looking_at_vae_z.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "from argparse import ArgumentParser\n",
    "import sys\n",
    "import json\n",
    "import random\n",
    "from collections import defaultdict\n",
    "\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import pandas as pd\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix\n",
    "from sklearn.manifold import TSNE\n",
    "\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from hcn.modules.entities import EntityTracker\n",
    "from hcn.modules.variational_hierarchical_lstm_v2 import VariationalHierarchicalLSTMv2\n",
    "from hcn.modules.actions import ActionTracker\n",
    "from hcn.modules.util import load_model, read_dialogs\n",
    "from hcn.modules.evaluation import mark_post_ood_turns\n",
    "from utils.training_utils import batch_generator\n",
    "from utils.preprocessing import (make_dataset_for_de_vhcn_v3,\n",
    "                                 make_variational_autoencoder_dataset,\n",
    "                                 generate_dropout_turns_for_de_vhcn_v3)\n",
    "from babi_tools.ood_augmentation import DEFAULT_CONFIG_FILE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "random.seed(273)\n",
    "np.random.seed(273)\n",
    "tf.set_random_seed(273)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "HCN_FOLDER = 'hcn'\n",
    "MODEL_FOLDER = os.path.join(HCN_FOLDER, 'models', 'babi_task6_model_08192018_variational_v2_unified_vocab_unk_turns')\n",
    "DATA_FOLDER = 'data'\n",
    "CLEAN_DATASET_FOLDER = os.path.join(HCN_FOLDER, DATA_FOLDER)\n",
    "NOISY_DATASET_FOLDER = os.path.join(DATA_FOLDER, 'babi_task6_ood_dataset_v2_0.2_0.4')\n",
    "\n",
    "with open(DEFAULT_CONFIG_FILE) as babi_config_in:\n",
    "    BABI_CONFIG = json.load(babi_config_in)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      ":: restoring checkpoint from hcn/models/babi_task6_model_08192018_variational_v2_unified_vocab_unk_turns/hcn.ckpt \n",
      "\n",
      "INFO:tensorflow:Restoring parameters from hcn/models/babi_task6_model_08192018_variational_v2_unified_vocab_unk_turns/hcn.ckpt\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "08/21/2018 08:29:20 PM Restoring parameters from hcn/models/babi_task6_model_08192018_variational_v2_unified_vocab_unk_turns/hcn.ckpt\n"
     ]
    }
   ],
   "source": [
    "config = tf.ConfigProto(device_count={'GPU': 0})\n",
    "sess = tf.Session(config=config)\n",
    "rev_vocab, kb, action_templates, config = load_model(MODEL_FOLDER)\n",
    "clean_dialogs, clean_indices = read_dialogs(os.path.join(CLEAN_DATASET_FOLDER, 'dialog-babi-task6-dstc2-tst.txt'),\n",
    "                                            with_indices=True)\n",
    "noisy_dialogs, noisy_indices = read_dialogs(os.path.join(NOISY_DATASET_FOLDER, 'dialog-babi-task6-dstc2-tst.txt'),\n",
    "                                            with_indices=True)\n",
    "\n",
    "max_noisy_dialog_length = max([item['end'] - item['start'] + 1 for item in noisy_indices])\n",
    "config['max_input_length'] = max_noisy_dialog_length\n",
    "post_ood_turns_clean, post_ood_turns_noisy = mark_post_ood_turns(noisy_dialogs, BABI_CONFIG['backoff_utterance'].lower())\n",
    "\n",
    "et = EntityTracker(kb)\n",
    "at = ActionTracker(None, et)\n",
    "at.set_action_templates(action_templates)\n",
    "\n",
    "vocab = {word: idx for idx, word in enumerate(rev_vocab)}\n",
    "dataset_clean = make_dataset_for_de_vhcn_v3(clean_dialogs, clean_indices, vocab, et, at, **config)\n",
    "dataset_noisy = make_dataset_for_de_vhcn_v3(noisy_dialogs, noisy_indices, vocab, et, at, **config)\n",
    "context_features_clean, action_masks_clean = dataset_clean[2:4]\n",
    "net = VariationalHierarchicalLSTMv2(rev_vocab, config, context_features_clean.shape[-1], action_masks_clean.shape[-1])\n",
    "net.restore(MODEL_FOLDER)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "clean_utterances = set([question for question, answer in clean_dialogs])\n",
    "noisy_utterances = set([question for question, answer in noisy_dialogs\n",
    "                        if answer == BABI_CONFIG['backoff_utterance'].lower()])\n",
    "random_utterances_vectorized = generate_dropout_turns_for_de_vhcn_v3(10000,\n",
    "                                                                     config['max_sequence_length'],\n",
    "                                                                     [utterance[0] for utterance in clean_dialogs],\n",
    "                                                                     vocab,\n",
    "                                                                     config['turn_word_dropout_prob'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Utterances\n",
    "=="
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['restaurant with corsica food',\n",
       " 'uh moderate',\n",
       " 'id like to find a restaurant serving crossover food',\n",
       " 'south',\n",
       " 'i am looking for a cheap restaurant that serves germ',\n",
       " 'australian food center part of town',\n",
       " 'breath no unusual food',\n",
       " 'doesnt matter but it should be moderately priced',\n",
       " 'yes in the south part of town',\n",
       " 'gastropub type food']"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(clean_utterances)[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['where is the closest tea house?',\n",
       " 'i need to know what time the sixty one c leaves the transportation center in mckeesport',\n",
       " 'what will the weather in fresno be in the next 48 hours?',\n",
       " 'give me the full weekly weather forecast',\n",
       " \"what's the weather like this week\",\n",
       " 'please show me the weekly weather report',\n",
       " \"hey there, i'm excited to book a trip somewhere for me and my colleagues.\",\n",
       " 'i would like to know if the sixty one c mcceesport go to kennywood park',\n",
       " 'i need a reminder set for my lab appointment at 5pm today. jeff is joining.',\n",
       " \"check calendar for when my doctor's appointment is\"]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(noisy_utterances)[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 368, 1777,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0],\n",
       "       [1976, 1305,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0],\n",
       "       [4151,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0],\n",
       "       [3291,   63,  765, 2308, 4238,   18,   17, 1444, 1434, 1288, 2229,\n",
       "         620,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0],\n",
       "       [  63,   26, 3477, 2405, 1634, 2242,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0,    0,    0,    0,    0,    0,    0,\n",
       "           0,    0,    0,    0,    0]], dtype=int32)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "random_utterances_vectorized[0][:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "def compute_z(in_dataset, in_model):\n",
    "    result = []\n",
    "    batch_gen = batch_generator(in_dataset, 64)\n",
    "\n",
    "    for batch in batch_gen:\n",
    "        enc_inp = batch[0]\n",
    "        z = in_model.sess.run(in_model.vrae.z, feed_dict={in_model.vrae.enc_inp: enc_inp})\n",
    "        result += z.tolist()\n",
    "    return np.array(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_mean_loss(in_dataset, in_model):\n",
    "    result = []\n",
    "    batch_gen = batch_generator(in_dataset, 64)\n",
    "\n",
    "    for batch in batch_gen:\n",
    "        enc_inp, dec_inp, dec_out, bow_targets = batch\n",
    "        dec_seq_length = np.ones(dec_out.shape[0]) * dec_out.shape[1]\n",
    "        losses = in_model.sess.run(in_model.vae_overall_loss, feed_dict={in_model.vrae.enc_inp: enc_inp,\n",
    "                                                                         in_model.vrae.dec_inp: dec_inp,\n",
    "                                                                         in_model.vrae.dec_out: dec_out,\n",
    "                                                                         in_model.vrae.bow_targets: bow_targets,\n",
    "                                                                         in_model.vrae.sequence_length_for_decoder: dec_seq_length})\n",
    "        result += losses.tolist()\n",
    "    return np.mean(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'clean_utterances_vectorized' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-16-690988ef22cd>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mclean_z\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompute_z\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mclean_utterances_vectorized\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnet\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mnoisy_z\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompute_z\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnoisy_utterances_vectorized\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnet\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mrandom_z\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompute_z\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrandom_utterances_vectorized\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnet\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'clean_utterances_vectorized' is not defined"
     ]
    }
   ],
   "source": [
    "clean_z = compute_z(clean_utterances_vectorized, net)\n",
    "noisy_z = compute_z(noisy_utterances_vectorized, net)\n",
    "random_z = compute_z(random_utterances_vectorized, net)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Reconstructions\n",
    "=="
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "IND uterance mean VAE loss: 20.026\n",
      "OOD uterance mean VAE loss: 54.452\n",
      "RND uterance mean VAE loss: 210.633\n"
     ]
    }
   ],
   "source": [
    "print('IND uterance mean VAE loss: {:.3f}'.format(compute_mean_loss(clean_utterances_vectorized, net)))\n",
    "print('OOD uterance mean VAE loss: {:.3f}'.format(compute_mean_loss(noisy_utterances_vectorized, net)))\n",
    "print('RND uterance mean VAE loss: {:.3f}'.format(compute_mean_loss(random_utterances_vectorized, net)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[t-SNE] Computing 121 nearest neighbors...\n",
      "[t-SNE] Indexed 6222 samples in 0.006s...\n",
      "[t-SNE] Computed neighbors for 6222 samples in 2.404s...\n",
      "[t-SNE] Computed conditional probabilities for sample 1000 / 6222\n",
      "[t-SNE] Computed conditional probabilities for sample 2000 / 6222\n",
      "[t-SNE] Computed conditional probabilities for sample 3000 / 6222\n",
      "[t-SNE] Computed conditional probabilities for sample 4000 / 6222\n",
      "[t-SNE] Computed conditional probabilities for sample 5000 / 6222\n",
      "[t-SNE] Computed conditional probabilities for sample 6000 / 6222\n",
      "[t-SNE] Computed conditional probabilities for sample 6222 / 6222\n",
      "[t-SNE] Mean sigma: 1.205183\n",
      "[t-SNE] KL divergence after 50 iterations with early exaggeration: 85.481155\n",
      "[t-SNE] Error after 300 iterations: 3.720408\n"
     ]
    }
   ],
   "source": [
    "tsne = TSNE(n_components=2, verbose=1, perplexity=40, n_iter=300, learning_rate=100.0)\n",
    "tsne_results = tsne.fit_transform(np.concatenate([clean_z, noisy_z, random_z], axis=0))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Scatter plot: IND (blue), OOD (red), RND (grey)\n",
    "=="
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fc4eaaea5c0>"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(tsne_results[:, 0], tsne_results[:, 1], color=['blue'] * clean_z.shape[0] + ['red'] * noisy_z.shape[0] + ['grey'] * random_z.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "clean_utterance_groups = [[utterance for utterance in clean_utterances if 'address' in utterance],\n",
    "                          [utterance for utterance in clean_utterances if 'phone' in utterance],\n",
    "                          [utterance for utterance in clean_utterances if 'part of' in utterance]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['could i have the address and phone number', 'okay whats their address', 'can i have the address and the price range', 'what is the address phone number and post code', 'okay ah wha whats the address', 'what is its address', 'may i have the address and price range please', 'what is prezzos address', 'may i get the address and phone number', 'give me the address phone number and price range']\n",
      "['uh could you repeat the phone number', 'could i have the address and phone number', 'what is the address phone number and post code', 'may i get the address and phone number', 'give me the address phone number and price range', 'may i have the phone number and post code please', 'phone number please', 'may i have he phone number please', 'and can you tell me the phone number', 'address phone number and postcode']\n",
      "['im looking for a moderately priced resturant in the east part of town', 'im looking for a restaurant in the south part of town that serves', 'im looking for a moderate priced restaurant in the east part of town', 'im looking for a restaurant in the south part of town and it should serve chinese food', 'is it in the north part of tow', 'yea id like a cheap restaurant in the north part of town', 'how about a restaurant in the west part of town serving vietnamese food', 'looking for romanian food in the east part of town', 'restaurant in the south part of town caribbean food', 'east part of town']\n"
     ]
    }
   ],
   "source": [
    "print(clean_utterance_groups[0][:10])\n",
    "print(clean_utterance_groups[1][:10])\n",
    "print(clean_utterance_groups[2][:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processed 5 out of 5 batches\r"
     ]
    }
   ],
   "source": [
    "utterance_group_z = [compute_z(make_variational_autoencoder_dataset(utterance_group, vocab, config['max_sequence_length']), net)\n",
    "                     for utterance_group in clean_utterance_groups]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[t-SNE] Computing 121 nearest neighbors...\n",
      "[t-SNE] Indexed 723 samples in 0.001s...\n",
      "[t-SNE] Computed neighbors for 723 samples in 0.039s...\n",
      "[t-SNE] Computed conditional probabilities for sample 723 / 723\n",
      "[t-SNE] Mean sigma: 1.682320\n",
      "[t-SNE] KL divergence after 50 iterations with early exaggeration: 61.956482\n",
      "[t-SNE] Error after 500 iterations: 2.051529\n"
     ]
    }
   ],
   "source": [
    "tsne = TSNE(n_components=2, verbose=1, perplexity=40, n_iter=500, learning_rate=20.0)\n",
    "tsne_results = tsne.fit_transform(np.concatenate(utterance_group_z, axis=0))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Scatter plot of clean utterance groups (address, phone, part of town)\n",
    "=="
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fc4e8688978>"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(tsne_results[:, 0],\n",
    "            tsne_results[:, 1],\n",
    "            color=['blue'] * len(clean_utterance_groups[0]) + ['red'] * len(clean_utterance_groups[1]) + ['green'] * len(clean_utterance_groups[2]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "post_ood_utterances_clean = [clean_dialogs[idx][0] for idx in sorted(post_ood_turns_clean)]\n",
    "post_ood_utterances_noisy = [noisy_dialogs[idx][0] for idx in sorted(post_ood_turns_noisy)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "thank you good bye:\t158\n",
      "yes:\t73\n",
      "phone number:\t59\n",
      "i dont care:\t44\n",
      "address:\t42\n",
      "whats the address:\t27\n",
      "what is the phone number:\t26\n",
      "thank you goodbye:\t24\n",
      "any:\t23\n",
      "moderate:\t22\n"
     ]
    }
   ],
   "source": [
    "freqdict = defaultdict(lambda: 0)\n",
    "for utt in post_ood_utterances_clean:\n",
    "    freqdict[utt] += 1\n",
    "\n",
    "top_post_ood_clean = []\n",
    "for utt, freq in sorted(freqdict.items(), key=lambda x: x[1], reverse=True)[:10]:\n",
    "    print('{}:\\t{}'.format(utt, freq))\n",
    "    top_post_ood_clean.append(utt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['so many mistakes :( yes', 'so many mistakes :( yes', 'ah, awful im sorry any', 'sorry sorry! phone number', 'whoop. my mistake. uh i dont care', 'i do not. sorry :/ i dont care', 'right sorry talking whats the address', 'right sorry talking whats the address', 'no idea, sorry whats the phone number talking', \"oh that's right. my mistake whats the phone number\", \"i can't sorry moderate\", 'big mistake yes that serves british food', 'sorry my dude any type of food', 'hey sorry i mentioned it what is the address and phone number', 'hey sorry i mentioned it what is the address and phone number', 'really? my mistake then. apologies. any price range', \"awww i'm sorry any type\", 'sorry thats not helpful yes', 'sorry, i misunderstood. yes', \"i'm sorry for the inconvenience. thank you good bye\", \"you're quite right, my mistake! whats their phone number\", '\"this was a big mistake\" yes', 'fair enough. my mistake. thank you good bye', 'oh right my mistake. whats the address', 'oh right my mistake. whats the address', 'yes dam auto correct. sorry how about indian type of food', \"sorry i'm exercising. (not really) yes\", 'no idea, sorry thank you good bye', 'ohmy im so sorry can i get the address', 'my mistake, cheers thank you good bye', 'ignore this, sorry mane address', 'yup. sorry, i was mistaken. thank you good bye', 'ah, yes.  my mistake! :) moderately priced restaurant in the south of town', 'ah, yes.  my mistake! :) moderately priced restaurant in the south of town', \"ugh!!! i'm so sorry :( thank you goodbye\", 'sorry :(((( yes', 'oh my mistake. moderate restaurant east part of town', 'mistake noted. address', \"oh no! i'm so sorry thank you good bye\", 'haha absolutely, my mistake :p anything else', 'sorry for my mistake,btw. is there anything else', 'sorry, yeah yes im looking for barbecue food', 'ah gotcha. my mistake! yes', 'ha. fair enough. sorry. phone number', 'ugh. so sorry man. thank you good bye', \"i've mayo mistake. uh anything else\", \"that's a mistake? the address\", 'ahhh my mistake thank you good bye', 'yes, sorry 😕 african', 'gotcha. my mistake. phone number', \"that's ridiculous! i'm so sorry! the address please\", 'ugh sorry buddy any price range', 'awww...so sorry are there any restaurants serving turkish food', 'yes ~ sorry russian', 'rookie mistake. lol can i have moderately priced restaurant in the west part of town', \"i'm sorry that was mean any kind of food\", 'oh! so sorry :( what is the phone number and post code', 'oh! so sorry :( what is the phone number and post code', 'oh okay, my mistake address', 'ive made a mistake moderate', 'my mistake (: yes', \"i'm so sorry. thank you goodbye\", \"apologies, i'm mistaken. phone number\", \"i'm sorry bro thank you goodbye\", \"i'm so sorry:( may i have the address\", 'wow dude so sorry thank you good bye', 'so sorry friend! :( i dont care', 'yes, my mistake i dont care', 'yes, my mistake i dont care', 'i hate you sorry address', 'wow. im so sorry yes', 'oooo, my mistake, my bad. what is the phone number', 'oooo, my mistake, my bad. what is the phone number', 'i’m sorry man. truly am. thank you good bye', 'thanks, my mistake. can i get the address', \"i'll be damned. my mistake. what is the phone number\", \"i'll be damned. my mistake. what is the phone number\", 'well. my mistake then! :) i want an indian food restaurant moderate priced', 'err my mistake is there anything else', 'i am sorry friend 😭 thank you good bye', 'no? my mistake what is the address', \"that's awful. sorry. im looking for vegetarian food in any price range\", 'lol sorry man any area', 'so very sorry... whats the address and post code', 'so very sorry... whats the address and post code', 'my mistake... lol phone number', 'sorry but not rlly what is the phone number', 'sorry but not rlly what is the phone number', \"i'm sorry 😩 moderate\", 'it was me, sorry. phone number', 'lol sorry. this. address', 'that was me,sorry. phone number', 'probably not, sorry bro lol thank you good bye', \"i'm sorry about that. thank you good bye\", \"i'm sorry :/ thank you good bye\", 'my mistake then. apologies. what is the address phone number and post code', 'my mistake then. apologies. what is the address phone number and post code', 'so sorry my friend! thank you goodbye', 'anyway, bizarre. sorry for this. give me the least', 'im so sorry dear i dont care', \"i'm sorry that was mean phone number\", 'my mistake shames me thank you good bye', 'yeah! sorry dude! and the phone number', 'that, was a mistake. thank you good bye', 'sorry, what did you say? yes indian food', 'thanks and sorry any type', 'so sorry :( a moderately priced restaurant that serves vietnamese food', 'that completely sucks, i’m sorry i want a moderately priced restaurant', 'so sorry friend! :( what is the phone number', 'so sorry friend! :( what is the phone number', \"woof. i'm so sorry. i dont care\", 'ah, my mistake! thank you good bye', \"you're right! my mistake! thank you good bye\", 'oh okay, my mistake whats its address', '😕sorry champ whats the phone number', 'sounds good, i was mistaken. im looking for a moderately priced restaurant that serves greek foo', 'an honest mistake, really phone number', 'ah gotcha. my mistake! dont care whats the phone number', 'sorry i was confused moderate', '😂 sorryyyyyy thank you good bye', \"my mistake, that's the one. can i have their address\", \"you're right - my mistake! and the phone number\", 'oh! so sorry :( thank you good bye', 'yeeaahhhhhh sorry about that thank you good bye', 'sorry the confusion around this! yes', 'mistakes happen haha any restaurant that serves asian_oriental food', 'im sorry that happened yes', 'sorry, brother... may i have its address', 'totally by mistake. address', 'sorry enough said! any', 'my mistake, thanks (: thank you good bye', \"i'm sorry, it's not address\", \"oh no! i'm so sorry!! phone number\", \"it's a simple mistake really any\", \"sorry all, it's taken address\", 'this was me. sorry phone number', \"oh boy i'm sorry thank you good bye\", 'yes sir, sorry sir. yes', 'sorry bro.. you vs yes', '😂 sorryyyyyy moderate', 'sorry but not rlly whats the address', 'sorry but not rlly whats the address', \"😭 i'm so sorry thank you good bye\", 'oh lordy...that was a mistake. moderate price range', 'sorry meant 💅 any', 'it was not really mistake moderate', 'hey sorry i mentioned it any price range', 'augh..yes sorry what is the type of food', 'haha! sorry about that! 😉 what is the phone number', 'haha! sorry about that! 😉 what is the phone number', 'tiny mistake there :/ what is the phone number', 'tiny mistake there :/ what is the phone number', \"that's me so sorry can i have the address and phone number\", \"that's me so sorry can i have the address and phone number\", 'indeed! (sorry, i phrased poorly). can i have the phone number', \"wow, didn't know. my mistake. can i have the phone number of the venue\", 'lol im sorry thank you good bye', 'yes, my mistake. cheap', \"noooo! i'm sorry sorry!! thank you good bye\", \"sorry, i'm listening now any\", 'ah, my mistake i guess. address', \"i'm really sorry :/ give me the address phone number and price range\", \"i'm really sorry :/ give me the address phone number and price range\", \"sorry can't help it yes\", 'sorry meant give me the address please', 'sorry for the misunderstanding! 😭 what is the phone number', 'sorry for the misunderstanding! 😭 what is the phone number', 'yeah! sorry dude! thank you good bye', 'my mistake- well spotted! yes', \"i know... i'm sorry. what is the phone number\", \"i know... i'm sorry. what is the phone number\", 'my mistake. new at this thank you goodbye', 'yep, sorry. =( thank you good bye', 'ok.  sorry for my mistake. oh i dont care', 'ahh my mistake! whats the phone number', 'this was me. sorry anything else', \"sorry! i couldn't help myself. and anything else\", 'sorry i meant this: do you have the address of the_lucky_star', 'i stand with my mistake whats the address', 'i stand with my mistake whats the address', 'lol sorry. this. address', 'a *big* mistake, really! haha! thank you good bye', 'my mistake.. thank you! what is the phone number', 'my mistake.. thank you! what is the phone number', 'hahahahahaha im so sorry i dont care', 'thank you! sorry for misunderstanding! moderate', 'aa sorry there whats the address', 'aa sorry there whats the address', 'fixed. sorry for the mistake! is there anything else', 'thanks, my mistake. thank you good bye', \"oh wow i'm so sorry any area\", \"sorry, can't help it... phone number\", 'ive made a mistake address', 'i am mistaken and ashamed. address and phone number', 'i am mistaken and ashamed. address and phone number', \"you're right i'm sorry address\", 'yes, mistakes were made. what is the address and phone number', 'yes, mistakes were made. what is the address and phone number', 'yes, mistakes were made. what is the address and phone number', 'ah, my mistake. i’m sorry. im looking for a moderately priced restaurant in the west part of town', 'excuse the title mistake what is the address and phone number', 'excuse the title mistake what is the address and phone number', 'oh oki, my mistake :d moderate', 'sorry man. no idea. thank you good bye', 'really very sorry. yes can i get an expensive restaurant in the south part of town', \"sorry! but omg it's infuriating! anything else\", 'ugh. sorry :( can i have the address', 'thanks man. sorry about that. yes can i have a vegetarian restaurant in the east part of town', 'everyone makes that mistake once. thank you goodbye', 'i didnt sorry is there anything else', 'lol im sorryy😂 west part of town moderately priced', \"i'm so sorry bud thank you good bye\", \"yes, i'm sorry. yes\", 'ah, yes.  my mistake! :) thank you good bye', 'ah, yes.  my mistake! :) thank you good bye', 'it was indeed.  my mistake. i dont care', 'oops just realized my mistake and the phone number', \"you're right, my mistake. i dont care\", \"that's weird,idk then sorry uh i dont care\", 'probably a translation mistake. thank you goodbye', \"i'm sorry about this! anything else\", 'aaaah, my mistake. yes', 'ah. my mistake. address', 'sorry misread address', 'yeah, fixed. my mistake phone number', 'ooooh, my mistake. sorry! thank you good bye', \"i'm sorry, it's not anything\", 'sorry for the typo ! yes', 'sorry no way !! phone number', 'aw right sorry aha phone number', 'sorry i know it sucks address', 'i made a mistake. any range', 'anyway, bizarre. sorry for this. thank you good bye', 'anyway, bizarre. sorry for this. thank you good bye', 'ah. got it. sorry. phone number', 'sorry, pre-coffee responding phone number', 'this is bad sorry any area', 'oops. sorry phone number', \"wait i'm sorry what. 😭 i dont care\", \"i'm sorry, what? 😒 anything else\", \"ugh i'm so sorry. &lt;3 anything else\", \"sorry, but that's absurd. whats the address\", \"sorry, but that's absurd. whats the address\", 'oh lordy...that was a mistake. and the phone number', 'oh yeah sorry my mistake. any', 'indeed! (sorry, i phrased poorly). thank you good bye', '😕sorry champ thank you good bye', \"you're right! my mistake! thank you goodbye\", 'sorry bout it 🙂 phone number', 'oof. sorry. can i have the address and the phone number', 'oof. sorry. can i have the address and the phone number', \":( i'm so sorry yes\", \"that sucks i'm sorry is there any thing else\", 'big mistake spanish food any area', \"i'm sorry, what? can i have the address and phone number\", \"i'm sorry, what? can i have the address and phone number\", \"you're right. my mistake. thank you good bye\", 'sorry have bothered you. address', 'lol honest mistake telephone number', 'sorry i’ll get on that moderate', 'sorry i was bored thank you good bye', 'oops, my mistake, fixed it moderate', 'sorry for the mistake..... what is the address', 'you are correct.  my mistake. thank you good bye', 'right. my mistake. corrected. thank you good bye', 'mistakes happen! thank you good bye', \"can't help you sorry what is the address\", 'i feel sorry for you what is the phone number', 'i feel sorry for you what is the phone number', 'ah, my mistake! thank you good bye', 'oh wow. so sorry. i would like the address and phone number', 'oh wow. so sorry. i would like the address and phone number', \"yep, my mistake. you're right. what is the phone number\", \"yep, my mistake. you're right. what is the phone number\", 'ohh yeah my mistake address', 'ooooh... my mistake :p address', 'sorry. autocowrong. thank you good bye', \"yes, i'm sorry. yea in the west part of town\", 'ya sorry about that bro yes', 'thats so weird, my mistake. whats the phone number', 'i am so sorry 😹 alright thank you good bye', \"i'm not, sorry :( is there anything else\", 'sorry new to this i am looking for a moderately priced restaurant that serves russian food', 'hey sorry i mentioned it i am looking for a restaurant in the moderate price range that serves russian', 'ah... my mistake. um anything else', \"we're sorry. address and food\", 'no! bad! i’m really sorry. no thank you goodbye', 'that stinks...sorry! :( what is the telephone number', \"oh i'm sorry my bad is there anything else\", 'oh 😕 sorry. thank you goodbye', 'whoops, my mistake moderate', 'aww, sorry :/ is their anything else', 'sorry meant 💅 phone number', 'that completely sucks, i’m sorry yes', '\"... by mistake.\" any', 'that sounds like a mistake. any type of food', \"i'm so so sorry any\", 'my mistake, you are correct moderately priced restaurant in', 'yeah sorry typo address phone number and post code', 'yeah sorry typo address phone number and post code', 'my mistake. fixed it yes', 'oooo, my mistake, my bad. thank you goodbye', 'sorry for the wait thank you good bye', 'i do, my mistake. looking for a restaurant in any area that serves irish food', 'hah lol, my mistake...edited.. thank you good bye', 'lol sorry man thank you good bye', 'sorry. my error. whats the address', 'sorry. my error. whats the address', 'probably a mistake. i dont care', 'i do, my mistake. phone number', 'sorry sorry! address', 'so this happened (sorry honey) phone number', 'hah! my mistake, fixed. is there anything else', 'sorry, brain malfunction. whats the phone number', 'sorry thanks for correction. i dont care', 'oooh. sorry, my mistake. may i have the address and phone number for that venue', 'oooh. sorry, my mistake. may i have the address and phone number for that venue', 'ya sorry about that bro what is the phone number', 'ya sorry about that bro what is the phone number', 'my mistake.  i was drinking. can i get the phone number', 'sorry for the confusion. could i have the address phone number', 'sorry for the confusion. could i have the address phone number', 'oh right. sorry. totally. whats the address', 'oh right. sorry. totally. whats the address', 'my mistake. thanks! whats the phone number', \"yea i'm sorry 😂 is there anything else\", 'i am so sorry 😥 yes', 'thank you! sorry for misunderstanding! thank you good bye', 'ah, sorry for my misunderstanding. i dont care', \"unreal, i'm sorry thank you good bye\", 'truly sorry 😭 thank you goodbye', 'yes dam auto correct. sorry yes id like a restaurant in the east part of town that serves indian food', \"but i'm sorry :/ and their address\", \"i'm so so sorry can i have their phone number please\", 'gotcha. my mistake. any area', 'that sucks sorry any price range', 'yes. i made a mistake. french food any french food', 'yes. i made a mistake. french food any french food', \"i'm sorry it's so hard. what is their phone number\", 'ah, sorry. my mistake. may i have their phone number please', 'not me, sorry! thank you goodbye', 'sorry but me no understanding thank you good bye', 'ooops yeah my mistake im looking for a moderately priced restaurant in the east part of town', 'aww! sorry! next time! i dont care', '...sorry... whats the address', '...sorry... whats the address', 'so sorry. yes. moderately priced', 'so sorry. yes. moderately priced', 'just mistake, ty ;) whats the address', 'just mistake, ty ;) whats the address', 'my mistake then! :d thank you good bye', 'people make mistakes everyday. whats the phone number', 'sorry read that wrong 😊 what is the phone number and address', 'sorry read that wrong 😊 what is the phone number and address', 'sorry read that wrong 😊 what is the phone number and address', 'sorry... lmao oops im looking for a restaurant in any area', 'ooo sorry .. what is the phone number and address', 'ooo sorry .. what is the phone number and address', 'ooo sorry .. what is the phone number and address', 'contact them. mistakes do happen. yes', 'sorry guys. yes', 'ahh my mistake! any', 'well, my mistake then. phone number', 'sorry for the misunderstanding! 😭 thank you good bye', 'oops. obviously not. sorry! i dont care the center part of town', 'yes, my mistake, thanks post code', 'ugh i am so sorry yes', 'sorry not sorry thank you good bye', 'oops my mistake lol north part of town moderately priced', '\"... by mistake.\" any', \"i'm sorry that was rash yes\", 'im sorry!!! 😫 😩 moderate', 'sorry that was me 😅 can i get the phone number', 'sorry about that !!! phone number', 'oops, my mistake, fixed it any', \"it's a mistake can i have the address\", \"i'm sorry lol moderate priced\", 'aww, so sorry moderate', 'i guess so.  my mistake. moderate', 'ah, my mistake!  sorry :) thank you good bye', \"sorry guys i'm bored haha is there anything else\", \"sorry didn't know whats the address and phone number\", \"sorry didn't know whats the address and phone number\", \"sorry didn't know whats the address and phone number\", '- very sorry. ❤️ uh yes can i get a cheap restaurant in the west part of town', 'im so sorry dear and the phone number', 'i didnt sorry can i have the address of the last one', \"i'm sorry what lol and the phone number\", 'omg so sorry. okay can i get the address', 'mistakes happen. thank you good bye', 'ah, sorry for my misunderstanding. moderately priced restaurant in the south part of town', 'right sorry i dont care', 'tiny mistake there :/ and the address', 'ah. my mistake. thank you good bye', 'i’m sorry man. truly am. i dont care', \"you're correct, my mistake. and the phone number\", \"you're right i'm sorry thank you good bye\", 'so sad and so sorry thank you good bye', 'whoops sorry! i dont care', 'ugh. sorry :( thank you good bye', \"ahh that's right my mistake whats the address\", \"ahh that's right my mistake whats the address\", 'lol, my mistake i dont care', 'lol sorry... may i have the address', 'i fixed my mistake =p okay thank you good bye', 'lol, my mistake what about a moderately priced restaurant in the south part of town', 'err my mistake thank you good bye', \"wow. that sucks. i'm sorry. phone number\", 'augh..yes sorry phone number', 'augh..yes sorry phone number', 'really? my mistake then. apologies. can i have the address', \"oh no! i'm so sorry!! yes\", 'my mistake sorry yes', \"i'm really sorry :/ yes\", 'yes, my mistake. breathing', 'oh ok, my mistake. i dont care', 'nope sorry bud whats the phone number', 'sorry for that i dont care', \"sorry, i'm listening now can i get the phone number and price range\", 'thanks - sorry about that any kind', 'sorry i was sleeping may i get the address and phone number', 'sorry i was sleeping may i get the address and phone number', 'oops, my mistake... is it moderately priced', 'was by mistake okay whats the address', 'was by mistake okay whats the address', \"i'm so sorry mate what is the phone number\", \"i'm so sorry mate what is the phone number\", \"...i'm so sorry i dont care\", 'sorry, wrong reply. thank you good bye', 'ah sorry about that. the address', 'sorry, i misunderstood. thank you good bye', \"then that's my mistake. any area\", 'my mistake. misunderstood thank you good bye', 'i misread that. my mistake. is there anything else', \"i'm sorry lol thank you good bye\", \"sorry 'bout it. moderate\", 'absolutely wrong!!! sorry but true. phone number', \"alright alright i'm sorry 😢 thank you good bye\", 'i have no idea! sorry! address', \"it's really my mistake. thank you good bye\", \"sorry, can't help it... phone number\", 'sorry... lmao oops can i have the phone number', 'sorry we gotchu fam address and post code', \"we're sorry. yes\", 'my mistake , sorry !! what is the address and post code', \"i'm sorry, what.... yes\", 'my mistake, thank you. what is the phone number', 'my mistake, thank you. what is the phone number', 'oops, sorry any area', 'oh, ok. my mistake. yes', 'yep hahaha, my mistake. and the phone number', 'so sorry boo😘😘😘 can i get the address', 'probably just an honest mistake thank you good bye', \"😂 i'm sorry &lt;3 what is the address\", 'it was not really mistake moderately priced', 'typo mistake xd thank you good bye', 'my mistake, edited. whats the address', 'my mistake, edited. whats the address', 'ah i gotcha. my mistake! thank you good bye', 'ah i misunderstood sorry x whats the address', 'ah i misunderstood sorry x whats the address', 'thank you! sorry for misunderstanding! thank you good bye', \":( i'm sorry. moderately priced\", 'sorry your right whats the address', 'sorry your right whats the address', '*that, sorry whats the phone number', \"i'm so sorry. why? whats the phone number\", 'sorry breh. that sucks and whats their phone number', \"sorry 'bout it great whats the phone number\", 'even louis vuitton makes mistakes thank you good bye', 'eh, we all make mistakes. i dont care', 'oh true. my mistake. thank you good bye', 'thats so weird, my mistake. whats their address', '\"i am so sorry!\" can i get the phone number', 'that was me,sorry. can i get the phone number', \"lmaooo i'm sorry whats the address\", \"lmaooo i'm sorry whats the address\", 'sorry for the mistake..... any', \"i'm sorry for the typos whats the address\", \"i'm sorry for the typos whats the address\", 'yep, sorry. =( any price range', \"😭 i'm so sorry im looking to find a moderately priced restaurant\", 'yes, sorry, my mistake! let me get the address', 'yes, sorry, my mistake! let me get the address', 'sorry we got excited. 😂 can i get the phone number', \"i'm so sorry :( thank you good bye\", 'sorry my dude anything else', 'my mistake, fixing it now any area', 'sorry, you are correct thank you good bye', 'yeah ... my mistake. apologies. thank you good bye', \"unreal, i'm sorry phone number\", 'gotcha. my mistake. phone number', 'oh, my mistake. sorry. moderate restaurant south part of town', 'ah right, my mistake yes', 'oops, my mistake phone number', 'this is bad sorry could i have the address', 'sorry, yeah i figured..... yes', 'sorry sorry thank you good bye', 'i feel sorry for you no moderately priced restaurant', \"i'm sorry. lol yes moderately priced\", \"i'm sorry. lol yes moderately priced\", 'sorry got a little frisky... thank you good bye', 'sorry... :( moderately priced', 'sorry, mistake corrected. any', \"this sucks, i'm so sorry any\", 'sorry. cousin. whatever. moderately', 'whoops sorry moderately priced restaurant', \"awwww i'm sorry any\", 'oh, i am mistaken then. phone number', 'oh, ok. my mistake. may i have the phone number and post code', 'that grammatical mistake got me may i have the address and price range', \"you're right. my mistake. i dont care\", 'sorry dude. thank you good bye', 'sorry if i misinterpreted! phone number', \"yes, i'm sorry. thank you\", 'sorry! my mistake. any', 'mistake noted. moderate', 'oh im very sorry moderately price', 'ive made a mistake phone number', 'haha! sorry about that! 😉 any kind of food', 'so sorry :( can i have the address please', 'ahhh my mistake yes', \"i'm sorry lol im looking for a moderately priced restaurant in the south\", 'sorry . i mean it. i dont care', 'oh, my mistake :/ what is the phone number and address', 'oh, my mistake :/ what is the phone number and address', 'oh, my mistake :/ what is the phone number and address', 'my mistake. fixed it i dont care', 'ahh gotcha, my mistake i dont care', \"yea i'm sorry bro moderately priced\", 'sorry meant what is the phone number and address', 'sorry meant what is the phone number and address', 'sorry meant what is the phone number and address', 'i feel sorry for you thank you good bye', 'ah sorry about that. address', 'ahh, my mistake address', 'so sorry my friend! thank you good bye', 'sorry confusion reigns.. yes', 'so sad and so sorry thank you good bye', 'ah, my mistake. i’m sorry. thank you goodbye', 'classic mistake can i get the phone number', 'clarify my mistakes. thank you good bye', 'sorry i meant this: the phone number', 'oh that! sorry. phone number', \"i'm so sorry. phone number\", 'yea, sorry, should have clarified address', 'ah, my mistake :) yes', 'no! bad! i’m really sorry. i want a moderately priced restaurant in the north part of town', 'hehe, *little mistake* what is the address', 'so sorry boo😘😘😘 yes indian food', 'tiny mistake there :/ thank you good bye', 'yea my mistake. thank you good bye', 'my mistake.  fixed yes', 'nope, sorry. nothing. where is the address', \"oops, that's my mistake. phone number\", \"i'm truly sorry! thank you good bye\", 'sorry, mistake corrected. yes', \"sorry but...i don't get it??😂 address\", 'yes, my mistake address', 'yes, my mistake address', 'duh, sorry :) address', 'ah yes, my mistake. south', 'yup, my mistake.  corrected. address', 'i’m sorry…. repeat that?????!!!! thank you good bye', 'oh that! sorry. yes', \"i'm so sorry! lol thank you good bye\", 'my mistake.  was on phone. i dont care', \"this sucks, i'm so sorry i dont care\", 'yeah dude. i misunderstood. sorry. what is the address and phone number', 'yeah dude. i misunderstood. sorry. what is the address and phone number', \"sorry 'bout it thank you good bye\", 'oh! lol toooootally misread, sorry! i dont care', \"sorry guys i'm bored haha what is the phone number and addre\", \"sorry guys i'm bored haha what is the phone number and addre\", 'right :-) my mistake :-) thank you good bye', 'oh, i am mistaken then. moderate', 'yup, sorry! my mistake! yes', 'im sorry bud. and can you tell me the phone number', \"awwwwwrrrrrr i'm sorry! okay thank you good bye\", 'ahh... my mistake whats the address', 'ahh... my mistake whats the address', \"😭 i'm so sorry tell me the phone number\", 'no idea, sorry yes', 'sorry , my fault 😅 thank you good bye', 'sorry new to this im looking for a moderately priced restaurant serving basque food', 'sorry my apology thank you good bye', 'sorry, that was me okay whats the address', 'sorry, that was me okay whats the address', 'my mistake, i apologize. how about the phone number', 'sorry i was bored thank you goodbye', \"i'm really sorry. thank you good bye\", \"that's the worst sorry 😓 whats the address\", \"that's the worst sorry 😓 whats the address\", 'im sorry whaaaa i dont care', 'mistake. sorry. thanks what is its address', 'i guess i was mistaken! what is its address', \"thank you..i'm sorry thank you good bye\", 'oops sorry...autocorrect i want a moderately priced restaurant in the south part of town', 'my mistake.  apologies. and the phone number', 'a *big* mistake, really! haha! post the phone number', 'oh that sucks. sorry. whats their address', 'oh! my mistake, thank you! may i have the address of it', \"by mistake. :'( thank you good bye\", 'that sucks. i’m sorry. could i have the address', \"that's weird,idk then sorry phone number\", 'too big.  what a mistake. anything else', 'ah, my mistake!  sorry :) any', 'sorry that was me 😅 address', 'a *big* mistake, really! haha! phone number', 'sorry we gotchu fam address', 'mistake! address', 'sorry. what were you saying? what is the address', 'oh, my mistake then! phone number', 'yeah sorry typo thank you good bye', 'i am so sorry i need moderately priced restaurant that serves spanish', 'my mistake, you are correct yes', 'sorry excuse moderate', \"i'm so sorry!! and the phone number\", \"omg i'm so sorry whats their address\", 'awh im sorry!! thank you good bye', 'horrible; so sorry. yes', 'im sorry bud. what is the phone number', 'im sorry bud. what is the phone number', 'ah my mistake :p what is the phone number', 'ah my mistake :p what is the phone number', 'oh, sorry my mistake. what is the address', 'sorry but... yes', 'yes, apologies. my semantics mistake! im looking for a expensive restaurant serving irish food', 'omg so sorry. can i have the address and the post code', 'ah, my mistake i guess. im looking for a moderately priced restaurant in the east part of', \"i'm sorry !! 😕 thank you good bye\", 'sorry i was mistaken... any area', 'shocking mistake! can i have the phone number and address of the venue', 'shocking mistake! can i have the phone number and address of the venue', 'lmfao im sorry moderately priced restaurant', 'sorry what. phone number', 'my mistake. thanks. phone number', 'sorry for any inconvenience moderately priced restaurant', 'sorry for any inconvenience moderately priced restaurant', 'im sorry what, address', 'my mistake, thanks (: phone number', 'my mistake. thank you find me a moderately priced restaurant in', \"my mistake, i'll fix it what is the phone number\", \"my mistake, i'll fix it what is the phone number\", 'yes sorry, my bad! expensive', 'i right sorry id like to know the phone number', 'mistakes happen, people mistype. whats the address', 'mistakes happen, people mistype. whats the address', 'sorry guys. yes', 'oh, my mistake :/ uh i dont care', 'i admit, my mistake :{ thank you good bye', 'ahh... my mistake im looking for a moderately priced in the south', 'my mistake (: what is the address and phone number', 'my mistake (: what is the address and phone number', 'im sorry was that bad? what is the address', 'that, was a mistake. i am looking for a moderate priced restaurant in the west of town', 'that was a mistake thank you good bye', 'my mistake :p can i have the address and phone number please', 'my mistake :p can i have the address and phone number please', 'aww, so sorry yes', 'ah ok my mistake sorry thank you good bye', 'mistakes happen! and can i have the address phone number and price range of the restaurant please', 'mistakes happen! and can i have the address phone number and price range of the restaurant please', 'ah true sorry missed that! thank you good bye', 'oh word, my mistake then what is the address and phone number', 'oh word, my mistake then what is the address and phone number', 'nooooooooo. sorry man! thank you good bye', 'sorry not sorry 😂😂 address', 'sorry you’re right, my bad thank you good bye', 'lmao =( sorry address and price range', 'really? my mistake then. apologies. what is the phone number', 'really? my mistake then. apologies. what is the phone number', 'haha absolutely, my mistake :p thank you good bye', 'oh oki, my mistake :d what is the phone number', 'oh oki, my mistake :d what is the phone number', 'good point, my mistake. what is the phone number', 'good point, my mistake. what is the phone number', \"i'm sorry !! 😕 what is the address and phone number\", \"i'm sorry !! 😕 what is the address and phone number\", 'honest mistake. i dont care', 'edit: nvm, was mistaken. thank you good bye', 'ahh my mistake thank you goodbye', 'sorry for the misunderstanding! 😭 and the phone number', 'oops typo again sorry yes can i have an expensive restaurant in the east', 'fixed. sorry for the mistake! thank you good bye', 'yup, sorry! my mistake! thank you good bye', 'so sad. so sorry. the address', 'it was me, sorry. anything else', 'sorry not sorry 😂😂 the address', 'yea, sorry, should have clarified thank you good bye', 'sorry brother... is there anything else', 'ok.  sorry for my mistake. what is the address and phone number', 'ok.  sorry for my mistake. what is the address and phone number', 'oh my mistake then, sorry i dont care', 'my mistake haha the address and phone number', 'my mistake haha the address and phone number', 'sorry. wish i new. thank you good bye', 'sorry for the grammar mistakes. the address', 'u right, im sorry :( thank you good bye', 'oof. sorry. id like a restaurant in any area that serves', \"i'm so sorry bud what is the address and phone number\", \"i'm so sorry bud what is the address and phone number\", 'oops. honest mistake. hi i need a restaurant in any area', ':( sorry bro. thank you good bye', 'my mistake- well spotted! hi im looking for a moderately priced restaurant', 'ah, my mistake.  thanks. thank you good bye', \"sorry i can't hear you....😉 i dont care\", \"sorry can't help it thank you good bye\", 'sorry my apology thank you good bye', 'mistakes happen, people mistype. thank you good bye', 'yeah, fixed. my mistake what is their phone number', 'fixed. sorry bout that what is there phone number', 'jup, hungover mistake! thank you good bye', 'wups my mistake. phone number', 'sorry, brain malfunction. is there anything else', 'oh i see. my mistake. anywhere', \"this sucks. i'm sorry. uh phone number\", \"really? ugh, i'm sorry okay thank you good bye\", 'my mistake.. thank you! address', \"i'm sorry about that. thank you good bye\", 'oh, i am mistaken then. moderately priced', 'aww, sorry :/ address', 'ah, yes.  my mistake! :) moderate', 'ah, yes.  my mistake! :) moderate', 'ooops yeah my mistake yes', 'ah, my mistake. thanks! any range', \"oops, that's my mistake. phone number\", 'ugh. that totally sucks. sorry! thank you good bye', 'ah i gotcha. my mistake! okay thank you good bye', 'yup! my mistake whats the address', 'yup! my mistake whats the address', 'right. sorry is there any restaurant', 'fixed, my mistake. thank you good bye', 'sorry , meant how about a moderately priced restauran', 'sorry not sorry 😂😂 what is the address', 'lol sorry. this. yes', 'might be a mistake? yes', 'my mistake then. apologies. may i have the phone number', \"you're right i'm sorry yes\", \"you're right! my mistake! what is the phone number\", \"you're right! my mistake! what is the phone number\", 'sorry :( ❤️ thank you good bye', 'ok.  sorry for my mistake. yes', \"awwwwwrrrrrr i'm sorry! i dont care\", 'that was a mistake is it moderately', \"wow, didn't know. my mistake. thank you goodbye\", \"i'm sorry ok yes\", 'sorry meant thank you goodbye', 'sorry. not sorry :) thank you good bye', 'sorry, man. yes', 'sorry about it yes', 'sorry for any inconvenience thank you good bye', 'sorry for any inconvenience thank you good bye', 'oh sorry about that anything else', 'sorry that’s me yes', 'yup! my mistake anything else', 'totally by mistake. address', 'oopsie, sorry phone number', 'sorry wrong text address', 'im so sorry... address', \"that's ridiculous! i'm so sorry! address\", 'i misread that. my mistake. thank you good bye', 'my mistake. thank you moderately priced', 'oh right my mistake. address', 'oh ok sorry about that thank you good bye', \"lmao i'm sorry. thank you good bye\", \"that sucks i'm sorry moderately priced restaurant in the west part of town\", 'sorry, you are correct moderately', \"i'm sorry what? address\", 'ok sorry :) yes spanish food', 'i have no idea! sorry! thank you goodbye', 'omg :((( im sorry !! what is the address', \"wait i'm sorry what. 😭 what is the phone number\", \"wait i'm sorry what. 😭 what is the phone number\", 'not sorry even at all yes', 'ahh rookie mistake, forgive me! thank you good bye', 'pst brother! sorry about that can i get the address and phone number', 'pst brother! sorry about that can i get the address and phone number', \"face palm i'm sorry hah great thank you good bye\", 'not sorry even at all i want a restaurant in any area that serves vietnamese food', 'oh yea. sorry. thank you good bye', 'oh, ok then.  my mistake. i dont care', 'my mistake. thanks! whats the phone number', 'im sorry i do that thank you good bye', \"i'm sorry about this! thank you good bye\", 'yea, sorry, should have clarified whats the address', 'yea, sorry, should have clarified whats the address', 'tough day dude sorry :( thank you good bye', \"sorry, i'm listening now thank you good bye\", 'whoops my mistake what is the phone number', 'whoops my mistake what is the phone number', \"i'm so sorry mate im looking for a moderately priced restaurant that serves\", 'totally a mistake thank you good bye', 'ah, sorry my mistake! thank you good bye', 'sorry, mistake corrected. any type', 'lmao im so sorry😢 is there anything else', 'ok yeah sorry is there anything else', 'sorry. now. lol!! mediterranean restaurant in any area', 'oh sorry, my mistake whats the address phone number and price range', 'oh sorry, my mistake whats the address phone number and price range', 'oh sorry, my mistake whats the address phone number and price range', 'so sorry lmfaooooo yes', \"this sucks, i'm so sorry i dont care\", 'my mistake, my friend whats the phone number', 'true, my mistake. thank you good bye', 'no idea man sorry :(( i am looking for a moderately priced restaurant that', ':( sorry bro. i dont care', 'sorry new to this ah thank you phone number', 'lol good catch.. rookie mistake any anything else', 'regret the mess, sorry. address phone number and area', 'regret the mess, sorry. address phone number and area', \"really? ugh, i'm sorry im looking for a moderately priced restaurant that serves european food\", 'oh, my mistake. thanks! is their anything else', \"i'm sorry man could you give me the phone number and post code\", 'sorry pal, honest mistake yes', 'that sucks. i’m sorry. yes', 'mistakes happen, people mistype. what is the address and phone number', 'mistakes happen, people mistype. what is the address and phone number', 'ah, thank you. my mistake. address and phone number', 'ah, thank you. my mistake. address and phone number', \"sorry i'm just excited thank you goodbye\", 'thanks! i was clearly mistaken. moderate', 'ok.  sorry for my mistake. thank you good bye', \"yes, that's right. my mistake. thank you good bye\", \"yes, that's right. my mistake. thank you good bye\", 'im sorry what, and phone number', 'my mistake.  was on phone. thank you good bye', 'okay sorry!!!!! :-( lol are there any that serve vietnamese food', 'mistakes happen! whats the address', 'mistakes happen! whats the address', 'im sorry my guy say thank you good bye', 'yes. i made a mistake. south', 'oh, i am mistaken then. whats the address', 'oh, i am mistaken then. whats the address', 'sorry comrade lol thank you good bye', 'typo mistake xd phone number', 'oops. honest mistake. thank you goodbye', 'sorry i meant this: i dont care', 'ah. my mistake then. may i have the address and postcode', 'oh, my mistake. thanks! i need the address and phone number', 'oh, my mistake. thanks! i need the address and phone number', 'ahh rookie mistake, forgive me! could i have the address and phone number', 'ahh rookie mistake, forgive me! could i have the address and phone number', 'ha. fair enough. sorry. phone number', 'honest mistake.  sorry. whats the address', 'honest mistake.  sorry. whats the address', 'sorry for the mistake..... thank you good bye', 'sorry for the confusion thank you good bye', 'sorry. not sorry :) yes in the west part of town', 'tough day dude sorry :( whats the phone number', 'ah. my mistake then. yes', 'sorry misread i dont care', 'rookie mistake by the keeper. thank you good bye', 'ah okay, my mistake. whats the address', 'ah okay, my mistake. whats the address', \"wow. that sucks. i'm sorry. whats the phone number\", 'ah i see, my mistake. thank you good bye', 'oh word, my mistake then uh i dont care', 'whoops sorry lmao is there anything else', 'sorry. i take full responsibility. thank you good bye', 'oops typo again sorry any part of town', 'not me, sorry! any price range', 'yup! my mistake yes', \"i think it's a mistake... what is the phone number\", \"i think it's a mistake... what is the phone number\", 'sorry if i misinterpreted! any type', 'i was slightly mistaken, updated. thank you good bye', \"that's so terrible. i'm sorry. yes\", 'sorry not sure any price range', 'sorry. cousin. whatever. yes', 'hahahahah, sorry about that.... hi id like a moderately priced restaurant in the north', \"omg i'm sorry i dont care\", 'well. my mistake then! :) what is the golden_woks phone number', \"really? hahaha. sorry, shouldn't laugh. moderately priced spanish restaurant\", 'well, see... mistakes were made. what is their phone number', 'ooooh... my mistake :p phone number', \"i'm sorry for the inconvenience. thank you good bye\", 'ah, my mistake i guess. phone number', 'oh yeah. mistake. thank you good bye', \"it's most likely a mistake. what is the address phone number and post code\", \"it's most likely a mistake. what is the address phone number and post code\", 'my mistake, sorry! thank you good bye', 'haha! sorry about that! can i have the address', 'ah, my mistake. thanks! thank you good bye', 'i was mistaken is there anything else', 'ha. fair enough. sorry. what is the address', 'good catch.  my mistake. yes', 'sorry for the grammar mistakes. thank you good bye', 'reasonable mistake actually whats the phone number', \"this mistake isn't really reasonable. i want a moderate priced restaurant in the north part of town\", 'sorry :(((( is there anything else', '*mistake!* thank you good bye', 'ohh yeah my mistake thank you goodbye', 'ah gotcha. my mistake. thanks! yes', \"oh man i'm so sorry thank you good bye\", 'sorry about it any', 'i think im mistaken, sorry i dont care', \"i'm sorry for everything thank you good bye\", 'so sad. so sorry. what is the phone number', 'so sad. so sorry. what is the phone number', 'oh my mistake. thank you. may i have the phone number', 'fair enough, sorry! may i have the phone number', 'i am so sorry. thank you goodbye', \"i'm so sorry about that any\", \"oh no!!! i'm so sorry any\", 'thank you! sorry for misunderstanding! thank you good bye', 'yeaahhh sorry ! im looking for a moderately priced restaurant in the west part of town', 'sorry actually this thank you good bye', 'yes, apologies. my semantics mistake! im looking for kosher food in the west side', 'im sorry bud. uh yes can i get the address', 'im sorry bud. uh yes can i get the address', 'sorry i was bored uh yes im looking for a restaurant that serves chinese food', 'that, was a mistake. uh yes can i find a cheap restaurant in the north part of town', 'sorry. im sorry uh yes can i get a moderate moderately priced restaurant in the west part of town', 'sorry. im sorry uh yes can i get a moderate moderately priced restaurant in the west part of town', 'no sorry! 💜 okay and can i get the address', 'my mistakes. uh can i get the address', 'okay, my mistake. yes', 'oh so sorry okay and can i get their address', 'oh my mistake then address', \"sorry i'm not more sorry. phone number\", 'sorry for the mistake..... thank you good bye', 'sorry , my fault 😅 whats the address', 'sorry , my fault 😅 whats the address', 'yeah ... my mistake. apologies. thank you goodbye', 'ah right, my mistake is there anything else', 'sorry this happened, dude. whats the address', 'sorry this happened, dude. whats the address', 'yeah - sorry for mistake. im looking for a restaurant in any area that servess', \"that's ridiculous! i'm so sorry! phone number\", '. oh i’m sorry! whoops. yes', \"sorry 'bout it yes\", 'oh man so sorry! can i have the address', 'gurus makes mistakes phone number', 'good point, my mistake. phone number', \"my mistake. that's the one. okay whats the phone number\", 'well, see... mistakes were made. okay may i have the address', 'whoops sorry! do you have any other restaurants', 'honest mistake.  sorry. im looking for a moderately priced restaurant and it should serve', 'wow. im so sorry can i have his phone number', 'yes, my mistake, thanks thank you goodbye', 'yes, my mistake, thanks thank you goodbye', 'yeah sorry about that 😂 i dont care', 'that sucks sorry yes', 'haha simple mistake. can i have the address', 'stupidest smart mistake ever! phone number', 'yes ~ sorry thank you good bye', 'yes ~ sorry thank you good bye', 'this is a purposeful \"mistake\". yes', 'ooooh, my mistake. sorry! phone number', 'really? my mistake then. apologies. yes', 'oh man im sorry!!! moderate', 'yes. i made a mistake. cantonese', 'sorry ,missed it address', \"i'm so sorry 😢 i dont care\", 'oops, sorry whats the address', 'oops, sorry whats the address', 'my mistake, then. whats the phone number', 'mistakes happen haha i dont care', 'so sorry my friend! moderate', 'you are correct.  my mistake. any', 'oops. sorry um thank you good bye', 'right. my mistake. corrected. phone number', 'sorry i was mistaken... phone number', \"sorry, i'm listening now what is the phone number\", \"sorry, i'm listening now what is the phone number\", 'lmao!!!! sorry 😂😂 i do i dont care', \"sorry, didn't know. i dont care\", 'sorry bout it 🙂 moderately priced', \"this sucks. i'm sorry. phone number\", 'im sorry bud. any', 'sorry sorry! anything else', 'ooops my mistake any', 'im sorry!!! 😫 😩 thank you good bye', 'whoops, sorry 💚 and the address', 'sorry what. thank you good bye']\n"
     ]
    }
   ],
   "source": [
    "top_post_ood_noisy = []\n",
    "for utt in post_ood_utterances_noisy:\n",
    "    for top_clean_utt in top_post_ood_utterances_clean:\n",
    "        if top_clean_utt in utt:\n",
    "            top_post_ood_noisy.append(utt)\n",
    "print(top_post_ood_noisy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processed 15 out of 15 batches\r"
     ]
    }
   ],
   "source": [
    "post_ood_clean_z = compute_z(make_variational_autoencoder_dataset(top_post_ood_clean, vocab, config['max_sequence_length']), net)\n",
    "post_ood_noisy_z = compute_z(make_variational_autoencoder_dataset(top_post_ood_noisy, vocab, config['max_sequence_length']), net)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[t-SNE] Computing 121 nearest neighbors...\n",
      "[t-SNE] Indexed 986 samples in 0.001s...\n",
      "[t-SNE] Computed neighbors for 986 samples in 0.086s...\n",
      "[t-SNE] Computed conditional probabilities for sample 986 / 986\n",
      "[t-SNE] Mean sigma: 1.420597\n",
      "[t-SNE] KL divergence after 50 iterations with early exaggeration: 64.462791\n",
      "[t-SNE] Error after 500 iterations: 2.293396\n"
     ]
    }
   ],
   "source": [
    "tsne = TSNE(n_components=2, verbose=1, perplexity=40, n_iter=500, learning_rate=10.0)\n",
    "tsne_results = tsne.fit_transform(np.concatenate([post_ood_clean_z, post_ood_noisy_z], axis=0))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Scatter plot of top clean post-ood utterances (blue) and their noisy variations (red)\n",
    "=="
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fc4cb6d6198>"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(tsne_results[:, 0],\n",
    "            tsne_results[:, 1],\n",
    "            color=['blue'] * len(top_post_ood_clean) + ['red'] * len(top_post_ood_noisy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}