Skip to content

Instantly share code, notes, and snippets.

@malkayo

malkayo/tensorflow-high-level-apis-notebook.ipynb

Created February 3, 2019 16:13
Show Gist options
  • Save malkayo/71e483c6f234dd5aca0eaaa52a5c24e0 to your computer and use it in GitHub Desktop.
Save malkayo/71e483c6f234dd5aca0eaaa52a5c24e0 to your computer and use it in GitHub Desktop.
Download ZIP
TensorFlow high-level APIs - Notebook.ipynb
Display the source blob
Display the rendered blob
Raw
tensorflow-high-level-apis-notebook.ipynb
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"name": "TensorFlow high-level APIs - Notebook.ipynb",
"version": "0.3.2",
"provenance": [],
"collapsed_sections": [],
"include_colab_link": true
},
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
}
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"<a href=\"https://colab.research.google.com/gist/malkayo/71e483c6f234dd5aca0eaaa52a5c24e0/tensorflow-high-level-apis-notebook.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
]
},
{
"metadata": {
"id": "Kx0YcyptFbTf",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"# TensorFlow high-level APIs - Notebook\n",
"This notebook is meant to help following along the series of video on the TF high-level API ([First video here](https://www.youtube.com/watch?v=oFFbKogYdfc))"
]
},
{
"metadata": {
"id": "YDw7b01EFbTp",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"## Getting the data\n",
"The Forest Cover Type Dataset set mentionned in the video can be found on [Kaggle](https://www.kaggle.com/uciml/forest-cover-type-dataset#covtype.csv) "
]
},
{
"metadata": {
"id": "D4oPDcuhFbTw",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"You can download the data directly on [Kaggle](https://www.kaggle.com/uciml/forest-cover-type-dataset#covtype.csv) or by running the following commands on you local machine with the [Kaggle API](https://github.com/Kaggle/kaggle-api)."
]
},
{
"metadata": {
"id": "-UVTx7RbFbT5",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"!kaggle datasets download -d uciml/forest-cover-type-dataset"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "7_qHH0j1FbUF",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"If you are using Google Collab, you can upload the forest-cover-type-dataset.zip file with the following"
]
},
{
"metadata": {
"colab_type": "code",
"id": "gDiRcPzcf61h",
"outputId": "086c08b6-7117-4c61-b94d-cdaa38534109",
"colab": {
"resources": {
"http://localhost:8080/nbextensions/google.colab/files.js": {
"data": "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",
"ok": true,
"headers": [
[
"content-type",
"application/javascript"
]
],
"status": 200,
"status_text": "OK"
}
},
"base_uri": "https://localhost:8080/",
"height": 77
}
},
"cell_type": "code",
"source": [
"from google.colab import files\n",
"uploaded = files.upload()"
],
"execution_count": 3,
"outputs": [
{
"output_type": "display_data",
"data": {
"text/html": [
"\n",
" <input type=\"file\" id=\"files-fdff2e91-e395-4bed-b1ca-8de594a63c78\" name=\"files[]\" multiple disabled />\n",
" <output id=\"result-fdff2e91-e395-4bed-b1ca-8de594a63c78\">\n",
" Upload widget is only available when the cell has been executed in the\n",
" current browser session. Please rerun this cell to enable.\n",
" </output>\n",
" <script src=\"/nbextensions/google.colab/files.js\"></script> "
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {
"tags": []
}
},
{
"output_type": "stream",
"text": [
"Saving forest-cover-type-dataset.zip to forest-cover-type-dataset.zip\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "ejBlx-mVGcr1",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"We can then unzip the csv file"
]
},
{
"metadata": {
"id": "dLuvliJ0GlIZ",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 69
},
"outputId": "74f61c41-48df-4f50-e61a-87cd3ddb6bee"
},
"cell_type": "code",
"source": [
"!unzip forest-cover-type-dataset.zip\n",
"!rm forest-cover-type-dataset.zip"
],
"execution_count": 4,
"outputs": [
{
"output_type": "stream",
"text": [
"Archive: forest-cover-type-dataset.zip\n",
"replace covtype.csv? [y]es, [n]o, [A]ll, [N]one, [r]ename: y\n",
" inflating: covtype.csv \n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "hKOppxCFFbUQ",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"## Splitting the data into the training and test sets"
]
},
{
"metadata": {
"colab_type": "code",
"id": "5Rar3bD8OAAl",
"colab": {}
},
"cell_type": "code",
"source": [
"import numpy as np\n",
"np.set_printoptions(suppress=True)\n",
"\n",
"from sklearn.model_selection import train_test_split\n",
"\n",
"data = np.genfromtxt(\"./covtype.csv\", delimiter=\",\", skip_header=1, dtype=np.int32)\n",
"\n",
"X = data[:, 0:-1]\n",
"y = data[:, -1]\n",
"\n",
"X_train, X_test, y_train, y_test = train_test_split(\n",
" X, y, test_size=0.33, random_state=42)\n",
"\n",
"training_data = np.column_stack((X_train, y_train))\n",
"np.savetxt(\"./covtype.csv.train\", training_data, \"%d\", delimiter=\",\")\n",
"\n",
"test_data = np.column_stack((X_test, y_test))\n",
"np.savetxt(\"./covtype.csv.test\", test_data, \"%d\", delimiter=\",\")"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "tSO3mtPwFbUY",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"# Part I - Loading data ([Video](https://www.youtube.com/watch?v=oFFbKogYdfc))"
]
},
{
"metadata": {
"colab_type": "code",
"id": "xz5vGSo9OAAt",
"colab": {}
},
"cell_type": "code",
"source": [
"import tensorflow as tf\n",
"import pprint\n",
"\n",
"tf.enable_eager_execution()"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "_B4LxOLmFbUh",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
},
"outputId": "53848eec-132d-4259-e54e-8107941a1444"
},
"cell_type": "code",
"source": [
"a = tf.constant(5)\n",
"b = a * 3\n",
"print(b)"
],
"execution_count": 7,
"outputs": [
{
"output_type": "stream",
"text": [
"tf.Tensor(15, shape=(), dtype=int32)\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"scrolled": true,
"id": "JHHKVZCgFbUr",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 972
},
"outputId": "7825feef-08df-4d1d-f164-de00b92be735"
},
"cell_type": "code",
"source": [
"defaults = [tf.int32] * 55\n",
"\n",
"# tf.contrib.data.CsvDataset is deprecated\n",
"dataset = tf.data.experimental.CsvDataset(['./covtype.csv.train'], defaults)\n",
"pprint.pprint(list(dataset.take(1)))"
],
"execution_count": 8,
"outputs": [
{
"output_type": "stream",
"text": [
"[(<tf.Tensor: id=72, shape=(), dtype=int32, numpy=3075>,\n",
" <tf.Tensor: id=73, shape=(), dtype=int32, numpy=241>,\n",
" <tf.Tensor: id=74, shape=(), dtype=int32, numpy=20>,\n",
" <tf.Tensor: id=75, shape=(), dtype=int32, numpy=382>,\n",
" <tf.Tensor: id=76, shape=(), dtype=int32, numpy=56>,\n",
" <tf.Tensor: id=77, shape=(), dtype=int32, numpy=5772>,\n",
" <tf.Tensor: id=78, shape=(), dtype=int32, numpy=179>,\n",
" <tf.Tensor: id=79, shape=(), dtype=int32, numpy=252>,\n",
" <tf.Tensor: id=80, shape=(), dtype=int32, numpy=207>,\n",
" <tf.Tensor: id=81, shape=(), dtype=int32, numpy=1849>,\n",
" <tf.Tensor: id=82, shape=(), dtype=int32, numpy=1>,\n",
" <tf.Tensor: id=83, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=84, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=85, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=86, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=87, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=88, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=89, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=90, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=91, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=92, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=93, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=94, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=95, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=96, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=97, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=98, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=99, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=100, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=101, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=102, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=103, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=104, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=105, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=106, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=107, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=108, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=109, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=110, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=111, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=112, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=113, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=114, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=115, shape=(), dtype=int32, numpy=1>,\n",
" <tf.Tensor: id=116, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=117, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=118, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=119, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=120, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=121, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=122, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=123, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=124, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=125, shape=(), dtype=int32, numpy=0>,\n",
" <tf.Tensor: id=126, shape=(), dtype=int32, numpy=2>)]\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "IkwNC1q0FbUx",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"col_names = [\"Elevation\",\"Aspect\",\"Slope\",\"Horizontal_Distance_To_Hydrology\",\n",
" \"Vertical_Distance_To_Hydrology\",\"Horizontal_Distance_To_Roadways\",\n",
" \"Hillshade_9am\",\"Hillshade_Noon\",\"Hillshade_3pm\",\n",
" \"Horizontal_Distance_To_Fire_Points\",\"Soil_Type\", \"Cover_Type\"]\n",
"\n",
"def _parse_csv_row(*vals):\n",
" soil_type_t = tf.convert_to_tensor(vals[14:54])\n",
" feat_vals = vals[:10] + (soil_type_t, vals[54])\n",
" features = dict(zip(col_names, feat_vals))\n",
"\n",
" class_label = tf.argmax(vals[10:14], axis=0) # wilderness area\n",
" return features, class_label"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"scrolled": true,
"id": "nkkX9fQ6FbU5",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 1493
},
"outputId": "5ef7c8e4-859a-41eb-fbc0-9f36e5f41405"
},
"cell_type": "code",
"source": [
"dataset = dataset.map(_parse_csv_row).batch(64)\n",
"pprint.pprint(list(dataset.take(1)))"
],
"execution_count": 10,
"outputs": [
{
"output_type": "stream",
"text": [
"[({'Aspect': <tf.Tensor: id=426, shape=(64,), dtype=int32, numpy=\n",
"array([241, 130, 122, 54, 335, 7, 90, 319, 124, 117, 216, 90, 345,\n",
" 45, 163, 347, 115, 217, 42, 17, 354, 109, 144, 350, 10, 360,\n",
" 107, 254, 85, 213, 5, 107, 57, 122, 206, 73, 264, 355, 19,\n",
" 237, 184, 27, 78, 89, 62, 288, 83, 341, 99, 38, 0, 51,\n",
" 202, 78, 193, 86, 34, 45, 346, 354, 92, 112, 159, 15],\n",
" dtype=int32)>,\n",
" 'Cover_Type': <tf.Tensor: id=427, shape=(64,), dtype=int32, numpy=\n",
"array([2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 2, 3, 2, 2, 1, 1, 3, 2, 1, 1, 1, 1,\n",
" 2, 2, 2, 2, 2, 2, 2, 2, 2, 6, 1, 1, 2, 1, 1, 1, 1, 3, 2, 1, 1, 7,\n",
" 2, 1, 2, 1, 2, 1, 2, 2, 2, 1, 2, 2, 2, 2, 2, 1, 6, 4, 7, 1],\n",
" dtype=int32)>,\n",
" 'Elevation': <tf.Tensor: id=428, shape=(64,), dtype=int32, numpy=\n",
"array([3075, 2705, 2926, 3162, 2794, 2800, 3283, 2891, 3437, 3156, 2877,\n",
" 2600, 3047, 2985, 3181, 3111, 2350, 2906, 3110, 3244, 2807, 3201,\n",
" 2926, 2933, 3006, 2515, 2606, 3025, 2551, 3068, 3038, 2429, 3152,\n",
" 2828, 2967, 2972, 3479, 2709, 3249, 2591, 2968, 2821, 2897, 3380,\n",
" 3100, 3191, 2845, 2839, 2896, 3160, 3161, 2785, 2551, 3041, 2997,\n",
" 2887, 2890, 3100, 2886, 3087, 2378, 2233, 3249, 3218], dtype=int32)>,\n",
" 'Hillshade_3pm': <tf.Tensor: id=429, shape=(64,), dtype=int32, numpy=\n",
"array([207, 125, 114, 104, 169, 151, 129, 187, 87, 114, 189, 71, 164,\n",
" 138, 150, 158, 86, 195, 116, 140, 157, 127, 147, 159, 141, 140,\n",
" 102, 191, 118, 183, 145, 124, 64, 134, 168, 75, 219, 149, 140,\n",
" 222, 157, 123, 73, 39, 142, 190, 143, 168, 89, 90, 156, 139,\n",
" 172, 118, 163, 134, 129, 120, 163, 157, 93, 97, 145, 139],\n",
" dtype=int32)>,\n",
" 'Hillshade_9am': <tf.Tensor: id=430, shape=(64,), dtype=int32, numpy=\n",
"array([179, 238, 242, 228, 195, 211, 232, 174, 250, 241, 193, 247, 181,\n",
" 222, 226, 155, 250, 180, 221, 210, 209, 235, 226, 190, 202, 172,\n",
" 244, 195, 235, 201, 199, 236, 227, 233, 215, 239, 161, 175, 212,\n",
" 148, 222, 211, 242, 248, 224, 189, 225, 183, 246, 211, 218, 223,\n",
" 212, 234, 214, 229, 218, 223, 191, 204, 244, 247, 229, 208],\n",
" dtype=int32)>,\n",
" 'Hillshade_Noon': <tf.Tensor: id=431, shape=(64,), dtype=int32, numpy=\n",
"array([252, 234, 229, 203, 225, 225, 228, 222, 220, 228, 254, 200, 208,\n",
" 225, 242, 182, 217, 253, 206, 214, 228, 231, 239, 212, 209, 182,\n",
" 221, 247, 222, 254, 209, 229, 175, 234, 248, 194, 244, 192, 216,\n",
" 248, 245, 202, 195, 180, 230, 239, 233, 213, 213, 178, 238, 227,\n",
" 252, 221, 251, 230, 213, 211, 217, 223, 213, 220, 242, 212],\n",
" dtype=int32)>,\n",
" 'Horizontal_Distance_To_Fire_Points': <tf.Tensor: id=432, shape=(64,), dtype=int32, numpy=\n",
"array([1849, 1651, 4341, 5246, 2114, 2794, 2279, 2313, 2818, 1934, 2047,\n",
" 878, 1382, 5409, 1008, 2088, 646, 726, 1343, 2350, 2744, 2546,\n",
" 1979, 2888, 2023, 1722, 624, 2142, 760, 1092, 752, 1366, 2501,\n",
" 1940, 774, 2589, 1622, 2442, 492, 1095, 4110, 3187, 3423, 484,\n",
" 2419, 1613, 2654, 313, 6199, 175, 741, 2738, 1146, 1329, 1774,\n",
" 2340, 1831, 3013, 6029, 1106, 1503, 1702, 2584, 2287], dtype=int32)>,\n",
" 'Horizontal_Distance_To_Hydrology': <tf.Tensor: id=433, shape=(64,), dtype=int32, numpy=\n",
"array([382, 212, 270, 0, 42, 30, 42, 85, 391, 212, 30, 551, 297,\n",
" 277, 0, 127, 124, 192, 120, 319, 42, 446, 240, 342, 175, 85,\n",
" 256, 212, 180, 330, 297, 60, 210, 153, 108, 192, 150, 342, 228,\n",
" 0, 433, 190, 170, 573, 247, 124, 150, 30, 85, 319, 210, 484,\n",
" 270, 90, 212, 0, 30, 256, 67, 60, 30, 60, 240, 573],\n",
" dtype=int32)>,\n",
" 'Horizontal_Distance_To_Roadways': <tf.Tensor: id=434, shape=(64,), dtype=int32, numpy=\n",
"array([5772, 2148, 5607, 2900, 2052, 3331, 1218, 1741, 2076, 690, 3421,\n",
" 1061, 366, 3598, 2468, 1047, 134, 859, 2954, 1485, 3300, 618,\n",
" 4806, 1846, 2330, 1728, 968, 1304, 108, 1342, 600, 451, 1583,\n",
" 1114, 1215, 1673, 3168, 2850, 5434, 2818, 5392, 1082, 4277, 603,\n",
" 2469, 722, 2759, 5060, 4293, 1366, 2288, 1717, 872, 1262, 4561,\n",
" 870, 3587, 4160, 4281, 1932, 658, 607, 2720, 1061], dtype=int32)>,\n",
" 'Slope': <tf.Tensor: id=435, shape=(64,), dtype=int32, numpy=\n",
"array([20, 11, 13, 16, 11, 8, 8, 17, 21, 13, 22, 22, 18, 7, 6, 28, 20,\n",
" 28, 15, 12, 7, 9, 4, 16, 15, 26, 15, 11, 10, 18, 15, 9, 26, 7,\n",
" 8, 21, 21, 23, 11, 32, 7, 17, 22, 29, 5, 11, 4, 17, 18, 25, 0,\n",
" 6, 15, 10, 20, 6, 12, 13, 14, 10, 16, 17, 8, 13], dtype=int32)>,\n",
" 'Soil_Type': <tf.Tensor: id=436, shape=(64, 40), dtype=int32, numpy=\n",
"array([[0, 0, 0, ..., 0, 0, 0],\n",
" [0, 0, 0, ..., 0, 0, 0],\n",
" [0, 0, 0, ..., 0, 0, 0],\n",
" ...,\n",
" [0, 0, 1, ..., 0, 0, 0],\n",
" [0, 0, 0, ..., 0, 0, 0],\n",
" [0, 0, 0, ..., 0, 0, 0]], dtype=int32)>,\n",
" 'Vertical_Distance_To_Hydrology': <tf.Tensor: id=437, shape=(64,), dtype=int32, numpy=\n",
"array([ 56, 44, 99, 0, 6, 3, 4, 21, -27, 39, 11, 69, 60,\n",
" 17, 0, 48, 28, 69, 16, 20, -4, 61, 10, 5, 37, 11,\n",
" 65, 76, 30, 1, 5, -4, 80, 25, -5, 36, 67, -14, 40,\n",
" 0, 55, 47, 40, 151, 34, 7, 21, 6, 13, 84, 35, -10,\n",
" 44, 24, 30, 0, 0, 14, 6, 8, 7, 12, 32, 101],\n",
" dtype=int32)>},\n",
" <tf.Tensor: id=438, shape=(64,), dtype=int64, numpy=\n",
"array([0, 0, 0, 0, 0, 0, 0, 2, 1, 2, 2, 2, 2, 0, 0, 0, 2, 2, 2, 0, 0, 0,\n",
" 0, 2, 2, 2, 0, 2, 3, 0, 2, 3, 2, 0, 2, 2, 0, 2, 0, 2, 0, 2, 0, 1,\n",
" 2, 2, 0, 0, 0, 0, 2, 0, 3, 2, 0, 2, 0, 0, 0, 2, 3, 3, 0, 2])>)]\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "mL1wlG7MFbVC",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"# Part 2 - Going deep on data and features ([Video](https://youtu.be/TOP2aLxcuu8))"
]
},
{
"metadata": {
"id": "Ql3m_M0kFbVF",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"# configuration objects that will be part of the model's graph\n",
"cover_type = tf.feature_column.categorical_column_with_identity(\"Cover_Type\", num_buckets=8)\n",
"cover_embedding = tf.feature_column.embedding_column(cover_type, dimension=10) # 1-hot vector"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "n5qU8dnGFbVK",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"numeric_cols = [\"Elevation\",\"Aspect\",\"Slope\",\"Horizontal_Distance_To_Hydrology\",\n",
" \"Vertical_Distance_To_Hydrology\",\"Horizontal_Distance_To_Roadways\",\n",
" \"Hillshade_9am\",\"Hillshade_Noon\",\"Hillshade_3pm\",\"Horizontal_Distance_To_Fire_Points\"]\n",
"numeric_features = [tf.feature_column.numeric_column(feat) for feat in numeric_cols]"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "PTV5dMmbFbVQ",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"soil_type = tf.feature_column.numeric_column(\"Soil_Type\", shape=(40, ))"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "Z1Z2t6RoFbVX",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"columns = numeric_features + [soil_type, cover_embedding]"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "0k-fCijHFbVd",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"As of January 2019, tf.keras.layers.DenseFeature layer does not exist in TF 1.12.0\n",
"Hence building the model with Keras as described in the video is not possible however the following code using tf.estimator achieves a similar outcomes."
]
},
{
"metadata": {
"id": "ox4aSgKzFbVf",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 193
},
"outputId": "94510c71-6c55-4ddc-ae75-30f6192bb235"
},
"cell_type": "code",
"source": [
"model = tf.estimator.DNNClassifier(feature_columns=columns, n_classes=4,\n",
" hidden_units=[256, 16, 8])"
],
"execution_count": 15,
"outputs": [
{
"output_type": "stream",
"text": [
"INFO:tensorflow:Using default config.\n",
"WARNING:tensorflow:Using temporary folder as model directory: /tmp/tmpoc5tjdcw\n",
"INFO:tensorflow:Using config: {'_model_dir': '/tmp/tmpoc5tjdcw', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true\n",
"graph_options {\n",
" rewrite_options {\n",
" meta_optimizer_iterations: ONE\n",
" }\n",
"}\n",
", '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fee4fac5748>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "1B5AbqnHFbVj",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"# Part 3 - Building and refining your models ([Video](https://www.youtube.com/watch?v=ChidCgtd1Lw))"
]
},
{
"metadata": {
"id": "2ExTSGcJFbVl",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"As described above, the example using the Keras layers has been adapted in order to work with the current version of Tensorflow \n",
"We use tf.estimator instead of a Keras model, more details can be found from 4:10 in the video\n",
"\n",
"As show below some methods like compiile are not available for estimators"
]
},
{
"metadata": {
"id": "NRh-s6RRFbVq",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 237
},
"outputId": "027c7480-3965-4f58-a20d-6c43116d4036"
},
"cell_type": "code",
"source": [
"model.compile(\n",
" optimizer=tf.train.AdamOptimizer(),\n",
" loss=\"sparse_categorical_crossentropy\",\n",
" metrics=[\"accuracy\"] \n",
")"
],
"execution_count": 16,
"outputs": [
{
"output_type": "error",
"ename": "AttributeError",
"evalue": "ignored",
"traceback": [
"\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
"\u001b[0;31mAttributeError\u001b[0m Traceback (most recent call last)",
"\u001b[0;32m<ipython-input-16-f752d44e3dd7>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m model.compile(\n\u001b[0m\u001b[1;32m 2\u001b[0m \u001b[0moptimizer\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mAdamOptimizer\u001b[0m\u001b[0;34m(\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[1;32m 3\u001b[0m \u001b[0mloss\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m\"sparse_categorical_crossentropy\"\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 4\u001b[0m \u001b[0mmetrics\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"accuracy\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 5\u001b[0m )\n",
"\u001b[0;31mAttributeError\u001b[0m: 'DNNClassifier' object has no attribute 'compile'"
]
}
]
},
{
"metadata": {
"id": "kU7RzMNWFbVw",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"def load_data(*filenames):\n",
" dataset = tf.data.experimental.CsvDataset(\n",
" filenames, defaults)\n",
" dataset = dataset.map(_parse_csv_row)\n",
" dataset = dataset.batch(64)\n",
" return dataset"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "hleE_dx0FbV2",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"### Training the model/estimator"
]
},
{
"metadata": {
"id": "wiVeKmE_FbV5",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 4550
},
"outputId": "6fd1c594-4f00-461f-aaf7-6cf6cca90e34"
},
"cell_type": "code",
"source": [
"nb_epoch = 2\n",
"for _ in range(nb_epoch):\n",
" model.train(lambda : load_data(\"covtype.csv.train\"))"
],
"execution_count": 18,
"outputs": [
{
"output_type": "stream",
"text": [
"INFO:tensorflow:Calling model_fn.\n",
"INFO:tensorflow:Done calling model_fn.\n",
"INFO:tensorflow:Create CheckpointSaverHook.\n",
"INFO:tensorflow:Graph was finalized.\n",
"INFO:tensorflow:Running local_init_op.\n",
"INFO:tensorflow:Done running local_init_op.\n",
"INFO:tensorflow:Saving checkpoints for 0 into /tmp/tmpoc5tjdcw/model.ckpt.\n",
"INFO:tensorflow:loss = 1563.7515, step = 1\n",
"INFO:tensorflow:global_step/sec: 125.934\n",
"INFO:tensorflow:loss = 54.0829, step = 101 (0.801 sec)\n",
"INFO:tensorflow:global_step/sec: 140.979\n",
"INFO:tensorflow:loss = 60.392498, step = 201 (0.705 sec)\n",
"INFO:tensorflow:global_step/sec: 140.561\n",
"INFO:tensorflow:loss = 50.694626, step = 301 (0.711 sec)\n",
"INFO:tensorflow:global_step/sec: 146.116\n",
"INFO:tensorflow:loss = 56.530968, step = 401 (0.684 sec)\n",
"INFO:tensorflow:global_step/sec: 142.203\n",
"INFO:tensorflow:loss = 52.510902, step = 501 (0.703 sec)\n",
"INFO:tensorflow:global_step/sec: 143.778\n",
"INFO:tensorflow:loss = 57.054874, step = 601 (0.695 sec)\n",
"INFO:tensorflow:global_step/sec: 139.115\n",
"INFO:tensorflow:loss = 56.987648, step = 701 (0.719 sec)\n",
"INFO:tensorflow:global_step/sec: 139.589\n",
"INFO:tensorflow:loss = 55.104744, step = 801 (0.716 sec)\n",
"INFO:tensorflow:global_step/sec: 140.143\n",
"INFO:tensorflow:loss = 53.28988, step = 901 (0.714 sec)\n",
"INFO:tensorflow:global_step/sec: 144.09\n",
"INFO:tensorflow:loss = 55.908768, step = 1001 (0.694 sec)\n",
"INFO:tensorflow:global_step/sec: 120.92\n",
"INFO:tensorflow:loss = 50.61058, step = 1101 (0.827 sec)\n",
"INFO:tensorflow:global_step/sec: 145.401\n",
"INFO:tensorflow:loss = 48.526443, step = 1201 (0.691 sec)\n",
"INFO:tensorflow:global_step/sec: 142.219\n",
"INFO:tensorflow:loss = 52.245483, step = 1301 (0.704 sec)\n",
"INFO:tensorflow:global_step/sec: 141.69\n",
"INFO:tensorflow:loss = 51.112408, step = 1401 (0.705 sec)\n",
"INFO:tensorflow:global_step/sec: 144.419\n",
"INFO:tensorflow:loss = 47.682133, step = 1501 (0.691 sec)\n",
"INFO:tensorflow:global_step/sec: 136.394\n",
"INFO:tensorflow:loss = 60.863514, step = 1601 (0.731 sec)\n",
"INFO:tensorflow:global_step/sec: 142.832\n",
"INFO:tensorflow:loss = 49.61512, step = 1701 (0.703 sec)\n",
"INFO:tensorflow:global_step/sec: 127.748\n",
"INFO:tensorflow:loss = 50.141735, step = 1801 (0.780 sec)\n",
"INFO:tensorflow:global_step/sec: 144.08\n",
"INFO:tensorflow:loss = 73.35028, step = 1901 (0.694 sec)\n",
"INFO:tensorflow:global_step/sec: 145.573\n",
"INFO:tensorflow:loss = 58.26995, step = 2001 (0.687 sec)\n",
"INFO:tensorflow:global_step/sec: 144.584\n",
"INFO:tensorflow:loss = 56.142826, step = 2101 (0.694 sec)\n",
"INFO:tensorflow:global_step/sec: 142.473\n",
"INFO:tensorflow:loss = 59.331448, step = 2201 (0.709 sec)\n",
"INFO:tensorflow:global_step/sec: 143.246\n",
"INFO:tensorflow:loss = 52.343704, step = 2301 (0.688 sec)\n",
"INFO:tensorflow:global_step/sec: 137.705\n",
"INFO:tensorflow:loss = 53.304695, step = 2401 (0.726 sec)\n",
"INFO:tensorflow:global_step/sec: 131.839\n",
"INFO:tensorflow:loss = 52.263824, step = 2501 (0.761 sec)\n",
"INFO:tensorflow:global_step/sec: 130.136\n",
"INFO:tensorflow:loss = 50.648746, step = 2601 (0.771 sec)\n",
"INFO:tensorflow:global_step/sec: 124.489\n",
"INFO:tensorflow:loss = 60.119858, step = 2701 (0.798 sec)\n",
"INFO:tensorflow:global_step/sec: 131.819\n",
"INFO:tensorflow:loss = 56.921307, step = 2801 (0.762 sec)\n",
"INFO:tensorflow:global_step/sec: 127.871\n",
"INFO:tensorflow:loss = 40.97254, step = 2901 (0.783 sec)\n",
"INFO:tensorflow:global_step/sec: 130.498\n",
"INFO:tensorflow:loss = 48.667816, step = 3001 (0.762 sec)\n",
"INFO:tensorflow:global_step/sec: 124.791\n",
"INFO:tensorflow:loss = 48.208115, step = 3101 (0.806 sec)\n",
"INFO:tensorflow:global_step/sec: 131.442\n",
"INFO:tensorflow:loss = 51.0486, step = 3201 (0.756 sec)\n",
"INFO:tensorflow:global_step/sec: 132.216\n",
"INFO:tensorflow:loss = 57.64761, step = 3301 (0.756 sec)\n",
"INFO:tensorflow:global_step/sec: 131.609\n",
"INFO:tensorflow:loss = 48.017967, step = 3401 (0.765 sec)\n",
"INFO:tensorflow:global_step/sec: 132.271\n",
"INFO:tensorflow:loss = 46.5169, step = 3501 (0.755 sec)\n",
"INFO:tensorflow:global_step/sec: 131.766\n",
"INFO:tensorflow:loss = 56.23243, step = 3601 (0.760 sec)\n",
"INFO:tensorflow:global_step/sec: 131.417\n",
"INFO:tensorflow:loss = 61.430374, step = 3701 (0.759 sec)\n",
"INFO:tensorflow:global_step/sec: 140.475\n",
"INFO:tensorflow:loss = 52.16754, step = 3801 (0.709 sec)\n",
"INFO:tensorflow:global_step/sec: 142.936\n",
"INFO:tensorflow:loss = 58.322666, step = 3901 (0.704 sec)\n",
"INFO:tensorflow:global_step/sec: 141.71\n",
"INFO:tensorflow:loss = 60.72654, step = 4001 (0.701 sec)\n",
"INFO:tensorflow:global_step/sec: 144.7\n",
"INFO:tensorflow:loss = 63.571835, step = 4101 (0.694 sec)\n",
"INFO:tensorflow:global_step/sec: 139.187\n",
"INFO:tensorflow:loss = 43.489136, step = 4201 (0.716 sec)\n",
"INFO:tensorflow:global_step/sec: 144.013\n",
"INFO:tensorflow:loss = 53.06375, step = 4301 (0.697 sec)\n",
"INFO:tensorflow:global_step/sec: 142.325\n",
"INFO:tensorflow:loss = 58.16475, step = 4401 (0.700 sec)\n",
"INFO:tensorflow:global_step/sec: 140.69\n",
"INFO:tensorflow:loss = 42.54741, step = 4501 (0.711 sec)\n",
"INFO:tensorflow:global_step/sec: 146.632\n",
"INFO:tensorflow:loss = 57.692978, step = 4601 (0.685 sec)\n",
"INFO:tensorflow:global_step/sec: 145.49\n",
"INFO:tensorflow:loss = 52.41902, step = 4701 (0.685 sec)\n",
"INFO:tensorflow:global_step/sec: 145.872\n",
"INFO:tensorflow:loss = 46.229134, step = 4801 (0.689 sec)\n",
"INFO:tensorflow:global_step/sec: 143.509\n",
"INFO:tensorflow:loss = 53.27777, step = 4901 (0.693 sec)\n",
"INFO:tensorflow:global_step/sec: 144.84\n",
"INFO:tensorflow:loss = 45.817062, step = 5001 (0.694 sec)\n",
"INFO:tensorflow:global_step/sec: 145.325\n",
"INFO:tensorflow:loss = 51.862717, step = 5101 (0.689 sec)\n",
"INFO:tensorflow:global_step/sec: 144.209\n",
"INFO:tensorflow:loss = 48.110218, step = 5201 (0.689 sec)\n",
"INFO:tensorflow:global_step/sec: 144.853\n",
"INFO:tensorflow:loss = 56.061565, step = 5301 (0.690 sec)\n",
"INFO:tensorflow:global_step/sec: 146.059\n",
"INFO:tensorflow:loss = 56.900913, step = 5401 (0.688 sec)\n",
"INFO:tensorflow:global_step/sec: 144.748\n",
"INFO:tensorflow:loss = 45.794617, step = 5501 (0.691 sec)\n",
"INFO:tensorflow:global_step/sec: 145.174\n",
"INFO:tensorflow:loss = 57.858955, step = 5601 (0.686 sec)\n",
"INFO:tensorflow:global_step/sec: 145.505\n",
"INFO:tensorflow:loss = 53.602364, step = 5701 (0.691 sec)\n",
"INFO:tensorflow:global_step/sec: 143.529\n",
"INFO:tensorflow:loss = 41.861874, step = 5801 (0.692 sec)\n",
"INFO:tensorflow:global_step/sec: 143.351\n",
"INFO:tensorflow:loss = 44.658897, step = 5901 (0.698 sec)\n",
"INFO:tensorflow:global_step/sec: 141.892\n",
"INFO:tensorflow:loss = 55.12124, step = 6001 (0.705 sec)\n",
"INFO:tensorflow:Saving checkpoints for 6083 into /tmp/tmpoc5tjdcw/model.ckpt.\n",
"INFO:tensorflow:Loss for final step: 19.818727.\n",
"INFO:tensorflow:Calling model_fn.\n",
"INFO:tensorflow:Done calling model_fn.\n",
"INFO:tensorflow:Create CheckpointSaverHook.\n",
"INFO:tensorflow:Graph was finalized.\n",
"INFO:tensorflow:Restoring parameters from /tmp/tmpoc5tjdcw/model.ckpt-6083\n",
"INFO:tensorflow:Running local_init_op.\n",
"INFO:tensorflow:Done running local_init_op.\n",
"INFO:tensorflow:Saving checkpoints for 6083 into /tmp/tmpoc5tjdcw/model.ckpt.\n",
"INFO:tensorflow:loss = 50.73872, step = 6084\n",
"INFO:tensorflow:global_step/sec: 124.892\n",
"INFO:tensorflow:loss = 46.179607, step = 6184 (0.806 sec)\n",
"INFO:tensorflow:global_step/sec: 145.727\n",
"INFO:tensorflow:loss = 57.620033, step = 6284 (0.686 sec)\n",
"INFO:tensorflow:global_step/sec: 146.717\n",
"INFO:tensorflow:loss = 47.93088, step = 6384 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 143.142\n",
"INFO:tensorflow:loss = 53.190434, step = 6484 (0.701 sec)\n",
"INFO:tensorflow:global_step/sec: 143.976\n",
"INFO:tensorflow:loss = 47.101044, step = 6584 (0.692 sec)\n",
"INFO:tensorflow:global_step/sec: 144.171\n",
"INFO:tensorflow:loss = 50.72544, step = 6684 (0.697 sec)\n",
"INFO:tensorflow:global_step/sec: 141.984\n",
"INFO:tensorflow:loss = 53.46061, step = 6784 (0.701 sec)\n",
"INFO:tensorflow:global_step/sec: 141.741\n",
"INFO:tensorflow:loss = 52.17375, step = 6884 (0.706 sec)\n",
"INFO:tensorflow:global_step/sec: 144.933\n",
"INFO:tensorflow:loss = 51.16942, step = 6984 (0.693 sec)\n",
"INFO:tensorflow:global_step/sec: 143.574\n",
"INFO:tensorflow:loss = 51.577843, step = 7084 (0.693 sec)\n",
"INFO:tensorflow:global_step/sec: 141.446\n",
"INFO:tensorflow:loss = 45.820217, step = 7184 (0.710 sec)\n",
"INFO:tensorflow:global_step/sec: 141.15\n",
"INFO:tensorflow:loss = 55.1689, step = 7284 (0.710 sec)\n",
"INFO:tensorflow:global_step/sec: 139.742\n",
"INFO:tensorflow:loss = 50.02367, step = 7384 (0.715 sec)\n",
"INFO:tensorflow:global_step/sec: 141.001\n",
"INFO:tensorflow:loss = 44.8162, step = 7484 (0.706 sec)\n",
"INFO:tensorflow:global_step/sec: 140.997\n",
"INFO:tensorflow:loss = 47.461018, step = 7584 (0.712 sec)\n",
"INFO:tensorflow:global_step/sec: 138.742\n",
"INFO:tensorflow:loss = 54.59662, step = 7684 (0.721 sec)\n",
"INFO:tensorflow:global_step/sec: 142.805\n",
"INFO:tensorflow:loss = 44.98487, step = 7784 (0.698 sec)\n",
"INFO:tensorflow:global_step/sec: 143.733\n",
"INFO:tensorflow:loss = 45.31526, step = 7884 (0.695 sec)\n",
"INFO:tensorflow:global_step/sec: 137.838\n",
"INFO:tensorflow:loss = 55.089863, step = 7984 (0.726 sec)\n",
"INFO:tensorflow:global_step/sec: 139.947\n",
"INFO:tensorflow:loss = 53.3806, step = 8084 (0.715 sec)\n",
"INFO:tensorflow:global_step/sec: 143.083\n",
"INFO:tensorflow:loss = 52.81919, step = 8184 (0.699 sec)\n",
"INFO:tensorflow:global_step/sec: 140.425\n",
"INFO:tensorflow:loss = 54.40442, step = 8284 (0.712 sec)\n",
"INFO:tensorflow:global_step/sec: 142.251\n",
"INFO:tensorflow:loss = 46.548256, step = 8384 (0.703 sec)\n",
"INFO:tensorflow:global_step/sec: 142.45\n",
"INFO:tensorflow:loss = 51.30117, step = 8484 (0.704 sec)\n",
"INFO:tensorflow:global_step/sec: 141.607\n",
"INFO:tensorflow:loss = 46.064327, step = 8584 (0.708 sec)\n",
"INFO:tensorflow:global_step/sec: 143.544\n",
"INFO:tensorflow:loss = 48.06105, step = 8684 (0.696 sec)\n",
"INFO:tensorflow:global_step/sec: 142.901\n",
"INFO:tensorflow:loss = 56.247322, step = 8784 (0.696 sec)\n",
"INFO:tensorflow:global_step/sec: 142.632\n",
"INFO:tensorflow:loss = 51.228027, step = 8884 (0.701 sec)\n",
"INFO:tensorflow:global_step/sec: 141.163\n",
"INFO:tensorflow:loss = 36.852295, step = 8984 (0.708 sec)\n",
"INFO:tensorflow:global_step/sec: 139.334\n",
"INFO:tensorflow:loss = 42.330986, step = 9084 (0.718 sec)\n",
"INFO:tensorflow:global_step/sec: 144.794\n",
"INFO:tensorflow:loss = 43.96254, step = 9184 (0.695 sec)\n",
"INFO:tensorflow:global_step/sec: 142.314\n",
"INFO:tensorflow:loss = 45.464436, step = 9284 (0.698 sec)\n",
"INFO:tensorflow:global_step/sec: 142.289\n",
"INFO:tensorflow:loss = 47.248096, step = 9384 (0.703 sec)\n",
"INFO:tensorflow:global_step/sec: 141.705\n",
"INFO:tensorflow:loss = 48.061234, step = 9484 (0.710 sec)\n",
"INFO:tensorflow:global_step/sec: 137.473\n",
"INFO:tensorflow:loss = 39.808403, step = 9584 (0.723 sec)\n",
"INFO:tensorflow:global_step/sec: 140.624\n",
"INFO:tensorflow:loss = 54.48674, step = 9684 (0.711 sec)\n",
"INFO:tensorflow:global_step/sec: 142.101\n",
"INFO:tensorflow:loss = 55.452187, step = 9784 (0.704 sec)\n",
"INFO:tensorflow:global_step/sec: 139.439\n",
"INFO:tensorflow:loss = 49.383556, step = 9884 (0.717 sec)\n",
"INFO:tensorflow:global_step/sec: 138.687\n",
"INFO:tensorflow:loss = 49.048637, step = 9984 (0.724 sec)\n",
"INFO:tensorflow:global_step/sec: 142.234\n",
"INFO:tensorflow:loss = 55.28456, step = 10084 (0.703 sec)\n",
"INFO:tensorflow:global_step/sec: 141.778\n",
"INFO:tensorflow:loss = 56.285095, step = 10184 (0.702 sec)\n",
"INFO:tensorflow:global_step/sec: 143.473\n",
"INFO:tensorflow:loss = 38.605423, step = 10284 (0.697 sec)\n",
"INFO:tensorflow:global_step/sec: 147.191\n",
"INFO:tensorflow:loss = 42.116234, step = 10384 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 139.867\n",
"INFO:tensorflow:loss = 49.800972, step = 10484 (0.715 sec)\n",
"INFO:tensorflow:global_step/sec: 143.772\n",
"INFO:tensorflow:loss = 34.982292, step = 10584 (0.696 sec)\n",
"INFO:tensorflow:global_step/sec: 141.879\n",
"INFO:tensorflow:loss = 48.09848, step = 10684 (0.710 sec)\n",
"INFO:tensorflow:global_step/sec: 141.511\n",
"INFO:tensorflow:loss = 45.812664, step = 10784 (0.701 sec)\n",
"INFO:tensorflow:global_step/sec: 144.204\n",
"INFO:tensorflow:loss = 35.634865, step = 10884 (0.693 sec)\n",
"INFO:tensorflow:global_step/sec: 143.072\n",
"INFO:tensorflow:loss = 44.835922, step = 10984 (0.699 sec)\n",
"INFO:tensorflow:global_step/sec: 143.269\n",
"INFO:tensorflow:loss = 35.582657, step = 11084 (0.700 sec)\n",
"INFO:tensorflow:global_step/sec: 143.555\n",
"INFO:tensorflow:loss = 42.163185, step = 11184 (0.694 sec)\n",
"INFO:tensorflow:global_step/sec: 146.598\n",
"INFO:tensorflow:loss = 35.836525, step = 11284 (0.682 sec)\n",
"INFO:tensorflow:global_step/sec: 146.434\n",
"INFO:tensorflow:loss = 46.57731, step = 11384 (0.683 sec)\n",
"INFO:tensorflow:global_step/sec: 143.294\n",
"INFO:tensorflow:loss = 51.163036, step = 11484 (0.701 sec)\n",
"INFO:tensorflow:global_step/sec: 144.755\n",
"INFO:tensorflow:loss = 39.366745, step = 11584 (0.688 sec)\n",
"INFO:tensorflow:global_step/sec: 142.696\n",
"INFO:tensorflow:loss = 46.037136, step = 11684 (0.706 sec)\n",
"INFO:tensorflow:global_step/sec: 143.905\n",
"INFO:tensorflow:loss = 44.876427, step = 11784 (0.690 sec)\n",
"INFO:tensorflow:global_step/sec: 143.408\n",
"INFO:tensorflow:loss = 34.15465, step = 11884 (0.699 sec)\n",
"INFO:tensorflow:global_step/sec: 142.664\n",
"INFO:tensorflow:loss = 38.461205, step = 11984 (0.700 sec)\n",
"INFO:tensorflow:global_step/sec: 143.557\n",
"INFO:tensorflow:loss = 48.154327, step = 12084 (0.697 sec)\n",
"INFO:tensorflow:Saving checkpoints for 12166 into /tmp/tmpoc5tjdcw/model.ckpt.\n",
"INFO:tensorflow:Loss for final step: 15.786152.\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "LQmpFUwNFbV-",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"### Validating the model"
]
},
{
"metadata": {
"colab_type": "code",
"id": "kxNpMGbDOAA1",
"outputId": "acf57836-59ff-463d-d844-7fe6655b1c5c",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 382
}
},
"cell_type": "code",
"source": [
"print(model.evaluate(lambda : load_data(\"covtype.csv.test\"), steps = 50))"
],
"execution_count": 19,
"outputs": [
{
"output_type": "stream",
"text": [
"INFO:tensorflow:Calling model_fn.\n",
"INFO:tensorflow:Done calling model_fn.\n",
"INFO:tensorflow:Starting evaluation at 2019-02-03-16:09:56\n",
"INFO:tensorflow:Graph was finalized.\n",
"INFO:tensorflow:Restoring parameters from /tmp/tmpoc5tjdcw/model.ckpt-12166\n",
"INFO:tensorflow:Running local_init_op.\n",
"INFO:tensorflow:Done running local_init_op.\n",
"INFO:tensorflow:Evaluation [5/50]\n",
"INFO:tensorflow:Evaluation [10/50]\n",
"INFO:tensorflow:Evaluation [15/50]\n",
"INFO:tensorflow:Evaluation [20/50]\n",
"INFO:tensorflow:Evaluation [25/50]\n",
"INFO:tensorflow:Evaluation [30/50]\n",
"INFO:tensorflow:Evaluation [35/50]\n",
"INFO:tensorflow:Evaluation [40/50]\n",
"INFO:tensorflow:Evaluation [45/50]\n",
"INFO:tensorflow:Evaluation [50/50]\n",
"INFO:tensorflow:Finished evaluation at 2019-02-03-16:09:57\n",
"INFO:tensorflow:Saving dict for global step 12166: accuracy = 0.7009375, average_loss = 0.6522408, global_step = 12166, loss = 41.743412\n",
"INFO:tensorflow:Saving 'checkpoint_path' summary for global step 12166: /tmp/tmpoc5tjdcw/model.ckpt-12166\n",
"{'accuracy': 0.7009375, 'average_loss': 0.6522408, 'loss': 41.743412, 'global_step': 12166}\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"colab_type": "code",
"id": "wR2LG5L-OAA8",
"colab": {}
},
"cell_type": "code",
"source": [
"### Exporting the model"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"colab_type": "code",
"id": "x2CDmMtWf4Up",
"colab": {}
},
"cell_type": "code",
"source": [
"features_sample = list(dataset.take(1))[0][0]\n",
"input_receiver_fn = tf.estimator.export.build_raw_serving_input_receiver_fn(\n",
" features_sample)"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "MXbFznYWFbWa",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 367
},
"outputId": "a46d1bee-ea22-49e1-cd63-4e4ed1bf6b48"
},
"cell_type": "code",
"source": [
"model.export_saved_model(\n",
" export_dir_base=\"models\",\n",
" serving_input_receiver_fn=input_receiver_fn\n",
")"
],
"execution_count": 22,
"outputs": [
{
"output_type": "stream",
"text": [
"INFO:tensorflow:Calling model_fn.\n",
"INFO:tensorflow:Done calling model_fn.\n",
"INFO:tensorflow:Signatures INCLUDED in export for Classify: None\n",
"INFO:tensorflow:Signatures INCLUDED in export for Regress: None\n",
"INFO:tensorflow:Signatures INCLUDED in export for Predict: ['predict']\n",
"INFO:tensorflow:Signatures INCLUDED in export for Train: None\n",
"INFO:tensorflow:Signatures INCLUDED in export for Eval: None\n",
"INFO:tensorflow:Signatures EXCLUDED from export because they cannot be be served via TensorFlow Serving APIs:\n",
"INFO:tensorflow:'serving_default' : Classification input must be a single string Tensor; got {'Elevation': <tf.Tensor 'Placeholder:0' shape=(?,) dtype=int32>, 'Aspect': <tf.Tensor 'Placeholder_1:0' shape=(?,) dtype=int32>, 'Slope': <tf.Tensor 'Placeholder_2:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Hydrology': <tf.Tensor 'Placeholder_3:0' shape=(?,) dtype=int32>, 'Vertical_Distance_To_Hydrology': <tf.Tensor 'Placeholder_4:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Roadways': <tf.Tensor 'Placeholder_5:0' shape=(?,) dtype=int32>, 'Hillshade_9am': <tf.Tensor 'Placeholder_6:0' shape=(?,) dtype=int32>, 'Hillshade_Noon': <tf.Tensor 'Placeholder_7:0' shape=(?,) dtype=int32>, 'Hillshade_3pm': <tf.Tensor 'Placeholder_8:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Fire_Points': <tf.Tensor 'Placeholder_9:0' shape=(?,) dtype=int32>, 'Soil_Type': <tf.Tensor 'Placeholder_10:0' shape=(?, 40) dtype=int32>, 'Cover_Type': <tf.Tensor 'Placeholder_11:0' shape=(?,) dtype=int32>}\n",
"INFO:tensorflow:'classification' : Classification input must be a single string Tensor; got {'Elevation': <tf.Tensor 'Placeholder:0' shape=(?,) dtype=int32>, 'Aspect': <tf.Tensor 'Placeholder_1:0' shape=(?,) dtype=int32>, 'Slope': <tf.Tensor 'Placeholder_2:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Hydrology': <tf.Tensor 'Placeholder_3:0' shape=(?,) dtype=int32>, 'Vertical_Distance_To_Hydrology': <tf.Tensor 'Placeholder_4:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Roadways': <tf.Tensor 'Placeholder_5:0' shape=(?,) dtype=int32>, 'Hillshade_9am': <tf.Tensor 'Placeholder_6:0' shape=(?,) dtype=int32>, 'Hillshade_Noon': <tf.Tensor 'Placeholder_7:0' shape=(?,) dtype=int32>, 'Hillshade_3pm': <tf.Tensor 'Placeholder_8:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Fire_Points': <tf.Tensor 'Placeholder_9:0' shape=(?,) dtype=int32>, 'Soil_Type': <tf.Tensor 'Placeholder_10:0' shape=(?, 40) dtype=int32>, 'Cover_Type': <tf.Tensor 'Placeholder_11:0' shape=(?,) dtype=int32>}\n",
"WARNING:tensorflow:Export includes no default signature!\n",
"INFO:tensorflow:Restoring parameters from /tmp/tmpoc5tjdcw/model.ckpt-12166\n",
"WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/estimator/estimator.py:1044: calling SavedModelBuilder.add_meta_graph_and_variables (from tensorflow.python.saved_model.builder_impl) with legacy_init_op is deprecated and will be removed in a future version.\n",
"Instructions for updating:\n",
"Pass your op to the equivalent parameter main_op instead.\n",
"INFO:tensorflow:Assets added to graph.\n",
"INFO:tensorflow:No assets to write.\n",
"INFO:tensorflow:SavedModel written to: models/temp-b'1549210197'/saved_model.pb\n"
],
"name": "stdout"
},
{
"output_type": "execute_result",
"data": {
"text/plain": [
"b'models/1549210197'"
]
},
"metadata": {
"tags": []
},
"execution_count": 22
}
]
},
{
"metadata": {
"id": "dhbRkuNYFbWd",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"## Swapping the model"
]
},
{
"metadata": {
"id": "AKe9EjVJFbWe",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"Since we did not use the Keras model in the previous section, swapping the model is completely straightforward."
]
},
{
"metadata": {
"scrolled": true,
"id": "aZ1UhrkIFbWf",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 5369
},
"outputId": "c445088a-1741-4f02-9c71-5c4fc347f05c"
},
"cell_type": "code",
"source": [
"# model definition\n",
"model = tf.estimator.DNNLinearCombinedClassifier(\n",
" linear_feature_columns=[cover_type, soil_type],\n",
" dnn_feature_columns=numeric_features,\n",
" dnn_hidden_units=[256, 16, 8],\n",
" n_classes=4\n",
")\n",
"\n",
"# model training\n",
"nb_epoch = 2\n",
"for _ in range(nb_epoch):\n",
" model.train(lambda : load_data(\"covtype.csv.train\"))\n",
"\n",
"# model validation\n",
"print(model.evaluate(lambda : load_data(\"covtype.csv.test\"), steps = 50))\n",
"\n",
"# exporting the model\n",
"model.export_saved_model(\n",
" export_dir_base=\"models\",\n",
" serving_input_receiver_fn=input_receiver_fn\n",
")"
],
"execution_count": 23,
"outputs": [
{
"output_type": "stream",
"text": [
"INFO:tensorflow:Using default config.\n",
"WARNING:tensorflow:Using temporary folder as model directory: /tmp/tmpx_4kh1pp\n",
"INFO:tensorflow:Using config: {'_model_dir': '/tmp/tmpx_4kh1pp', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true\n",
"graph_options {\n",
" rewrite_options {\n",
" meta_optimizer_iterations: ONE\n",
" }\n",
"}\n",
", '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fee4e8b7eb8>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n",
"INFO:tensorflow:Calling model_fn.\n",
"INFO:tensorflow:Done calling model_fn.\n",
"INFO:tensorflow:Create CheckpointSaverHook.\n",
"INFO:tensorflow:Graph was finalized.\n",
"INFO:tensorflow:Running local_init_op.\n",
"INFO:tensorflow:Done running local_init_op.\n",
"INFO:tensorflow:Saving checkpoints for 0 into /tmp/tmpx_4kh1pp/model.ckpt.\n",
"INFO:tensorflow:loss = 21178.918, step = 1\n",
"INFO:tensorflow:global_step/sec: 124.621\n",
"INFO:tensorflow:loss = 695.51685, step = 101 (0.804 sec)\n",
"INFO:tensorflow:global_step/sec: 144.57\n",
"INFO:tensorflow:loss = 80.324326, step = 201 (0.693 sec)\n",
"INFO:tensorflow:global_step/sec: 148.545\n",
"INFO:tensorflow:loss = 82.16355, step = 301 (0.672 sec)\n",
"INFO:tensorflow:global_step/sec: 147.094\n",
"INFO:tensorflow:loss = 72.09811, step = 401 (0.680 sec)\n",
"INFO:tensorflow:global_step/sec: 140.449\n",
"INFO:tensorflow:loss = 64.45215, step = 501 (0.712 sec)\n",
"INFO:tensorflow:global_step/sec: 145.797\n",
"INFO:tensorflow:loss = 55.31878, step = 601 (0.686 sec)\n",
"INFO:tensorflow:global_step/sec: 146.884\n",
"INFO:tensorflow:loss = 57.930946, step = 701 (0.683 sec)\n",
"INFO:tensorflow:global_step/sec: 147.77\n",
"INFO:tensorflow:loss = 58.940742, step = 801 (0.675 sec)\n",
"INFO:tensorflow:global_step/sec: 132.308\n",
"INFO:tensorflow:loss = 52.013374, step = 901 (0.760 sec)\n",
"INFO:tensorflow:global_step/sec: 146.264\n",
"INFO:tensorflow:loss = 50.14777, step = 1001 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 143.131\n",
"INFO:tensorflow:loss = 46.58689, step = 1101 (0.698 sec)\n",
"INFO:tensorflow:global_step/sec: 143.793\n",
"INFO:tensorflow:loss = 47.136314, step = 1201 (0.695 sec)\n",
"INFO:tensorflow:global_step/sec: 145.61\n",
"INFO:tensorflow:loss = 46.79516, step = 1301 (0.687 sec)\n",
"INFO:tensorflow:global_step/sec: 143.846\n",
"INFO:tensorflow:loss = 48.593464, step = 1401 (0.699 sec)\n",
"INFO:tensorflow:global_step/sec: 140.033\n",
"INFO:tensorflow:loss = 44.56623, step = 1501 (0.710 sec)\n",
"INFO:tensorflow:global_step/sec: 123.601\n",
"INFO:tensorflow:loss = 46.604683, step = 1601 (0.809 sec)\n",
"INFO:tensorflow:global_step/sec: 146.05\n",
"INFO:tensorflow:loss = 40.529945, step = 1701 (0.685 sec)\n",
"INFO:tensorflow:global_step/sec: 150.624\n",
"INFO:tensorflow:loss = 42.553047, step = 1801 (0.664 sec)\n",
"INFO:tensorflow:global_step/sec: 148.968\n",
"INFO:tensorflow:loss = 58.286285, step = 1901 (0.671 sec)\n",
"INFO:tensorflow:global_step/sec: 149.646\n",
"INFO:tensorflow:loss = 47.92729, step = 2001 (0.668 sec)\n",
"INFO:tensorflow:global_step/sec: 148.996\n",
"INFO:tensorflow:loss = 47.835617, step = 2101 (0.671 sec)\n",
"INFO:tensorflow:global_step/sec: 143.741\n",
"INFO:tensorflow:loss = 42.22122, step = 2201 (0.696 sec)\n",
"INFO:tensorflow:global_step/sec: 150.339\n",
"INFO:tensorflow:loss = 41.930344, step = 2301 (0.668 sec)\n",
"INFO:tensorflow:global_step/sec: 150.075\n",
"INFO:tensorflow:loss = 41.02935, step = 2401 (0.667 sec)\n",
"INFO:tensorflow:global_step/sec: 135.393\n",
"INFO:tensorflow:loss = 43.472324, step = 2501 (0.736 sec)\n",
"INFO:tensorflow:global_step/sec: 145.757\n",
"INFO:tensorflow:loss = 40.552483, step = 2601 (0.690 sec)\n",
"INFO:tensorflow:global_step/sec: 148.831\n",
"INFO:tensorflow:loss = 40.73275, step = 2701 (0.668 sec)\n",
"INFO:tensorflow:global_step/sec: 144.073\n",
"INFO:tensorflow:loss = 44.880142, step = 2801 (0.694 sec)\n",
"INFO:tensorflow:global_step/sec: 146.563\n",
"INFO:tensorflow:loss = 35.04409, step = 2901 (0.686 sec)\n",
"INFO:tensorflow:global_step/sec: 148.58\n",
"INFO:tensorflow:loss = 36.582817, step = 3001 (0.672 sec)\n",
"INFO:tensorflow:global_step/sec: 147.986\n",
"INFO:tensorflow:loss = 40.33685, step = 3101 (0.672 sec)\n",
"INFO:tensorflow:global_step/sec: 148.604\n",
"INFO:tensorflow:loss = 36.125923, step = 3201 (0.673 sec)\n",
"INFO:tensorflow:global_step/sec: 147.71\n",
"INFO:tensorflow:loss = 44.038193, step = 3301 (0.677 sec)\n",
"INFO:tensorflow:global_step/sec: 146.841\n",
"INFO:tensorflow:loss = 37.700592, step = 3401 (0.681 sec)\n",
"INFO:tensorflow:global_step/sec: 147.767\n",
"INFO:tensorflow:loss = 36.90215, step = 3501 (0.677 sec)\n",
"INFO:tensorflow:global_step/sec: 146.638\n",
"INFO:tensorflow:loss = 44.456184, step = 3601 (0.682 sec)\n",
"INFO:tensorflow:global_step/sec: 146.123\n",
"INFO:tensorflow:loss = 43.691116, step = 3701 (0.685 sec)\n",
"INFO:tensorflow:global_step/sec: 147.202\n",
"INFO:tensorflow:loss = 34.283455, step = 3801 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 143.189\n",
"INFO:tensorflow:loss = 37.454475, step = 3901 (0.699 sec)\n",
"INFO:tensorflow:global_step/sec: 150.086\n",
"INFO:tensorflow:loss = 49.208107, step = 4001 (0.666 sec)\n",
"INFO:tensorflow:global_step/sec: 142.35\n",
"INFO:tensorflow:loss = 43.246468, step = 4101 (0.706 sec)\n",
"INFO:tensorflow:global_step/sec: 133.637\n",
"INFO:tensorflow:loss = 35.596638, step = 4201 (0.745 sec)\n",
"INFO:tensorflow:global_step/sec: 146.987\n",
"INFO:tensorflow:loss = 34.981503, step = 4301 (0.684 sec)\n",
"INFO:tensorflow:global_step/sec: 148.917\n",
"INFO:tensorflow:loss = 39.480564, step = 4401 (0.667 sec)\n",
"INFO:tensorflow:global_step/sec: 147.352\n",
"INFO:tensorflow:loss = 29.907627, step = 4501 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 128.565\n",
"INFO:tensorflow:loss = 45.20193, step = 4601 (0.778 sec)\n",
"INFO:tensorflow:global_step/sec: 142.212\n",
"INFO:tensorflow:loss = 36.020462, step = 4701 (0.706 sec)\n",
"INFO:tensorflow:global_step/sec: 130.839\n",
"INFO:tensorflow:loss = 29.850979, step = 4801 (0.765 sec)\n",
"INFO:tensorflow:global_step/sec: 144.53\n",
"INFO:tensorflow:loss = 36.98899, step = 4901 (0.688 sec)\n",
"INFO:tensorflow:global_step/sec: 144.85\n",
"INFO:tensorflow:loss = 30.55624, step = 5001 (0.690 sec)\n",
"INFO:tensorflow:global_step/sec: 113.157\n",
"INFO:tensorflow:loss = 30.075686, step = 5101 (0.884 sec)\n",
"INFO:tensorflow:global_step/sec: 130.163\n",
"INFO:tensorflow:loss = 29.997782, step = 5201 (0.769 sec)\n",
"INFO:tensorflow:global_step/sec: 143.481\n",
"INFO:tensorflow:loss = 40.189583, step = 5301 (0.696 sec)\n",
"INFO:tensorflow:global_step/sec: 136.485\n",
"INFO:tensorflow:loss = 38.054596, step = 5401 (0.733 sec)\n",
"INFO:tensorflow:global_step/sec: 147.387\n",
"INFO:tensorflow:loss = 30.590908, step = 5501 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 146.656\n",
"INFO:tensorflow:loss = 36.124546, step = 5601 (0.685 sec)\n",
"INFO:tensorflow:global_step/sec: 147.313\n",
"INFO:tensorflow:loss = 34.145, step = 5701 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 146.127\n",
"INFO:tensorflow:loss = 28.415867, step = 5801 (0.685 sec)\n",
"INFO:tensorflow:global_step/sec: 145.423\n",
"INFO:tensorflow:loss = 31.236776, step = 5901 (0.684 sec)\n",
"INFO:tensorflow:global_step/sec: 146.627\n",
"INFO:tensorflow:loss = 38.86322, step = 6001 (0.682 sec)\n",
"INFO:tensorflow:Saving checkpoints for 6083 into /tmp/tmpx_4kh1pp/model.ckpt.\n",
"INFO:tensorflow:Loss for final step: 10.247061.\n",
"INFO:tensorflow:Calling model_fn.\n",
"INFO:tensorflow:Done calling model_fn.\n",
"INFO:tensorflow:Create CheckpointSaverHook.\n",
"INFO:tensorflow:Graph was finalized.\n",
"INFO:tensorflow:Restoring parameters from /tmp/tmpx_4kh1pp/model.ckpt-6083\n",
"INFO:tensorflow:Running local_init_op.\n",
"INFO:tensorflow:Done running local_init_op.\n",
"INFO:tensorflow:Saving checkpoints for 6083 into /tmp/tmpx_4kh1pp/model.ckpt.\n",
"INFO:tensorflow:loss = 35.453964, step = 6084\n",
"INFO:tensorflow:global_step/sec: 124.084\n",
"INFO:tensorflow:loss = 29.9327, step = 6184 (0.808 sec)\n",
"INFO:tensorflow:global_step/sec: 149.675\n",
"INFO:tensorflow:loss = 33.40795, step = 6284 (0.668 sec)\n",
"INFO:tensorflow:global_step/sec: 149.71\n",
"INFO:tensorflow:loss = 33.848015, step = 6384 (0.668 sec)\n",
"INFO:tensorflow:global_step/sec: 146.496\n",
"INFO:tensorflow:loss = 30.754347, step = 6484 (0.684 sec)\n",
"INFO:tensorflow:global_step/sec: 147.329\n",
"INFO:tensorflow:loss = 34.508507, step = 6584 (0.678 sec)\n",
"INFO:tensorflow:global_step/sec: 147.116\n",
"INFO:tensorflow:loss = 33.75431, step = 6684 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 142.934\n",
"INFO:tensorflow:loss = 35.52318, step = 6784 (0.703 sec)\n",
"INFO:tensorflow:global_step/sec: 142.388\n",
"INFO:tensorflow:loss = 36.761726, step = 6884 (0.699 sec)\n",
"INFO:tensorflow:global_step/sec: 141.821\n",
"INFO:tensorflow:loss = 33.967083, step = 6984 (0.709 sec)\n",
"INFO:tensorflow:global_step/sec: 143.195\n",
"INFO:tensorflow:loss = 39.754387, step = 7084 (0.699 sec)\n",
"INFO:tensorflow:global_step/sec: 115.992\n",
"INFO:tensorflow:loss = 30.153181, step = 7184 (0.861 sec)\n",
"INFO:tensorflow:global_step/sec: 144.986\n",
"INFO:tensorflow:loss = 35.46028, step = 7284 (0.687 sec)\n",
"INFO:tensorflow:global_step/sec: 145.373\n",
"INFO:tensorflow:loss = 32.308365, step = 7384 (0.688 sec)\n",
"INFO:tensorflow:global_step/sec: 146.769\n",
"INFO:tensorflow:loss = 37.288086, step = 7484 (0.681 sec)\n",
"INFO:tensorflow:global_step/sec: 145.405\n",
"INFO:tensorflow:loss = 29.26889, step = 7584 (0.688 sec)\n",
"INFO:tensorflow:global_step/sec: 144.189\n",
"INFO:tensorflow:loss = 34.921825, step = 7684 (0.693 sec)\n",
"INFO:tensorflow:global_step/sec: 141.528\n",
"INFO:tensorflow:loss = 28.719109, step = 7784 (0.713 sec)\n",
"INFO:tensorflow:global_step/sec: 144.528\n",
"INFO:tensorflow:loss = 27.658295, step = 7884 (0.686 sec)\n",
"INFO:tensorflow:global_step/sec: 147.17\n",
"INFO:tensorflow:loss = 43.41176, step = 7984 (0.682 sec)\n",
"INFO:tensorflow:global_step/sec: 148.705\n",
"INFO:tensorflow:loss = 37.003395, step = 8084 (0.670 sec)\n",
"INFO:tensorflow:global_step/sec: 146.518\n",
"INFO:tensorflow:loss = 31.534742, step = 8184 (0.682 sec)\n",
"INFO:tensorflow:global_step/sec: 145.21\n",
"INFO:tensorflow:loss = 27.774275, step = 8284 (0.688 sec)\n",
"INFO:tensorflow:global_step/sec: 145.796\n",
"INFO:tensorflow:loss = 31.540936, step = 8384 (0.686 sec)\n",
"INFO:tensorflow:global_step/sec: 144.975\n",
"INFO:tensorflow:loss = 30.644445, step = 8484 (0.690 sec)\n",
"INFO:tensorflow:global_step/sec: 145.579\n",
"INFO:tensorflow:loss = 29.438581, step = 8584 (0.687 sec)\n",
"INFO:tensorflow:global_step/sec: 142.772\n",
"INFO:tensorflow:loss = 30.256893, step = 8684 (0.704 sec)\n",
"INFO:tensorflow:global_step/sec: 145.99\n",
"INFO:tensorflow:loss = 29.146053, step = 8784 (0.682 sec)\n",
"INFO:tensorflow:global_step/sec: 148.302\n",
"INFO:tensorflow:loss = 36.77201, step = 8884 (0.674 sec)\n",
"INFO:tensorflow:global_step/sec: 147.015\n",
"INFO:tensorflow:loss = 27.146791, step = 8984 (0.680 sec)\n",
"INFO:tensorflow:global_step/sec: 148.159\n",
"INFO:tensorflow:loss = 28.759703, step = 9084 (0.675 sec)\n",
"INFO:tensorflow:global_step/sec: 147.061\n",
"INFO:tensorflow:loss = 32.43556, step = 9184 (0.680 sec)\n",
"INFO:tensorflow:global_step/sec: 140.112\n",
"INFO:tensorflow:loss = 24.901987, step = 9284 (0.716 sec)\n",
"INFO:tensorflow:global_step/sec: 147.965\n",
"INFO:tensorflow:loss = 35.65362, step = 9384 (0.678 sec)\n",
"INFO:tensorflow:global_step/sec: 146.533\n",
"INFO:tensorflow:loss = 30.539782, step = 9484 (0.682 sec)\n",
"INFO:tensorflow:global_step/sec: 146.142\n",
"INFO:tensorflow:loss = 26.402475, step = 9584 (0.680 sec)\n",
"INFO:tensorflow:global_step/sec: 135.711\n",
"INFO:tensorflow:loss = 35.693443, step = 9684 (0.737 sec)\n",
"INFO:tensorflow:global_step/sec: 146.079\n",
"INFO:tensorflow:loss = 36.578415, step = 9784 (0.688 sec)\n",
"INFO:tensorflow:global_step/sec: 131.246\n",
"INFO:tensorflow:loss = 25.770401, step = 9884 (0.758 sec)\n",
"INFO:tensorflow:global_step/sec: 147.088\n",
"INFO:tensorflow:loss = 27.556356, step = 9984 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 148.006\n",
"INFO:tensorflow:loss = 42.780975, step = 10084 (0.676 sec)\n",
"INFO:tensorflow:global_step/sec: 146.281\n",
"INFO:tensorflow:loss = 34.690804, step = 10184 (0.689 sec)\n",
"INFO:tensorflow:global_step/sec: 146.065\n",
"INFO:tensorflow:loss = 28.994417, step = 10284 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 140.057\n",
"INFO:tensorflow:loss = 27.364697, step = 10384 (0.714 sec)\n",
"INFO:tensorflow:global_step/sec: 144.54\n",
"INFO:tensorflow:loss = 31.856724, step = 10484 (0.692 sec)\n",
"INFO:tensorflow:global_step/sec: 146.205\n",
"INFO:tensorflow:loss = 25.337952, step = 10584 (0.684 sec)\n",
"INFO:tensorflow:global_step/sec: 144.729\n",
"INFO:tensorflow:loss = 35.8079, step = 10684 (0.695 sec)\n",
"INFO:tensorflow:global_step/sec: 142.37\n",
"INFO:tensorflow:loss = 31.155233, step = 10784 (0.703 sec)\n",
"INFO:tensorflow:global_step/sec: 134.903\n",
"INFO:tensorflow:loss = 23.521465, step = 10884 (0.742 sec)\n",
"INFO:tensorflow:global_step/sec: 145.907\n",
"INFO:tensorflow:loss = 30.1797, step = 10984 (0.680 sec)\n",
"INFO:tensorflow:global_step/sec: 149.019\n",
"INFO:tensorflow:loss = 23.848671, step = 11084 (0.671 sec)\n",
"INFO:tensorflow:global_step/sec: 146.881\n",
"INFO:tensorflow:loss = 24.006592, step = 11184 (0.684 sec)\n",
"INFO:tensorflow:global_step/sec: 146.897\n",
"INFO:tensorflow:loss = 23.731112, step = 11284 (0.681 sec)\n",
"INFO:tensorflow:global_step/sec: 147.004\n",
"INFO:tensorflow:loss = 34.677574, step = 11384 (0.677 sec)\n",
"INFO:tensorflow:global_step/sec: 146.701\n",
"INFO:tensorflow:loss = 31.434517, step = 11484 (0.683 sec)\n",
"INFO:tensorflow:global_step/sec: 147.075\n",
"INFO:tensorflow:loss = 25.500505, step = 11584 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 148.152\n",
"INFO:tensorflow:loss = 30.964556, step = 11684 (0.675 sec)\n",
"INFO:tensorflow:global_step/sec: 148.226\n",
"INFO:tensorflow:loss = 26.70921, step = 11784 (0.674 sec)\n",
"INFO:tensorflow:global_step/sec: 148.127\n",
"INFO:tensorflow:loss = 29.678047, step = 11884 (0.675 sec)\n",
"INFO:tensorflow:global_step/sec: 147.249\n",
"INFO:tensorflow:loss = 24.557219, step = 11984 (0.679 sec)\n",
"INFO:tensorflow:global_step/sec: 149\n",
"INFO:tensorflow:loss = 33.76008, step = 12084 (0.671 sec)\n",
"INFO:tensorflow:Saving checkpoints for 12166 into /tmp/tmpx_4kh1pp/model.ckpt.\n",
"INFO:tensorflow:Loss for final step: 7.664049.\n",
"INFO:tensorflow:Calling model_fn.\n",
"INFO:tensorflow:Done calling model_fn.\n",
"INFO:tensorflow:Starting evaluation at 2019-02-03-16:11:28\n",
"INFO:tensorflow:Graph was finalized.\n",
"INFO:tensorflow:Restoring parameters from /tmp/tmpx_4kh1pp/model.ckpt-12166\n",
"INFO:tensorflow:Running local_init_op.\n",
"INFO:tensorflow:Done running local_init_op.\n",
"INFO:tensorflow:Evaluation [5/50]\n",
"INFO:tensorflow:Evaluation [10/50]\n",
"INFO:tensorflow:Evaluation [15/50]\n",
"INFO:tensorflow:Evaluation [20/50]\n",
"INFO:tensorflow:Evaluation [25/50]\n",
"INFO:tensorflow:Evaluation [30/50]\n",
"INFO:tensorflow:Evaluation [35/50]\n",
"INFO:tensorflow:Evaluation [40/50]\n",
"INFO:tensorflow:Evaluation [45/50]\n",
"INFO:tensorflow:Evaluation [50/50]\n",
"INFO:tensorflow:Finished evaluation at 2019-02-03-16:11:29\n",
"INFO:tensorflow:Saving dict for global step 12166: accuracy = 0.8528125, average_loss = 0.46170443, global_step = 12166, loss = 29.549084\n",
"INFO:tensorflow:Saving 'checkpoint_path' summary for global step 12166: /tmp/tmpx_4kh1pp/model.ckpt-12166\n",
"{'accuracy': 0.8528125, 'average_loss': 0.46170443, 'loss': 29.549084, 'global_step': 12166}\n",
"INFO:tensorflow:Calling model_fn.\n",
"INFO:tensorflow:Done calling model_fn.\n",
"INFO:tensorflow:Signatures INCLUDED in export for Classify: None\n",
"INFO:tensorflow:Signatures INCLUDED in export for Regress: None\n",
"INFO:tensorflow:Signatures INCLUDED in export for Predict: ['predict']\n",
"INFO:tensorflow:Signatures INCLUDED in export for Train: None\n",
"INFO:tensorflow:Signatures INCLUDED in export for Eval: None\n",
"INFO:tensorflow:Signatures EXCLUDED from export because they cannot be be served via TensorFlow Serving APIs:\n",
"INFO:tensorflow:'serving_default' : Classification input must be a single string Tensor; got {'Elevation': <tf.Tensor 'Placeholder:0' shape=(?,) dtype=int32>, 'Aspect': <tf.Tensor 'Placeholder_1:0' shape=(?,) dtype=int32>, 'Slope': <tf.Tensor 'Placeholder_2:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Hydrology': <tf.Tensor 'Placeholder_3:0' shape=(?,) dtype=int32>, 'Vertical_Distance_To_Hydrology': <tf.Tensor 'Placeholder_4:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Roadways': <tf.Tensor 'Placeholder_5:0' shape=(?,) dtype=int32>, 'Hillshade_9am': <tf.Tensor 'Placeholder_6:0' shape=(?,) dtype=int32>, 'Hillshade_Noon': <tf.Tensor 'Placeholder_7:0' shape=(?,) dtype=int32>, 'Hillshade_3pm': <tf.Tensor 'Placeholder_8:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Fire_Points': <tf.Tensor 'Placeholder_9:0' shape=(?,) dtype=int32>, 'Soil_Type': <tf.Tensor 'Placeholder_10:0' shape=(?, 40) dtype=int32>, 'Cover_Type': <tf.Tensor 'Placeholder_11:0' shape=(?,) dtype=int32>}\n",
"INFO:tensorflow:'classification' : Classification input must be a single string Tensor; got {'Elevation': <tf.Tensor 'Placeholder:0' shape=(?,) dtype=int32>, 'Aspect': <tf.Tensor 'Placeholder_1:0' shape=(?,) dtype=int32>, 'Slope': <tf.Tensor 'Placeholder_2:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Hydrology': <tf.Tensor 'Placeholder_3:0' shape=(?,) dtype=int32>, 'Vertical_Distance_To_Hydrology': <tf.Tensor 'Placeholder_4:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Roadways': <tf.Tensor 'Placeholder_5:0' shape=(?,) dtype=int32>, 'Hillshade_9am': <tf.Tensor 'Placeholder_6:0' shape=(?,) dtype=int32>, 'Hillshade_Noon': <tf.Tensor 'Placeholder_7:0' shape=(?,) dtype=int32>, 'Hillshade_3pm': <tf.Tensor 'Placeholder_8:0' shape=(?,) dtype=int32>, 'Horizontal_Distance_To_Fire_Points': <tf.Tensor 'Placeholder_9:0' shape=(?,) dtype=int32>, 'Soil_Type': <tf.Tensor 'Placeholder_10:0' shape=(?, 40) dtype=int32>, 'Cover_Type': <tf.Tensor 'Placeholder_11:0' shape=(?,) dtype=int32>}\n",
"WARNING:tensorflow:Export includes no default signature!\n",
"INFO:tensorflow:Restoring parameters from /tmp/tmpx_4kh1pp/model.ckpt-12166\n",
"INFO:tensorflow:Assets added to graph.\n",
"INFO:tensorflow:No assets to write.\n",
"INFO:tensorflow:SavedModel written to: models/temp-b'1549210289'/saved_model.pb\n"
],
"name": "stdout"
},
{
"output_type": "execute_result",
"data": {
"text/plain": [
"b'models/1549210289'"
]
},
"metadata": {
"tags": []
},
"execution_count": 23
}
]
},
{
"metadata": {
"id": "Z7lVjHrdFbWw",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
""
],
"execution_count": 0,
"outputs": []
}
]
}
@andresviana
Copy link

Thanks!!!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment