notebooks/variational_inference/convolutional_vae_keras_advi.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Convolutional variational autoencoder with PyMC3 and Keras"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this document, I will show how autoencoding variational Bayes (AEVB) works in PyMC3's automatic differentiation variational inference (ADVI). The example here is borrowed from [Keras example](https://github.com/fchollet/keras/blob/master/examples/variational_autoencoder_deconv.py), where convolutional variational autoencoder is applied to the MNIST dataset. The network architecture of the encoder and decoder are the same. However, PyMC3 allows us to define a probabilistic model, which combines the encoder and decoder, in the same way as other probabilistic models (e.g., generalized linear models), rather than directly implementing of Monte Carlo sampling and the loss function, as is done in the Keras example. Thus the framework of AEVB in PyMC3 can be extended to more complex models such as [latent dirichlet allocation](https://taku-y.github.io/notebook/20160928/lda-advi-ae.html). "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Notebook Written by Taku Yoshioka (c) 2016"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To use Keras with PyMC3, we need to choose [Theano](http://deeplearning.net/software/theano/) as the backend for Keras. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "IPython.notebook.set_autosave_interval(0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Autosave disabled\n",
      "env: KERAS_BACKEND=theano\n",
      "env: THEANO_FLAGS=device=cuda3,floatX=float32,optimizer=fast_run\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using Theano backend.\n",
      "ERROR (theano.gpuarray): pygpu was configured but could not be imported or is too old (version 0.7 or higher required)\n",
      "NoneType: None\n"
     ]
    }
   ],
   "source": [
    "%autosave 0\n",
    "%env KERAS_BACKEND=theano\n",
    "%env THEANO_FLAGS=device=cuda3,floatX=float32,optimizer=fast_run\n",
    "\n",
    "import os\n",
    "import sys\n",
    "\n",
    "from collections import OrderedDict\n",
    "\n",
    "import arviz as az\n",
    "import keras\n",
    "import matplotlib\n",
    "import matplotlib.gridspec as gridspec\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pymc3 as pm\n",
    "import theano.tensor as tt\n",
    "\n",
    "from keras import backend as K\n",
    "from keras.layers import (\n",
    "    Activation,\n",
    "    BatchNormalization,\n",
    "    Conv2D,\n",
    "    Deconv2D,\n",
    "    Dense,\n",
    "    Flatten,\n",
    "    InputLayer,\n",
    "    Reshape,\n",
    ")\n",
    "from theano import clone, config, function, pp, shared\n",
    "\n",
    "print(f\"Running on PyMC3 v{pm.__version__}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%config InlineBackend.figure_format = 'retina'\n",
    "az.style.use(\"arviz-darkgrid\")\n",
    "K.set_image_dim_ordering(\"th\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load images\n",
    "MNIST dataset can be obtained by [scikit-learn API](http://scikit-learn.org/stable/datasets/) or from [Keras datasets](https://keras.io/datasets/). The dataset contains images of digits. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://s3.amazonaws.com/img-datasets/mnist.npz\n",
      "11493376/11490434 [==============================] - 13s 1us/step\n"
     ]
    }
   ],
   "source": [
    "from keras.datasets import mnist\n",
    "\n",
    "(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
    "data = pm.floatX(x_train.reshape(-1, 1, 28, 28))\n",
    "data /= np.max(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use Keras\n",
    "We define a utility function to get parameters from Keras models. Since we have set the backend to Theano, parameter objects are obtained as shared variables of Theano. \n",
    "\n",
    "In the code, 'updates' are expected to include update objects (dictionary of pairs of shared variables and update equation) of scaling parameters of batch normalization. While not using batch normalization in this example, if we want to use it, we need to pass these update objects as an argument of `theano.function()` inside the PyMC3 ADVI function. The current version of PyMC3 does not support it, it is easy to modify (I want to send PR in future). \n",
    "\n",
    "The learning phase below is used for Keras to known the learning phase, training or test. This information is important also for batch normalization. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.layers import BatchNormalization, Dense\n",
    "from keras.models import Sequential\n",
    "\n",
    "\n",
    "def get_params(model):\n",
    "    \"\"\"Get parameters and updates from Keras model\"\"\"\n",
    "    shared_in_updates = list()\n",
    "    params = list()\n",
    "    updates = dict()\n",
    "\n",
    "    for l in model.layers:\n",
    "        attrs = dir(l)\n",
    "        # Updates\n",
    "        if \"updates\" in attrs:\n",
    "            updates.update(l.updates)\n",
    "            shared_in_updates += [e[0] for e in l.updates]\n",
    "\n",
    "        # Shared variables\n",
    "        for attr_str in attrs:\n",
    "            attr = getattr(l, attr_str)\n",
    "            if isinstance(attr, tt.compile.SharedVariable):\n",
    "                if attr is not model.get_input_at(0):\n",
    "                    params.append(attr)\n",
    "\n",
    "    return list(set(params) - set(shared_in_updates)), updates\n",
    "\n",
    "\n",
    "# This code is required when using BatchNormalization layer\n",
    "keras.backend.theano_backend._LEARNING_PHASE = shared(np.uint8(1), name=\"keras_learning_phase\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Encoder and decoder"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, we define the convolutional neural network for encoder using the Keras API. This function returns a CNN model given the shared variable representing observations (images of digits), the dimension of latent space, and the parameters of the model architecture. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cnn_enc(xs, latent_dim, nb_filters=64, nb_conv=3, intermediate_dim=128):\n",
    "    \"\"\"Returns a CNN model of Keras.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    xs : theano.TensorVariable\n",
    "        Input tensor.\n",
    "    latent_dim : int\n",
    "        Dimension of latent vector.\n",
    "    \"\"\"\n",
    "    input_layer = InputLayer(input_tensor=xs, batch_input_shape=xs.tag.test_value.shape)\n",
    "    model = Sequential()\n",
    "    model.add(input_layer)\n",
    "\n",
    "    cp1 = {\"padding\": \"same\", \"activation\": \"relu\"}\n",
    "    cp2 = {\"padding\": \"same\", \"activation\": \"relu\", \"strides\": (2, 2)}\n",
    "    cp3 = {\"padding\": \"same\", \"activation\": \"relu\", \"strides\": (1, 1)}\n",
    "    cp4 = cp3\n",
    "\n",
    "    model.add(Conv2D(1, (2, 2), **cp1))\n",
    "    model.add(Conv2D(nb_filters, (2, 2), **cp2))\n",
    "    model.add(Conv2D(nb_filters, (nb_conv, nb_conv), **cp3))\n",
    "    model.add(Conv2D(nb_filters, (nb_conv, nb_conv), **cp4))\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(intermediate_dim, activation=\"relu\"))\n",
    "    model.add(Dense(2 * latent_dim))\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we define a utility class for encoders. This class does not depend on the architecture of the encoder except for input shape (`tensor4` for images), so we can use this class for various encoding networks. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Encoder:\n",
    "    \"\"\"Encode observed images to variational parameters (mean/std of Gaussian).\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    xs : theano.tensor.sharedvar.TensorSharedVariable\n",
    "        Placeholder of input images.\n",
    "    dim_hidden : int\n",
    "        The number of hidden variables.\n",
    "    net : Function\n",
    "        Returns\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, xs, dim_hidden, net):\n",
    "        model = net(xs, dim_hidden)\n",
    "\n",
    "        self.model = model\n",
    "        self.xs = xs\n",
    "        self.out = model.get_output_at(-1)\n",
    "        self.means = self.out[:, :dim_hidden]\n",
    "        self.rhos = self.out[:, dim_hidden:]\n",
    "        self.params, self.updates = get_params(model)\n",
    "        self.enc_func = None\n",
    "        self.dim_hidden = dim_hidden\n",
    "\n",
    "    def _get_enc_func(self):\n",
    "        if self.enc_func is None:\n",
    "            xs = tt.tensor4()\n",
    "            means = clone(self.means, {self.xs: xs})\n",
    "            rhos = clone(self.rhos, {self.xs: xs})\n",
    "            self.enc_func = function([xs], [means, rhos])\n",
    "\n",
    "        return self.enc_func\n",
    "\n",
    "    def encode(self, xs):\n",
    "        # Used in test phase\n",
    "        keras.backend.theano_backend._LEARNING_PHASE.set_value(np.uint8(0))\n",
    "\n",
    "        enc_func = self._get_enc_func()\n",
    "        means, _ = enc_func(xs)\n",
    "\n",
    "        return means\n",
    "\n",
    "    def draw_samples(self, xs, n_samples=1):\n",
    "        \"\"\"Draw samples of hidden variables based on variational parameters encoded.\n",
    "\n",
    "        Parameters\n",
    "        ----------\n",
    "        xs : numpy.ndarray, shape=(n_images, 1, height, width)\n",
    "            Images.\n",
    "        \"\"\"\n",
    "        # Used in test phase\n",
    "        keras.backend.theano_backend._LEARNING_PHASE.set_value(np.uint8(0))\n",
    "\n",
    "        enc_func = self._get_enc_func()\n",
    "        means, rhos = enc_func(xs)\n",
    "        means = np.repeat(means, n_samples, axis=0)\n",
    "        rhos = np.repeat(rhos, n_samples, axis=0)\n",
    "        ns = np.random.randn(len(xs) * n_samples, self.dim_hidden)\n",
    "        zs = means + pm.distributions.dist_math.rho2sd(rhos) * ns\n",
    "\n",
    "        return zs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In a similar way, we define the decoding network and a utility class for decoders. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cnn_dec(zs, nb_filters=64, nb_conv=3, output_shape=(1, 28, 28)):\n",
    "    \"\"\"Returns a CNN model of Keras.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    zs : theano.tensor.var.TensorVariable\n",
    "        Input tensor.\n",
    "    \"\"\"\n",
    "    minibatch_size, dim_hidden = zs.tag.test_value.shape\n",
    "    input_layer = InputLayer(input_tensor=zs, batch_input_shape=zs.tag.test_value.shape)\n",
    "    model = Sequential()\n",
    "    model.add(input_layer)\n",
    "\n",
    "    model.add(Dense(dim_hidden, activation=\"relu\"))\n",
    "    model.add(Dense(nb_filters * 14 * 14, activation=\"relu\"))\n",
    "\n",
    "    cp1 = {\"padding\": \"same\", \"activation\": \"relu\", \"strides\": (1, 1)}\n",
    "    cp2 = cp1\n",
    "    cp3 = {\"padding\": \"valid\", \"activation\": \"relu\", \"strides\": (2, 2)}\n",
    "    cp4 = {\"padding\": \"same\", \"activation\": \"sigmoid\"}\n",
    "\n",
    "    output_shape_ = (minibatch_size, nb_filters, 14, 14)\n",
    "    model.add(Reshape(output_shape_[1:]))\n",
    "    model.add(Deconv2D(nb_filters, (nb_conv, nb_conv), data_format=\"channels_first\", **cp1))\n",
    "    model.add(Deconv2D(nb_filters, (nb_conv, nb_conv), data_format=\"channels_first\", **cp2))\n",
    "    output_shape_ = (minibatch_size, nb_filters, 29, 29)\n",
    "    model.add(Deconv2D(nb_filters, (2, 2), data_format=\"channels_first\", **cp3))\n",
    "    model.add(Conv2D(1, (2, 2), **cp4))\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Decoder:\n",
    "    \"\"\"Decode hidden variables to images.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    zs : Theano tensor\n",
    "        Hidden variables.\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, zs, net):\n",
    "        model = net(zs)\n",
    "        self.model = model\n",
    "        self.zs = zs\n",
    "        self.out = model.get_output_at(-1)\n",
    "        self.params, self.updates = get_params(model)\n",
    "        self.dec_func = None\n",
    "\n",
    "    def _get_dec_func(self):\n",
    "        if self.dec_func is None:\n",
    "            zs = tt.matrix()\n",
    "            xs = clone(self.out, {self.zs: zs})\n",
    "            self.dec_func = function([zs], xs)\n",
    "\n",
    "        return self.dec_func\n",
    "\n",
    "    def decode(self, zs):\n",
    "        \"\"\"Decode hidden variables to images.\n",
    "\n",
    "        An image consists of the mean parameters of the observation noise.\n",
    "\n",
    "        Parameters\n",
    "        ----------\n",
    "        zs : numpy.ndarray, shape=(n_samples, dim_hidden)\n",
    "            Hidden variables.\n",
    "        \"\"\"\n",
    "        # Used in test phase\n",
    "        keras.backend.theano_backend._LEARNING_PHASE.set_value(np.uint8(0))\n",
    "\n",
    "        return self._get_dec_func()(zs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generative model\n",
    "We can construct the generative model with the PyMC3 API and the functions and classes defined above. We set the size of mini-batches to 100 and the dimension of the latent space to 2 for visualization. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Constants\n",
    "minibatch_size = 100\n",
    "dim_hidden = 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We require a placeholder for images, into which mini-batches of images will be placed during ADVI inference. It is also the input for the encoder. Below, `enc.model` is a Keras model of the encoder network and we can check the model architecture using the method `summary()`. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (100, 1, 28, 28)          5         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (100, 64, 14, 14)         320       \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (100, 64, 14, 14)         36928     \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (100, 64, 14, 14)         36928     \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (100, 12544)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (100, 128)                1605760   \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (100, 4)                  516       \n",
      "=================================================================\n",
      "Total params: 1,680,457\n",
      "Trainable params: 1,680,457\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Placeholder of images\n",
    "xs_t = tt.tensor4(name=\"xs_t\")\n",
    "xs_t.tag.test_value = np.zeros((minibatch_size, 1, 28, 28)).astype(\"float32\")\n",
    "# Encoder\n",
    "enc = Encoder(xs_t, dim_hidden, net=cnn_enc)\n",
    "enc.model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The probabilistic model involves only two random variables; latent variable $\\mathbf{z}$ and observation $\\mathbf{x}$. We put a Normal prior on $\\mathbf{z}$, decode the variational parameters of $q(\\mathbf{z}|\\mathbf{x})$ and define the likelihood of the observation $\\mathbf{x}$. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with pm.Model() as model:\n",
    "    # Hidden variables\n",
    "    zs = pm.Normal(\n",
    "        \"zs\",\n",
    "        mu=0,\n",
    "        sigma=1,\n",
    "        shape=(minibatch_size, dim_hidden),\n",
    "        dtype=\"float32\",\n",
    "        total_size=len(data),\n",
    "    )\n",
    "\n",
    "    # Decoder and its parameters\n",
    "    dec = Decoder(zs, net=cnn_dec)\n",
    "\n",
    "    # Observation model\n",
    "    xs_ = pm.Normal(\n",
    "        \"xs_\", mu=dec.out, sigma=0.1, observed=xs_t, dtype=\"float32\", total_size=len(data)\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the generative model above, we do not know how the decoded variational parameters are passed to $q(\\mathbf{z}|\\mathbf{x})$. To do this, we will set the argument `local_RVs` in the ADVI function of PyMC3. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "local_RVs = OrderedDict({zs: dict(mu=enc.means, rho=enc.rhos)})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This argument is an `OrderedDict` whose keys are random variables to which the decoded variational parameters are set (`zs` in this model). Each value of the dictionary contains two Theano expressions representing variational mean (`enc.means`) and rhos (`enc.rhos`). A scaling constant (`len(data) / float(minibatch_size)`) is set automatically (as we specified it in the model saying what's the `total_size`) to compensate for the size of mini-batches of the corresponding log probability terms in the evidence lower bound (ELBO), the objective of the variational inference. \n",
    "\n",
    "The scaling constant for the observed random variables is set in the same way. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can also check the architecture of the decoding network, as we did for the encoding network. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_3 (Dense)              (100, 2)                  6         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (100, 12544)              37632     \n",
      "_________________________________________________________________\n",
      "reshape_1 (Reshape)          (100, 64, 14, 14)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_1 (Conv2DTr (100, 64, 14, 14)         36928     \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_2 (Conv2DTr (100, 64, 14, 14)         36928     \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_3 (Conv2DTr (100, 64, 28, 28)         16448     \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (100, 1, 28, 28)          257       \n",
      "=================================================================\n",
      "Total params: 128,199\n",
      "Trainable params: 128,199\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "dec.model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Inference"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's use ADVI to fit the model. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/twiecki/working/projects/pymc/pymc3/data.py:244: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.\n",
      "  self.shared = theano.shared(data[in_memory_slc])\n",
      "Average Loss = 31,566: 100%|██████████| 15000/15000 [3:35:45<00:00,  1.16it/s]   \n",
      "Finished [100%]: Average Loss = 31,552\n"
     ]
    }
   ],
   "source": [
    "# In memory Minibatches for better speed\n",
    "xs_t_minibatch = pm.Minibatch(data, minibatch_size)\n",
    "\n",
    "with model:\n",
    "    approx = pm.fit(\n",
    "        15000,\n",
    "        local_rv=local_RVs,\n",
    "        more_obj_params=enc.params + dec.params,\n",
    "        obj_optimizer=pm.rmsprop(learning_rate=0.001),\n",
    "        more_replacements={xs_t: xs_t_minibatch},\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can plot the trace of the negative ELBO obtained during optimization, to verify convergence. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(approx.hist);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, we can plot the distribution of the images in the latent space. To do this, we make a 2-dimensional grid of points and feed them into the decoding network. The mean of $p(\\mathbf{x}|\\mathbf{z})$ is the image corresponding to the samples on the grid. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "nn = 10\n",
    "zs = np.array([(z1, z2) for z1 in np.linspace(-2, 2, nn) for z2 in np.linspace(-2, 2, nn)]).astype(\n",
    "    \"float32\"\n",
    ")\n",
    "xs = dec.decode(zs)[:, 0, :, :]\n",
    "xs = np.bmat([[xs[i + j * nn] for i in range(nn)] for j in range(nn)])\n",
    "matplotlib.rc(\"axes\", **{\"grid\": False})\n",
    "plt.figure(figsize=(10, 10))\n",
    "plt.imshow(xs, interpolation=\"none\", cmap=\"gray\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "pymc3 3.8\n",
      "arviz 0.8.3\n",
      "numpy 1.17.5\n",
      "last updated: Thu Jun 11 2020 \n",
      "\n",
      "CPython 3.8.2\n",
      "IPython 7.11.0\n",
      "watermark 2.0.2\n"
     ]
    }
   ],
   "source": [
    "%load_ext watermark\n",
    "%watermark -n -u -v -iv -w"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.2"
  },
  "latex_envs": {
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 0
  },
  "nav_menu": {},
  "toc": {
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 6,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}