AISummit - Segunda Aula InferSentTest.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "!pip3 install --upgrade torch torchvision googledrivedownloader\n",
        "\n",
        "from IPython.display import clear_output\n",
        "from google.colab import files\n",
        "from random import randint\n",
        "import nltk\n",
        "import seaborn as sns\n",
        "import numpy as np\n",
        "import torch\n",
        "from nltk.tokenize import RegexpTokenizer\n",
        "from string import punctuation\n",
        "\n",
        "tokenizer = RegexpTokenizer(r'\\w+')\n",
        "%load_ext autoreload\n",
        "%autoreload 2\n",
        "%matplotlib inline\n",
        "\n",
        "nltk.download('punkt')\n",
        "nltk.download('stopwords')\n",
        "pt_stop = nltk.corpus.stopwords.words('english')\n",
        "pt_stop.append(\"<s>\")\n",
        "clear_output()"
      ],
      "outputs": [],
      "execution_count": 0,
      "metadata": {
        "id": "uReUVN59LTLO",
        "colab_type": "code",
        "colab": {}
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Baixar os pesos do modelo pre treinado do InferSent ( se fossemos treinar demoraria mais de 1 dia )\n",
        "!rm -rf infersent1.pkl\n",
        "!rm -rf glove.840B.300d.zip\n",
        "!curl -Lo infersent1.pkl https://s3.amazonaws.com/senteval/infersent/infersent1.pkl\n",
        "\n",
        "# Baixar os embeddingss do GloVe e extrair\n",
        "from google_drive_downloader import GoogleDriveDownloader as gdd\n",
        "\n",
        "gdd.download_file_from_google_drive(file_id='1aVh7pYsiVovgTzeLDAdeLU2qpqtLXPTU', dest_path='./glove.840B.300d.txt', unzip=False)  \n",
        "\n!unzip glove.840B.300d.zip"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\n",
            "                                 Dload  Upload   Total   Spent    Left  Speed\n",
            "100  146M  100  146M    0     0  37.4M      0  0:00:03  0:00:03 --:--:-- 37.4M\n",
            "Downloading 1IMmKCFOXiKxvNUn6rF_H3CvRZ0JKmsiP into ./glove.840B.300d.zip... Done.\n",
            "Archive:  glove.840B.300d.zip\n",
            "  inflating: glove.840B.300d.txt     \n"
          ]
        }
      ],
      "execution_count": 11,
      "metadata": {
        "id": "WRsXYMDBYnxg",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        },
        "outputId": "f912d2a1-fa63-470e-b794-5defc8b4dfbd",
        "executionInfo": {
          "status": "ok",
          "timestamp": 1539628982853,
          "user_tz": 180,
          "elapsed": 279726,
          "user": {
            "displayName": "Marco Oliveira",
            "photoUrl": "",
            "userId": "08693184838751300851"
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Classe modificada pra facilitar a inferencia no modelo\n",
        "!rm -rf InferSentClass.py\n",
        "!wget https://raw.githubusercontent.com/marcoaleixo/aisummit/master/InferSentClass.py\n",
        "from InferSentClass import InferSent\n",
        "\n\n",
        "MODEL_PATH =  'infersent1.pkl'\n",
        "\n",
        "# Hiperparametros citados no paper\n",
        "params_model = {'bsize': 64, 'word_emb_dim': 300, 'enc_lstm_dim': 2048,'pool_type': 'max', 'dpout_model': 0.0, 'version': 1}\n",
        "\n",
        "model = InferSent(params_model)\n",
        "model.load_state_dict(torch.load(MODEL_PATH))"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2018-10-15 18:43:06--  https://raw.githubusercontent.com/marcoaleixo/aisummit/master/InferSentClass.py\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.0.133, 151.101.64.133, 151.101.128.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.0.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 33982 (33K) [text/plain]\n",
            "Saving to: ‘InferSentClass.py’\n",
            "\n\rInferSentClass.py     0%[                    ]       0  --.-KB/s               \r",
            "InferSentClass.py   100%[===================>]  33.19K  --.-KB/s    in 0.009s  \n",
            "\n",
            "2018-10-15 18:43:07 (3.54 MB/s) - ‘InferSentClass.py’ saved [33982/33982]\n",
            "\n"
          ]
        }
      ],
      "execution_count": 12,
      "metadata": {
        "id": "mcwSXJ1DZATp",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "ae336c10-eec8-4509-ed71-3e481eda7d70",
        "executionInfo": {
          "status": "ok",
          "timestamp": 1539628995605,
          "user_tz": 180,
          "elapsed": 12097,
          "user": {
            "displayName": "Marco Oliveira",
            "photoUrl": "",
            "userId": "08693184838751300851"
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.set_w2v_path('glove.840B.300d.txt')"
      ],
      "outputs": [],
      "execution_count": 0,
      "metadata": {
        "id": "K9ZW0T8faJmj",
        "colab_type": "code",
        "colab": {}
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Carregando os embeddings no modelo\n",
        "model.build_vocab_k_words(K=100000)"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Vocab size : 100000\n"
          ]
        }
      ],
      "execution_count": 14,
      "metadata": {
        "id": "TojuhJ2faOsB",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "a10f979f-8f95-40c0-881a-d2411a8b2615",
        "executionInfo": {
          "status": "ok",
          "timestamp": 1539629062695,
          "user_tz": 180,
          "elapsed": 65320,
          "user": {
            "displayName": "Marco Oliveira",
            "photoUrl": "",
            "userId": "08693184838751300851"
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "input = 'machine learning is subfield of deep learning.'\n",
        "\n",
        "# Grafico da importancia de cada palavra.\n",
        "model.visualize(input)"
      ],
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": [
              "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEfCAYAAACtRRYAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\n",
              "AAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBo\n",
              "dHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzt3Xu8pWP9//HXmO3QjKGNLSH8nN5i\n",
              "+iojUjkVEomcKuczSULRwTn6kb6SLypCkm86qSSKjDPpoH7K6eNUyXkwGCaHYX5/XNcey549e8+e\n",
              "2ffBvt7Px8PD2muvua/3Wvtan3Wt677u+x41ffp0zMysDPM0HcDMzOrjom9mVhAXfTOzgrjom5kV\n",
              "xEXfzKwgLvpmZgVx0R+BJN0gaf9+7t8n/24tSZdX0O4Gku7Nt0+QtN9cbGvvjtsTJa0xHBkHaXNL\n",
              "SY9I+nbVbfVp90pJuw3h8edL2qLCSLNqV5LWq7tdG15dTQewSpwH7AF8q8/9OwPnRcQfgQ9VGSAi\n",
              "vjSn/1bSaODrwHfztj44XLkG8VHg7Ig4sqb25khE7NJQ0x8j1YzrGmrfhoGL/sj0E+BUSctHxP0A\n",
              "kpYD3gVsLmkDUnFbUdJ4UnFdCJgPODUiTpd0HnBvRByf//2MnyWtA5wOjAVeBQ6MiCs7A/R5/AHA\n",
              "p4FRwLPA7hFx+wDb+R2wsKS7gA8DVwM7RcQNkrYDjib13YeBvSPiPknHAIsBSwGrA08AW0bEI31y\n",
              "zQMcB2yT77o5Z9sL2BZ4SdISEdH5TeN64NiIuFLS2vnfrBwR90jaGtgN2Kq/7UbE85KuAW4Etgb2\n",
              "BB4FLsx5b6bjfSjpeGC7/Fo9mJ/3w32ewzX573eBpOnAPsCBwJuBXYG9gfcCdwBbAEsDf8v5dgUW\n",
              "AT4VERfP6vXoJ/ePgEPy69MdEZ+TdCSwU85/Z8769EB/C0nLkwYlSwKTgX0j4i+Slga+DSjn+GxE\n",
              "/AYbdp7eGYEi4lngF6Q3ZK8dgV/m33U6GvhORKwGrANsJGn+QZo4C/h6RKwCnAh8Z1YPlDSOVFTW\n",
              "yo//OrD5INvZA3glIlaJiH90bGsZ0gfUVvnfXAqc2dHcdsBBwArA43k7fW1P+iCZAKxGKpQHR8Sp\n",
              "pNfs1M6Cn11Nem0A1iMVxvfln9cFJs5qux3bmACsFhE35ec6MSJWAE7t3Zak1fJ2xkfEyjnPRv08\n",
              "h74Wi4h3AD8GLiL9TVcG3gGsnx8zDpgeEeOBXYCzJXUNIfexHa/P5yRNAA4A3g2sBMyff+41q7/F\n",
              "WcCFEbEi8FXgB/n+7wP/Lz/vzYALJC06G8/dhshFf+Q6j9cX/Z3yfX09DmyT58yfjIitIuLFQbb9\n",
              "TtK3CYDrgeUHeOwLwHRgT0lviYifRsRJc7AdgI2BqyPi3vzz2cCGuXgBXBcR/4qI6cBfgWX62cbm\n",
              "wPcj4vmIeAX4HrDJIO12Fv11SSPS3qL/flLRH2y7l0XEq/n2eqQCTZ5quyvf/zTQA+yYR9OnRcT5\n",
              "g2QD+GX+/9+B+yLi7vw3vIc0ou51Tm7zSmBeUrEeSu4ZIuIW4G0R8Wz+/U28/u83099C0gLAhqRv\n",
              "OQAXA2tLGpvvPyVv+15Sf9gcG3Yu+iPXVcACktaWtCZpCuWqfh73BeA2UvH9d387gPuxI/BHSUGa\n",
              "ihk1qwdGxMvAB0lF8m5J10t6x1C3k/WQpgR6t/1M/jeL5bue6XjsK8DowbaRby8+SLu/B96Z9zWs\n",
              "RHqt3i1pQeCtEXHbbGz3qY7bi/TJOjk/n4dIUynbAQ9IulTS2wbJBjAl//8V4LmO+ztfg+kR0Znv\n",
              "aaB7iLlnkDQGOE1S5L/f/ry+nvT3t1gkP+YZgIiYHhHPAQuT/o43SborT+utSfrWYcPMRX+EyqOv\n",
              "84FP5v/On8WI7bmI+HL+uv0x4DhJKzNz0ewGkLQUaYplr4gQaWpgsCx/jYjtSAXmcuA7c7Id4DFg\n",
              "xld+Sd2kfQFPzMa/7Xcb+fZjg+R/AQhSQb4z//wf0jRE707NoWx3MqnQ9erpaOvqiNgcWAJ4gDQV\n",
              "NBxG9Zku6SYV9CG/HtlBpA/ACfnvd9Zs/JsnSd/6FgWQNErSisAkUn9bM0/prRIRS0fE/8zGNm2I\n",
              "XPRHtvNIK1K2pP+pHSRdkueSIY34nyG9MR8h7YQj73x7f35MD/A8cFeeVtknP2bBWWz/HZJ+Kmm+\n",
              "iHgJ+HPe/kDbeRmYJ+8P6PQ7YL2cB2A/4IqImDZbr0bya2AnSWNyu3uS9g0M5mrgc6Qdm5Dm9Q8m\n",
              "Te0Mdbu/J33AIum9wIr59iaSzpA0T0Q8D9xKeq2Gyw697ZA+tO4eYu6XeW30vThwV0Q8J2lZ0gdg\n",
              "v32gV55yuoK04xvSCrLL8rfBS0l/T3KWc2fzW44NkYv+CJbnRh8GHu2YB+/rNOCHku4E/gJ8KyLu\n",
              "IY3Cl5N0D3AC8LP8+FuBy0gF4/fAJaQCeO0stn8b8A/gdkm3A8cAnx1kO48AN5CmON7b8XweJK2y\n",
              "uThPAawH7DuEl4T8PC4DbsnZ/g3MzojyamBt0tw1+f/v4bUps6Fs9zBgC0n3kXZ+/i7ffx0whjQN\n",
              "djvwceCo2X1ig3gFmC9v9/ukb1ivDjH3JcB+kn5G2um+fp7aOZm0sueDkg4aJMdepOd+P3A8+YMI\n",
              "+FTe3l2kfnh/RPx7zp6qDWSUz6dvNrLl5br3RoSXaJtH+mZmJXHRNzMriKd3zMwK4pG+mVlBXPTN\n",
              "zArS6r35kyZNaWTuqbt7DJMnT22i6VbmgPZkaUsOaE+WtuQAZ2lLjp6ecbM8ut0j/X50dfV39H79\n",
              "2pID2pOlLTmgPVnakgOcpT9tydHLRd/MrCAu+mZmBXHRNzMriIu+mVlBXPTNzAriom9mVhAXfTOz\n",
              "grjom5kVpNVH5JqVZo8T+7uMcTXO/eIHamvL2sMjfTOzgrjom5kVpNLpHUnjgYuBUyLidEnzkq7P\n",
              "uSIwBdg2IiZXmcHMzF5T2Uhf0ljSRbcndty9NzApItYCfgysW1X7ZmY2sypH+i8CmwFf6LhvC+Bo\n",
              "gIg4q8K2zcysH5UV/YiYBkyT1Hn3csCHJZ0EPArsHxFPzWob3d1jGjstaU/PuEba7astOaA9WdqS\n",
              "A9qVZaiqyt6m16QtWdqSA+pfsjkKiIg4VtIRwJeAQ2f14KYugNDTM45Jk6Y00nYbc0B7srQlB7Qr\n",
              "y5yoInubXpO2ZGkix0AfMnWv3nkMuDbfvhxYreb2zcyKVnfR/w2wab49AYia2zczK1pl0zuSJgAn\n",
              "k+bxX5a0LbADcKqkPYHngF2rat/MzGZW5Y7cW4AN+vnVdlW12VZ1HloPPrzezGbNR+SamRXEJ1wz\n",
              "s5n42+nI5ZG+mVlBXPTNzAriom9mVhAXfTOzgrjom5kVxEXfzKwgLvpmZgVx0TczK4iLvplZQVz0\n",
              "zcwK4qJvZlYQF30zs4K46JuZFcRF38ysIJUWfUnjJd0n6YA+939I0vQq2zYzs5lVVvQljQVOAyb2\n",
              "uX8B4EvAI1W1bWZm/atypP8isBnwcJ/7vwycAbxUYdtmZtaPKq+ROw2YJmnGfZJWBlaPiKMkfX2w\n",
              "bXR3j6Gra3RVEQfU0zOukXaHQ1XZ2/KatCUHtCvLULUpu/tsfeq+XOIpwIGz++DJk6dWGGXWenrG\n",
              "MWnSlEbaHg5VZG/La9KWHNCuLHOiTdndZ4e/zVmpbfWOpKWAVYD/lXQz8FZJ19bVvpmZ1TjSj4iH\n",
              "gBV6f5b0z4hYv672zcyswqIvaQJwMrAc8LKkbYGtI+Kpqto0M7OBVbkj9xZggwF+v1xVbZuZWf98\n",
              "RK6ZWUFc9M3MCuKib2ZWEBd9M7OCuOibmRXERd/MrCAu+mZmBXHRNzMriIu+mVlBXPTNzAriom9m\n",
              "VhAXfTOzgrjom5kVxEXfzKwgLvpmZgVx0TczK0ill0uUNB64GDglIk6X9Dbge8C8wMvAThHxaJUZ\n",
              "zMzsNZWN9CWNBU4DJnbcfTxwVr427i+AQ6pq38zMZlbl9M6LwGbAwx337Q9clG9PAhatsH0zM+uj\n",
              "ymvkTgOmSeq873kASaOBTwNfGWgb3d1j6OoaXVXEAfX0jGuk3eFQVfa2vCZtyQHtyjJUbcruPluf\n",
              "Suf0+5ML/g+AqyJi4kCPnTx5aj2h+ujpGcekSVMaaXs4VJG9La9JW3JAu7LMiTZld58d/jZnpYnV\n",
              "O98D7omIYxto28ysaLUWfUk7Ai9FxNF1tmtmZkll0zuSJgAnA8sBL0vaFlgceEHSNflhd0TE/lVl\n",
              "MDOz16tyR+4twAZVbd/MzIbOR+SamRXERd/MrCAu+mZmBXHRNzMriIu+mVlBXPTNzAriom9mVhAX\n",
              "fTOzgrjom5kVxEXfzKwgLvpmZgVx0TczK4iLvplZQVz0zcwK4qJvZlYQF30zs4JUemF0SeOBi4FT\n",
              "IuJ0SW8jXRR9NPAIsHNEvFhlBjMze01lI31JY4HTgIkdd38FOCMi1gXuBfaoqn0zM5tZldM7LwKb\n",
              "AQ933LcB8Kt8+xJgowrbNzOzPqq8Ru40YJqkzrvHdkznPA68daBtdHePoatrdEUJB9bTM66RdodD\n",
              "Vdnb8pq0JQe0K8tQtSm7+2x9Kp3TH8SowR4wefLUOnLMpKdnHJMmTWmk7eFQRfa2vCZtyQHtyjIn\n",
              "2pTdfXb425yVulfvPCfpTfn2Urx+6sfMzCpWd9G/Etgm394G+G3N7ZuZFa2y6R1JE4CTgeWAlyVt\n",
              "C+wInCdpX+BfwPerat/MzGZW5Y7cW0irdfrauKo2zebEHideVWt7537xA7W2Z9bJR+SamRWkydU7\n",
              "1gCPau2Nps4+W0J/9UjfzKwgLvpmZgVx0TczK4iLvplZQVz0zcwK4qJvZlaQIS3ZlLQGsALwGHB9\n",
              "REyvJJWZmVVitkf6ko4FtgMWJp0H/+dVhTIzs2oMONKX9GXgaxHxCrAMsEfv6F7S72vIZ2Zmw2iw\n",
              "6Z0HgSslHQn8L3BFvijK/MD3Ks5mZmbDbMCiHxHnS7oUOBGYDmwfEZNrSWZmZsNu0Dn9iHgyIvYG\n",
              "zgcukrRj9bHMzKwKg83pTwAOABYH7gf2A7aRdAlwcETcW31EMzMbLoPN6Z8OfAJ4CHg7cGpEfFjS\n",
              "8sB/A1tXnM/MzIbRYEX/VWBZYDRp9c5LABFxP3NQ8CUtSJom6ibtDD42Ii4f6nbMzGzODDanvzPp\n",
              "SleHAeOBPeayvd2AiIgNgW2BU+dye2ZmNgSDrd65HzhyGNt7AvivfLs7/2wF8sVc7I1mpPTZWs+9\n",
              "ExE/ApaRdC9wHfD5Ots3MytdrZdLlLQT8EBEbCppdeAcYM1ZPb67ewxdXaNry9epp2dcI+0OhzZl\n",
              "b0uWtuSA9mRpSw5oT5a25IDqstR9jdz3AZcDRMStkpaUNDqf5mEmkydPrTVcr56ecUyaNKWRtodD\n",
              "m7K3JUtbckB7srQlB7QnS1tywNxlGegDo+5TK98LrA0gaVnguVkVfDMzG351j/TPBM6VdG1ue7+a\n",
              "2zczK1qtRT8ingO2r6u9Ove2e3WImb0R+MpZZmYFcdE3MyuIi76ZWUFc9M3MCuKib2ZWEBd9M7OC\n",
              "uOibmRXERd/MrCAu+mZmBXHRNzMriIu+mVlBXPTNzAriom9mVhAXfTOzgrjom5kVxEXfzKwgdV85\n",
              "C0k7AocB04CjIuLSujOYmZWq1pG+pEWBo4H3Ax8BtqyzfTOz0tU90t8IuDIipgBTgH1qbt/MrGh1\n",
              "F/3lgDGSfgV0A8dExMSaM5iZFavuoj8KWBT4GLAscLWkZSNien8P7u4eQ1fX6DrzzbGennFNR5jB\n",
              "WWbWlhzQnixtyQHtydKWHFBdlrqL/mPATRExDbhP0hSgB3i8vwdPnjy1zmxzZdKkKU1HmMFZZtaW\n",
              "HNCeLG3JAe3J0pYcMHdZBvrAqHvJ5hXAByTNk3fqLgg8UXMGM7Ni1Vr0I+Ih4GfAzcBvgM9ExKt1\n",
              "ZjAzK1nt6/Qj4kzgzLrbNTMzH5FrZlYUF30zs4K46JuZFcRF38ysIC76ZmYFcdE3MyuIi76ZWUFc\n",
              "9M3MCuKib2ZWEBd9M7OCuOibmRXERd/MrCAu+mZmBXHRNzMriIu+mVlBXPTNzArSSNGX9CZJ90na\n",
              "rYn2zcxK1dRI/wjgqYbaNjMrVu1FX9IqwKrApXW3bWZWuiZG+icDhzTQrplZ8Wq9MLqkXYDfR8Q/\n",
              "JA36+O7uMXR1ja4+2DDo6RnXdIQZnGVmbckB7cnSlhzQnixtyQHVZam16AObA8tL+giwNPCipAcj\n",
              "4sr+Hjx58tRaw82NSZOmNB1hBmeZWVtyQHuytCUHtCdLW3LA3GUZ6AOj1qIfER/vvS3pGOCfsyr4\n",
              "ZmY2/LxO38ysIHVP78wQEcc01baZWak80jczK4iLvplZQVz0zcwK4qJvZlYQF30zs4K46JuZFcRF\n",
              "38ysIC76ZmYFcdE3MyuIi76ZWUFc9M3MCuKib2ZWEBd9M7OCuOibmRXERd/MrCAu+mZmBXHRNzMr\n",
              "SO1XzpJ0ErBubvuEiPh53RnMzEpV60hf0obA+IhYB9gU+Gad7ZuZla7u6Z3rgO3y7aeBsZJG15zB\n",
              "zKxYtU7vRMQrwPP5xz2By/J9/eruHkNX1xvjM6GnZ1zTEWZwlpm1JQe0J0tbckB7srQlB1SXpfY5\n",
              "fQBJW5KK/iYDPW7y5Kn1BBoGkyZNaTrCDM4ys7bkgPZkaUsOaE+WtuSAucsy0AdGEztyPwQcDmwa\n",
              "Ec/U3b6ZWclqLfqSFga+DmwUEU/V2baZmdU/0v84sBjwE0m99+0SEQ/UnMPMrEh178g9CzirzjbN\n",
              "zOw1PiLXzKwgLvpmZgVx0TczK4iLvplZQVz0zcwK4qJvZlYQF30zs4K46JuZFcRF38ysIC76ZmYF\n",
              "cdE3MyuIi76ZWUFc9M3MCuKib2ZWEBd9M7OCuOibmRWkiWvkngK8B5gOfDYi/lR3BjOzUtU60pe0\n",
              "PrBSRKwD7An8T53tm5mVru7pnQ8CvwSIiDuBbkkL1ZzBzKxYo6ZPn15bY5LOAi6NiIvzz9cDe0bE\n",
              "3bWFMDMrWNM7ckc13L6ZWVHqLvoPA0t0/Lwk8EjNGczMilV30b8C2BZA0hrAwxExpeYMZmbFqnVO\n",
              "H0DSicB6wKvApyPi1loDmJkVrPaib2ZmzWl6R66ZmdXIRd/MrCAu+mZDIOmdkpZsOofZnHLRt2El\n",
              "aViPvZDUmj4qaRngQmAxSQs0neeNbrj7is2e1ryh2qRNhaY/nW8WSaObzNJL0nIAETF9ON/MEfFq\n",
              "3v7GktaTNGa4tj0HXgZ+B+wFHNJgjlmS9K78bWT5/HPrCmtVfaXt+r5Xm6ozrS5uTYmIVyUtKumj\n",
              "klZpW8eMiOkAknYGDpO0TxM5el+XXGDOl3Rob765fc0krSxppXx7H+BYYBFgwY7H1NJ/e9uJiEeA\n",
              "PwE7A89IGltH+7NL0rrAj0kfShdK2rAthbXKvvJGERGvSFooD2AW6R3Q1M1LNvuQtCPpqOHNSAXm\n",
              "tojYs9lUM5O0E7ALcBhwM7BDRPy8gRybAjsALwCrAJdFxIn5d6N6P6CGuM15ga1Jo+qVgMOBTwIL\n",
              "Ae8E1gG+FRGPDsuTmP1cuwDzAfsD1wC3AL+LiMfrzNEfSeOBTwE/jIgbJW0PHA3sExE3NpsuqaKv\n",
              "vBHkQcPHgLeS+nU36cDUzZvI45F+Jmnh/Eb5BGk0txNwMOmN3biOkdI8kt4EvBs4AlgG+DXwC0mr\n",
              "15hndJ5qORS4OCL2IX0AvV/SITDno7iIeBn4KbAw6RTcK+SfTwI+QLoOxFF1Tm1J2oE0gr4TeIj0\n",
              "5t0IWE/SEgP925psDawBLC9pgYj4CfBNYMumR9JV9pXZaPstknqGe7uz0W7v+3VBUh/u7Tu7AweR\n",
              "vpE1wkX/NaOBS4BtI+I60kVejgHuajIUzDQKelNE/Ae4ETgT2Dcits2/P1jSW6vOkm/OExFTgZuA\n",
              "+SR1AX8ELgV2kHQQvDYVNVT5q+9U4J/AH4C7I2LniDgU+BVptFh5MZM0Kj+3TYBv5FHz1qTnujGw\n",
              "BbBu3fOzHUVlxTwNdjzwc2BC/g/gCeD/0ND7vK6+MohlgFr3A0map+O5jAbOAbaLiInAS8BxwD/q\n",
              "zNTJRR+QtDfwI2CtjrvnBW6OiKuaSZV0diBJnwYukXQgMD9wPXBt/pbyYdLXx6kV5RgFM0ZkG5Hn\n",
              "jIF/A9sB6+RCfSfpHEvrSXrP3LQZEY8BZwF/BaZL2l3SUcAJwDkRMW1utj+bGabndv4IrCJpyfxN\n",
              "ZEfSh8+vgRvrnp/Nf4fNgPNJ01+/An4CPE/68P9aznhORLxSZ7Ym+sqsRMSfIuJfVWx7gDZ7Fx/s\n",
              "CZwLbBgRz+VfL0ya1rq+zkydii76eWft9qQdc3sBtwGLSnozsDnw3ED/voZ8ozo60ATSlM63gTcD\n",
              "y5NGwdNIheezwOcj4pkKcnR+8IwnvVZ/B75Imur4E3CgpFNzvh8A95I+mOZKRDxBWib5OLArsDKw\n",
              "V0TcPrfbHqLLSKPmjSQtSxr5rwhcEhEP15wFSUsBewMfAi4AloiIf0XE4aR9PCsBP4qI39acq7G+\n",
              "0jRJK0g6IN/emrTP7RDgzo5v4LvS8Cnli96RK2krUtFckbQcbzHgI6RR020R8Yv8uNp3Mkl6O3BQ\n",
              "ROwraU3gDOC8iPi20hlKNyG9UX4F3Ad0RcRTFeRYlDQn+U3SDu7rgS9ExI8kfYS07+MHwCukr65v\n",
              "AZ4m7W/YJSLuG6YcPaQztF7U1I5TSSsDu5H6y5tIr8MdDWUZA3yNdOLCFUk7cRcC3gecDRxJ+luc\n",
              "Dfyljv7blr7SlLxPbVPStOtupCnIx4HVgI8D3yddRKqRPtOr2JG+pC2Bz5Dm7pck7SD8NamwvAJM\n",
              "6n1sE6sK8uUkvyRpPeAvwMXA5pKWjoi/AL8hFZ4tgOerKPjZvKSRdg/wKHARcJKk+SLi16SOvD/p\n",
              "TX4L8ED+eZ/hfBNHxCTgrCZXyuQrvB0LHEC64lttb948D46k1fOH/uLAraQpyW9FxAOkb4Cr5umc\n",
              "E0gj6wdr7L+t6CsNehFYGvgwqZZsQaonV5E+DFYmvTaNHj9R3Eg/f/18VdLuwGMRcVmf328P7EPq\n",
              "iPc3lHF07zyspAtIy9vWIo3eVgUOjYgHJK1Geg5PVJxnHuCrpA69K3AiqROvFREv5LnlR/OHEZJ6\n",
              "dzbbXJLUkz/wkPRR0kDlb8ACpB3cywEijSg3AA7v26frVHJfkbQ56UJRJwNHRMRN+f4FSMuMDyPV\n",
              "lX83l7LAog9peSZpCeBngP+Q1s1uDzxJ+pQ+MI+06861WG8Bz2+OZyPiBkn/F1gfWBf4MvBeUud5\n",
              "sIZM44FnSF/FNyRNJewPfIW0dn7ViHih6hwlygX0UuCRiNhD0i+APUhTBR+NiM3y/oUlgPcAf4iI\n",
              "mxvMW2xfyTMHh0TE+pLWIa3s+i5pefHhpFVEhzawL2omXU0HaIiAsaQdo/uRvoY+TZqLO6eKnaGD\n",
              "BkqHpu8v6QrSm+ZI0sqcg0lvmOOAq0lvps+TpqWqzrQu8MPc7t3ADaQpwdNJH5hdpNewsZUII1Xv\n",
              "TvxcTK6XdAbp0qLbkRYZHKB0LqA18r6nPzQYt9i+0jtzQDpa/KR8932kfW5LRcSfJR0DvBgRkxuK\n",
              "+TqljvR7SMX+DuCBiPhTw5GQ1E0aFS1E2qF8REQ8IukrpB1BOwD/DawcER+qIc9qwFbAb0n7OLYi\n",
              "7TS8lnTls+V6j1RuYkd3KST9F2np5QdI6+8fBjaNiNskbUM6mHB3YGrdy0Y7MhbdVzpmDg4g7bzt\n",
              "An4GvA0Yn5cet0aRRb9NOt8Eee5vX9Io4ULSyKmLNKXzPtII781V7czszaJ0fpRvktYUfyYi/ibp\n",
              "/cAHSachmAj8s6l9HqWQ9C7ScsxdSHP4J5Hmhm8k9Y/9gIPzQT91Z3NfySStBZxCWoL6KdIHwPOk\n",
              "FUoT65iGHQoX/ZbI63q7gHtIo7oe4JqI+K2k+YHPAedX3YEkrU/aIXgB8CXSwV5HRcTT+Xcbk5aO\n",
              "3ltlDgNJ7yBN8+0VEc8qne7hAtLO/J1JUwY3NJjPfYV2zhwMxEW/BZTOIrkt6cjT84H7SVM9bwb+\n",
              "HBEX15BhHtIS0N5THmytdBbJc0jL746LiCcldbdlbnKkk7QQaV/OzcDVEfGopM+TCsymTRVT95WB\n",
              "tX0Kq9h1+i2zOnBMRHwhIm7PS9j+QDpgbE1VeArfjvXCYyLiedK+g40k7ZN/3p10JOpxkrpKfBM3\n",
              "JSKeJU0ZbAB8QunMqqsAmzdR8N1XZk+bCz54pF+7PnP480fEi5JOBB6PiG/k+zcE3gWcCixU9ZtH\n",
              "0iakw8X/TDoHEaTTGh8ZEWfnoz8VEX+tMof1L8+bb0E65cJpEfGbBrO4r7zBueg3ROlkTG8nzeFf\n",
              "RDpq74yIOFPSx0lrsXeoel1zPuL3q6QDR3YlrSc+EniKdM6UAyPi3Coz2OzpGCQ0Mn3gvjIylLpO\n",
              "v1FKF+PYgXSStL+RpnE2Bs6UtCrpQJtdajqQZSzpTJY9pA+hnwFfAM4jLUFr1cqDwr0EjU4fuK+M\n",
              "AB7p16DPlE7vzrnvkE6t8DFgbdJRi38nnb9/ekQ8WVGW3tNQLEpabTEf6Zwgh5OuH3AHaTXGGsA2\n",
              "EXFr23dMWTXcV0YmF/0aSfoM6ci9hUhH7W0QEdvnr80Xkk47u19UcBnA3qmBfHsL0gE/00lv2j+Q\n",
              "vrafl+/bADiz1B1xpXNfGdm8eqdCnWfSy9M2WwKPkd4sawNvyQdkLUga/e9dUcHvJl1ecA2l83p/\n",
              "kXQK3JeBnfL5fu4hXcbte8AdfhOXyX1l5PNIvyJ9pnT2ApYlHSp/Qj4h0ydJJ8/6I2nN8yeioiv8\n",
              "5K/nB5DmZB8inebhWtIqjH1JV9y6m3Rq3DER8S9/TS+T+8rI56JfsXw63M8B15BOlvatSBeVWIN0\n",
              "zvPnSGffq/QwdUmLkY7iXBX4KOnSdVtFxIOSjiSdyfHsKjPYG4P7ysjm6Z0K5cPodwFOj4ijgW8A\n",
              "m0n6ZD6f+BGkC5tXfl6S/LX8AtJX8ymkE2LtJGkD0pkbb6s6g70xuK+MbB7pV0jS4sCBpKMoj42I\n",
              "v+dT5e4O/CAiLmogUzdpqeg7SRe6uJx0Ye/GLrxh7eS+MjK56FdM0iKkHWFLAd+NiNuVrrDz12jg\n",
              "gto506Kka3guBXytbad+tfZbNKzFAAAAaUlEQVRwXxl5XPRrkOdIdyUd0HJyNHBVrr7yt5CxEfGP\n",
              "prNYu7mvjCwu+jXJb5xPAD/2aMnMmuKiXyN1XPDczKwJLvpmZgXxkk0zs4K46JuZFcRF38ysIC76\n",
              "ZmYFcdE3MyuIi76ZWUH+Py0dO5vYhy5dAAAAAElFTkSuQmCC\n"
            ],
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7ff492186e80>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "execute_result",
          "execution_count": 15,
          "data": {
            "text/plain": [
              "(tensor([[ 0.0357,  0.1055, -0.0027,  ..., -0.0149, -0.0381,  0.0526]],\n",
              "        grad_fn=<MaxBackward0>), array([[3, 4, 7, ..., 6, 8, 2]]))"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ],
      "execution_count": 15,
      "metadata": {
        "id": "BUWfsfsOpjcZ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 338
        },
        "outputId": "7440eb1e-d356-4263-a422-1e50e17331a5",
        "executionInfo": {
          "status": "ok",
          "timestamp": 1539629064803,
          "user_tz": 180,
          "elapsed": 1384,
          "user": {
            "displayName": "Marco Oliveira",
            "photoUrl": "",
            "userId": "08693184838751300851"
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Gera o vetor que representa a frase\n",
        "model.encode(input, tokenize=True)"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 22,
          "data": {
            "text/plain": [
              "array([[ 0.02408349, -0.03946692, -0.03044511, ..., -0.03055104,\n",
              "        -0.03814263, -0.02892261],\n",
              "       [ 0.0024533 , -0.01888101, -0.00417832, ..., -0.02155335,\n",
              "        -0.03814263, -0.02892261],\n",
              "       [ 0.02491982, -0.01791508, -0.03044511, ..., -0.03926747,\n",
              "        -0.03814263,  0.03790347],\n",
              "       ...,\n",
              "       [ 0.05712789, -0.08939895, -0.03044511, ..., -0.02854737,\n",
              "        -0.03814263, -0.02892261],\n",
              "       [ 0.00894748,  0.02972344, -0.03044511, ...,  0.00111258,\n",
              "        -0.03814263,  0.00537444],\n",
              "       [ 0.01152656, -0.08591913,  0.00797681, ..., -0.03926747,\n",
              "        -0.03814263, -0.02892261]], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ],
      "execution_count": 22,
      "metadata": {
        "id": "g-RRu3bK-1rL",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 238
        },
        "outputId": "6044b576-3b5d-40f6-e654-95641dc38342",
        "executionInfo": {
          "status": "ok",
          "timestamp": 1539629160429,
          "user_tz": 180,
          "elapsed": 815,
          "user": {
            "displayName": "Marco Oliveira",
            "photoUrl": "",
            "userId": "08693184838751300851"
          }
        }
      }
    }
  ],
  "metadata": {
    "colab": {
      "name": "AISummit - Segunda Aula InferSentTest.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "language": "python",
      "display_name": "Python 3"
    },
    "kernel_info": {
      "name": "python3"
    },
    "language_info": {
      "name": "python",
      "version": "3.6.6",
      "mimetype": "text/x-python",
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "pygments_lexer": "ipython3",
      "nbconvert_exporter": "python",
      "file_extension": ".py"
    },
    "nteract": {
      "version": "0.11.7"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}