UnitAndWeightBasedPruning.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "UnitAndWeightBasedPruning.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Snehal-Reddy/NeuralNetPruning/blob/master/UnitAndWeightBasedPruning.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "id": "pjvzFrTQqV1H",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "e5d8fc16-6ee8-4727-c510-95f37e84c2a0"
      },
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "import matplotlib.pyplot as plt\n",
        "import time\n",
        "import keras\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Flatten\n",
        "tf.enable_eager_execution()"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "VRQ_VYR0rA_R",
        "colab_type": "code",
        "outputId": "feb259bc-571d-49de-97ea-798148293244",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 786
        }
      },
      "cell_type": "code",
      "source": [
        "mnist = tf.keras.datasets.mnist\n",
        "(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
        "# features, labels = next(iter(mnist))\n",
        "\n",
        "fig = plt.figure()\n",
        "for i in range(9):\n",
        "    plt.subplot(3,3,i+1)\n",
        "    plt.tight_layout()\n",
        "    plt.imshow(x_train[i], cmap='gray', interpolation='none')\n",
        "    plt.title(\"Digit: {}\".format(y_train[i]))\n",
        "    plt.xticks([])\n",
        "    plt.yticks([])\n",
        "fig"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n",
            "11493376/11490434 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdcAAAFvCAYAAAAYHYHZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAgBg2f8Dxy59It9JnnEAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 2
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdcAAAFvCAYAAAAYHYHZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAgBg2f8Dxy59It9JnnEAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "i4offW0fcJjy",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "dataset_train = tf.data.Dataset.from_tensor_slices((\n",
        "    tf.cast(x_train/255, tf.float32),\n",
        "    tf.cast(y_train, tf.int64)\n",
        ")).shuffle(1000).batch(64)\n",
        "\n",
        "dataset_test = tf.data.Dataset.from_tensor_slices((\n",
        "    tf.cast(x_test/255, tf.float32),\n",
        "    tf.cast(y_test, tf.int64)\n",
        ")).batch(64)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "vdPxsuP_rknL",
        "colab_type": "code",
        "outputId": "718f0294-afbd-43df-cb95-ce34ed511348",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 89
        }
      },
      "cell_type": "code",
      "source": [
        "naive_model = tf.keras.Sequential([\n",
        "    tf.keras.layers.Flatten(input_shape=(28, 28)),\n",
        "    tf.keras.layers.Dense(1000, activation=tf.nn.relu, use_bias=False),\n",
        "    tf.keras.layers.Dense(1000, activation=tf.nn.relu, use_bias=False),\n",
        "    tf.keras.layers.Dense(500, activation=tf.nn.relu, use_bias=False),\n",
        "    tf.keras.layers.Dense(200, activation=tf.nn.relu, use_bias=False),\n",
        "    tf.keras.layers.Dense(10, use_bias=False)\n",
        "])"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:642: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "uJCN0eOVgNxY",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def loss(naive_model, x, y):\n",
        "    y_ = naive_model(x)\n",
        "    return tf.losses.sparse_softmax_cross_entropy(labels=y, logits=y_)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "m9lQhH_ngjNB",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def grad(naive_model, inputs, targets):\n",
        "    with tf.GradientTape() as tape:\n",
        "        loss_value = loss(naive_model, inputs, targets)\n",
        "    return loss_value, tape.gradient(loss_value, naive_model.trainable_variables)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "WjYgres0hLzZ",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "optimizer = tf.train.AdamOptimizer()\n",
        "global_step = tf.train.get_or_create_global_step()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "_Fp8l-I9eR3S",
        "colab_type": "code",
        "outputId": "ffbf9a89-ed74-45d8-ab3b-3401afae8af5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 228
        }
      },
      "cell_type": "code",
      "source": [
        "from tensorflow import contrib\n",
        "tfe = contrib.eager\n",
        "\n",
        "train_loss_results = []\n",
        "train_accuracy_results = []\n",
        "\n",
        "epoch_num = 5\n",
        "\n",
        "for epoch in range(epoch_num):\n",
        "    epoch_loss_avg = tfe.metrics.Mean()\n",
        "    epoch_accuracy = tfe.metrics.Accuracy()\n",
        "    \n",
        "    for x, y in (dataset_train):\n",
        "        outputs = naive_model(x)\n",
        "        loss_value, grads = grad(naive_model, x, y)\n",
        "        optimizer.apply_gradients(zip(grads, naive_model.trainable_variables),global_step)  \n",
        "        \n",
        "        epoch_loss_avg(loss_value)  \n",
        "#         epoch_accuracy(tf.argmax(naive_model(x), axis=1, output_type=tf.int32), y)\n",
        "        epoch_accuracy(tf.argmax(outputs, axis=1, output_type=tf.int64), y)\n",
        "    \n",
        "    train_loss_results.append(epoch_loss_avg.result())\n",
        "    train_accuracy_results.append(epoch_accuracy.result())\n",
        "\n",
        "    if epoch % 50 == 0:\n",
        "        print(\"Epoch {:}: Loss: {:.3f}, Accuracy: {:.3%}\".format(epoch,\n",
        "                                                                epoch_loss_avg.result(),\n",
        "                                                                epoch_accuracy.result()))\n"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "WARNING: The TensorFlow contrib module will not be included in TensorFlow 2.0.\n",
            "For more information, please see:\n",
            "  * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md\n",
            "  * https://github.com/tensorflow/addons\n",
            "If you depend on functionality not listed there, please file an issue.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/losses/losses_impl.py:209: to_float (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Epoch 000: Loss: 0.199, Accuracy: 94.003%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "HKN5FBZaxpak",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        " def test(model, data_test):  \n",
        "    epoch_loss_avg = tfe.metrics.Mean()\n",
        "    epoch_accuracy = tfe.metrics.Accuracy()\n",
        "    for x, y in dataset_test:\n",
        "        outputs = model(x)\n",
        "        loss = tf.losses.softmax_cross_entropy(tf.one_hot(y, 10), outputs)\n",
        "        epoch_loss_avg(loss)\n",
        "        epoch_accuracy(tf.argmax(outputs, axis=1, output_type=tf.int64), y)\n",
        "    result = epoch_loss_avg.result().numpy(), epoch_accuracy.result().numpy()\n",
        "    return result\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "A2tnF4l9swnf",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import time\n",
        "list_percentile = [0,1,5,10,15,25,50,80,95]\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "3HIRfUiS2qGX",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        ""
      ]
    },
    {
      "metadata": {
        "id": "NYVAJlnlwU7p",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "class NewModel():\n",
        "    def __init__(self, sparsify=True):\n",
        "        self.sparsify = sparsify\n",
        "        self.flatten = tf.layers.Flatten()\n",
        "        self.weights = []\n",
        "        \n",
        "    def add_layer(self, weights):\n",
        "        if self.sparsify:\n",
        "            weights = tf.contrib.layers.dense_to_sparse(weights)\n",
        "        self.weights.append(weights)\n",
        "        \n",
        "    def __call__(self, dense_x):\n",
        "        dense_x = self.flatten(dense_x)\n",
        "        if self.sparsify:\n",
        "            dense_x = tf.transpose(dense_x) \n",
        "        for i, weights in enumerate(self.weights):\n",
        "            if self.sparsify:\n",
        "                dense_x = tf.sparse.matmul(weights, dense_x, adjoint_a=True) \n",
        "            else:\n",
        "                dense_x = tf.matmul(dense_x, weights)\n",
        "            if i < len(self.weights)-1:\n",
        "                dense_x = tf.nn.relu(dense_x)\n",
        "        if self.sparsify:\n",
        "            dense_x = tf.transpose(dense_x) # X == (X^T)^T\n",
        "        return dense_x\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "6p97KmB6mG2l",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def weights_zero(model, percentile, sparsify=True):\n",
        "    new_model = NewModel(sparsify)\n",
        "\n",
        "    for num_layer, weights in enumerate(model.trainable_weights):\n",
        "        weights_np = weights.numpy()\n",
        "        if num_layer < len(model.trainable_weights)-1:\n",
        "            all_greater = np.abs(weights_np) < np.percentile(np.abs(weights_np), percentile)\n",
        "            weights_np[all_greater] = 0\n",
        "\n",
        "        new_model.add_layer(tf.convert_to_tensor(weights_np))\n",
        "\n",
        "    return new_model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "_QcCJuv-2-kf",
        "colab_type": "code",
        "outputId": "da16c87f-4677-4f64-9ec4-e14159ef4963",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 173
        }
      },
      "cell_type": "code",
      "source": [
        "test_losses = []\n",
        "test_accuracies = []\n",
        "sparse_timings = []\n",
        "timings = []\n",
        "\n",
        "for percentile in list_percentile:\n",
        "    sparsified_model = weights_zero(naive_model, percentile, True)\n",
        "    start_time_1 = time.clock()\n",
        "    l, a = test(sparsified_model, dataset_test)\n",
        "    end_time_1 = time.clock()\n",
        "\n",
        "    pruned_model = weights_zero(naive_model, percentile, False)\n",
        "    start_time_2 = time.clock()\n",
        "    l2, a2 = test(pruned_model, dataset_test)\n",
        "    end_time_2 = time.clock()\n",
        "    \n",
        "    np.testing.assert_almost_equal(l, l2)\n",
        "    np.testing.assert_almost_equal(a, a2)\n",
        "    \n",
        "    print(f\"{percentile} -> Loss: {l:.4f},  Accuracy: {a:.4f}, Sparse Time {end_time_1-start_time_1:.2f}s, Prune Time {end_time_2-start_time_2:.2f}s)\")\n",
        "    test_losses.append(l)\n",
        "    test_accuracies.append(a)\n",
        "    sparse_timings.append(end_time_1-start_time_1)\n",
        "    timings.append(end_time_2-start_time_2)"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0 -> Loss: 0.0931,  Accuracy: 0.9730, Sparse Time 33.95s, Prune Time 2.91s)\n",
            "1 -> Loss: 0.0931,  Accuracy: 0.9730, Sparse Time 33.61s, Prune Time 2.93s)\n",
            "5 -> Loss: 0.0932,  Accuracy: 0.9729, Sparse Time 32.26s, Prune Time 2.94s)\n",
            "10 -> Loss: 0.0930,  Accuracy: 0.9731, Sparse Time 30.60s, Prune Time 2.92s)\n",
            "15 -> Loss: 0.0931,  Accuracy: 0.9732, Sparse Time 28.99s, Prune Time 2.84s)\n",
            "25 -> Loss: 0.0916,  Accuracy: 0.9731, Sparse Time 25.65s, Prune Time 2.94s)\n",
            "50 -> Loss: 0.0838,  Accuracy: 0.9754, Sparse Time 17.41s, Prune Time 2.88s)\n",
            "80 -> Loss: 0.2835,  Accuracy: 0.9726, Sparse Time 7.45s, Prune Time 2.94s)\n",
            "95 -> Loss: 2.0756,  Accuracy: 0.8624, Sparse Time 2.45s, Prune Time 3.10s)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "Tjknbp_iscMF",
        "colab_type": "code",
        "outputId": "2fd4f304-be80-4f2e-8135-e380e1364082",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        }
      },
      "cell_type": "code",
      "source": [
        "plt.plot(list_percentile, test_losses, color='g')\n",
        "plt.xlabel('k-percentile')\n",
        "plt.ylabel('Loss')\n",
        "plt.title('Loss vs percentile plot for weight pruning')\n",
        "plt.show()"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCIY6ERGRjfj/4Ybk19kVKvcAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "8yEgVoLKtmnn",
        "colab_type": "code",
        "outputId": "98bcb125-5fc5-49f5-9eba-2a2fe0e6c7d7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        }
      },
      "cell_type": "code",
      "source": [
        "plt.plot(list_percentile, test_accuracies, color='g')\n",
        "plt.xlabel('k-percentile')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.title('Accuraccy vs percentile plot for weight pruning')\n",
        "plt.show()"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nPw8npMZQuJi2AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "935-rSukyE6G",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "###########################################################################################################################################"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "jWb84-2qyHzx",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def neuron_prune(model, percentile):\n",
        "    new_model = Sequential()\n",
        "    new_model.add(Flatten(input_shape=(28, 28)))\n",
        "    cols_prev = None\n",
        "\n",
        "    for layer_num, weights in enumerate(model.trainable_weights):\n",
        "        wts_np = weights.numpy()\n",
        "        if layer_num < len(model.trainable_weights)-1: # Do not prune last layer\n",
        "            min_val = np.percentile(np.linalg.norm(wts_np, ord=2, axis=0), percentile)\n",
        "            all_greater = np.linalg.norm(wts_np, ord=2, axis=0) >= min_val\n",
        "            wts_np = wts_np[:, all_greater]\n",
        "            \n",
        "        if cols_prev is not None:\n",
        "            wts_np = wts_np[cols_prev, :]\n",
        "        \n",
        "        if layer_num < len(model.trainable_weights)-1: \n",
        "            cols_prev = np.argwhere(all_greater).reshape(-1)\n",
        "\n",
        "        new_layer = tf.keras.layers.Dense(wts_np.shape[1], activation=tf.nn.relu, use_bias=False)\n",
        "        new_model.add(new_layer)\n",
        "        new_layer.set_weights([wts_np])\n",
        "\n",
        "    return new_model\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "kEhGJtDO91TN",
        "colab_type": "code",
        "outputId": "8768c266-d3e3-4933-b62f-89321f492819",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 173
        }
      },
      "cell_type": "code",
      "source": [
        "neuron_test_losses = []\n",
        "neuron_test_accuracies = []\n",
        "unit_timings = []\n",
        "\n",
        "for percentile in list_percentile:\n",
        "    sparse_model = neuron_prune(naive_model, percentile)\n",
        "    start_t = time.clock()\n",
        "    loss, acc = test(sparse_model, dataset_test)\n",
        "    end_t = time.clock()\n",
        "    print(f\"{percentile} ->  Loss: {loss:.4f},  Accuracy: {acc:.4f}, Timing: {start_t - end_t:.2f}s\")\n",
        "    neuron_test_losses.append(loss)\n",
        "    neuron_test_accuracies.append(acc)\n",
        "    unit_timings.append(start_t - end_t)"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0 ->  Loss: 0.0950,  Accuracy: 0.9730, Timing: -3.00s\n",
            "1 ->  Loss: 0.0950,  Accuracy: 0.9730, Timing: -3.42s\n",
            "5 ->  Loss: 0.0948,  Accuracy: 0.9731, Timing: -3.19s\n",
            "10 ->  Loss: 0.0949,  Accuracy: 0.9731, Timing: -3.05s\n",
            "15 ->  Loss: 0.0937,  Accuracy: 0.9730, Timing: -2.86s\n",
            "25 ->  Loss: 0.0915,  Accuracy: 0.9736, Timing: -2.52s\n",
            "50 ->  Loss: 0.2258,  Accuracy: 0.9686, Timing: -1.55s\n",
            "80 ->  Loss: 2.0074,  Accuracy: 0.7042, Timing: -1.04s\n",
            "95 ->  Loss: 2.3023,  Accuracy: 0.1378, Timing: -0.65s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "D_s_-E3z-hUa",
        "colab_type": "code",
        "outputId": "9f86eb68-9c09-457b-b055-85c0162d050c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        }
      },
      "cell_type": "code",
      "source": [
        "plt.plot(list_percentile, neuron_test_losses, color='g')\n",
        "plt.xlabel('k-percentile')\n",
        "plt.ylabel('Loss')\n",
        "plt.title('Loss vs percentile plot for neuron pruning')\n",
        "plt.show()"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "PCP8Ykdg-yLq",
        "colab_type": "code",
        "outputId": "1ae86b7d-164b-42b3-80a6-c926ecbecaee",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        }
      },
      "cell_type": "code",
      "source": [
        "plt.plot(list_percentile, neuron_test_accuracies, color='g')\n",
        "plt.xlabel('k-percentile')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.title('Accuracy vs percentile plot for neuron_pruning')\n",
        "plt.show()"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUXmLiIg4mP8Hu/ic2l2CkMcAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "VTSr4bFkzUL_",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        },
        "outputId": "94a18294-cb4e-4950-942c-d8626a656a69"
      },
      "cell_type": "code",
      "source": [
        "plt.plot(list_percentile, neuron_test_accuracies, color='b', label = 'neuron pruning')\n",
        "plt.plot(list_percentile, test_accuracies, color='r', label = 'weight pruning')\n",
        "plt.xlabel('k-percentile')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.title('Accuracy vs percentile plot ')\n",
        "plt.show()"
      ],
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "LqYeTNVFzfz0",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        },
        "outputId": "304ed6a8-02d9-4607-8be9-c81e1f8c5995"
      },
      "cell_type": "code",
      "source": [
        "plt.plot(list_percentile, neuron_test_losses, color='b', label = 'neuron pruning')\n",
        "plt.plot(list_percentile, test_losses, color='r', label = 'weight pruning')\n",
        "plt.xlabel('k-percentile')\n",
        "plt.ylabel('Loss')\n",
        "plt.title('Loss vs percentile plot ')\n",
        "plt.show()"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nM/8fS3FvZOue2RUAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}