.ckpt.\n",
"INFO:tensorflow:Loss for final step: 1.62302e+06.\n"
]
},
{
"data": {
"text/plain": [
"DNNRegressor(params={'head': <tensorflow.contrib.learn.python.learn.estimators.head._RegressionHead object at 0x7fc2d60a98d0>, 'hidden_units': [50, 30, 10], 'feature_columns': (_NumericColumn(key='symboling', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='normalized-losses', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='wheel-base', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='length', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='width', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='height', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='curb-weight', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='engine-size', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='bore', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='stroke', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='compression-ratio', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='horsepower', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='peak-rpm', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='city-mpg', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _NumericColumn(key='highway-mpg', shape=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), _IndicatorColumn(categorical_column=_HashedCategoricalColumn(key='make', hash_bucket_size=50, dtype=tf.string)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='fuel-type', vocabulary_list=('diesel', 'gas'), dtype=tf.string, default_value=-1)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='aspiration', vocabulary_list=('std', 'turbo'), dtype=tf.string, default_value=-1)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='num-of-doors', vocabulary_list=('two', 'four'), dtype=tf.string, default_value=-1)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='body-style', vocabulary_list=('hardtop', 'wagon', 'sedan', 'hatchback', 'convertible'), dtype=tf.string, default_value=-1)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='drive-wheels', vocabulary_list=('4wd', 'rwd', 'fwd'), dtype=tf.string, default_value=-1)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='engine-location', vocabulary_list=('front', 'rear'), dtype=tf.string, default_value=-1)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='engine-type', vocabulary_list=('dohc', 'dohcv', 'l', 'ohc', 'ohcf', 'ohcv', 'rotor'), dtype=tf.string, default_value=-1)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='num-of-cylinders', vocabulary_list=('eight', 'five', 'four', 'six', 'three', 'twelve', 'two'), dtype=tf.string, default_value=-1)), _IndicatorColumn(categorical_column=_VocabularyListCategoricalColumn(key='fuel-system', vocabulary_list=('1bbl', '2bbl', '4bbl', 'idi', 'mfi', 'mpfi', 'spdi', 'spfi'), dtype=tf.string, default_value=-1))), 'embedding_lr_multipliers': None, 'optimizer': None, 'dropout': None, 'gradient_clip_norm': None, 'activation_fn': <function relu at 0x7fc2d9ed87d0>, 'input_layer_min_slice_size': None})"
]
},
"execution_count": 17,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"dnnregressor.fit(input_fn=training_input_fn, steps=10000)"
]
},
{
"cell_type": "code",
"execution_count": 18,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"WARNING:tensorflow:From /usr/local/tensorflow/local/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py:625: scalar_summary (from tensorflow.python.ops.logging_ops) is deprecated and will be removed after 2016-11-30.\n",
"Instructions for updating:\n",
"Please switch to tf.summary.scalar. Note that tf.summary.scalar uses the node name instead of the tag. This means that TensorFlow will automatically de-duplicate summary names based on the scope they are created in. Also, passing a tensor or list of tags to a scalar summary op is no longer supported.\n",
"INFO:tensorflow:Starting evaluation at 2017-05-18-02:27:00\n",
"INFO:tensorflow:Restoring parameters from /tmp/tmpnqPn3Q/model.ckpt-10000\n",
"INFO:tensorflow:Finished evaluation at 2017-05-18-02:27:00\n",
"INFO:tensorflow:Saving dict for global step 10000: global_step = 10000, loss = 9.60459e+06\n"
]
},
{
"data": {
"text/plain": [
"{'global_step': 10000, 'loss': 9604591.0}"
]
},
"execution_count": 18,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"dnnregressor.evaluate(input_fn=eval_input_fn)"
]
},
{
"cell_type": "code",
"execution_count": 19,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"def experiment_fn(run_config, params):\n",
" # This function makes an Experiment, containing an Estimator and inputs for training and evaluation.\n",
" # You can use params and config here to customize the Estimator depending on the cluster or to use\n",
" # hyperparameter tuning.\n",
"\n",
" # Collect information for training\n",
" return tf.contrib.learn.Experiment(estimator=tf.contrib.learn.LinearRegressor(\n",
" feature_columns=linear_features, config=run_config),\n",
" train_input_fn=training_input_fn,\n",
" train_steps=10000,\n",
" eval_input_fn=eval_input_fn)"
]
},
{
"cell_type": "code",
"execution_count": 22,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"WARNING:tensorflow:uid (from tensorflow.contrib.learn.python.learn.estimators.run_config) is experimental and may change or be removed at any time, and without warning.\n",
"INFO:tensorflow:Using config: {'_model_dir': '/tmp/output_dir', '_save_checkpoints_secs': 600, '_num_ps_replicas': 0, '_keep_checkpoint_max': 5, '_tf_random_seed': None, '_task_type': None, '_environment': 'local', '_is_chief': True, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fc2d5195410>, '_tf_config': gpu_options {\n",
" per_process_gpu_memory_fraction: 1\n",
"}\n",
", '_num_worker_replicas': 0, '_task_id': 0, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_evaluation_master': '', '_keep_checkpoint_every_n_hours': 10000, '_master': '', '_session_config': None}\n",
"WARNING:tensorflow:uid (from tensorflow.contrib.learn.python.learn.estimators.run_config) is experimental and may change or be removed at any time, and without warning.\n",
"WARNING:tensorflow:From /usr/local/tensorflow/local/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/monitors.py:268: __init__ (from tensorflow.contrib.learn.python.learn.monitors) is deprecated and will be removed after 2016-12-05.\n",
"Instructions for updating:\n",
"Monitors are deprecated. Please use tf.train.SessionRunHook.\n",
"WARNING:tensorflow:From /usr/local/tensorflow/local/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py:625: scalar_summary (from tensorflow.python.ops.logging_ops) is deprecated and will be removed after 2016-11-30.\n",
"Instructions for updating:\n",
"Please switch to tf.summary.scalar. Note that tf.summary.scalar uses the node name instead of the tag. This means that TensorFlow will automatically de-duplicate summary names based on the scope they are created in. Also, passing a tensor or list of tags to a scalar summary op is no longer supported.\n",
"INFO:tensorflow:Create CheckpointSaverHook.\n",
"INFO:tensorflow:Saving checkpoints for 1 into /tmp/output_dir/model.ckpt.\n",
"INFO:tensorflow:loss = 3.18835e+08, step = 1\n",
"WARNING:tensorflow:From /usr/local/tensorflow/local/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py:625: scalar_summary (from tensorflow.python.ops.logging_ops) is deprecated and will be removed after 2016-11-30.\n",
"Instructions for updating:\n",
"Please switch to tf.summary.scalar. Note that tf.summary.scalar uses the node name instead of the tag. This means that TensorFlow will automatically de-duplicate summary names based on the scope they are created in. Also, passing a tensor or list of tags to a scalar summary op is no longer supported.\n",
"INFO:tensorflow:Starting evaluation at 2017-05-18-02:28:37\n",
"INFO:tensorflow:Restoring parameters from /tmp/output_dir/model.ckpt-1\n",
"INFO:tensorflow:Evaluation [1/100]\n",
"INFO:tensorflow:Finished evaluation at 2017-05-18-02:28:37\n",
"INFO:tensorflow:Saving dict for global step 1: global_step = 1, loss = 1.43582e+08\n",
"INFO:tensorflow:Validation (step 1): loss = 1.43582e+08, global_step = 1\n",
"INFO:tensorflow:global_step/sec: 20.5355\n",
"INFO:tensorflow:loss = 4.21164e+07, step = 101 (4.870 sec)\n",
"INFO:tensorflow:global_step/sec: 118.296\n",
"INFO:tensorflow:loss = 3.19501e+07, step = 201 (0.845 sec)\n",
"INFO:tensorflow:global_step/sec: 129.594\n",
"INFO:tensorflow:loss = 8.01906e+07, step = 301 (0.778 sec)\n",
"INFO:tensorflow:global_step/sec: 112.126\n",
"INFO:tensorflow:loss = 4.19203e+07, step = 401 (0.885 sec)\n",
"INFO:tensorflow:global_step/sec: 123.946\n",
"INFO:tensorflow:loss = 4.79577e+07, step = 501 (0.807 sec)\n",
"INFO:tensorflow:global_step/sec: 123.068\n",
"INFO:tensorflow:loss = 6.31445e+07, step = 601 (0.813 sec)\n",
"INFO:tensorflow:global_step/sec: 124.585\n",
"INFO:tensorflow:loss = 4.18562e+07, step = 701 (0.802 sec)\n",
"INFO:tensorflow:global_step/sec: 124.402\n",
"INFO:tensorflow:loss = 4.22422e+07, step = 801 (0.804 sec)\n",
"INFO:tensorflow:global_step/sec: 127.149\n",
"INFO:tensorflow:loss = 3.59709e+07, step = 901 (0.786 sec)\n",
"INFO:tensorflow:global_step/sec: 124.612\n",
"INFO:tensorflow:loss = 3.05073e+07, step = 1001 (0.802 sec)\n",
"INFO:tensorflow:global_step/sec: 119.977\n",
"INFO:tensorflow:loss = 4.06392e+07, step = 1101 (0.843 sec)\n",
"INFO:tensorflow:global_step/sec: 118.837\n",
"INFO:tensorflow:loss = 3.96273e+07, step = 1201 (0.833 sec)\n",
"INFO:tensorflow:global_step/sec: 124.73\n",
"INFO:tensorflow:loss = 5.83794e+07, step = 1301 (0.801 sec)\n",
"INFO:tensorflow:global_step/sec: 117.193\n",
"INFO:tensorflow:loss = 2.31712e+07, step = 1401 (0.854 sec)\n",
"INFO:tensorflow:global_step/sec: 123.772\n",
"INFO:tensorflow:loss = 2.47854e+07, step = 1501 (0.808 sec)\n",
"INFO:tensorflow:global_step/sec: 121.554\n",
"INFO:tensorflow:loss = 2.69959e+07, step = 1601 (0.823 sec)\n",
"INFO:tensorflow:global_step/sec: 125.984\n",
"INFO:tensorflow:loss = 2.44264e+07, step = 1701 (0.794 sec)\n",
"INFO:tensorflow:global_step/sec: 121.942\n",
"INFO:tensorflow:loss = 3.27822e+07, step = 1801 (0.820 sec)\n",
"INFO:tensorflow:global_step/sec: 124.062\n",
"INFO:tensorflow:loss = 2.68355e+07, step = 1901 (0.806 sec)\n",
"INFO:tensorflow:global_step/sec: 126.735\n",
"INFO:tensorflow:loss = 3.12895e+07, step = 2001 (0.789 sec)\n",
"INFO:tensorflow:global_step/sec: 119.837\n",
"INFO:tensorflow:loss = 1.82017e+07, step = 2101 (0.835 sec)\n",
"INFO:tensorflow:global_step/sec: 130.215\n",
"INFO:tensorflow:loss = 3.53857e+07, step = 2201 (0.767 sec)\n",
"INFO:tensorflow:global_step/sec: 122.283\n",
"INFO:tensorflow:loss = 4.33676e+07, step = 2301 (0.818 sec)\n",
"INFO:tensorflow:global_step/sec: 121.716\n",
"INFO:tensorflow:loss = 2.25773e+07, step = 2401 (0.821 sec)\n",
"INFO:tensorflow:global_step/sec: 132.131\n",
"INFO:tensorflow:loss = 2.51927e+07, step = 2501 (0.757 sec)\n",
"INFO:tensorflow:global_step/sec: 121.687\n",
"INFO:tensorflow:loss = 1.48666e+07, step = 2601 (0.826 sec)\n",
"INFO:tensorflow:global_step/sec: 121.38\n",
"INFO:tensorflow:loss = 1.77214e+07, step = 2701 (0.821 sec)\n",
"INFO:tensorflow:global_step/sec: 120.194\n",
"INFO:tensorflow:loss = 2.67903e+07, step = 2801 (0.830 sec)\n",
"INFO:tensorflow:global_step/sec: 118.984\n",
"INFO:tensorflow:loss = 1.97603e+07, step = 2901 (0.841 sec)\n",
"INFO:tensorflow:global_step/sec: 121.729\n",
"INFO:tensorflow:loss = 3.27341e+07, step = 3001 (0.822 sec)\n",
"INFO:tensorflow:global_step/sec: 123.155\n",
"INFO:tensorflow:loss = 1.98052e+07, step = 3101 (0.812 sec)\n",
"INFO:tensorflow:global_step/sec: 122.353\n",
"INFO:tensorflow:loss = 2.83327e+07, step = 3201 (0.817 sec)\n",
"INFO:tensorflow:global_step/sec: 120.479\n",
"INFO:tensorflow:loss = 3.48686e+07, step = 3301 (0.830 sec)\n",
"INFO:tensorflow:global_step/sec: 119.914\n",
"INFO:tensorflow:loss = 1.57687e+07, step = 3401 (0.833 sec)\n",
"INFO:tensorflow:global_step/sec: 119.723\n",
"INFO:tensorflow:loss = 3.37787e+07, step = 3501 (0.836 sec)\n",
"INFO:tensorflow:global_step/sec: 117.654\n",
"INFO:tensorflow:loss = 1.89048e+07, step = 3601 (0.850 sec)\n",
"INFO:tensorflow:global_step/sec: 121.744\n",
"INFO:tensorflow:loss = 1.54444e+07, step = 3701 (0.826 sec)\n",
"INFO:tensorflow:global_step/sec: 125.56\n",
"INFO:tensorflow:loss = 2.89857e+07, step = 3801 (0.794 sec)\n",
"INFO:tensorflow:global_step/sec: 122.2\n",
"INFO:tensorflow:loss = 2.58254e+07, step = 3901 (0.816 sec)\n",
"INFO:tensorflow:global_step/sec: 119.617\n",
"INFO:tensorflow:loss = 2.41199e+07, step = 4001 (0.836 sec)\n",
"INFO:tensorflow:global_step/sec: 118.098\n",
"INFO:tensorflow:loss = 1.89337e+07, step = 4101 (0.846 sec)\n",
"INFO:tensorflow:global_step/sec: 119.541\n",
"INFO:tensorflow:loss = 2.81653e+07, step = 4201 (0.837 sec)\n",
"INFO:tensorflow:global_step/sec: 118.926\n",
"INFO:tensorflow:loss = 1.05359e+07, step = 4301 (0.841 sec)\n",
"INFO:tensorflow:global_step/sec: 117.887\n",
"INFO:tensorflow:loss = 2.06676e+07, step = 4401 (0.848 sec)\n",
"INFO:tensorflow:global_step/sec: 119.209\n",
"INFO:tensorflow:loss = 2.89711e+07, step = 4501 (0.839 sec)\n",
"INFO:tensorflow:global_step/sec: 121.083\n",
"INFO:tensorflow:loss = 1.81756e+07, step = 4601 (0.825 sec)\n",
"INFO:tensorflow:global_step/sec: 123.458\n",
"INFO:tensorflow:loss = 2.29722e+07, step = 4701 (0.811 sec)\n",
"INFO:tensorflow:global_step/sec: 119.553\n",
"INFO:tensorflow:loss = 2.18481e+07, step = 4801 (0.836 sec)\n",
"INFO:tensorflow:global_step/sec: 125.019\n",
"INFO:tensorflow:loss = 1.68649e+07, step = 4901 (0.799 sec)\n",
"INFO:tensorflow:global_step/sec: 123.658\n",
"INFO:tensorflow:loss = 2.78938e+07, step = 5001 (0.809 sec)\n",
"INFO:tensorflow:global_step/sec: 121.055\n",
"INFO:tensorflow:loss = 2.37118e+07, step = 5101 (0.825 sec)\n",
"INFO:tensorflow:global_step/sec: 124.397\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"INFO:tensorflow:loss = 1.80479e+07, step = 5201 (0.805 sec)\n",
"INFO:tensorflow:global_step/sec: 124.082\n",
"INFO:tensorflow:loss = 1.42421e+07, step = 5301 (0.806 sec)\n",
"INFO:tensorflow:global_step/sec: 115.894\n",
"INFO:tensorflow:loss = 2.45195e+07, step = 5401 (0.863 sec)\n",
"INFO:tensorflow:global_step/sec: 121.967\n",
"INFO:tensorflow:loss = 2.10585e+07, step = 5501 (0.820 sec)\n",
"INFO:tensorflow:global_step/sec: 121.443\n",
"INFO:tensorflow:loss = 1.59946e+07, step = 5601 (0.824 sec)\n",
"INFO:tensorflow:global_step/sec: 118.556\n",
"INFO:tensorflow:loss = 1.98039e+07, step = 5701 (0.845 sec)\n",
"INFO:tensorflow:global_step/sec: 117.499\n",
"INFO:tensorflow:loss = 1.51192e+07, step = 5801 (0.849 sec)\n",
"INFO:tensorflow:global_step/sec: 115.655\n",
"INFO:tensorflow:loss = 3.23047e+07, step = 5901 (0.864 sec)\n",
"INFO:tensorflow:global_step/sec: 119.006\n",
"INFO:tensorflow:loss = 2.65075e+07, step = 6001 (0.841 sec)\n",
"INFO:tensorflow:global_step/sec: 121.712\n",
"INFO:tensorflow:loss = 2.03057e+07, step = 6101 (0.822 sec)\n",
"INFO:tensorflow:global_step/sec: 122.988\n",
"INFO:tensorflow:loss = 1.98623e+07, step = 6201 (0.813 sec)\n",
"INFO:tensorflow:global_step/sec: 119.101\n",
"INFO:tensorflow:loss = 1.64578e+07, step = 6301 (0.840 sec)\n",
"INFO:tensorflow:global_step/sec: 118.79\n",
"INFO:tensorflow:loss = 1.26528e+07, step = 6401 (0.842 sec)\n",
"INFO:tensorflow:global_step/sec: 115.122\n",
"INFO:tensorflow:loss = 1.27203e+07, step = 6501 (0.869 sec)\n",
"INFO:tensorflow:global_step/sec: 122\n",
"INFO:tensorflow:loss = 2.51296e+07, step = 6601 (0.820 sec)\n",
"INFO:tensorflow:global_step/sec: 117.994\n",
"INFO:tensorflow:loss = 3.57162e+07, step = 6701 (0.848 sec)\n",
"INFO:tensorflow:global_step/sec: 122.801\n",
"INFO:tensorflow:loss = 2.07559e+07, step = 6801 (0.814 sec)\n",
"INFO:tensorflow:global_step/sec: 118.899\n",
"INFO:tensorflow:loss = 2.02879e+07, step = 6901 (0.842 sec)\n",
"INFO:tensorflow:global_step/sec: 120.236\n",
"INFO:tensorflow:loss = 8.61477e+06, step = 7001 (0.831 sec)\n",
"INFO:tensorflow:global_step/sec: 120.851\n",
"INFO:tensorflow:loss = 2.15346e+07, step = 7101 (0.827 sec)\n",
"INFO:tensorflow:global_step/sec: 129.997\n",
"INFO:tensorflow:loss = 2.13486e+07, step = 7201 (0.771 sec)\n",
"INFO:tensorflow:global_step/sec: 117.11\n",
"INFO:tensorflow:loss = 1.58869e+07, step = 7301 (0.854 sec)\n",
"INFO:tensorflow:global_step/sec: 128.525\n",
"INFO:tensorflow:loss = 2.14093e+07, step = 7401 (0.779 sec)\n",
"INFO:tensorflow:global_step/sec: 126.143\n",
"INFO:tensorflow:loss = 2.37592e+07, step = 7501 (0.791 sec)\n",
"INFO:tensorflow:global_step/sec: 120.863\n",
"INFO:tensorflow:loss = 2.27634e+07, step = 7601 (0.828 sec)\n",
"INFO:tensorflow:global_step/sec: 113.427\n",
"INFO:tensorflow:loss = 3.10648e+07, step = 7701 (0.881 sec)\n",
"INFO:tensorflow:global_step/sec: 119.571\n",
"INFO:tensorflow:loss = 2.08745e+07, step = 7801 (0.836 sec)\n",
"INFO:tensorflow:global_step/sec: 128.025\n",
"INFO:tensorflow:loss = 2.24044e+07, step = 7901 (0.781 sec)\n",
"INFO:tensorflow:global_step/sec: 121.197\n",
"INFO:tensorflow:loss = 2.12339e+07, step = 8001 (0.828 sec)\n",
"INFO:tensorflow:global_step/sec: 124.79\n",
"INFO:tensorflow:loss = 2.23047e+07, step = 8101 (0.801 sec)\n",
"INFO:tensorflow:global_step/sec: 124.69\n",
"INFO:tensorflow:loss = 2.48484e+07, step = 8201 (0.800 sec)\n",
"INFO:tensorflow:global_step/sec: 122.447\n",
"INFO:tensorflow:loss = 3.00606e+07, step = 8301 (0.819 sec)\n",
"INFO:tensorflow:global_step/sec: 125.032\n",
"INFO:tensorflow:loss = 2.62958e+07, step = 8401 (0.802 sec)\n",
"INFO:tensorflow:global_step/sec: 124.911\n",
"INFO:tensorflow:loss = 2.19767e+07, step = 8501 (0.797 sec)\n",
"INFO:tensorflow:global_step/sec: 120.855\n",
"INFO:tensorflow:loss = 2.58696e+07, step = 8601 (0.828 sec)\n",
"INFO:tensorflow:global_step/sec: 122.078\n",
"INFO:tensorflow:loss = 3.12598e+07, step = 8701 (0.819 sec)\n",
"INFO:tensorflow:global_step/sec: 123.366\n",
"INFO:tensorflow:loss = 2.26468e+07, step = 8801 (0.811 sec)\n",
"INFO:tensorflow:global_step/sec: 115.682\n",
"INFO:tensorflow:loss = 2.3405e+07, step = 8901 (0.863 sec)\n",
"INFO:tensorflow:global_step/sec: 118.603\n",
"INFO:tensorflow:loss = 1.89545e+07, step = 9001 (0.844 sec)\n",
"INFO:tensorflow:global_step/sec: 121.888\n",
"INFO:tensorflow:loss = 1.80466e+07, step = 9101 (0.821 sec)\n",
"INFO:tensorflow:global_step/sec: 116.303\n",
"INFO:tensorflow:loss = 2.86328e+07, step = 9201 (0.859 sec)\n",
"INFO:tensorflow:global_step/sec: 119.984\n",
"INFO:tensorflow:loss = 1.39935e+07, step = 9301 (0.832 sec)\n",
"INFO:tensorflow:global_step/sec: 120.686\n",
"INFO:tensorflow:loss = 1.27423e+07, step = 9401 (0.832 sec)\n",
"INFO:tensorflow:global_step/sec: 127.189\n",
"INFO:tensorflow:loss = 1.76635e+07, step = 9501 (0.784 sec)\n",
"INFO:tensorflow:global_step/sec: 120.85\n",
"INFO:tensorflow:loss = 1.24075e+07, step = 9601 (0.828 sec)\n",
"INFO:tensorflow:global_step/sec: 127.394\n",
"INFO:tensorflow:loss = 2.11293e+07, step = 9701 (0.785 sec)\n",
"INFO:tensorflow:global_step/sec: 129.532\n",
"INFO:tensorflow:loss = 2.34291e+07, step = 9801 (0.772 sec)\n",
"INFO:tensorflow:global_step/sec: 119.468\n",
"INFO:tensorflow:loss = 2.34618e+07, step = 9901 (0.837 sec)\n",
"INFO:tensorflow:Saving checkpoints for 10000 into /tmp/output_dir/model.ckpt.\n",
"INFO:tensorflow:Loss for final step: 2.59231e+07.\n",
"WARNING:tensorflow:From /usr/local/tensorflow/local/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py:625: scalar_summary (from tensorflow.python.ops.logging_ops) is deprecated and will be removed after 2016-11-30.\n",
"Instructions for updating:\n",
"Please switch to tf.summary.scalar. Note that tf.summary.scalar uses the node name instead of the tag. This means that TensorFlow will automatically de-duplicate summary names based on the scope they are created in. Also, passing a tensor or list of tags to a scalar summary op is no longer supported.\n",
"INFO:tensorflow:Starting evaluation at 2017-05-18-02:30:04\n",
"INFO:tensorflow:Restoring parameters from /tmp/output_dir/model.ckpt-10000\n",
"INFO:tensorflow:Evaluation [1/100]\n",
"INFO:tensorflow:Finished evaluation at 2017-05-18-02:30:05\n",
"INFO:tensorflow:Saving dict for global step 10000: global_step = 10000, loss = 8.2764e+06\n"
]
},
{
"data": {
"text/plain": [
"({'global_step': 10000, 'loss': 8276404.5}, [])"
]
},
"execution_count": 22,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"import shutil\n",
"shutil.rmtree(\"/tmp/output_dir\", ignore_errors=True)\n",
"tf.contrib.learn.learn_runner.run(experiment_fn, run_config=tf.contrib.learn.RunConfig(model_dir=\"/tmp/output_dir\"))"
]
},
{
"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 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": "2.7.6"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment