Variables_Solutions.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "# Variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "from __future__ import print_function\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "datetime.date(2017, 2, 24)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from datetime import date\n",
    "date.today()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "author = \"kyubyong. https://github.com/Kyubyong/tensorflow-exercises\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.0.0'"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.12.0'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.__version__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "NOTE on notation\n",
    "\n",
    "    _x, _y, _z, ...: NumPy 0-d or 1-d arrays\n",
    "    _X, _Y, _Z, ...: NumPy 2-d or higer dimensional arrays\n",
    "    x, y, z, ...: 0-d or 1-d tensors\n",
    "    X, Y, Z, ...: 2-d or higher dimensional tensors\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Variables "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q0. Create a variable `w` with an initial value of 1.0 and name `weight`.\n",
    "Then, print out the value of `w`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.0\n"
     ]
    }
   ],
   "source": [
    "w = tf.Variable(1.0, name=\"weight\")\n",
    "with tf.Session() as sess:\n",
    "    sess.run(w.initializer)\n",
    "    print(sess.run(w))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q1. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.0 =>2.0 =>3.0 =>4.0 =>5.0 =>6.0 =>7.0 =>8.0 =>9.0 =>10.0 =>"
     ]
    }
   ],
   "source": [
    "# Create a variable w.\n",
    "w = tf.Variable(1.0, name=\"Weight\")\n",
    "\n",
    "# Q. Add 1 to w and assign the value to w.\n",
    "assign_op = w.assign(w + 1.0)\n",
    "# Or assign_op = w.assign_add(1.0)\n",
    "# Or assgin_op = tf.assgin(w, w + 1.0)\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    sess.run(w.initializer)\n",
    "    for _ in range(10):\n",
    "        print(sess.run(w), \"=>\", end=\"\")\n",
    "        sess.run(assign_op)\n",
    "        \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q2. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6.0\n"
     ]
    }
   ],
   "source": [
    "w1 = tf.Variable(1.0)\n",
    "w2 = tf.Variable(2.0)\n",
    "w3 = tf.Variable(3.0)\n",
    "\n",
    "out = w1 + w2 + w3\n",
    "\n",
    "# Q. Add an Op to initialize global variables.\n",
    "init_op = tf.global_variables_initializer()\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    sess.run(init_op) # Initialize all variables.\n",
    "    print(sess.run(out))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q3-4. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[-1.77987027  0.15173827  0.29789153 -0.0687403  -0.82471323  0.74895507\n",
      "   0.17162903  1.33624291  0.49532315  1.82508922]]\n",
      "[[-3.55974054  0.30347654  0.59578305 -0.1374806  -1.64942646  1.49791014\n",
      "   0.34325805  2.67248583  0.9906463   3.65017843]]\n"
     ]
    }
   ],
   "source": [
    "V = tf.Variable(tf.truncated_normal([1, 10]))\n",
    "# Q3. Initialize `W` with 2 * W\n",
    "W = tf.Variable(V.initialized_value() * 2.0)\n",
    "\n",
    "# Q4. Add an Op to initialize global variables.\n",
    "init_op = tf.global_variables_initializer()\n",
    "with tf.Session() as sess:\n",
    "    sess.run(init_op) # Initialize all variables.\n",
    "    _V, _W = sess.run([V, W])\n",
    "    print(_V)\n",
    "    print(_W)\n",
    "    assert np.array_equiv(_V * 2.0, _W)\n",
    "\n",
    "# Compare the value of `W` with Z = 2 * V. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q5-8. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Q5. Weight:0\n",
      "Q6. Weight\n",
      "Q7. <dtype: 'float32_ref'>\n",
      "Q8. [2, 2]\n",
      "Q9. 2\n",
      "Q10. True\n"
     ]
    }
   ],
   "source": [
    "g = tf.Graph()\n",
    "with g.as_default():\n",
    "    W = tf.Variable([[0,1],[2,3]], name=\"Weight\", dtype=tf.float32)\n",
    "    # Q5. Print the name of `W`.\n",
    "    print(\"Q5.\", W.name)\n",
    "    # Q6. Print the name of the op of `W`.\n",
    "    print(\"Q6.\", W.op.name)\n",
    "    # Q7. Print the data type of `w`.\n",
    "    print(\"Q7.\", W.dtype)\n",
    "    # Q8. Print the shape of `w`.\n",
    "    print(\"Q8.\", W.get_shape().as_list())\n",
    "    # Q9. Print the rank (or ndims) of `W`.\n",
    "    print(\"Q9.\", W.get_shape().ndims)\n",
    "    # Q10. Check if the graph of `W` is the same as `g`.\n",
    "    print(\"Q10.\", W.graph == g)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Variable helper functions "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q11-15. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "global variable => weight1:0\n",
      "global variable => weight2:0\n",
      "global variable => weight3:0\n",
      "trainable_variable => weight1:0\n",
      "trainable_variable => weight3:0\n"
     ]
    }
   ],
   "source": [
    "tf.reset_default_graph()\n",
    "\n",
    "w1 = tf.Variable(1.0, name=\"weight1\")\n",
    "w2 = tf.Variable(2.0, name=\"weight2\", trainable=False)\n",
    "w3 = tf.Variable(3.0, name=\"weight3\")\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    # Q11. Initialize the variables w1 and w2.\n",
    "    sess.run(tf.variables_initializer([w1, w2]))\n",
    "    \n",
    "    # Q12. Print the name of all global variables \n",
    "    for v in tf.global_variables():\n",
    "        print(\"global variable =>\", v.name)\n",
    "    \n",
    "    # Q13. Print the name of all trainable variables\n",
    "    for v in tf.trainable_variables():\n",
    "        print(\"trainable_variable =>\", v.name)\n",
    "    \n",
    "        \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Saving and Restoring Variables "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q14-15. This is a simple example code to find the coefficient of a linear function. (Here y = 2x).\n",
    "Complete the code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.06684 =>Saved successfully\n",
      "1.5197 =>Saved successfully\n",
      "1.74107 =>Saved successfully\n",
      "1.87005 =>Saved successfully\n",
      "1.93196 =>Saved successfully\n",
      "1.96433 =>Saved successfully\n",
      "1.98112 =>Saved successfully\n",
      "1.99044 =>Saved successfully\n",
      "1.9952 =>Saved successfully\n",
      "1.99749 =>Saved successfully\n",
      "['my-model-7000.data-00000-of-00001', 'my-model-10000.data-00000-of-00001', 'my-model-6000.meta', 'my-model-8000.data-00000-of-00001', 'my-model-10000.index', 'checkpoint', 'my-model-10000.meta', 'my-model-8000.meta', 'my-model-7000.index', 'my-model-9000.meta', 'my-model-6000.index', 'my-model-9000.index', 'my-model-8000.index', 'my-model-7000.meta', 'my-model-9000.data-00000-of-00001', 'my-model-6000.data-00000-of-00001']\n",
      "model/my-model-10000\n",
      "Restored succesfully!\n"
     ]
    }
   ],
   "source": [
    "tf.reset_default_graph()\n",
    "w = tf.Variable(0.2, 'weight') # <- This is what we want to find. The true value is 2.\n",
    "x = tf.random_uniform([1])\n",
    "y = 2. * x # Let's pretend we don't know the coefficient 2 here.\n",
    "y_hat = w * x\n",
    "loss = tf.squared_difference(y, y_hat)\n",
    "train_op = tf.train.GradientDescentOptimizer(0.001).minimize(loss)\n",
    "\n",
    "init_op = tf.global_variables_initializer()\n",
    "with tf.Session() as sess:\n",
    "    sess.run(init_op)\n",
    "    saver = tf.train.Saver()\n",
    "    for step in range(1, 10001):\n",
    "        sess.run(train_op)\n",
    "        \n",
    "        import os\n",
    "        if not os.path.exists('model'): os.mkdir('model') # Let's make a folder in order not to discrupt our current folder.\n",
    "            \n",
    "        if step % 1000 == 0:\n",
    "            print(sess.run(w), \"=>\", end=\"\")\n",
    "            \n",
    "            # Q14. Save the checkpoint file, giving it a base name of `model/my-model`\n",
    "            save_path = saver.save(sess, 'model/my-model', global_step=step)\n",
    "            print(\"Saved successfully\")\n",
    "    \n",
    "    print(os.listdir('model'))\n",
    "    # Q15. Find the filename of the latest saved checkpoint file.\n",
    "    # And if it exists, restore it.\n",
    "    ckpt = tf.train.latest_checkpoint('model')\n",
    "    print(ckpt)\n",
    "    if ckpt is not None:\n",
    "        saver.restore(sess, ckpt)\n",
    "        print(\"Restored succesfully!\")\n",
    "    \n",
    "    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Sharing Variables "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q16. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "g = tf.Graph()\n",
    "with g.as_default():\n",
    "    with tf.variable_scope(\"foo\"):\n",
    "        v = tf.get_variable(\"vv\", [1,])  # v.name == \"foo/vv:0\"\n",
    "        \n",
    "    #Q. Get the existing variable `v`\n",
    "    with tf.variable_scope(\"foo\", reuse=True):\n",
    "        v1 = tf.get_variable(\"vv\")  # The same as v above.\n",
    "\n",
    "assert v1 == v    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q17. Predict the result of this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "v.name = foo/bar/vv:0\n"
     ]
    }
   ],
   "source": [
    "with tf.variable_scope(\"foo\"):\n",
    "    with tf.variable_scope(\"bar\"):\n",
    "        v = tf.get_variable(\"vv\", [1])\n",
    "        print(\"v.name =\", v.name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q18. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x =\n",
      " [[ 0.  1.  2.  3.]\n",
      " [ 4.  5.  6.  7.]]\n"
     ]
    }
   ],
   "source": [
    "value = [0, 1, 2, 3, 4, 5, 6, 7]\n",
    "# Q. Create an initializer with `value`.\n",
    "init = tf.constant_initializer(value)\n",
    "\n",
    "tf.reset_default_graph()\n",
    "x = tf.get_variable('x', shape=[2, 4], initializer=init)\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    sess.run(x.initializer)\n",
    "    print(\"x =\\n\", sess.run(x))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q19. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Make sure the mean -0.0291916 is close to 0\n",
      "Make sure the standard deviation 1.99621 is close to 2\n"
     ]
    }
   ],
   "source": [
    "# Q. Create an initializer with a normal distribution of mean equals 0 and standard deviation equals 2.\n",
    "init = tf.random_normal_initializer(mean=0, stddev=2)\n",
    "\n",
    "tf.reset_default_graph()\n",
    "x = tf.get_variable('x', shape=[10, 1000], initializer=init)\n",
    "\n",
    "with tf.Session():\n",
    "    x.initializer.run()\n",
    "    _x = x.eval()\n",
    "    print(\"Make sure the mean\", np.mean(_x), \"is close to 0\" )\n",
    "    print(\"Make sure the standard deviation\", np.std(_x), \"is close to 2\" )\n",
    "    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q20. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAD8CAYAAABjAo9vAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1388392990>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Q. Create an initializer with a truncated normal distribution of mean equals 0 and standard deviation equals 2.\n",
    "init = tf.truncated_normal_initializer(mean=0, stddev=2)\n",
    "\n",
    "tf.reset_default_graph()\n",
    "x = tf.get_variable('x', shape=[1000,], initializer=init)\n",
    "\n",
    "with tf.Session():\n",
    "    x.initializer.run()\n",
    "    _x = x.eval()\n",
    "    plt.scatter(np.arange(1000), _x)\n",
    "    _avg = np.array([np.mean(_x)] * 1000)\n",
    "    _std = np.array([np.std(_x)] * 1000)\n",
    "    plt.plot(np.arange(1000), _avg, 'r-')\n",
    "    plt.plot(np.arange(1000), _avg + 2*_std, 'g-')\n",
    "    plt.plot(np.arange(1000), _avg - 2*_std, 'k-')\n",
    "    plt.legend(['mean', 'upper 2*std', 'lower 2*std'])\n",
    "    plt.show()\n",
    "# Note that few data points exceed the bounaries.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q21. Complete this code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f138828e090>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Q. Create an initializer with a random normal distribution of minimum 0 and maximum 1.\n",
    "init = tf.random_uniform_initializer(0, 1)\n",
    "\n",
    "tf.reset_default_graph()\n",
    "x = tf.get_variable('x', shape=[5000,], initializer=init)\n",
    "\n",
    "with tf.Session():\n",
    "    x.initializer.run()\n",
    "    _x = x.eval()\n",
    "    count, bins, ignored = plt.hist(_x, 20, normed=True)\n",
    "    plt.plot(bins, np.ones_like(bins), linewidth=2, color='r')\n",
    "    plt.show()\n",
    "# Note that few data points exceed the bounaries.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "# Check other initializers such as zeros_initializer, ones_initializer, or orthogonal_initializer, as well."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Exporting and Importing Meta Graphs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Q22. Complete the code. Make sure you've done questions 14-15."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Of course, there're no variables since we reset the graph. See []\n",
      "Now we have a variable Variable:0\n"
     ]
    }
   ],
   "source": [
    "tf.reset_default_graph()\n",
    "print(\"Of course, there're no variables since we reset the graph. See\", tf.global_variables())\n",
    "with tf.Session() as sess:\n",
    "    # Q. Import the saved graph of `model/my-model-10000`.\n",
    "    new_saver = tf.train.import_meta_graph('model/my-model-10000.meta')\n",
    "    \n",
    "    new_saver.restore(sess, 'model/my-model-10000')\n",
    "    \n",
    "    for v in tf.global_variables():\n",
    "        print(\"Now we have a variable\", v.name)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}