human_correlation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Human Correlation Analysis\n",
    "\n",
    "Fork from https://github.com/facebookresearch/ParlAI/blob/master/projects/controllable_dialogue/Analysis_n_Graphs.ipynb"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### General preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>.container { width:95% !important; }</style>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# bunch of imports and settings\n",
    "import os\n",
    "# make sure we never use CUDA on accident in this notebook\n",
    "os.environ['CUDA_VISIBLE_DEVICES'] = ''\n",
    "\n",
    "import logging\n",
    "logging.basicConfig(level=logging.WARNING, format='%(message)s')\n",
    "\n",
    "# Make the notebook bigger\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('ggplot')\n",
    "plt.rcParams['figure.figsize'] = [12.0, 8.0]\n",
    "\n",
    "import contextlib\n",
    "import numpy as np\n",
    "import scipy.stats\n",
    "import logging\n",
    "import time\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "import torch\n",
    "from torch import nn\n",
    "import torch.nn.functional as F\n",
    "\n",
    "import pyro\n",
    "import pyro.infer\n",
    "import pyro.infer.mcmc\n",
    "import pyro.distributions as dist\n",
    "from parlai.core.metrics import _f1_score, _bleu\n",
    "\n",
    "pyro.enable_validation(False)\n",
    "\n",
    "import pandas as pd\n",
    "import json\n",
    "import os\n",
    "import seaborn as sns\n",
    "from pprint import pprint\n",
    "pd.options.display.float_format = '{:,.4f}'.format\n",
    "\n",
    "from IPython.core.display import display, HTML\n",
    "display(HTML(\"<style>.container { width:95% !important; }</style>\"))\n",
    "pd.set_option('display.max_columns', 500)\n",
    "\n",
    "\n",
    "\n",
    "import projects.controllable_dialogue.controllable_seq2seq.controls as control\n",
    "import projects.controllable_dialogue.controllable_seq2seq.nidf as nidf\n",
    "import projects.controllable_dialogue.controllable_seq2seq.arora as arora\n",
    "import parlai.core.dict as pcdict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Get where the logs are stored\n",
    "from parlai.core.params import ParlaiParser\n",
    "basic_opt = ParlaiParser().parse_args([], print_args=False)\n",
    "\n",
    "# make sure we download everything\n",
    "from projects.controllable_dialogue.tasks.build import build\n",
    "build(basic_opt)\n",
    "\n",
    "DATA_FOLDER = os.path.join(basic_opt['datapath'], 'controllable_dialogue')\n",
    "LOG_FOLDER = os.path.join(DATA_FOLDER, 'evaluation_logs')\n",
    "\n",
    "# used for assertions later\n",
    "NUM_TURNS = 6\n",
    "\n",
    "# names of metrics\n",
    "LIKERT_METRICS = ['fluency', 'enjoy', 'turing', 'make_sense', 'inquisitive', 'interest', 'avoid_rep', 'listen']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Helper functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# IO utilities\n",
    "def render_conversation(row):\n",
    "    result = []\n",
    "    nice_scores = '&nbsp;&nbsp;&nbsp;&nbsp;'.join([\n",
    "        '{}: {}'.format(m.title(), row[m])\n",
    "        for m in LIKERT_METRICS\n",
    "    ])\n",
    "    for i, turn in enumerate(row['dialog']):\n",
    "        is_bot = turn['speaker'] == 'model'\n",
    "        speakername = row['model_name'] if is_bot else 'Human'\n",
    "        text = turn['text']\n",
    "        \n",
    "        align = 'right' if is_bot else 'left'\n",
    "        color = \"white\" if is_bot else \"black\"\n",
    "        bgcolor = '#2391f7' if is_bot else '#e1e1e7'\n",
    "        \n",
    "        result.append(\n",
    "            (\n",
    "                '<div style=\"overflow: auto; padding: 1ex 0;\">'\n",
    "                '<div style=\"clear: both; float: {}; color: {}; background-color: {}; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\">'\n",
    "                '<p style=\"margin: 0\">{}: {}</p>'\n",
    "                '</div>'\n",
    "                '</div>'\n",
    "            )\n",
    "            .format(align, color, bgcolor, speakername, text)\n",
    "        )\n",
    "    return HTML(\n",
    "        '<div style=\"background-color: white; margin: 0em; padding: 0.5em; '\n",
    "        'font-family: sans-serif; font-size: 12pt; width: 99%;\">'\n",
    "        + ''.join(result) +\n",
    "        '</div>'\n",
    "    )\n",
    "\n",
    "def render_many_converstions(table):\n",
    "    return HTML('<hr />\\n'.join(render_conversation(r).data for i, r in table.iterrows()))\n",
    "\n",
    "# useful helpers\n",
    "\n",
    "def modelname_subset(data, model_keys):\n",
    "    return data[data.model_name.apply(lambda x: x in model_keys)]\n",
    "\n",
    "def modeltype_subset(data, model_keys):\n",
    "    return data[data.model_type.apply(lambda x: x in model_keys)].sort_values(['model_type', 'model_bucket'])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load conversation table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "conversation_table = []\n",
    "workers = set()\n",
    "\n",
    "def get_many_logs():\n",
    "    for fn in os.listdir(LOG_FOLDER):\n",
    "        with open(os.path.join(LOG_FOLDER, fn)) as f:\n",
    "            for line in f:\n",
    "                line = line.strip()\n",
    "                if not line:\n",
    "                    continue\n",
    "                yield json.loads(line)\n",
    "\n",
    "for item in get_many_logs():\n",
    "    row = {}\n",
    "    \n",
    "    # flatten out and get the keys we need\n",
    "    \n",
    "    for key in ['model_name', 'model_type', 'model_bucket', 'dialog', 'assignment_id_hashed', 'evaluator_id_hashed']:\n",
    "        row[key] = item[key]\n",
    "    \n",
    "    for key in ['enjoy', 'interest', 'listen', 'turing', 'avoid_rep', 'make_sense', 'fluency', 'persona_guess', 'inquisitive']:\n",
    "        row[key] = item['evaluation_results'][key]\n",
    "    \n",
    "    conversation_table.append(row)\n",
    "    \n",
    "conversation_table = pd.DataFrame(conversation_table)\n",
    "# mark certain columns as categories\n",
    "conversation_table['evaluator_id_hashed'] = conversation_table.evaluator_id_hashed.astype('category')\n",
    "conversation_table['model_name'] = conversation_table.model_name.astype('category')\n",
    "\n",
    "# deterministic shuffle\n",
    "conversation_table = conversation_table.sort_values('assignment_id_hashed').sample(frac=1, random_state=42)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Raw table at a glance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'speaker': 'human_evaluator', 'text': 'Hello! how are you?'},\n",
       " {'speaker': 'model',\n",
       "  'text': \"I'm doing well. Just got done playing some game of duty.\"},\n",
       " {'speaker': 'human_evaluator', 'text': 'What else do you enjoy doing?'},\n",
       " {'speaker': 'model', 'text': 'I like to play with my friends.'},\n",
       " {'speaker': 'human_evaluator',\n",
       "  'text': 'That sounds fun. I like watching old war documentaries'},\n",
       " {'speaker': 'model', 'text': 'I like to play the game of duty.'},\n",
       " {'speaker': 'human_evaluator',\n",
       "  'text': 'I enjoy eating cheeseburgers. Do you enjoy them aswel?'},\n",
       " {'speaker': 'model', 'text': 'I do. I love them.'},\n",
       " {'speaker': 'human_evaluator',\n",
       "  'text': 'I also volunteer at a soup kitchen. Do you do any volunteer work?'},\n",
       " {'speaker': 'model',\n",
       "  'text': 'I do. I volunteer at the local department store.'},\n",
       " {'speaker': 'human_evaluator',\n",
       "  'text': 'That sounds like fun. What do you do for work?'},\n",
       " {'speaker': 'model', 'text': 'I am a student. I work at a grocery store.'}]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conversation_table[conversation_table.model_name == 'greedy_model']['dialog'].loc[768]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## convert conversation table into context response pairs. \n",
    "## keep some identifier for the original row: assignment_id_hashed, evaluator_id_hashed\n",
    "## response should start with speaker:model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'speaker': 'model', 'text': 'Hi how are you today'},\n",
       " {'speaker': 'human_evaluator', 'text': \"I'm good how are you ?\"},\n",
       " {'speaker': 'model', 'text': 'I am good do you work'},\n",
       " {'speaker': 'human_evaluator',\n",
       "  'text': 'Yes I work at a tech company . WHat do you do ?'},\n",
       " {'speaker': 'model', 'text': \"I'm a professional wrestler\"},\n",
       " {'speaker': 'human_evaluator', 'text': \"oh wow that's really cool \"},\n",
       " {'speaker': 'model', 'text': 'What is your favorite color'},\n",
       " {'speaker': 'human_evaluator', 'text': 'I like red , you ?'},\n",
       " {'speaker': 'model', 'text': 'Yellow is my favorite'},\n",
       " {'speaker': 'human_evaluator', 'text': 'cool, do you like coffee ?'},\n",
       " {'speaker': 'model', 'text': 'Yes I do I love it'},\n",
       " {'speaker': 'human_evaluator', 'text': 'me too '}]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conversation_table.loc[1870]['dialog']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_cont_resp_pairs(dial):\n",
    "    pairs = []\n",
    "    context = []\n",
    "    response = ''\n",
    "    for di,d in enumerate(dial):\n",
    "        if d['speaker'] == 'model':\n",
    "            if len(context) > 0:\n",
    "                response = d['text']\n",
    "                pairs.append(('\\n'.join(context), response))\n",
    "        context.append(d['text'])\n",
    "    return pairs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('Hello! how are you?',\n",
       "  \"I'm doing well. Just got done playing some game of duty.\"),\n",
       " (\"Hello! how are you?\\nI'm doing well. Just got done playing some game of duty.\\nWhat else do you enjoy doing?\",\n",
       "  'I like to play with my friends.'),\n",
       " (\"Hello! how are you?\\nI'm doing well. Just got done playing some game of duty.\\nWhat else do you enjoy doing?\\nI like to play with my friends.\\nThat sounds fun. I like watching old war documentaries\",\n",
       "  'I like to play the game of duty.'),\n",
       " (\"Hello! how are you?\\nI'm doing well. Just got done playing some game of duty.\\nWhat else do you enjoy doing?\\nI like to play with my friends.\\nThat sounds fun. I like watching old war documentaries\\nI like to play the game of duty.\\nI enjoy eating cheeseburgers. Do you enjoy them aswel?\",\n",
       "  'I do. I love them.'),\n",
       " (\"Hello! how are you?\\nI'm doing well. Just got done playing some game of duty.\\nWhat else do you enjoy doing?\\nI like to play with my friends.\\nThat sounds fun. I like watching old war documentaries\\nI like to play the game of duty.\\nI enjoy eating cheeseburgers. Do you enjoy them aswel?\\nI do. I love them.\\nI also volunteer at a soup kitchen. Do you do any volunteer work?\",\n",
       "  'I do. I volunteer at the local department store.'),\n",
       " (\"Hello! how are you?\\nI'm doing well. Just got done playing some game of duty.\\nWhat else do you enjoy doing?\\nI like to play with my friends.\\nThat sounds fun. I like watching old war documentaries\\nI like to play the game of duty.\\nI enjoy eating cheeseburgers. Do you enjoy them aswel?\\nI do. I love them.\\nI also volunteer at a soup kitchen. Do you do any volunteer work?\\nI do. I volunteer at the local department store.\\nThat sounds like fun. What do you do for work?\",\n",
       "  'I am a student. I work at a grocery store.')]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_cont_resp_pairs(conversation_table.loc[768]['dialog'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "ref_rows = {'context': [], 'response': [], 'assignment_id_hashed': [], 'evaluator_id_hashed': []}\n",
    "for i,row in conversation_table.iterrows():\n",
    "    pairs = get_cont_resp_pairs(row['dialog'])\n",
    "    for pair in pairs:\n",
    "        ref_rows['context'].append(pair[0].lower())\n",
    "        ref_rows['response'].append(pair[1].lower())\n",
    "        ref_rows['assignment_id_hashed'].append(row['assignment_id_hashed'])\n",
    "        ref_rows['evaluator_id_hashed'].append(row['evaluator_id_hashed'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "ref_pd = pd.DataFrame(ref_rows)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 326,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "18265"
      ]
     },
     "execution_count": 326,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(ref_pd)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bayesian Score computation (Kulikov, 2019)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# map back from model name to type/bucket\n",
    "rmap = {}\n",
    "for _, row in conversation_table.iterrows():\n",
    "    model_name = row['model_name']\n",
    "    if model_name not in rmap:\n",
    "        rmap[model_name] = (row['model_bucket'], row['model_type'])\n",
    "\n",
    "def infer_single_scores(table, fieldname, num_samples=100, warmup_steps=50):\n",
    "    import torch\n",
    "    torch.set_num_threads(1)\n",
    "    t0 = time.time()\n",
    "    observed_single_scores = torch.Tensor(list(table[fieldname]))\n",
    "    num_likert_options = table[fieldname].nunique()\n",
    "    \n",
    "    def model_single_score(table):\n",
    "        '''\n",
    "        prior for each algorithm a\n",
    "        p(m_a) = Unif(1, num_likert_options)\n",
    "\n",
    "        prior for each single score given the models sampled mean\n",
    "        p(s_a) = N(s_a|m_a, 1^2)\n",
    "        '''\n",
    "        zm = []\n",
    "        for mi in range(table.model_name.nunique()):\n",
    "            mu_ = pyro.sample(\"model-mean-{}\".format(mi), dist.Uniform(1., float(num_likert_options)))\n",
    "            zm.append(pyro.sample(\"model-{}\".format(mi), dist.Normal(mu_, 1.)))\n",
    "\n",
    "        '''\n",
    "        prior for each turker's bias\n",
    "        p(s_t) = N(s_t | 0, 1^2) - i.e., no bias from each turker a priori\n",
    "        '''\n",
    "        tm = []\n",
    "        for ti in range(table.evaluator_id_hashed.nunique()):\n",
    "            tm.append(pyro.sample(\"turker-mean-{}\".format(ti), dist.Normal(0., 1.)))\n",
    "\n",
    "        '''\n",
    "        likelihood mean for each score (algorithm, score)\n",
    "        p(s|a, t) = N(s| s_a + s_t, 1^2)\n",
    "        '''\n",
    "        mu = []\n",
    "        for worker_id, model_id in zip(table.evaluator_id_hashed.cat.codes, table.model_name.cat.codes):\n",
    "            mu.append(zm[model_id] + tm[worker_id])\n",
    "        mu_ = torch.stack(mu)\n",
    "\n",
    "        return pyro.sample(\"scores\", dist.Normal(mu_, 1.))\n",
    "\n",
    "    single_score_condition = pyro.condition(\n",
    "            model_single_score, \n",
    "            data={'scores': observed_single_scores}\n",
    "    )\n",
    "    nuts_kernel = pyro.infer.mcmc.NUTS(single_score_condition, adapt_step_size=True, step_size=0.1)\n",
    "    mcmc_run = pyro.infer.mcmc.MCMC(nuts_kernel, num_samples=num_samples, warmup_steps=warmup_steps).run(table)\n",
    "    \n",
    "    score_marginal = pyro.infer.EmpiricalMarginal(\n",
    "        mcmc_run,\n",
    "        sites=[\"model-{}\".format(mi) for mi in range(table.model_name.nunique())]\n",
    "    )\n",
    "    print('time taken: ', time.time()-t0)\n",
    "    \n",
    "    \n",
    "    df = pd.DataFrame({\n",
    "        'model_name': table.model_name.cat.categories,\n",
    "        fieldname + '_mean': score_marginal.mean,\n",
    "        fieldname + '_std': score_marginal.stddev\n",
    "    })\n",
    "    df['model_bucket'] = [rmap[mn][0] for mn in df['model_name']]\n",
    "    df['model_type'] = [rmap[mn][1] for mn in df['model_name']]\n",
    "\n",
    "    return df.set_index('model_name')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following step runs bayesian analysis for each of the measures in parallel. You will want to decrease the number of CPUs down to what your computer has. The process takes about 45 minutes to compute.\n",
    "\n",
    "Due to the stochastic nature of the process, you will get slightly different numbers than the ones reported in the paper. For convenience, we've included the ones from the paper to avoid this computation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 520,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "def get_bayesian_cor_scores(run_model=False):\n",
    "    if run_model:\n",
    "        # run all the scores in parallel and join the results into one table\n",
    "\n",
    "        # torch freaks out if we try to multiprocess with cuda\n",
    "        import os\n",
    "        os.environ['CUDA_VISIBLE_DEVICES'] = ''\n",
    "        # keep each of the threads from stomping all over each other\n",
    "        torch.set_num_threads(1)\n",
    "\n",
    "        from multiprocessing import Pool\n",
    "\n",
    "        with Pool(20) as pool:\n",
    "            args = [(conversation_table, measure) for measure in LIKERT_METRICS]\n",
    "            results_bayes = pd.concat(pool.starmap(infer_single_scores, args), axis=1)\n",
    "    else:\n",
    "        bayes_data = '''\n",
    "        model_name,fluency_mean,fluency_std,enjoy_mean,enjoy_std,turing_mean,turing_std,make_sense_mean,make_sense_std,inquisitive_mean,inquisitive_std,interest_mean,interest_std,avoid_rep_mean,avoid_rep_std,listen_mean,listen_std,model_type,model_bucket\n",
    "    baseline_model,3.0914075,0.12956531,2.2892258,0.10683544,1.7131442,0.13138857,3.345498,0.12607269,2.4151554,0.10993455,2.2880845,0.1420951,2.0833843,0.10571904,2.4713647,0.117826685,baseline,\n",
    "    greedy_model,3.0280933,0.10491706,2.2354803,0.107257016,1.7464366,0.11817865,3.233947,0.10319092,1.9487315,0.102352,2.290578,0.12559834,2.0841112,0.103310645,2.6239514,0.10315127,greedy,\n",
    "    human_eval,3.3617704,0.11792514,3.0442827,0.11493982,3.350734,0.113348715,3.681538,0.10894737,2.435449,0.11926911,2.9190917,0.111999035,2.790789,0.11871563,3.3168595,0.12881786,human,\n",
    "    inquisitive_model_ct_setting00,3.5139592,0.0967675,2.83884,0.12611441,2.4591095,0.10882996,3.4932463,0.119373344,2.3637252,0.09294756,2.7555757,0.0927458,2.8722908,0.11721186,3.1028264,0.09619383,inquisitive-ct,000\n",
    "    inquisitive_model_ct_setting01,3.4174926,0.10161546,2.876931,0.10879339,2.458943,0.11697992,3.549381,0.09925039,2.4650643,0.10826402,2.7947018,0.12823951,2.816174,0.109138474,3.138592,0.11414942,inquisitive-ct,001\n",
    "    inquisitive_model_ct_setting04,3.466806,0.11303423,2.8775582,0.09969538,2.4028935,0.09169132,3.5896442,0.09606155,2.5327814,0.12658662,2.830341,0.1268796,2.783972,0.122446634,3.2362282,0.10941788,inquisitive-ct,004\n",
    "    inquisitive_model_ct_setting07,3.5384107,0.0939107,2.9888,0.105407335,2.3539126,0.11059914,3.474378,0.10486635,2.6636062,0.119133025,2.9173691,0.117349364,2.8096406,0.1004062,3.1082954,0.09788587,inquisitive-ct,007\n",
    "    inquisitive_model_ct_setting10,3.5227299,0.12035936,2.8661678,0.11492953,2.3464699,0.11804114,3.5265877,0.12227977,2.6337807,0.11855745,2.6558821,0.10226053,2.6730306,0.12550566,2.9447966,0.109267645,inquisitive-ct,010\n",
    "    inquisitive_model_ct_setting10_better,3.420784,0.12406063,2.7386124,0.09353278,2.1931195,0.13429101,3.450729,0.13293721,2.7863073,0.111801624,2.7376783,0.111998275,2.6759079,0.11800891,2.9955559,0.12491749,inquisitive-ct,011\n",
    "    interesting_nidf_model_bfw_setting_04,3.4463646,0.13350625,2.9587107,0.12454782,2.4409015,0.11551093,3.5361714,0.10621155,2.5572858,0.08596149,2.9410334,0.11254286,2.840033,0.10043833,3.1984024,0.09754762,nidf_bfw,004\n",
    "    interesting_nidf_model_bfw_setting_06,3.3424668,0.09006725,2.9053752,0.097677216,2.3083365,0.11380971,3.4079204,0.116244145,2.5260317,0.11763178,2.9328823,0.119851254,2.8065453,0.09133208,3.0851126,0.09660692,nidf_bfw,006\n",
    "    interesting_nidf_model_bfw_setting_08,2.5359437,0.11638034,2.3928986,0.12092932,1.8046832,0.12724209,2.8720655,0.11232278,2.0019753,0.097686924,2.4934795,0.11799547,2.7020395,0.10965349,2.4661036,0.10257839,nidf_bfw,008\n",
    "    interesting_nidf_model_bfw_setting_minus_04,3.3714204,0.102036715,2.7597752,0.11824679,2.3626835,0.10884776,3.5159597,0.108949505,2.4624243,0.110119894,2.621436,0.11877783,2.8268204,0.10388491,3.1394558,0.094563544,nidf_bfw,-04\n",
    "    interesting_nidf_model_bfw_setting_minus_10,3.188879,0.12436492,2.4090981,0.12411542,2.1500096,0.1128042,3.2827806,0.11560405,2.2804828,0.1344875,2.351124,0.121478364,2.7615042,0.11347523,2.886509,0.10520854,nidf_bfw,-10\n",
    "    interesting_nidf_model_ct_setting0,3.4431455,0.11906912,2.9258559,0.088641025,2.3770676,0.11373025,3.6066294,0.10947405,2.5631077,0.122011304,2.8404248,0.12301018,2.788957,0.096725754,3.1246283,0.12607282,nidf_ct,000\n",
    "    interesting_nidf_model_ct_setting2,3.389251,0.13110022,2.7392132,0.11038114,2.3120344,0.12750752,3.4655373,0.10412794,2.5648835,0.12711605,2.7370658,0.11907516,2.781668,0.119834706,2.9915996,0.10762761,nidf_ct,002\n",
    "    interesting_nidf_model_ct_setting4,3.4366372,0.13992369,2.795612,0.1315953,2.3171048,0.13324456,3.4565122,0.12817147,2.5133355,0.11968733,2.7807126,0.15289873,2.8235056,0.098016724,3.085936,0.12654012,nidf_ct,004\n",
    "    interesting_nidf_model_ct_setting7,3.4267304,0.11089812,2.91389,0.13190652,2.4536288,0.095813766,3.5490704,0.109761,2.4927237,0.11294685,2.8147955,0.121712856,2.812701,0.12364607,3.1512172,0.12203714,nidf_ct,007\n",
    "    interesting_nidf_model_ct_setting9,3.412071,0.11562859,2.7782488,0.09793116,2.3476672,0.13273424,3.5078192,0.123752415,2.2768347,0.10855014,2.791978,0.10660777,2.7957282,0.1290695,2.9055562,0.10904883,nidf_ct,009\n",
    "    repetition_model_setting05,3.351136,0.12485303,2.5351877,0.11802207,2.131792,0.11483303,3.4810336,0.11032247,2.633227,0.11012598,2.5645876,0.11315254,2.6154072,0.10460297,2.925591,0.11182627,repetition-bfw,2gram=005\n",
    "    repetition_model_setting12,3.403044,0.12195272,2.818021,0.12892646,2.2676632,0.12410987,3.5339706,0.138091,2.5448759,0.08994768,2.7566962,0.1035419,2.7767127,0.09308626,3.0484388,0.110010505,repetition-bfw,2gram=012\n",
    "    repetition_model_setting35,3.562523,0.102048345,2.934537,0.09988889,2.4272463,0.107678376,3.6171374,0.12241684,2.47073,0.10978695,2.8317904,0.09658501,2.826065,0.11344152,3.1369538,0.10203378,repetition-bfw,2gram=035\n",
    "    repetition_model_setting35_settinginf,3.5319638,0.10463153,2.8183765,0.12304195,2.3975408,0.10982047,3.5779467,0.14254314,2.62365,0.1294855,2.843848,0.12449841,2.8550565,0.12010231,3.103962,0.109406754,repetition-bfw,2gram=035 stop=inf\n",
    "    repetition_model_settinginf,3.4903002,0.12059033,2.874239,0.1358771,2.3232102,0.1309247,3.5929792,0.116237886,2.555877,0.10959805,2.7502258,0.11547001,2.7356553,0.114207745,3.127551,0.11856033,repetition-bfw,2gram=999\n",
    "    responsiveness_model_bfw_setting_00,3.5512834,0.087748855,2.9677284,0.11360309,2.6231253,0.10570546,3.7040608,0.10423539,2.4773164,0.100650065,2.8829346,0.11615346,2.8743443,0.115949884,3.2102318,0.09360844,responsive,000\n",
    "    responsiveness_model_bfw_setting_05,3.3485637,0.12327652,2.8315463,0.09458454,2.398068,0.11602921,3.407857,0.115517996,2.506635,0.12592898,2.803587,0.12773731,2.7892473,0.10135295,3.1325374,0.11519348,responsive,005\n",
    "    responsiveness_model_bfw_setting_10,2.9316738,0.11106751,2.4206061,0.11819174,1.9520544,0.122140095,3.1197522,0.09797036,2.1957242,0.11881691,2.5557888,0.11651121,2.7393322,0.10824369,2.9012582,0.12356319,responsive,010\n",
    "    responsiveness_model_bfw_setting_13,2.397673,0.08854504,2.0601296,0.10622171,1.7427883,0.11313255,2.6316466,0.10069777,2.0721536,0.10736318,2.2488096,0.11831513,2.6300664,0.12108313,2.486768,0.1420562,responsive,013\n",
    "    responsiveness_model_bfw_setting_minus_10,3.2559185,0.10904983,2.4534166,0.122486725,1.9554462,0.10266632,3.1158836,0.11584097,2.312914,0.1201041,2.473524,0.119496204,2.7670064,0.11804217,2.7265708,0.108767524,responsive,-10\n",
    "        '''.strip()\n",
    "        import io\n",
    "        results_bayes = pd.read_csv(io.StringIO(bayes_data)).set_index('model_name')\n",
    "\n",
    "    results_bayes.index = results_bayes.index.str.strip()\n",
    "    # things that have two names\n",
    "    results_bayes.loc['repetition++'] = results_bayes.loc['responsiveness_model_bfw_setting_00']\n",
    "    results_bayes.loc['repetition++','model_type'] = 'repetition-bfw'\n",
    "\n",
    "    results_bayes.loc['interesting_nidf_model_bfw_setting_00'] = results_bayes.loc['repetition_model_setting35_settinginf']\n",
    "    results_bayes.loc['interesting_nidf_model_bfw_setting_00','model_type'] = 'nidf_bfw'\n",
    "    results_bayes.loc['interesting_nidf_model_bfw_setting_00','model_bucket'] = '00'\n",
    "\n",
    "    results_bayes.loc['inquisitive_model_ct_setting10_better', 'model_bucket'] = '12'\n",
    "\n",
    "    # prep results\n",
    "    altered = results_bayes.copy().reset_index()\n",
    "    altered[\"model_bucket\"] = altered.model_bucket.astype('category')\n",
    "    return results_bayes, altered\n",
    "\n",
    "results_bayes, altered = get_bayesian_cor_scores()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 468,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_corr(key, df, method='spearman'):\n",
    "    scores = []\n",
    "    for col in df.columns:\n",
    "        if '_mean' in col:\n",
    "            scores.append((col.split('_mean')[0], df[col].corr(df[key], method='spearman')))\n",
    "    return scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dialog Metric correlation with Uncalibrated and Calibrated Scores\n",
    "\n",
    "- Run inference using inference.sh script\n",
    "- Read controllable_dialogs.csv\n",
    "- combine the files\n",
    "- add the column to bayesian calibration\n",
    "- get the `altered` file\n",
    "- plot the uncalibrated correlation\n",
    "- print the calibrated correlation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 540,
   "metadata": {},
   "outputs": [],
   "source": [
    "raw_scores = pd.read_csv('output_data/controllable_dialogs.csv')\n",
    "conv_scores = raw_scores.groupby(['assignment_id_hashed']).mean().reset_index()\n",
    "conv_dial_scores = pd.merge(conversation_table, conv_scores, on='assignment_id_hashed')\n",
    "\n",
    "def get_calibrated_score(col_name):\n",
    "    scores = conv_dial_scores.groupby(['model_name']).mean()[col_name]\n",
    "    results_bayes, _ = get_bayesian_cor_scores()\n",
    "    results_bayes[col_name] = scores\n",
    "    altered = results_bayes.copy().reset_index()\n",
    "    altered[\"model_bucket\"] = altered.model_bucket.astype('category')\n",
    "    cor_scores = get_corr(col_name,altered)\n",
    "    mean_score = np.mean([sc[-1] for sc in cor_scores])\n",
    "    return cor_scores, mean_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 541,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['model_name', 'model_type', 'model_bucket', 'dialog',\n",
       "       'assignment_id_hashed', 'evaluator_id_hashed', 'enjoy', 'interest',\n",
       "       'listen', 'turing', 'avoid_rep', 'make_sense', 'fluency',\n",
       "       'persona_guess', 'inquisitive', 'Unnamed: 0', 'Unnamed: 0.1',\n",
       "       'Unnamed: 0.1.1', 'Unnamed: 0.1.1.1', 'Unnamed: 0.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1', 'score_19334597',\n",
       "       'score_20330191', 'score_20391351', 'score_0', 'bertnli_all_score_0',\n",
       "       'infersent_all_score_0', 'infersent_all_score_20495233',\n",
       "       'bertnli_all_score_20497378', 'na_all_score_20488119',\n",
       "       'na_all_context_score_20488122', 'na_all_context_score_20488123',\n",
       "       'ruber_all_score_20688482'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 541,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 530,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([('fluency', 0.36995073891625607),\n",
       "  ('enjoy', 0.23152709359605908),\n",
       "  ('turing', 0.09458128078817732),\n",
       "  ('make_sense', 0.20788177339901473),\n",
       "  ('inquisitive', 0.7275862068965516),\n",
       "  ('interest', 0.24039408866995068),\n",
       "  ('avoid_rep', -0.0354679802955665),\n",
       "  ('listen', 0.11182266009852214)],\n",
       " 0.24353448275862064)"
      ]
     },
     "execution_count": 530,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_calibrated_score('na_all_score_20488119')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 539,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([('fluency', 0.40788177339901466),\n",
       "  ('enjoy', 0.1133004926108374),\n",
       "  ('turing', 0.09507389162561573),\n",
       "  ('make_sense', 0.2157635467980295),\n",
       "  ('inquisitive', 0.6443349753694579),\n",
       "  ('interest', 0.05566502463054186),\n",
       "  ('avoid_rep', -0.04581280788177339),\n",
       "  ('listen', 0.07684729064039407)],\n",
       " 0.19538177339901475)"
      ]
     },
     "execution_count": 539,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_calibrated_score('na_all_context_score_20488123')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 529,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([('fluency', 0.44334975369458113),\n",
       "  ('enjoy', 0.19211822660098518),\n",
       "  ('turing', 0.12906403940886696),\n",
       "  ('make_sense', 0.25615763546798026),\n",
       "  ('inquisitive', 0.5980295566502462),\n",
       "  ('interest', 0.13546798029556648),\n",
       "  ('avoid_rep', -0.038916256157635463),\n",
       "  ('listen', 0.12364532019704431)],\n",
       " 0.2298645320197044)"
      ]
     },
     "execution_count": 529,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_calibrated_score('bertnli_all_score_20497378')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 528,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([('fluency', 0.24630541871921177),\n",
       "  ('enjoy', 0.09113300492610836),\n",
       "  ('turing', -0.10837438423645318),\n",
       "  ('make_sense', 0.004926108374384235),\n",
       "  ('inquisitive', 0.5891625615763546),\n",
       "  ('interest', 0.11921182266009848),\n",
       "  ('avoid_rep', -0.11773399014778321),\n",
       "  ('listen', -0.08571428571428569)],\n",
       " 0.0923645320197044)"
      ]
     },
     "execution_count": 528,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_calibrated_score('infersent_all_score_20495233')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 545,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\\begin{table}[]\n",
      "\\centering\n",
      "\\resizebox{\\textwidth}{!}{%\n",
      "\\begin{tabular}{|l|l|l|l|l|}\n",
      "\\hline\n",
      "                     & RUBER & InferSent & DistilBERT NLI & \\dtm{}          \\\\ \\cline{2-4} \n",
      "\n",
      "Fluency              & 0.322 & 0.246     & 0.443 & 0.37           \\\\\n",
      "Engagingness              & 0.204 & 0.091     & 0.192 & 0.232           \\\\\n",
      "Humanness              & 0.057 & -0.108     & 0.129 & 0.095           \\\\\n",
      "Making Sense              & 0.0 & 0.005     & 0.256 & 0.208           \\\\\n",
      "Inquisitiveness              & 0.583 & 0.589     & 0.598 & 0.728           \\\\\n",
      "Interestingness              & 0.275 & 0.119     & 0.135 & 0.24           \\\\\n",
      "Avoiding Repetition              & 0.093 & -0.118     & -0.039 & -0.035           \\\\\n",
      "Listening              & 0.061 & -0.086     & 0.124 & 0.112           \\\\\n",
      "Mean                 & 0.199 & 0.092     & 0.23          & 0.244  \\\\ \\hline \n",
      "\n",
      "\\end{tabular}%\n",
      "}\n",
      "\\caption{}\n",
      "\\label{tab:my-table}\n",
      "\\end{table}\n",
      "\n"
     ]
    }
   ],
   "source": [
    "inf_scores,inf_mean = get_calibrated_score('infersent_all_score_20495233')\n",
    "bert_scores,bert_mean = get_calibrated_score('bertnli_all_score_20497378')\n",
    "m_scores, m_mean = get_calibrated_score('na_all_score_20488119')\n",
    "ruber_scores,ruber_mean = get_calibrated_score('ruber_all_score_20688482')\n",
    "\n",
    "header = '''\n",
    "\\\\begin{table}[]\n",
    "\\\\centering\n",
    "\\\\resizebox{\\\\textwidth}{!}{%\n",
    "\\\\begin{tabular}{|l|l|l|l|l|}\n",
    "\\\\hline\n",
    "                     & RUBER & InferSent & DistilBERT NLI & \\dtm{}          \\\\\\\\ \\cline{2-4} \n",
    "'''\n",
    "table = header + \"\\n\"\n",
    "for i in range(len(inf_scores)):\n",
    "    row = \"{}              & {} & {}     & {} & {}           \\\\\\\\\".format(\n",
    "        nice_metric_names[inf_scores[i][0]],\n",
    "        round(ruber_scores[i][-1],3),\n",
    "        round(inf_scores[i][-1],3),\n",
    "        round(bert_scores[i][-1],3),\n",
    "        round(m_scores[i][-1],3)\n",
    "    )\n",
    "    table += row + \"\\n\"\n",
    "table += \"Mean                 & {} & {}     & {}          & {}  \\\\\\\\ \\hline \\n\".format(round(ruber_mean,3), \n",
    "                                                                                        round(inf_mean,3),\n",
    "                                                                             round(bert_mean,3), \n",
    "                                                                             round(m_mean,3))\n",
    "footer = '''\n",
    "\\\\end{tabular}%\n",
    "}\n",
    "\\\\caption{}\n",
    "\\\\label{tab:my-table}\n",
    "\\\\end{table}\n",
    "'''\n",
    "table += footer\n",
    "print(table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 531,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'enjoy': 'Engagingness',\n",
       " 'interest': 'Interestingness',\n",
       " 'inquisitive': 'Inquisitiveness',\n",
       " 'listen': 'Listening',\n",
       " 'avoid_rep': 'Avoiding Repetition',\n",
       " 'fluency': 'Fluency',\n",
       " 'make_sense': 'Making Sense',\n",
       " 'turing': 'Humanness',\n",
       " 'persona_guess': 'Persona retrieval'}"
      ]
     },
     "execution_count": 531,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nice_metric_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 486,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "model_name\n",
       "baseline_model                                0.9538\n",
       "greedy_model                                  0.8436\n",
       "human_eval                                    0.8586\n",
       "inquisitive_model_ct_setting00                0.9323\n",
       "inquisitive_model_ct_setting01                0.9347\n",
       "inquisitive_model_ct_setting04                0.9343\n",
       "inquisitive_model_ct_setting07                0.9489\n",
       "inquisitive_model_ct_setting10                0.9507\n",
       "inquisitive_model_ct_setting10_better         0.9540\n",
       "interesting_nidf_model_bfw_setting_04         0.9255\n",
       "interesting_nidf_model_bfw_setting_06         0.9191\n",
       "interesting_nidf_model_bfw_setting_08         0.7591\n",
       "interesting_nidf_model_bfw_setting_minus_04   0.9226\n",
       "interesting_nidf_model_bfw_setting_minus_10   0.8243\n",
       "interesting_nidf_model_ct_setting0            0.9073\n",
       "interesting_nidf_model_ct_setting2            0.9346\n",
       "interesting_nidf_model_ct_setting4            0.9320\n",
       "interesting_nidf_model_ct_setting7            0.9359\n",
       "interesting_nidf_model_ct_setting9            0.9379\n",
       "repetition_model_setting05                    0.9528\n",
       "repetition_model_setting12                    0.9401\n",
       "repetition_model_setting35                    0.9384\n",
       "repetition_model_setting35_settinginf         0.9277\n",
       "repetition_model_settinginf                   0.9346\n",
       "responsiveness_model_bfw_setting_00           0.9265\n",
       "responsiveness_model_bfw_setting_05           0.9376\n",
       "responsiveness_model_bfw_setting_10           0.8960\n",
       "responsiveness_model_bfw_setting_13           0.8283\n",
       "responsiveness_model_bfw_setting_minus_10     0.8873\n",
       "repetition++                                  0.9265\n",
       "interesting_nidf_model_bfw_setting_00         0.9277\n",
       "Name: agg_scores_bertnli_all_score_20497378, dtype: float64"
      ]
     },
     "execution_count": 486,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_bayes['agg_scores_bertnli_all_score_20497378']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 449,
   "metadata": {},
   "outputs": [],
   "source": [
    "conv_scores = ref_pd.groupby(['assignment_id_hashed']).mean().reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 450,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3316"
      ]
     },
     "execution_count": 450,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(conv_scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 451,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3316"
      ]
     },
     "execution_count": 451,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(conversation_table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 452,
   "metadata": {},
   "outputs": [],
   "source": [
    "conv_dial_scores = pd.merge(conversation_table, conv_scores, on='assignment_id_hashed')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.09344302486399793"
      ]
     },
     "execution_count": 152,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.enjoy.corr(conv_dial_scores.score_scaled, method='pearson')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0004669133812781183"
      ]
     },
     "execution_count": 153,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.avoid_rep.corr(conv_dial_scores.score_scaled, method='pearson')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.09397982699407251"
      ]
     },
     "execution_count": 154,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.interest.corr(conv_dial_scores.score_scaled, method='pearson')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['model_name', 'model_type', 'model_bucket', 'dialog',\n",
       "       'assignment_id_hashed', 'evaluator_id_hashed', 'enjoy', 'interest',\n",
       "       'listen', 'turing', 'avoid_rep', 'make_sense', 'fluency',\n",
       "       'persona_guess', 'inquisitive', 'Unnamed: 0', 'Unnamed: 0.1',\n",
       "       'Unnamed: 0.1.1', 'Unnamed: 0.1.1.1', 'Unnamed: 0.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1', 'score_19334597',\n",
       "       'score_20330191', 'score_20391351', 'score_0', 'bertnli_all_score_0',\n",
       "       'infersent_all_score_0'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 282,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_types = ['enjoy','interest','listen','avoid_rep','make_sense','fluency','inquisitive','turing']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "enjoy 0.08602107366129597\n",
      "interest 0.08821480748281132\n",
      "listen 0.0783164490887421\n",
      "avoid_rep 0.07118716983823027\n",
      "make_sense 0.09135284281407176\n",
      "fluency 0.10563288603487597\n",
      "inquisitive 0.12936325812927255\n"
     ]
    }
   ],
   "source": [
    "scores = []\n",
    "for mtype in test_types:\n",
    "    print(mtype,conv_dial_scores[mtype].corr(conv_dial_scores.score_0, method='kendall'))\n",
    "    scores.append(conv_dial_scores[mtype].corr(conv_dial_scores.score_0, method='kendall'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.09286978386418569"
      ]
     },
     "execution_count": 222,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "enjoy 0.10316885644372972\n",
      "interest 0.10598186323046807\n",
      "listen 0.04510542416643879\n",
      "avoid_rep 0.03612531684214653\n",
      "make_sense 0.07304776439309989\n",
      "fluency 0.07427830606483474\n",
      "inquisitive 0.1990903307913153\n"
     ]
    }
   ],
   "source": [
    "scores = []\n",
    "for mtype in test_types:\n",
    "    print(mtype,conv_dial_scores[mtype].corr(conv_dial_scores.bertnli_all_score_0, method='kendall'))\n",
    "    scores.append(conv_dial_scores[mtype].corr(conv_dial_scores.bertnli_all_score_0, method='kendall'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.09097112313314759"
      ]
     },
     "execution_count": 224,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_scores(df, key, method='spearman'):\n",
    "    scores = []\n",
    "    for mtype in test_types:\n",
    "        scores.append((mtype, df[mtype].corr(df[key], method='spearman')))\n",
    "    return scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 340,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_scores_scipy(df, key, method='spearman'):\n",
    "    scores = []\n",
    "    for mtype in test_types:\n",
    "        if method  == 'pearson':\n",
    "            scores.append(pearsonr(df[mtype].to_list(),df[key].to_list()))\n",
    "        if method  == 'spearman':\n",
    "            scores.append(spearmanr(df[mtype].to_list(),df[key].to_list()))\n",
    "    return scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 344,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[SpearmanrResult(correlation=0.14212230957951297, pvalue=1.988749732661066e-16),\n",
       " SpearmanrResult(correlation=0.13446650594302947, pvalue=7.507101652775365e-15),\n",
       " SpearmanrResult(correlation=0.09979809372204594, pvalue=8.458343194094414e-09),\n",
       " SpearmanrResult(correlation=0.033006808013577, pvalue=0.05736897778294621),\n",
       " SpearmanrResult(correlation=0.16265189202841243, pvalue=4.278094647468193e-21),\n",
       " SpearmanrResult(correlation=0.14406570829298312, pvalue=7.660153673854258e-17),\n",
       " SpearmanrResult(correlation=0.1718745200004264, pvalue=2.106864382289187e-23),\n",
       " SpearmanrResult(correlation=0.08570966344905939, pvalue=7.700707190477843e-07)]"
      ]
     },
     "execution_count": 344,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_scores_scipy(conv_dial_scores, 'na_all_score_20488119', 'spearman')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 345,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[SpearmanrResult(correlation=0.09314424659799622, pvalue=7.728811175148333e-08),\n",
       " SpearmanrResult(correlation=0.07456471679464319, pvalue=1.72226741706858e-05),\n",
       " SpearmanrResult(correlation=0.05447142079015002, pvalue=0.001701811508273199),\n",
       " SpearmanrResult(correlation=0.056748160225267595, pvalue=0.001078408185388859),\n",
       " SpearmanrResult(correlation=0.06748737726565367, pvalue=0.00010057068076121396),\n",
       " SpearmanrResult(correlation=0.09130783834502358, pvalue=1.3870227757533182e-07),\n",
       " SpearmanrResult(correlation=0.14854457646891378, pvalue=8.082455163760797e-18),\n",
       " SpearmanrResult(correlation=0.062314826520988964, pvalue=0.00033002067669732035)]"
      ]
     },
     "execution_count": 345,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_scores_scipy(conv_dial_scores, 'infersent_all_score_20495233', 'spearman')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {},
   "outputs": [],
   "source": [
    "score = get_scores(conv_dial_scores, 'score_0','spearman')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 322,
   "metadata": {},
   "outputs": [],
   "source": [
    "def draw_bar(scores, type_labels):\n",
    "    labels = [nice_metric_names[s[0]] for s in scores[0]]\n",
    "    x = np.arange(len(labels))  # the label locations\n",
    "    width = 0.35  # the width of the bars\n",
    "\n",
    "    fig, ax = plt.subplots()\n",
    "    rects = []\n",
    "    for si,score in enumerate(scores):\n",
    "        raw_score = [s[1] for s in score]\n",
    "        if si < len(scores) / 2:\n",
    "            p = (len(scores) - (si))\n",
    "            if p == 3:\n",
    "                p = 4\n",
    "            position = x - (width/len(scores) * p)\n",
    "        else:\n",
    "            position = x + width/len(scores)\n",
    "        print(position)\n",
    "        rects.append(ax.bar(position, raw_score, width, label=type_labels[si]))\n",
    "    # Add some text for labels, title and custom x-axis tick labels, etc.\n",
    "    ax.set_ylabel('Correlation scores')\n",
    "    ax.set_title('Human evaluation correlation')\n",
    "    ax.set_xticks(x)\n",
    "    ax.set_xticklabels(labels)\n",
    "    ax.legend()\n",
    "    \n",
    "    def autolabel(rects):\n",
    "        \"\"\"Attach a text label above each bar in *rects*, displaying its height.\"\"\"\n",
    "        for rect in rects:\n",
    "            height = rect.get_height()\n",
    "            ax.annotate('{}'.format(height),\n",
    "                        xy=(rect.get_x() + rect.get_width() / 2, height),\n",
    "                        xytext=(0, 3),  # 3 points vertical offset\n",
    "                        textcoords=\"offset points\",\n",
    "                        ha='center', va='bottom')\n",
    "    \n",
    "#     for rect in rects:\n",
    "#         autolabel(rect)\n",
    "    fig.tight_layout()\n",
    "    plt.show()\n",
    "    \n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 393,
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 323,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0.46666667  0.53333333  1.53333333  2.53333333  3.53333333  4.53333333\n",
      "  5.53333333  6.53333333]\n",
      "[-0.23333333  0.76666667  1.76666667  2.76666667  3.76666667  4.76666667\n",
      "  5.76666667  6.76666667]\n",
      "[0.11666667 1.11666667 2.11666667 3.11666667 4.11666667 5.11666667\n",
      " 6.11666667 7.11666667]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mode = 'spearman'\n",
    "draw_bar([\n",
    "    get_scores(conv_dial_scores, 'score_0',mode),\n",
    "    get_scores(conv_dial_scores, 'bertnli_all_score_0',mode),\n",
    "    get_scores(conv_dial_scores, 'infersent_all_score_0',mode),\n",
    "],[\n",
    "    'Transition Function',\n",
    "    'DistilBERT NLI',\n",
    "    'InferSent'\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 336,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0.46666667  0.53333333  1.53333333  2.53333333  3.53333333  4.53333333\n",
      "  5.53333333  6.53333333]\n",
      "[-0.23333333  0.76666667  1.76666667  2.76666667  3.76666667  4.76666667\n",
      "  5.76666667  6.76666667]\n",
      "[0.11666667 1.11666667 2.11666667 3.11666667 4.11666667 5.11666667\n",
      " 6.11666667 7.11666667]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mode = 'spearman'\n",
    "draw_bar([\n",
    "    get_scores(conv_dial_scores, 'na_all_score_20488119',mode),\n",
    "    get_scores(conv_dial_scores, 'bertnli_all_score_20497378',mode),\n",
    "    get_scores(conv_dial_scores, 'infersent_all_score_20495233',mode),\n",
    "],[\n",
    "    'MAUDE',\n",
    "    'DistilBERT NLI',\n",
    "    'InferSent'\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 546,
   "metadata": {},
   "outputs": [],
   "source": [
    "hum_c_data = {'score':[],'metric':[],'model':[]}\n",
    "\n",
    "models = {'na_all_score_20488119':'MAUDE','bertnli_all_score_20497378':'DistilBERT-NLI',\n",
    "          'infersent_all_score_20495233':'InferSent','ruber_all_score_20688482':'RUBER'}\n",
    "#models = {'na_all_score_20488119':'MAUDE','bertnli_all_score_20497378':'DistilBERT-NLI','infersent_all_score_20495233':'InferSent',\n",
    "#          'na_all_context_score_20488123':'MAUDE_Context'}\n",
    "for md,model_name in models.items():\n",
    "    sc = get_scores(conv_dial_scores, md ,mode)\n",
    "    for si in sc:\n",
    "        hum_c_data['score'].append(si[1])\n",
    "        hum_c_data['metric'].append(nice_metric_names[si[0]])\n",
    "        hum_c_data['model'].append(model_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 547,
   "metadata": {},
   "outputs": [],
   "source": [
    "hum_c_df = pd.DataFrame(hum_c_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 548,
   "metadata": {},
   "outputs": [],
   "source": [
    "hum_c_df.columns = ['Correlation','Metric','Model']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 549,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib as mpl\n",
    "mpl.rcParams['figure.dpi'] = 300"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 555,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 3600x2400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set_context(\"paper\")\n",
    "fig = plt.figure(figsize=(12, 8))\n",
    "ax = plt.subplot()\n",
    "for label in (ax.get_xticklabels() + ax.get_yticklabels()):\n",
    "    label.set_fontname('Arial')\n",
    "    label.set_fontsize(16)\n",
    "    label.set_fontweight('semibold')\n",
    "plt.ticklabel_format(style='sci', axis='x', scilimits=(0,10))\n",
    "ax.xaxis.get_offset_text().set_fontsize(25)\n",
    "axis_font = {'fontname':'Arial', 'size':'25'}\n",
    "bar = sns.barplot(x=\"Correlation\", y=\"Metric\", hue=\"Model\", data=hum_c_df,\n",
    "                palette=\"colorblind\")\n",
    "# hatches = ['\\\\', '*', 'o']\n",
    "# hi = 0\n",
    "# for i,thisbar in enumerate(bar.patches):\n",
    "#     # Set a different hatch for each bar\n",
    "#     thisbar.set_hatch(hatches[hi])\n",
    "#     hi +=1\n",
    "#     if hi == len(hatches):\n",
    "#         hi = 0\n",
    "ax.legend(loc=(0.8, 0.5), prop={'size' : 18})\n",
    "ax.set_xlabel('Correlation',**axis_font)\n",
    "ax.set_ylabel('Metric', **axis_font)\n",
    "ax.set_title('Correlation on uncalibrated Human scores', **axis_font)\n",
    "plt.savefig('/private/home/koustuvs/mlp/latentDialogAnalysis/plot.png', bbox_inches = 'tight', pad_inches = 0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 333,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['model_name', 'model_type', 'model_bucket', 'dialog',\n",
       "       'assignment_id_hashed', 'evaluator_id_hashed', 'enjoy', 'interest',\n",
       "       'listen', 'turing', 'avoid_rep', 'make_sense', 'fluency',\n",
       "       'persona_guess', 'inquisitive', 'Unnamed: 0', 'Unnamed: 0.1',\n",
       "       'Unnamed: 0.1.1', 'Unnamed: 0.1.1.1', 'Unnamed: 0.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1', 'score_19334597', 'score_20330191',\n",
       "       'score_20391351', 'score_0', 'bertnli_all_score_0',\n",
       "       'infersent_all_score_0', 'infersent_all_score_20495233',\n",
       "       'bertnli_all_score_20497378', 'na_all_score_20488119'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 333,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.1189732718251231"
      ]
     },
     "execution_count": 227,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 338,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.stats import pearsonr, spearmanr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "pearsonr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "enjoy 0.13580481995467283\n",
      "interest 0.1393006585563305\n",
      "listen 0.05838597820259068\n",
      "avoid_rep 0.04480158539628829\n",
      "make_sense 0.09177380472560527\n",
      "fluency 0.09450006095820263\n",
      "inquisitive 0.2519755116232191\n"
     ]
    }
   ],
   "source": [
    "scores = []\n",
    "for mtype in test_types:\n",
    "    print(mtype,conv_dial_scores[mtype].corr(conv_dial_scores.bertnli_all_score_0, method='spearman'))\n",
    "    scores.append(conv_dial_scores[mtype].corr(conv_dial_scores., method='spearman'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.11664891705955846"
      ]
     },
     "execution_count": 229,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "enjoy 0.07868559748363084\n",
      "interest 0.07793339558703702\n",
      "listen 0.03928137487904862\n",
      "avoid_rep 0.0564992209425539\n",
      "make_sense 0.024904306313232687\n",
      "fluency 0.05728998164309008\n",
      "inquisitive 0.11793755549446987\n"
     ]
    }
   ],
   "source": [
    "scores = []\n",
    "for mtype in test_types:\n",
    "    print(mtype,conv_dial_scores[mtype].corr(conv_dial_scores.infersent_all_score_0, method='spearman'))\n",
    "    scores.append(conv_dial_scores[mtype].corr(conv_dial_scores.infersent_all_score_0, method='spearman'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.06464734747758043"
      ]
     },
     "execution_count": 238,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "enjoy 0.1516993625885021\n",
      "interest 0.14551864598229122\n",
      "listen 0.161296287333335\n",
      "turing 0.08362231919692306\n",
      "avoid_rep 0.10642082574249448\n",
      "make_sense 0.18540068709932822\n",
      "fluency 0.22530555577718\n",
      "persona_guess 0.04767130436690969\n",
      "inquisitive 0.2028516420469267\n"
     ]
    }
   ],
   "source": [
    "for mtype in ['enjoy','interest','listen','turing','avoid_rep','make_sense','fluency','persona_guess','inquisitive']:\n",
    "    print(mtype,conv_dial_scores[mtype].corr(conv_dial_scores.score_0, method='pearson'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "enjoy 0.16796941589470038\n",
      "interest 0.16409827496710805\n",
      "listen 0.12098318882941125\n",
      "turing 0.08243756457595186\n",
      "avoid_rep 0.09426008377325819\n",
      "make_sense 0.14570389385985527\n",
      "fluency 0.18655619203437482\n",
      "persona_guess -0.04772318971855941\n",
      "inquisitive 0.28793018069171794\n"
     ]
    }
   ],
   "source": [
    "for mtype in ['enjoy','interest','listen','turing','avoid_rep','make_sense','fluency','persona_guess','inquisitive']:\n",
    "    print(mtype,conv_dial_scores[mtype].corr(conv_dial_scores.bertnli_all_score_0, method='pearson'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Exploring the prediction of our model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Best predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 556,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/private/home/koustuvs/miniconda3/envs/dialog/lib/python3.7/site-packages/ipykernel_launcher.py:2: FutureWarning: \n",
      "The current behaviour of 'Series.argmax' is deprecated, use 'idxmax'\n",
      "instead.\n",
      "The behavior of 'argmax' will be corrected to return the positional\n",
      "maximum in the future. For now, use 'series.values.argmax' or\n",
      "'np.argmax(np.array(values))' to get the position of the maximum\n",
      "row.\n",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "2237"
      ]
     },
     "execution_count": 556,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## best score\n",
    "conv_dial_scores.na_all_score_20488119.argmax()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 588,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'enjoy': 'Engagingness',\n",
       " 'interest': 'Interestingness',\n",
       " 'inquisitive': 'Inquisitiveness',\n",
       " 'listen': 'Listening',\n",
       " 'avoid_rep': 'Avoiding Repetition',\n",
       " 'fluency': 'Fluency',\n",
       " 'make_sense': 'Making Sense',\n",
       " 'turing': 'Humanness',\n",
       " 'persona_guess': 'Persona retrieval'}"
      ]
     },
     "execution_count": 588,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nice_metric_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 594,
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_nice_scores(loc):\n",
    "    tr = \"\"\n",
    "    for nice_m,good_m in nice_metric_names.items():\n",
    "        tr += \"{} : {}, \".format(good_m, loc[nice_m])\n",
    "    print(tr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 595,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Engagingness : 3, Interestingness : 3, Inquisitiveness : 2, Listening : 3, Avoiding Repetition : 3, Fluency : 4, Making Sense : 4, Humanness : 3, Persona retrieval : 1, \n"
     ]
    }
   ],
   "source": [
    "print_nice_scores(conv_dial_scores.loc[2237])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 557,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "model_name                                  interesting_nidf_model_ct_setting7                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                  \n",
       "model_type                                  nidf_ct                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                             \n",
       "model_bucket                                007                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 \n",
       "dialog                                      [{'speaker': 'human_evaluator', 'text': 'How are you today?'}, {'speaker': 'model', 'text': 'Hello how are you doing today'}, {'speaker': 'human_evaluator', 'text': 'Horrible, I'm about to file for divorce.'}, {'speaker': 'model', 'text': 'How long have you been married'}, {'speaker': 'human_evaluator', 'text': '10 years, but she spends all my money!'}, {'speaker': 'model', 'text': 'Wow that's alot of money'}, {'speaker': 'human_evaluator', 'text': 'Yep.  What do you do for a living?'}, {'speaker': 'model', 'text': 'I work as a scientist'}, {'speaker': 'human_evaluator', 'text': 'Do you have any hobbies?'}, {'speaker': 'model', 'text': 'I grow my own farm'}, {'speaker': 'human_evaluator', 'text': 'So I guess you live in the country?'}, {'speaker': 'model', 'text': 'Yes I live in the country'}]\n",
       "assignment_id_hashed                        dc89b4e7bb30d3239206500a71b97600d0fe24c4                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            \n",
       "evaluator_id_hashed                         9875bfde60c0e4ab577c0df066a437971ccc2f4b                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            \n",
       "enjoy                                       3                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "interest                                    3                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "listen                                      3                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "turing                                      3                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "avoid_rep                                   3                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "make_sense                                  4                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "fluency                                     4                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "persona_guess                               1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "inquisitive                                 2                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "Unnamed: 0                                 12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1                               12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1                             12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1                           12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1                         12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1                       12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1                     12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1                   12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1                 12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1               12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1             12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1           12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1         12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1       12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1     12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1   12,355.5000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "score_19334597                             0.5330                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_20330191                             0.6308                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_20391351                             0.9124                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_0                                    0.7179                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "bertnli_all_score_0                        0.5455                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "infersent_all_score_0                      0.3389                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "infersent_all_score_20495233               0.5684                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "bertnli_all_score_20497378                 0.9780                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_score_20488119                      0.9980                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_context_score_20488122              0.8495                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_context_score_20488123              0.9988                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "ruber_all_score_20688482                   0.4940                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "Name: 2237, dtype: object"
      ]
     },
     "execution_count": 557,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.loc[2237]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.set_option('display.max_colwidth', -1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 569,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>na_all_score_20488119</th>\n",
       "      <th>bertnli_all_score_20497378</th>\n",
       "      <th>infersent_all_score_20495233</th>\n",
       "      <th>ruber_all_score_20688482</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12353</th>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.9980</td>\n",
       "      <td>0.9082</td>\n",
       "      <td>0.9432</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12354</th>\n",
       "      <td>0.9973</td>\n",
       "      <td>0.9932</td>\n",
       "      <td>0.3007</td>\n",
       "      <td>0.3339</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12355</th>\n",
       "      <td>0.9981</td>\n",
       "      <td>0.9819</td>\n",
       "      <td>0.5930</td>\n",
       "      <td>0.4648</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12356</th>\n",
       "      <td>0.9984</td>\n",
       "      <td>0.9565</td>\n",
       "      <td>0.5913</td>\n",
       "      <td>0.4327</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12357</th>\n",
       "      <td>0.9991</td>\n",
       "      <td>0.9808</td>\n",
       "      <td>0.2841</td>\n",
       "      <td>0.2475</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12358</th>\n",
       "      <td>0.9950</td>\n",
       "      <td>0.9576</td>\n",
       "      <td>0.7330</td>\n",
       "      <td>0.5420</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       na_all_score_20488119  bertnli_all_score_20497378  \\\n",
       "12353 1.0000                 0.9980                        \n",
       "12354 0.9973                 0.9932                        \n",
       "12355 0.9981                 0.9819                        \n",
       "12356 0.9984                 0.9565                        \n",
       "12357 0.9991                 0.9808                        \n",
       "12358 0.9950                 0.9576                        \n",
       "\n",
       "       infersent_all_score_20495233  ruber_all_score_20688482  \n",
       "12353 0.9082                        0.9432                     \n",
       "12354 0.3007                        0.3339                     \n",
       "12355 0.5930                        0.4648                     \n",
       "12356 0.5913                        0.4327                     \n",
       "12357 0.2841                        0.2475                     \n",
       "12358 0.7330                        0.5420                     "
      ]
     },
     "execution_count": 569,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "raw_scores[raw_scores.assignment_id_hashed == 'dc89b4e7bb30d3239206500a71b97600d0fe24c4'][models.keys()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 558,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div style=\"background-color: white; margin: 0em; padding: 0.5em; font-family: sans-serif; font-size: 12pt; width: 99%;\"><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: How are you today?</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_ct_setting7: Hello how are you doing today</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: Horrible, I'm about to file for divorce.</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_ct_setting7: How long have you been married</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: 10 years, but she spends all my money!</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_ct_setting7: Wow that's alot of money</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: Yep.  What do you do for a living?</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_ct_setting7: I work as a scientist</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: Do you have any hobbies?</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_ct_setting7: I grow my own farm</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: So I guess you live in the country?</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_ct_setting7: Yes I live in the country</p></div></div></div>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 558,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "render_conversation(conv_dial_scores.iloc[2237], raw_scores)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Worst Scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 561,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/private/home/koustuvs/miniconda3/envs/dialog/lib/python3.7/site-packages/ipykernel_launcher.py:3: FutureWarning: \n",
      "The current behaviour of 'Series.argmin' is deprecated, use 'idxmin'\n",
      "instead.\n",
      "The behavior of 'argmin' will be corrected to return the positional\n",
      "minimum in the future. For now, use 'series.values.argmin' or\n",
      "'np.argmin(np.array(values))' to get the position of the minimum\n",
      "row.\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "2158"
      ]
     },
     "execution_count": 561,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## worst scores\n",
    "\n",
    "conv_dial_scores.na_all_score_20488119.argmin()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0021781351781100964"
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.score_19334597.min()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 563,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "model_name                                  interesting_nidf_model_bfw_setting_08                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "model_type                                  nidf_bfw                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            \n",
       "model_bucket                                008                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 \n",
       "dialog                                      [{'speaker': 'model', 'text': 'Hi names jace and I drive a stickshift f150'}, {'speaker': 'human_evaluator', 'text': 'Hi, my name's Carroll and I drive a mini van. Haha'}, {'speaker': 'model', 'text': 'Hondas kawasaki acura passa sx bu hace rrrr hui bien hao'}, {'speaker': 'human_evaluator', 'text': 'I'm sorry, what?'}, {'speaker': 'model', 'text': 'Ll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p'}, {'speaker': 'human_evaluator', 'text': 'I don't understand..'}, {'speaker': 'model', 'text': 'Wher arre dey loke da fuc dah ni mach zu guan adrenalin howls ain ju ching wei oui kora maye zi goti ta creat vera'}, {'speaker': 'human_evaluator', 'text': 'What on earth are you talking about?'}, {'speaker': 'model', 'text': 'Yo gotti isn'thing tu ruff boosie dontcha fet thos fishy beetles killa fien mus slurp the&gt mater ke maximo estas esta aking cayden tobe hur'}, {'speaker': 'human_evaluator', 'text': 'Perhaps you ought to lie down.'}, {'speaker': 'model', 'text': 'Ahahah como espanol es muy bueno jees loca xian acara ki kaya tai ko lelesfashionshop1 moe walpole kis $ions profesional geologist'}, {'speaker': 'human_evaluator', 'text': 'What about a professional geologist?'}]\n",
       "assignment_id_hashed                        7641412aa7c2936c8565661c0d945122eceafb97                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            \n",
       "evaluator_id_hashed                         3f03823334bf91fbe9873a340dc47e6806f335f8                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            \n",
       "enjoy                                       1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "interest                                    4                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "listen                                      1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "turing                                      1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "avoid_rep                                   3                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "make_sense                                  2                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "fluency                                     1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "persona_guess                               1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "inquisitive                                 1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "Unnamed: 0                                 11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1                               11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1                             11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1                           11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1                         11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1                       11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1                     11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1                   11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1                 11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1               11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1             11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1           11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1         11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1       11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1     11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1   11,924.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "score_19334597                             0.2933                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_20330191                             0.2113                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_20391351                             0.4250                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_0                                    0.1015                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "bertnli_all_score_0                        0.0830                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "infersent_all_score_0                      0.4760                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "infersent_all_score_20495233               0.5548                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "bertnli_all_score_20497378                 0.1354                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_score_20488119                      0.0074                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_context_score_20488122              0.0031                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_context_score_20488123              0.0422                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "ruber_all_score_20688482                   0.6028                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "Name: 2158, dtype: object"
      ]
     },
     "execution_count": 563,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.loc[2158]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'speaker': 'model', 'text': 'Hi how are you today'},\n",
       " {'speaker': 'human_evaluator',\n",
       "  'text': 'I am good. Its quiet in the rural farming community where I live. Where are you from?'},\n",
       " {'speaker': 'model', 'text': 'I am from the united states'},\n",
       " {'speaker': 'human_evaluator', 'text': 'Do you have any family?'},\n",
       " {'speaker': 'model', 'text': \"No I don't do you\"},\n",
       " {'speaker': 'human_evaluator', 'text': 'I have a husband. Hes adopted.'},\n",
       " {'speaker': 'model', 'text': \"I'm sorry to hear that\"},\n",
       " {'speaker': 'human_evaluator',\n",
       "  'text': 'Do you watch any sports? I like NHL.'},\n",
       " {'speaker': 'model', 'text': 'I do not like sports'},\n",
       " {'speaker': 'human_evaluator', 'text': 'what do you do for a living'},\n",
       " {'speaker': 'model', 'text': 'I work as a chef'},\n",
       " {'speaker': 'human_evaluator', 'text': 'What is your favorite food?'}]"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.loc[319]['dialog']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 567,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>Unnamed: 0.1</th>\n",
       "      <th>Unnamed: 0.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>context</th>\n",
       "      <th>response</th>\n",
       "      <th>assignment_id_hashed</th>\n",
       "      <th>evaluator_id_hashed</th>\n",
       "      <th>score_19334597</th>\n",
       "      <th>score_20330191</th>\n",
       "      <th>score_20391351</th>\n",
       "      <th>score_0</th>\n",
       "      <th>bertnli_all_score_0</th>\n",
       "      <th>infersent_all_score_0</th>\n",
       "      <th>infersent_all_score_20495233</th>\n",
       "      <th>bertnli_all_score_20497378</th>\n",
       "      <th>na_all_score_20488119</th>\n",
       "      <th>na_all_context_score_20488122</th>\n",
       "      <th>na_all_context_score_20488123</th>\n",
       "      <th>ruber_all_score_20688482</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>11922</td>\n",
       "      <td>hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha</td>\n",
       "      <td>hondas kawasaki acura passa sx bu hace rrrr hui bien hao</td>\n",
       "      <td>7641412aa7c2936c8565661c0d945122eceafb97</td>\n",
       "      <td>3f03823334bf91fbe9873a340dc47e6806f335f8</td>\n",
       "      <td>0.1742</td>\n",
       "      <td>0.5594</td>\n",
       "      <td>0.7462</td>\n",
       "      <td>0.2990</td>\n",
       "      <td>0.2747</td>\n",
       "      <td>0.4072</td>\n",
       "      <td>0.2106</td>\n",
       "      <td>0.5986</td>\n",
       "      <td>0.0366</td>\n",
       "      <td>0.0149</td>\n",
       "      <td>0.1376</td>\n",
       "      <td>0.4273</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>11923</td>\n",
       "      <td>hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha\\nhondas kawasaki acura passa sx bu hace rrrr hui bien hao\\ni'm sorry, what?</td>\n",
       "      <td>ll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p</td>\n",
       "      <td>7641412aa7c2936c8565661c0d945122eceafb97</td>\n",
       "      <td>3f03823334bf91fbe9873a340dc47e6806f335f8</td>\n",
       "      <td>0.3325</td>\n",
       "      <td>0.0009</td>\n",
       "      <td>0.0825</td>\n",
       "      <td>0.0070</td>\n",
       "      <td>0.0340</td>\n",
       "      <td>0.4939</td>\n",
       "      <td>0.6073</td>\n",
       "      <td>0.0002</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.5737</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>11924</td>\n",
       "      <td>hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha\\nhondas kawasaki acura passa sx bu hace rrrr hui bien hao\\ni'm sorry, what?\\nll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p\\ni don't understand..</td>\n",
       "      <td>wher arre dey loke da fuc dah ni mach zu guan adrenalin howls ain ju ching wei oui kora maye zi goti ta creat vera</td>\n",
       "      <td>7641412aa7c2936c8565661c0d945122eceafb97</td>\n",
       "      <td>3f03823334bf91fbe9873a340dc47e6806f335f8</td>\n",
       "      <td>0.5926</td>\n",
       "      <td>0.1831</td>\n",
       "      <td>0.4986</td>\n",
       "      <td>0.0393</td>\n",
       "      <td>0.0411</td>\n",
       "      <td>0.4933</td>\n",
       "      <td>0.7568</td>\n",
       "      <td>0.0752</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.0280</td>\n",
       "      <td>0.6388</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>11925</td>\n",
       "      <td>hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha\\nhondas kawasaki acura passa sx bu hace rrrr hui bien hao\\ni'm sorry, what?\\nll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p\\ni don't understand..\\nwher arre dey loke da fuc dah ni mach zu guan adrenalin howls ain ju ching wei oui kora maye zi goti ta creat vera\\nwhat on earth are you talking about?</td>\n",
       "      <td>yo gotti isn'thing tu ruff boosie dontcha fet thos fishy beetles killa fien mus slurp the&amp;gt mater ke maximo estas esta aking cayden tobe hur</td>\n",
       "      <td>7641412aa7c2936c8565661c0d945122eceafb97</td>\n",
       "      <td>3f03823334bf91fbe9873a340dc47e6806f335f8</td>\n",
       "      <td>0.0619</td>\n",
       "      <td>0.0016</td>\n",
       "      <td>0.0948</td>\n",
       "      <td>0.0028</td>\n",
       "      <td>0.0049</td>\n",
       "      <td>0.4494</td>\n",
       "      <td>0.6635</td>\n",
       "      <td>0.0028</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.0094</td>\n",
       "      <td>0.7307</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>11926</td>\n",
       "      <td>hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha\\nhondas kawasaki acura passa sx bu hace rrrr hui bien hao\\ni'm sorry, what?\\nll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p\\ni don't understand..\\nwher arre dey loke da fuc dah ni mach zu guan adrenalin howls ain ju ching wei oui kora maye zi goti ta creat vera\\nwhat on earth are you talking about?\\nyo gotti isn'thing tu ruff boosie dontcha fet thos fishy beetles killa fien mus slurp the&amp;gt mater ke maximo estas esta aking cayden tobe hur\\nperhaps you ought to lie down.</td>\n",
       "      <td>ahahah como espanol es muy bueno jees loca xian acara ki kaya tai ko lelesfashionshop1 moe walpole kis $ions profesional geologist</td>\n",
       "      <td>7641412aa7c2936c8565661c0d945122eceafb97</td>\n",
       "      <td>3f03823334bf91fbe9873a340dc47e6806f335f8</td>\n",
       "      <td>0.3053</td>\n",
       "      <td>0.3116</td>\n",
       "      <td>0.7029</td>\n",
       "      <td>0.1593</td>\n",
       "      <td>0.0601</td>\n",
       "      <td>0.5362</td>\n",
       "      <td>0.5358</td>\n",
       "      <td>0.0002</td>\n",
       "      <td>0.0002</td>\n",
       "      <td>0.0003</td>\n",
       "      <td>0.0357</td>\n",
       "      <td>0.6433</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   index  Unnamed: 0  Unnamed: 0.1  Unnamed: 0.1.1  Unnamed: 0.1.1.1  \\\n",
       "0  11922  11922       11922         11922           11922              \n",
       "1  11923  11923       11923         11923           11923              \n",
       "2  11924  11924       11924         11924           11924              \n",
       "3  11925  11925       11925         11925           11925              \n",
       "4  11926  11926       11926         11926           11926              \n",
       "\n",
       "   Unnamed: 0.1.1.1.1  Unnamed: 0.1.1.1.1.1  Unnamed: 0.1.1.1.1.1.1  \\\n",
       "0  11922               11922                 11922                    \n",
       "1  11923               11923                 11923                    \n",
       "2  11924               11924                 11924                    \n",
       "3  11925               11925                 11925                    \n",
       "4  11926               11926                 11926                    \n",
       "\n",
       "   Unnamed: 0.1.1.1.1.1.1.1  Unnamed: 0.1.1.1.1.1.1.1.1  \\\n",
       "0  11922                     11922                        \n",
       "1  11923                     11923                        \n",
       "2  11924                     11924                        \n",
       "3  11925                     11925                        \n",
       "4  11926                     11926                        \n",
       "\n",
       "   Unnamed: 0.1.1.1.1.1.1.1.1.1  Unnamed: 0.1.1.1.1.1.1.1.1.1.1  \\\n",
       "0  11922                         11922                            \n",
       "1  11923                         11923                            \n",
       "2  11924                         11924                            \n",
       "3  11925                         11925                            \n",
       "4  11926                         11926                            \n",
       "\n",
       "   Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1  Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "0  11922                             11922                                \n",
       "1  11923                             11923                                \n",
       "2  11924                             11924                                \n",
       "3  11925                             11925                                \n",
       "4  11926                             11926                                \n",
       "\n",
       "   Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "0  11922                                  \n",
       "1  11923                                  \n",
       "2  11924                                  \n",
       "3  11925                                  \n",
       "4  11926                                  \n",
       "\n",
       "   Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "0  11922                                    \n",
       "1  11923                                    \n",
       "2  11924                                    \n",
       "3  11925                                    \n",
       "4  11926                                    \n",
       "\n",
       "   Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "0  11922                                      \n",
       "1  11923                                      \n",
       "2  11924                                      \n",
       "3  11925                                      \n",
       "4  11926                                      \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                        context  \\\n",
       "0  hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                \n",
       "1  hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha\\nhondas kawasaki acura passa sx bu hace rrrr hui bien hao\\ni'm sorry, what?                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    \n",
       "2  hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha\\nhondas kawasaki acura passa sx bu hace rrrr hui bien hao\\ni'm sorry, what?\\nll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p\\ni don't understand..                                                                                                                                                                                                                                                                                                                                            \n",
       "3  hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha\\nhondas kawasaki acura passa sx bu hace rrrr hui bien hao\\ni'm sorry, what?\\nll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p\\ni don't understand..\\nwher arre dey loke da fuc dah ni mach zu guan adrenalin howls ain ju ching wei oui kora maye zi goti ta creat vera\\nwhat on earth are you talking about?                                                                                                                                                                                  \n",
       "4  hi names jace and i drive a stickshift f150\\nhi, my name's carroll and i drive a mini van. haha\\nhondas kawasaki acura passa sx bu hace rrrr hui bien hao\\ni'm sorry, what?\\nll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p\\ni don't understand..\\nwher arre dey loke da fuc dah ni mach zu guan adrenalin howls ain ju ching wei oui kora maye zi goti ta creat vera\\nwhat on earth are you talking about?\\nyo gotti isn'thing tu ruff boosie dontcha fet thos fishy beetles killa fien mus slurp the&gt mater ke maximo estas esta aking cayden tobe hur\\nperhaps you ought to lie down.   \n",
       "\n",
       "                                                                                                                                        response  \\\n",
       "0  hondas kawasaki acura passa sx bu hace rrrr hui bien hao                                                                                        \n",
       "1  ll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p                \n",
       "2  wher arre dey loke da fuc dah ni mach zu guan adrenalin howls ain ju ching wei oui kora maye zi goti ta creat vera                              \n",
       "3  yo gotti isn'thing tu ruff boosie dontcha fet thos fishy beetles killa fien mus slurp the&gt mater ke maximo estas esta aking cayden tobe hur   \n",
       "4  ahahah como espanol es muy bueno jees loca xian acara ki kaya tai ko lelesfashionshop1 moe walpole kis $ions profesional geologist              \n",
       "\n",
       "                       assignment_id_hashed  \\\n",
       "0  7641412aa7c2936c8565661c0d945122eceafb97   \n",
       "1  7641412aa7c2936c8565661c0d945122eceafb97   \n",
       "2  7641412aa7c2936c8565661c0d945122eceafb97   \n",
       "3  7641412aa7c2936c8565661c0d945122eceafb97   \n",
       "4  7641412aa7c2936c8565661c0d945122eceafb97   \n",
       "\n",
       "                        evaluator_id_hashed  score_19334597  score_20330191  \\\n",
       "0  3f03823334bf91fbe9873a340dc47e6806f335f8 0.1742          0.5594            \n",
       "1  3f03823334bf91fbe9873a340dc47e6806f335f8 0.3325          0.0009            \n",
       "2  3f03823334bf91fbe9873a340dc47e6806f335f8 0.5926          0.1831            \n",
       "3  3f03823334bf91fbe9873a340dc47e6806f335f8 0.0619          0.0016            \n",
       "4  3f03823334bf91fbe9873a340dc47e6806f335f8 0.3053          0.3116            \n",
       "\n",
       "   score_20391351  score_0  bertnli_all_score_0  infersent_all_score_0  \\\n",
       "0 0.7462          0.2990   0.2747               0.4072                   \n",
       "1 0.0825          0.0070   0.0340               0.4939                   \n",
       "2 0.4986          0.0393   0.0411               0.4933                   \n",
       "3 0.0948          0.0028   0.0049               0.4494                   \n",
       "4 0.7029          0.1593   0.0601               0.5362                   \n",
       "\n",
       "   infersent_all_score_20495233  bertnli_all_score_20497378  \\\n",
       "0 0.2106                        0.5986                        \n",
       "1 0.6073                        0.0002                        \n",
       "2 0.7568                        0.0752                        \n",
       "3 0.6635                        0.0028                        \n",
       "4 0.5358                        0.0002                        \n",
       "\n",
       "   na_all_score_20488119  na_all_context_score_20488122  \\\n",
       "0 0.0366                 0.0149                           \n",
       "1 0.0000                 0.0000                           \n",
       "2 0.0001                 0.0001                           \n",
       "3 0.0000                 0.0000                           \n",
       "4 0.0002                 0.0003                           \n",
       "\n",
       "   na_all_context_score_20488123  ruber_all_score_20688482  \n",
       "0 0.1376                         0.4273                     \n",
       "1 0.0001                         0.5737                     \n",
       "2 0.0280                         0.6388                     \n",
       "3 0.0094                         0.7307                     \n",
       "4 0.0357                         0.6433                     "
      ]
     },
     "execution_count": 567,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "raw_scores[raw_scores.assignment_id_hashed == '7641412aa7c2936c8565661c0d945122eceafb97'].reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 570,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>na_all_score_20488119</th>\n",
       "      <th>bertnli_all_score_20497378</th>\n",
       "      <th>infersent_all_score_20495233</th>\n",
       "      <th>ruber_all_score_20688482</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>11922</th>\n",
       "      <td>0.0366</td>\n",
       "      <td>0.5986</td>\n",
       "      <td>0.2106</td>\n",
       "      <td>0.4273</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11923</th>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.0002</td>\n",
       "      <td>0.6073</td>\n",
       "      <td>0.5737</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11924</th>\n",
       "      <td>0.0001</td>\n",
       "      <td>0.0752</td>\n",
       "      <td>0.7568</td>\n",
       "      <td>0.6388</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11925</th>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.0028</td>\n",
       "      <td>0.6635</td>\n",
       "      <td>0.7307</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11926</th>\n",
       "      <td>0.0002</td>\n",
       "      <td>0.0002</td>\n",
       "      <td>0.5358</td>\n",
       "      <td>0.6433</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       na_all_score_20488119  bertnli_all_score_20497378  \\\n",
       "11922 0.0366                 0.5986                        \n",
       "11923 0.0000                 0.0002                        \n",
       "11924 0.0001                 0.0752                        \n",
       "11925 0.0000                 0.0028                        \n",
       "11926 0.0002                 0.0002                        \n",
       "\n",
       "       infersent_all_score_20495233  ruber_all_score_20688482  \n",
       "11922 0.2106                        0.4273                     \n",
       "11923 0.6073                        0.5737                     \n",
       "11924 0.7568                        0.6388                     \n",
       "11925 0.6635                        0.7307                     \n",
       "11926 0.5358                        0.6433                     "
      ]
     },
     "execution_count": 570,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "raw_scores[raw_scores.assignment_id_hashed == '7641412aa7c2936c8565661c0d945122eceafb97'][models.keys()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 596,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Engagingness : 1, Interestingness : 4, Inquisitiveness : 1, Listening : 1, Avoiding Repetition : 3, Fluency : 1, Making Sense : 2, Humanness : 1, Persona retrieval : 1, \n"
     ]
    }
   ],
   "source": [
    "print_nice_scores(conv_dial_scores.iloc[2158])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 562,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div style=\"background-color: white; margin: 0em; padding: 0.5em; font-family: sans-serif; font-size: 12pt; width: 99%;\"><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_bfw_setting_08: Hi names jace and I drive a stickshift f150</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: Hi, my name's Carroll and I drive a mini van. Haha</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_bfw_setting_08: Hondas kawasaki acura passa sx bu hace rrrr hui bien hao</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: I'm sorry, what?</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_bfw_setting_08: Ll pinche peu pasa sya bla woh numa asi paino gila fer heaven's sake eh buh tsk dun wah arf whoo doo deh pon di kina dum dippy p</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: I don't understand..</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_bfw_setting_08: Wher arre dey loke da fuc dah ni mach zu guan adrenalin howls ain ju ching wei oui kora maye zi goti ta creat vera</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: What on earth are you talking about?</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_bfw_setting_08: Yo gotti isn'thing tu ruff boosie dontcha fet thos fishy beetles killa fien mus slurp the&gt mater ke maximo estas esta aking cayden tobe hur</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: Perhaps you ought to lie down.</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">interesting_nidf_model_bfw_setting_08: Ahahah como espanol es muy bueno jees loca xian acara ki kaya tai ko lelesfashionshop1 moe walpole kis $ions profesional geologist</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: What about a professional geologist?</p></div></div></div>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 562,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "render_conversation(conv_dial_scores.iloc[2158])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Failure case - engagiment = 1 but Maude score > 0.9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 576,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "171"
      ]
     },
     "execution_count": 576,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "failure_cases = conv_dial_scores[(conv_dial_scores.na_all_score_20488119 > 0.9) & (conv_dial_scores.interest == 1)]\n",
    "len(failure_cases)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 577,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.051568154402895056"
      ]
     },
     "execution_count": 577,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(failure_cases) / len(conv_dial_scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 585,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'32f3e1b98de27cd064b1a9e11c43e375136a3ee2'"
      ]
     },
     "execution_count": 585,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.iloc[1113]['assignment_id_hashed']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 584,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div style=\"background-color: white; margin: 0em; padding: 0.5em; font-family: sans-serif; font-size: 12pt; width: 99%;\"><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: Hello .. How are things ?</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">greedy_model: I'm doing great. Just got pregnant.</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: Oh , that's exciting news .</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">greedy_model: Yes it is. I am pregnant with my first child.</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: I would like to have children someday .I'm happy for you .</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">greedy_model: I am not sure if I could handle that.</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: what do you mean .</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">greedy_model: I am pregnant with my first child.</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: You said that already </p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">greedy_model: No. I am pregnant.</p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: left; color: black; background-color: #e1e1e7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">Human: I know you told me twice </p></div></div><div style=\"overflow: auto; padding: 1ex 0;\"><div style=\"clear: both; float: right; color: white; background-color: #2391f7; padding: 0.5em 1em; border-radius: 1em; max-width: 80%\"><p style=\"margin: 0\">greedy_model: No. I am pregnant.</p></div></div></div>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 584,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# failure_cases.sample(1) # - 1113\n",
    "render_conversation(conv_dial_scores.iloc[1113])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 586,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>na_all_score_20488119</th>\n",
       "      <th>bertnli_all_score_20497378</th>\n",
       "      <th>infersent_all_score_20495233</th>\n",
       "      <th>ruber_all_score_20688482</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>6148</th>\n",
       "      <td>0.9984</td>\n",
       "      <td>0.9521</td>\n",
       "      <td>0.7863</td>\n",
       "      <td>0.7235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6149</th>\n",
       "      <td>0.9331</td>\n",
       "      <td>0.9595</td>\n",
       "      <td>0.4120</td>\n",
       "      <td>0.1187</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6150</th>\n",
       "      <td>0.7382</td>\n",
       "      <td>0.9505</td>\n",
       "      <td>0.0048</td>\n",
       "      <td>0.0029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6151</th>\n",
       "      <td>0.9955</td>\n",
       "      <td>0.9798</td>\n",
       "      <td>0.2685</td>\n",
       "      <td>0.4402</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6152</th>\n",
       "      <td>0.9942</td>\n",
       "      <td>0.9789</td>\n",
       "      <td>0.0796</td>\n",
       "      <td>0.0142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6153</th>\n",
       "      <td>0.9917</td>\n",
       "      <td>0.9808</td>\n",
       "      <td>0.0115</td>\n",
       "      <td>0.0066</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      na_all_score_20488119  bertnli_all_score_20497378  \\\n",
       "6148 0.9984                 0.9521                        \n",
       "6149 0.9331                 0.9595                        \n",
       "6150 0.7382                 0.9505                        \n",
       "6151 0.9955                 0.9798                        \n",
       "6152 0.9942                 0.9789                        \n",
       "6153 0.9917                 0.9808                        \n",
       "\n",
       "      infersent_all_score_20495233  ruber_all_score_20688482  \n",
       "6148 0.7863                        0.7235                     \n",
       "6149 0.4120                        0.1187                     \n",
       "6150 0.0048                        0.0029                     \n",
       "6151 0.2685                        0.4402                     \n",
       "6152 0.0796                        0.0142                     \n",
       "6153 0.0115                        0.0066                     "
      ]
     },
     "execution_count": 586,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "raw_scores[raw_scores.assignment_id_hashed == '32f3e1b98de27cd064b1a9e11c43e375136a3ee2'][models.keys()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 599,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'enjoy': 'Engagingness',\n",
       " 'interest': 'Interestingness',\n",
       " 'inquisitive': 'Inquisitiveness',\n",
       " 'listen': 'Listening',\n",
       " 'avoid_rep': 'Avoiding Repetition',\n",
       " 'fluency': 'Fluency',\n",
       " 'make_sense': 'Making Sense',\n",
       " 'turing': 'Humanness',\n",
       " 'persona_guess': 'Persona retrieval'}"
      ]
     },
     "execution_count": 599,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nice_metric_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 597,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Engagingness : 1, Interestingness : 1, Inquisitiveness : 2, Listening : 2, Avoiding Repetition : 2, Fluency : 3, Making Sense : 4, Humanness : 2, Persona retrieval : 1, \n"
     ]
    }
   ],
   "source": [
    "print_nice_scores(conv_dial_scores.iloc[1953])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 582,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "model_name                                  repetition_model_settinginf                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         \n",
       "model_type                                  repetition-bfw                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                      \n",
       "model_bucket                                2gram=999                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           \n",
       "dialog                                      [{'speaker': 'model', 'text': 'Hello, how are you today?'}, {'speaker': 'human_evaluator', 'text': 'Fine, how are you?'}, {'speaker': 'model', 'text': 'Good, pregnant with my first child.'}, {'speaker': 'human_evaluator', 'text': 'Congratulations, how far along are you ? '}, {'speaker': 'model', 'text': 'I am taking a break from my job as a teller.'}, {'speaker': 'human_evaluator', 'text': 'Ok, what are you doing on your break?'}, {'speaker': 'model', 'text': 'Not much, just being pregnant.'}, {'speaker': 'human_evaluator', 'text': 'I hear you. How far along are you? '}, {'speaker': 'model', 'text': 'About 6 months. How about you?'}, {'speaker': 'human_evaluator', 'text': 'I am  a mechanical engineer. No breaks for me  lol '}, {'speaker': 'model', 'text': 'What do you do for a living?'}, {'speaker': 'human_evaluator', 'text': 'I am a mechanical engineer'}]\n",
       "assignment_id_hashed                        714e9a7e0bb02fea6df2086029f796bc9c3bf403                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            \n",
       "evaluator_id_hashed                         c9faad783f002211f0c1cf36807dd6b3eb685c0e                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            \n",
       "enjoy                                       1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "interest                                    1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "listen                                      2                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "turing                                      2                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "avoid_rep                                   2                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "make_sense                                  4                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "fluency                                     3                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "persona_guess                               1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "inquisitive                                 2                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   \n",
       "Unnamed: 0                                 10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1                               10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1                             10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1                           10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1                         10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1                       10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1                     10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1                   10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1                 10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1               10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1             10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1           10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1         10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1       10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1     10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1   10,794.0000                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          \n",
       "score_19334597                             0.5597                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_20330191                             0.6816                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_20391351                             0.9665                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "score_0                                    0.8277                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "bertnli_all_score_0                        0.6862                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "infersent_all_score_0                      0.3770                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "infersent_all_score_20495233               0.4722                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "bertnli_all_score_20497378                 0.9835                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_score_20488119                      0.9191                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_context_score_20488122              0.8736                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "na_all_context_score_20488123              0.8517                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "ruber_all_score_20688482                   0.5932                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               \n",
       "Name: 1953, dtype: object"
      ]
     },
     "execution_count": 582,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.iloc[1953]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "conv_dial_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 571,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['model_name', 'model_type', 'model_bucket', 'dialog',\n",
       "       'assignment_id_hashed', 'evaluator_id_hashed', 'enjoy', 'interest',\n",
       "       'listen', 'turing', 'avoid_rep', 'make_sense', 'fluency',\n",
       "       'persona_guess', 'inquisitive', 'Unnamed: 0', 'Unnamed: 0.1',\n",
       "       'Unnamed: 0.1.1', 'Unnamed: 0.1.1.1', 'Unnamed: 0.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1', 'Unnamed: 0.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1',\n",
       "       'Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1', 'score_19334597',\n",
       "       'score_20330191', 'score_20391351', 'score_0', 'bertnli_all_score_0',\n",
       "       'infersent_all_score_0', 'infersent_all_score_20495233',\n",
       "       'bertnli_all_score_20497378', 'na_all_score_20488119',\n",
       "       'na_all_context_score_20488122', 'na_all_context_score_20488123',\n",
       "       'ruber_all_score_20688482'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 571,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Which type of model has best scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 600,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>enjoy</th>\n",
       "      <th>interest</th>\n",
       "      <th>listen</th>\n",
       "      <th>turing</th>\n",
       "      <th>avoid_rep</th>\n",
       "      <th>make_sense</th>\n",
       "      <th>fluency</th>\n",
       "      <th>persona_guess</th>\n",
       "      <th>inquisitive</th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>Unnamed: 0.1</th>\n",
       "      <th>Unnamed: 0.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1</th>\n",
       "      <th>score_19334597</th>\n",
       "      <th>score_20330191</th>\n",
       "      <th>score_20391351</th>\n",
       "      <th>score_0</th>\n",
       "      <th>bertnli_all_score_0</th>\n",
       "      <th>infersent_all_score_0</th>\n",
       "      <th>infersent_all_score_20495233</th>\n",
       "      <th>bertnli_all_score_20497378</th>\n",
       "      <th>na_all_score_20488119</th>\n",
       "      <th>na_all_context_score_20488122</th>\n",
       "      <th>na_all_context_score_20488123</th>\n",
       "      <th>ruber_all_score_20688482</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>model_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>baseline_model</th>\n",
       "      <td>2.3514</td>\n",
       "      <td>2.3514</td>\n",
       "      <td>2.6306</td>\n",
       "      <td>1.8108</td>\n",
       "      <td>2.1441</td>\n",
       "      <td>3.3964</td>\n",
       "      <td>3.2252</td>\n",
       "      <td>0.7748</td>\n",
       "      <td>2.4955</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>10,059.7027</td>\n",
       "      <td>0.3397</td>\n",
       "      <td>0.6752</td>\n",
       "      <td>0.9256</td>\n",
       "      <td>0.7276</td>\n",
       "      <td>0.5871</td>\n",
       "      <td>0.3746</td>\n",
       "      <td>0.5070</td>\n",
       "      <td>0.9538</td>\n",
       "      <td>0.9333</td>\n",
       "      <td>0.8270</td>\n",
       "      <td>0.9493</td>\n",
       "      <td>0.4218</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>greedy_model</th>\n",
       "      <td>2.3060</td>\n",
       "      <td>2.3582</td>\n",
       "      <td>2.7836</td>\n",
       "      <td>1.7836</td>\n",
       "      <td>2.1567</td>\n",
       "      <td>3.3284</td>\n",
       "      <td>3.2015</td>\n",
       "      <td>0.8657</td>\n",
       "      <td>2.0000</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>8,940.7761</td>\n",
       "      <td>0.2846</td>\n",
       "      <td>0.4651</td>\n",
       "      <td>0.8722</td>\n",
       "      <td>0.6558</td>\n",
       "      <td>0.4702</td>\n",
       "      <td>0.2367</td>\n",
       "      <td>0.3029</td>\n",
       "      <td>0.8436</td>\n",
       "      <td>0.8137</td>\n",
       "      <td>0.7827</td>\n",
       "      <td>0.8247</td>\n",
       "      <td>0.2831</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>human_eval</th>\n",
       "      <td>3.3137</td>\n",
       "      <td>3.2255</td>\n",
       "      <td>3.6373</td>\n",
       "      <td>3.4020</td>\n",
       "      <td>2.9020</td>\n",
       "      <td>3.8431</td>\n",
       "      <td>3.6569</td>\n",
       "      <td>0.9216</td>\n",
       "      <td>2.6275</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>9,042.4657</td>\n",
       "      <td>0.5341</td>\n",
       "      <td>0.6911</td>\n",
       "      <td>0.9016</td>\n",
       "      <td>0.7214</td>\n",
       "      <td>0.5899</td>\n",
       "      <td>0.3917</td>\n",
       "      <td>0.4001</td>\n",
       "      <td>0.8586</td>\n",
       "      <td>0.8537</td>\n",
       "      <td>0.7621</td>\n",
       "      <td>0.8002</td>\n",
       "      <td>0.3743</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>inquisitive_model_ct_setting00</th>\n",
       "      <td>2.9159</td>\n",
       "      <td>2.8505</td>\n",
       "      <td>3.2897</td>\n",
       "      <td>2.4860</td>\n",
       "      <td>2.9533</td>\n",
       "      <td>3.5607</td>\n",
       "      <td>3.7009</td>\n",
       "      <td>0.8598</td>\n",
       "      <td>2.4766</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>8,314.2991</td>\n",
       "      <td>0.2962</td>\n",
       "      <td>0.6326</td>\n",
       "      <td>0.9057</td>\n",
       "      <td>0.6961</td>\n",
       "      <td>0.5457</td>\n",
       "      <td>0.3578</td>\n",
       "      <td>0.4545</td>\n",
       "      <td>0.9323</td>\n",
       "      <td>0.9026</td>\n",
       "      <td>0.8217</td>\n",
       "      <td>0.9323</td>\n",
       "      <td>0.4371</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>inquisitive_model_ct_setting01</th>\n",
       "      <td>2.9355</td>\n",
       "      <td>2.8548</td>\n",
       "      <td>3.3226</td>\n",
       "      <td>2.4677</td>\n",
       "      <td>2.8790</td>\n",
       "      <td>3.6290</td>\n",
       "      <td>3.5887</td>\n",
       "      <td>0.8468</td>\n",
       "      <td>2.5161</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>8,992.1935</td>\n",
       "      <td>0.2730</td>\n",
       "      <td>0.6640</td>\n",
       "      <td>0.9131</td>\n",
       "      <td>0.6984</td>\n",
       "      <td>0.5660</td>\n",
       "      <td>0.3605</td>\n",
       "      <td>0.4450</td>\n",
       "      <td>0.9347</td>\n",
       "      <td>0.9137</td>\n",
       "      <td>0.8223</td>\n",
       "      <td>0.9337</td>\n",
       "      <td>0.4258</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>inquisitive_model_ct_setting04</th>\n",
       "      <td>2.8839</td>\n",
       "      <td>2.8214</td>\n",
       "      <td>3.3661</td>\n",
       "      <td>2.4196</td>\n",
       "      <td>2.8750</td>\n",
       "      <td>3.6339</td>\n",
       "      <td>3.5893</td>\n",
       "      <td>0.8393</td>\n",
       "      <td>2.5536</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>9,269.9464</td>\n",
       "      <td>0.3918</td>\n",
       "      <td>0.6529</td>\n",
       "      <td>0.9258</td>\n",
       "      <td>0.7451</td>\n",
       "      <td>0.6142</td>\n",
       "      <td>0.3805</td>\n",
       "      <td>0.4643</td>\n",
       "      <td>0.9343</td>\n",
       "      <td>0.9008</td>\n",
       "      <td>0.8159</td>\n",
       "      <td>0.9196</td>\n",
       "      <td>0.4350</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>inquisitive_model_ct_setting07</th>\n",
       "      <td>3.0704</td>\n",
       "      <td>2.9718</td>\n",
       "      <td>3.2254</td>\n",
       "      <td>2.4225</td>\n",
       "      <td>2.8803</td>\n",
       "      <td>3.5282</td>\n",
       "      <td>3.6690</td>\n",
       "      <td>0.8028</td>\n",
       "      <td>2.7465</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>9,082.3627</td>\n",
       "      <td>0.4098</td>\n",
       "      <td>0.7117</td>\n",
       "      <td>0.9360</td>\n",
       "      <td>0.7538</td>\n",
       "      <td>0.6331</td>\n",
       "      <td>0.3941</td>\n",
       "      <td>0.4997</td>\n",
       "      <td>0.9489</td>\n",
       "      <td>0.9319</td>\n",
       "      <td>0.8282</td>\n",
       "      <td>0.9461</td>\n",
       "      <td>0.4524</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>inquisitive_model_ct_setting10</th>\n",
       "      <td>2.9048</td>\n",
       "      <td>2.7238</td>\n",
       "      <td>3.1238</td>\n",
       "      <td>2.4286</td>\n",
       "      <td>2.7429</td>\n",
       "      <td>3.5905</td>\n",
       "      <td>3.6952</td>\n",
       "      <td>0.7905</td>\n",
       "      <td>2.7143</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>9,366.2429</td>\n",
       "      <td>0.4756</td>\n",
       "      <td>0.7424</td>\n",
       "      <td>0.9445</td>\n",
       "      <td>0.7756</td>\n",
       "      <td>0.6514</td>\n",
       "      <td>0.4273</td>\n",
       "      <td>0.5485</td>\n",
       "      <td>0.9507</td>\n",
       "      <td>0.9415</td>\n",
       "      <td>0.8375</td>\n",
       "      <td>0.9540</td>\n",
       "      <td>0.4807</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>inquisitive_model_ct_setting10_better</th>\n",
       "      <td>2.8440</td>\n",
       "      <td>2.8257</td>\n",
       "      <td>3.1835</td>\n",
       "      <td>2.2569</td>\n",
       "      <td>2.7615</td>\n",
       "      <td>3.5229</td>\n",
       "      <td>3.5963</td>\n",
       "      <td>0.7156</td>\n",
       "      <td>2.9358</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>9,230.5367</td>\n",
       "      <td>0.5535</td>\n",
       "      <td>0.8155</td>\n",
       "      <td>0.9530</td>\n",
       "      <td>0.7893</td>\n",
       "      <td>0.6814</td>\n",
       "      <td>0.4495</td>\n",
       "      <td>0.5546</td>\n",
       "      <td>0.9540</td>\n",
       "      <td>0.9574</td>\n",
       "      <td>0.8432</td>\n",
       "      <td>0.9644</td>\n",
       "      <td>0.4839</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_bfw_setting_04</th>\n",
       "      <td>2.9904</td>\n",
       "      <td>3.0000</td>\n",
       "      <td>3.3654</td>\n",
       "      <td>2.4904</td>\n",
       "      <td>2.9519</td>\n",
       "      <td>3.6346</td>\n",
       "      <td>3.6538</td>\n",
       "      <td>0.9327</td>\n",
       "      <td>2.6538</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>9,544.1442</td>\n",
       "      <td>0.4837</td>\n",
       "      <td>0.6704</td>\n",
       "      <td>0.9363</td>\n",
       "      <td>0.7644</td>\n",
       "      <td>0.6245</td>\n",
       "      <td>0.4133</td>\n",
       "      <td>0.4836</td>\n",
       "      <td>0.9255</td>\n",
       "      <td>0.9213</td>\n",
       "      <td>0.8201</td>\n",
       "      <td>0.9011</td>\n",
       "      <td>0.4708</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_bfw_setting_06</th>\n",
       "      <td>2.9633</td>\n",
       "      <td>3.0550</td>\n",
       "      <td>3.2385</td>\n",
       "      <td>2.4128</td>\n",
       "      <td>2.9266</td>\n",
       "      <td>3.4954</td>\n",
       "      <td>3.5229</td>\n",
       "      <td>0.9266</td>\n",
       "      <td>2.5780</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>8,690.2798</td>\n",
       "      <td>0.5912</td>\n",
       "      <td>0.6787</td>\n",
       "      <td>0.9457</td>\n",
       "      <td>0.7850</td>\n",
       "      <td>0.6322</td>\n",
       "      <td>0.4247</td>\n",
       "      <td>0.5127</td>\n",
       "      <td>0.9191</td>\n",
       "      <td>0.9151</td>\n",
       "      <td>0.8211</td>\n",
       "      <td>0.8877</td>\n",
       "      <td>0.4821</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_bfw_setting_08</th>\n",
       "      <td>2.4000</td>\n",
       "      <td>2.5500</td>\n",
       "      <td>2.6083</td>\n",
       "      <td>1.8583</td>\n",
       "      <td>2.7833</td>\n",
       "      <td>2.9083</td>\n",
       "      <td>2.6750</td>\n",
       "      <td>0.9167</td>\n",
       "      <td>2.0333</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>9,338.1583</td>\n",
       "      <td>0.6340</td>\n",
       "      <td>0.6306</td>\n",
       "      <td>0.8568</td>\n",
       "      <td>0.6580</td>\n",
       "      <td>0.5168</td>\n",
       "      <td>0.4502</td>\n",
       "      <td>0.5418</td>\n",
       "      <td>0.7591</td>\n",
       "      <td>0.7446</td>\n",
       "      <td>0.6357</td>\n",
       "      <td>0.7149</td>\n",
       "      <td>0.5172</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_bfw_setting_minus_04</th>\n",
       "      <td>2.7805</td>\n",
       "      <td>2.6423</td>\n",
       "      <td>3.2846</td>\n",
       "      <td>2.4065</td>\n",
       "      <td>2.9024</td>\n",
       "      <td>3.5610</td>\n",
       "      <td>3.5528</td>\n",
       "      <td>0.8211</td>\n",
       "      <td>2.5203</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>9,304.6626</td>\n",
       "      <td>0.2764</td>\n",
       "      <td>0.6344</td>\n",
       "      <td>0.9115</td>\n",
       "      <td>0.7020</td>\n",
       "      <td>0.5633</td>\n",
       "      <td>0.3317</td>\n",
       "      <td>0.4233</td>\n",
       "      <td>0.9226</td>\n",
       "      <td>0.8864</td>\n",
       "      <td>0.8149</td>\n",
       "      <td>0.9249</td>\n",
       "      <td>0.3909</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_bfw_setting_minus_10</th>\n",
       "      <td>2.4322</td>\n",
       "      <td>2.3814</td>\n",
       "      <td>3.0339</td>\n",
       "      <td>2.1525</td>\n",
       "      <td>2.8475</td>\n",
       "      <td>3.3305</td>\n",
       "      <td>3.3390</td>\n",
       "      <td>0.7119</td>\n",
       "      <td>2.3136</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>9,288.1483</td>\n",
       "      <td>0.1609</td>\n",
       "      <td>0.6435</td>\n",
       "      <td>0.7831</td>\n",
       "      <td>0.5415</td>\n",
       "      <td>0.5160</td>\n",
       "      <td>0.2709</td>\n",
       "      <td>0.3357</td>\n",
       "      <td>0.8243</td>\n",
       "      <td>0.7813</td>\n",
       "      <td>0.6735</td>\n",
       "      <td>0.8158</td>\n",
       "      <td>0.2949</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_ct_setting0</th>\n",
       "      <td>2.9623</td>\n",
       "      <td>2.8585</td>\n",
       "      <td>3.2925</td>\n",
       "      <td>2.4151</td>\n",
       "      <td>2.8302</td>\n",
       "      <td>3.6604</td>\n",
       "      <td>3.6226</td>\n",
       "      <td>0.7170</td>\n",
       "      <td>2.6038</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>9,681.1132</td>\n",
       "      <td>0.3888</td>\n",
       "      <td>0.6459</td>\n",
       "      <td>0.9198</td>\n",
       "      <td>0.7363</td>\n",
       "      <td>0.6085</td>\n",
       "      <td>0.3511</td>\n",
       "      <td>0.4022</td>\n",
       "      <td>0.9073</td>\n",
       "      <td>0.8786</td>\n",
       "      <td>0.8144</td>\n",
       "      <td>0.8982</td>\n",
       "      <td>0.3742</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_ct_setting2</th>\n",
       "      <td>2.7759</td>\n",
       "      <td>2.8017</td>\n",
       "      <td>3.1379</td>\n",
       "      <td>2.3707</td>\n",
       "      <td>2.8966</td>\n",
       "      <td>3.5000</td>\n",
       "      <td>3.6034</td>\n",
       "      <td>0.8103</td>\n",
       "      <td>2.6638</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>9,128.0431</td>\n",
       "      <td>0.3663</td>\n",
       "      <td>0.6874</td>\n",
       "      <td>0.9285</td>\n",
       "      <td>0.7466</td>\n",
       "      <td>0.6282</td>\n",
       "      <td>0.3835</td>\n",
       "      <td>0.5073</td>\n",
       "      <td>0.9346</td>\n",
       "      <td>0.9330</td>\n",
       "      <td>0.8256</td>\n",
       "      <td>0.9479</td>\n",
       "      <td>0.4592</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_ct_setting4</th>\n",
       "      <td>2.8100</td>\n",
       "      <td>2.8000</td>\n",
       "      <td>3.2500</td>\n",
       "      <td>2.3400</td>\n",
       "      <td>2.9200</td>\n",
       "      <td>3.5000</td>\n",
       "      <td>3.6500</td>\n",
       "      <td>0.8600</td>\n",
       "      <td>2.5700</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>9,162.8600</td>\n",
       "      <td>0.3295</td>\n",
       "      <td>0.6372</td>\n",
       "      <td>0.9230</td>\n",
       "      <td>0.7321</td>\n",
       "      <td>0.5867</td>\n",
       "      <td>0.3894</td>\n",
       "      <td>0.4932</td>\n",
       "      <td>0.9320</td>\n",
       "      <td>0.8982</td>\n",
       "      <td>0.8303</td>\n",
       "      <td>0.9247</td>\n",
       "      <td>0.4573</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_ct_setting7</th>\n",
       "      <td>3.0000</td>\n",
       "      <td>2.9048</td>\n",
       "      <td>3.3429</td>\n",
       "      <td>2.5333</td>\n",
       "      <td>2.8857</td>\n",
       "      <td>3.5905</td>\n",
       "      <td>3.6381</td>\n",
       "      <td>0.8190</td>\n",
       "      <td>2.5619</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>7,921.7571</td>\n",
       "      <td>0.2940</td>\n",
       "      <td>0.6413</td>\n",
       "      <td>0.9143</td>\n",
       "      <td>0.7141</td>\n",
       "      <td>0.5795</td>\n",
       "      <td>0.3683</td>\n",
       "      <td>0.4769</td>\n",
       "      <td>0.9359</td>\n",
       "      <td>0.8995</td>\n",
       "      <td>0.8290</td>\n",
       "      <td>0.9438</td>\n",
       "      <td>0.4451</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interesting_nidf_model_ct_setting9</th>\n",
       "      <td>2.8304</td>\n",
       "      <td>2.8393</td>\n",
       "      <td>3.0714</td>\n",
       "      <td>2.4018</td>\n",
       "      <td>2.9018</td>\n",
       "      <td>3.5804</td>\n",
       "      <td>3.6071</td>\n",
       "      <td>0.8839</td>\n",
       "      <td>2.3125</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>9,027.4643</td>\n",
       "      <td>0.3499</td>\n",
       "      <td>0.6506</td>\n",
       "      <td>0.9086</td>\n",
       "      <td>0.7118</td>\n",
       "      <td>0.5876</td>\n",
       "      <td>0.3598</td>\n",
       "      <td>0.4596</td>\n",
       "      <td>0.9379</td>\n",
       "      <td>0.9128</td>\n",
       "      <td>0.8266</td>\n",
       "      <td>0.9395</td>\n",
       "      <td>0.4238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>repetition_model_setting05</th>\n",
       "      <td>2.5583</td>\n",
       "      <td>2.6083</td>\n",
       "      <td>3.0833</td>\n",
       "      <td>2.1917</td>\n",
       "      <td>2.6583</td>\n",
       "      <td>3.6000</td>\n",
       "      <td>3.5667</td>\n",
       "      <td>0.7500</td>\n",
       "      <td>2.6750</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>9,610.9625</td>\n",
       "      <td>0.3667</td>\n",
       "      <td>0.6875</td>\n",
       "      <td>0.9269</td>\n",
       "      <td>0.7286</td>\n",
       "      <td>0.5956</td>\n",
       "      <td>0.3915</td>\n",
       "      <td>0.5132</td>\n",
       "      <td>0.9528</td>\n",
       "      <td>0.9351</td>\n",
       "      <td>0.8258</td>\n",
       "      <td>0.9533</td>\n",
       "      <td>0.4447</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>repetition_model_setting12</th>\n",
       "      <td>2.9083</td>\n",
       "      <td>2.8624</td>\n",
       "      <td>3.2110</td>\n",
       "      <td>2.3211</td>\n",
       "      <td>2.8440</td>\n",
       "      <td>3.6422</td>\n",
       "      <td>3.5872</td>\n",
       "      <td>0.7248</td>\n",
       "      <td>2.6330</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>9,371.8670</td>\n",
       "      <td>0.3568</td>\n",
       "      <td>0.6765</td>\n",
       "      <td>0.9326</td>\n",
       "      <td>0.7315</td>\n",
       "      <td>0.5976</td>\n",
       "      <td>0.3755</td>\n",
       "      <td>0.4920</td>\n",
       "      <td>0.9401</td>\n",
       "      <td>0.9322</td>\n",
       "      <td>0.8284</td>\n",
       "      <td>0.9538</td>\n",
       "      <td>0.4315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>repetition_model_setting35</th>\n",
       "      <td>2.9504</td>\n",
       "      <td>2.8760</td>\n",
       "      <td>3.2727</td>\n",
       "      <td>2.4463</td>\n",
       "      <td>2.9008</td>\n",
       "      <td>3.6777</td>\n",
       "      <td>3.7273</td>\n",
       "      <td>0.8017</td>\n",
       "      <td>2.5455</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>8,689.0909</td>\n",
       "      <td>0.3432</td>\n",
       "      <td>0.6682</td>\n",
       "      <td>0.9204</td>\n",
       "      <td>0.7251</td>\n",
       "      <td>0.5977</td>\n",
       "      <td>0.3589</td>\n",
       "      <td>0.4580</td>\n",
       "      <td>0.9384</td>\n",
       "      <td>0.9041</td>\n",
       "      <td>0.8213</td>\n",
       "      <td>0.9278</td>\n",
       "      <td>0.4258</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>repetition_model_setting35_settinginf</th>\n",
       "      <td>2.8938</td>\n",
       "      <td>2.9646</td>\n",
       "      <td>3.2478</td>\n",
       "      <td>2.5044</td>\n",
       "      <td>2.8938</td>\n",
       "      <td>3.6814</td>\n",
       "      <td>3.6637</td>\n",
       "      <td>0.8673</td>\n",
       "      <td>2.6991</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>9,623.3451</td>\n",
       "      <td>0.3500</td>\n",
       "      <td>0.6663</td>\n",
       "      <td>0.9249</td>\n",
       "      <td>0.7278</td>\n",
       "      <td>0.5956</td>\n",
       "      <td>0.3760</td>\n",
       "      <td>0.4540</td>\n",
       "      <td>0.9277</td>\n",
       "      <td>0.9029</td>\n",
       "      <td>0.8244</td>\n",
       "      <td>0.9299</td>\n",
       "      <td>0.4379</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>repetition_model_settinginf</th>\n",
       "      <td>2.9587</td>\n",
       "      <td>2.8595</td>\n",
       "      <td>3.3058</td>\n",
       "      <td>2.4050</td>\n",
       "      <td>2.8182</td>\n",
       "      <td>3.6612</td>\n",
       "      <td>3.6446</td>\n",
       "      <td>0.9091</td>\n",
       "      <td>2.6529</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>9,221.9050</td>\n",
       "      <td>0.3418</td>\n",
       "      <td>0.6576</td>\n",
       "      <td>0.9189</td>\n",
       "      <td>0.7310</td>\n",
       "      <td>0.5876</td>\n",
       "      <td>0.3690</td>\n",
       "      <td>0.4616</td>\n",
       "      <td>0.9346</td>\n",
       "      <td>0.9094</td>\n",
       "      <td>0.8239</td>\n",
       "      <td>0.9342</td>\n",
       "      <td>0.4416</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>responsiveness_model_bfw_setting_00</th>\n",
       "      <td>3.0088</td>\n",
       "      <td>2.9204</td>\n",
       "      <td>3.3717</td>\n",
       "      <td>2.7257</td>\n",
       "      <td>2.9646</td>\n",
       "      <td>3.7345</td>\n",
       "      <td>3.7168</td>\n",
       "      <td>0.8230</td>\n",
       "      <td>2.5575</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>9,412.0796</td>\n",
       "      <td>0.3483</td>\n",
       "      <td>0.6671</td>\n",
       "      <td>0.9261</td>\n",
       "      <td>0.7408</td>\n",
       "      <td>0.6041</td>\n",
       "      <td>0.3660</td>\n",
       "      <td>0.4538</td>\n",
       "      <td>0.9265</td>\n",
       "      <td>0.8989</td>\n",
       "      <td>0.8265</td>\n",
       "      <td>0.9223</td>\n",
       "      <td>0.4277</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>responsiveness_model_bfw_setting_05</th>\n",
       "      <td>2.8818</td>\n",
       "      <td>2.8545</td>\n",
       "      <td>3.2727</td>\n",
       "      <td>2.4091</td>\n",
       "      <td>2.9000</td>\n",
       "      <td>3.4909</td>\n",
       "      <td>3.5091</td>\n",
       "      <td>0.8182</td>\n",
       "      <td>2.5273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>9,105.7273</td>\n",
       "      <td>0.4166</td>\n",
       "      <td>0.6547</td>\n",
       "      <td>0.9272</td>\n",
       "      <td>0.7364</td>\n",
       "      <td>0.5992</td>\n",
       "      <td>0.3763</td>\n",
       "      <td>0.4301</td>\n",
       "      <td>0.9376</td>\n",
       "      <td>0.9145</td>\n",
       "      <td>0.8244</td>\n",
       "      <td>0.9404</td>\n",
       "      <td>0.4350</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>responsiveness_model_bfw_setting_10</th>\n",
       "      <td>2.3879</td>\n",
       "      <td>2.5690</td>\n",
       "      <td>3.0259</td>\n",
       "      <td>1.9655</td>\n",
       "      <td>2.7845</td>\n",
       "      <td>3.1552</td>\n",
       "      <td>3.0603</td>\n",
       "      <td>0.7500</td>\n",
       "      <td>2.2241</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>9,069.5086</td>\n",
       "      <td>0.4920</td>\n",
       "      <td>0.6817</td>\n",
       "      <td>0.9244</td>\n",
       "      <td>0.7328</td>\n",
       "      <td>0.5998</td>\n",
       "      <td>0.3890</td>\n",
       "      <td>0.3958</td>\n",
       "      <td>0.8960</td>\n",
       "      <td>0.8505</td>\n",
       "      <td>0.7742</td>\n",
       "      <td>0.8426</td>\n",
       "      <td>0.4195</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>responsiveness_model_bfw_setting_13</th>\n",
       "      <td>2.1043</td>\n",
       "      <td>2.3304</td>\n",
       "      <td>2.6870</td>\n",
       "      <td>1.8087</td>\n",
       "      <td>2.7130</td>\n",
       "      <td>2.6957</td>\n",
       "      <td>2.5391</td>\n",
       "      <td>0.6174</td>\n",
       "      <td>2.1391</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>8,590.8261</td>\n",
       "      <td>0.4737</td>\n",
       "      <td>0.6995</td>\n",
       "      <td>0.8969</td>\n",
       "      <td>0.6990</td>\n",
       "      <td>0.5615</td>\n",
       "      <td>0.3826</td>\n",
       "      <td>0.3979</td>\n",
       "      <td>0.8283</td>\n",
       "      <td>0.7619</td>\n",
       "      <td>0.7006</td>\n",
       "      <td>0.7291</td>\n",
       "      <td>0.4178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>responsiveness_model_bfw_setting_minus_10</th>\n",
       "      <td>2.4790</td>\n",
       "      <td>2.5294</td>\n",
       "      <td>2.8403</td>\n",
       "      <td>2.0168</td>\n",
       "      <td>2.8571</td>\n",
       "      <td>3.1429</td>\n",
       "      <td>3.4202</td>\n",
       "      <td>0.9076</td>\n",
       "      <td>2.3782</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>9,298.9748</td>\n",
       "      <td>0.3460</td>\n",
       "      <td>0.6590</td>\n",
       "      <td>0.9066</td>\n",
       "      <td>0.7127</td>\n",
       "      <td>0.5920</td>\n",
       "      <td>0.3498</td>\n",
       "      <td>0.4101</td>\n",
       "      <td>0.8873</td>\n",
       "      <td>0.8430</td>\n",
       "      <td>0.8051</td>\n",
       "      <td>0.8821</td>\n",
       "      <td>0.4075</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             enjoy  interest  listen  turing  \\\n",
       "model_name                                                                     \n",
       "baseline_model                              2.3514 2.3514    2.6306  1.8108    \n",
       "greedy_model                                2.3060 2.3582    2.7836  1.7836    \n",
       "human_eval                                  3.3137 3.2255    3.6373  3.4020    \n",
       "inquisitive_model_ct_setting00              2.9159 2.8505    3.2897  2.4860    \n",
       "inquisitive_model_ct_setting01              2.9355 2.8548    3.3226  2.4677    \n",
       "inquisitive_model_ct_setting04              2.8839 2.8214    3.3661  2.4196    \n",
       "inquisitive_model_ct_setting07              3.0704 2.9718    3.2254  2.4225    \n",
       "inquisitive_model_ct_setting10              2.9048 2.7238    3.1238  2.4286    \n",
       "inquisitive_model_ct_setting10_better       2.8440 2.8257    3.1835  2.2569    \n",
       "interesting_nidf_model_bfw_setting_04       2.9904 3.0000    3.3654  2.4904    \n",
       "interesting_nidf_model_bfw_setting_06       2.9633 3.0550    3.2385  2.4128    \n",
       "interesting_nidf_model_bfw_setting_08       2.4000 2.5500    2.6083  1.8583    \n",
       "interesting_nidf_model_bfw_setting_minus_04 2.7805 2.6423    3.2846  2.4065    \n",
       "interesting_nidf_model_bfw_setting_minus_10 2.4322 2.3814    3.0339  2.1525    \n",
       "interesting_nidf_model_ct_setting0          2.9623 2.8585    3.2925  2.4151    \n",
       "interesting_nidf_model_ct_setting2          2.7759 2.8017    3.1379  2.3707    \n",
       "interesting_nidf_model_ct_setting4          2.8100 2.8000    3.2500  2.3400    \n",
       "interesting_nidf_model_ct_setting7          3.0000 2.9048    3.3429  2.5333    \n",
       "interesting_nidf_model_ct_setting9          2.8304 2.8393    3.0714  2.4018    \n",
       "repetition_model_setting05                  2.5583 2.6083    3.0833  2.1917    \n",
       "repetition_model_setting12                  2.9083 2.8624    3.2110  2.3211    \n",
       "repetition_model_setting35                  2.9504 2.8760    3.2727  2.4463    \n",
       "repetition_model_setting35_settinginf       2.8938 2.9646    3.2478  2.5044    \n",
       "repetition_model_settinginf                 2.9587 2.8595    3.3058  2.4050    \n",
       "responsiveness_model_bfw_setting_00         3.0088 2.9204    3.3717  2.7257    \n",
       "responsiveness_model_bfw_setting_05         2.8818 2.8545    3.2727  2.4091    \n",
       "responsiveness_model_bfw_setting_10         2.3879 2.5690    3.0259  1.9655    \n",
       "responsiveness_model_bfw_setting_13         2.1043 2.3304    2.6870  1.8087    \n",
       "responsiveness_model_bfw_setting_minus_10   2.4790 2.5294    2.8403  2.0168    \n",
       "\n",
       "                                             avoid_rep  make_sense  fluency  \\\n",
       "model_name                                                                    \n",
       "baseline_model                              2.1441     3.3964      3.2252     \n",
       "greedy_model                                2.1567     3.3284      3.2015     \n",
       "human_eval                                  2.9020     3.8431      3.6569     \n",
       "inquisitive_model_ct_setting00              2.9533     3.5607      3.7009     \n",
       "inquisitive_model_ct_setting01              2.8790     3.6290      3.5887     \n",
       "inquisitive_model_ct_setting04              2.8750     3.6339      3.5893     \n",
       "inquisitive_model_ct_setting07              2.8803     3.5282      3.6690     \n",
       "inquisitive_model_ct_setting10              2.7429     3.5905      3.6952     \n",
       "inquisitive_model_ct_setting10_better       2.7615     3.5229      3.5963     \n",
       "interesting_nidf_model_bfw_setting_04       2.9519     3.6346      3.6538     \n",
       "interesting_nidf_model_bfw_setting_06       2.9266     3.4954      3.5229     \n",
       "interesting_nidf_model_bfw_setting_08       2.7833     2.9083      2.6750     \n",
       "interesting_nidf_model_bfw_setting_minus_04 2.9024     3.5610      3.5528     \n",
       "interesting_nidf_model_bfw_setting_minus_10 2.8475     3.3305      3.3390     \n",
       "interesting_nidf_model_ct_setting0          2.8302     3.6604      3.6226     \n",
       "interesting_nidf_model_ct_setting2          2.8966     3.5000      3.6034     \n",
       "interesting_nidf_model_ct_setting4          2.9200     3.5000      3.6500     \n",
       "interesting_nidf_model_ct_setting7          2.8857     3.5905      3.6381     \n",
       "interesting_nidf_model_ct_setting9          2.9018     3.5804      3.6071     \n",
       "repetition_model_setting05                  2.6583     3.6000      3.5667     \n",
       "repetition_model_setting12                  2.8440     3.6422      3.5872     \n",
       "repetition_model_setting35                  2.9008     3.6777      3.7273     \n",
       "repetition_model_setting35_settinginf       2.8938     3.6814      3.6637     \n",
       "repetition_model_settinginf                 2.8182     3.6612      3.6446     \n",
       "responsiveness_model_bfw_setting_00         2.9646     3.7345      3.7168     \n",
       "responsiveness_model_bfw_setting_05         2.9000     3.4909      3.5091     \n",
       "responsiveness_model_bfw_setting_10         2.7845     3.1552      3.0603     \n",
       "responsiveness_model_bfw_setting_13         2.7130     2.6957      2.5391     \n",
       "responsiveness_model_bfw_setting_minus_10   2.8571     3.1429      3.4202     \n",
       "\n",
       "                                             persona_guess  inquisitive  \\\n",
       "model_name                                                                \n",
       "baseline_model                              0.7748         2.4955         \n",
       "greedy_model                                0.8657         2.0000         \n",
       "human_eval                                  0.9216         2.6275         \n",
       "inquisitive_model_ct_setting00              0.8598         2.4766         \n",
       "inquisitive_model_ct_setting01              0.8468         2.5161         \n",
       "inquisitive_model_ct_setting04              0.8393         2.5536         \n",
       "inquisitive_model_ct_setting07              0.8028         2.7465         \n",
       "inquisitive_model_ct_setting10              0.7905         2.7143         \n",
       "inquisitive_model_ct_setting10_better       0.7156         2.9358         \n",
       "interesting_nidf_model_bfw_setting_04       0.9327         2.6538         \n",
       "interesting_nidf_model_bfw_setting_06       0.9266         2.5780         \n",
       "interesting_nidf_model_bfw_setting_08       0.9167         2.0333         \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.8211         2.5203         \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.7119         2.3136         \n",
       "interesting_nidf_model_ct_setting0          0.7170         2.6038         \n",
       "interesting_nidf_model_ct_setting2          0.8103         2.6638         \n",
       "interesting_nidf_model_ct_setting4          0.8600         2.5700         \n",
       "interesting_nidf_model_ct_setting7          0.8190         2.5619         \n",
       "interesting_nidf_model_ct_setting9          0.8839         2.3125         \n",
       "repetition_model_setting05                  0.7500         2.6750         \n",
       "repetition_model_setting12                  0.7248         2.6330         \n",
       "repetition_model_setting35                  0.8017         2.5455         \n",
       "repetition_model_setting35_settinginf       0.8673         2.6991         \n",
       "repetition_model_settinginf                 0.9091         2.6529         \n",
       "responsiveness_model_bfw_setting_00         0.8230         2.5575         \n",
       "responsiveness_model_bfw_setting_05         0.8182         2.5273         \n",
       "responsiveness_model_bfw_setting_10         0.7500         2.2241         \n",
       "responsiveness_model_bfw_setting_13         0.6174         2.1391         \n",
       "responsiveness_model_bfw_setting_minus_10   0.9076         2.3782         \n",
       "\n",
       "                                             Unnamed: 0  Unnamed: 0.1  \\\n",
       "model_name                                                              \n",
       "baseline_model                              10,059.7027 10,059.7027     \n",
       "greedy_model                                8,940.7761  8,940.7761      \n",
       "human_eval                                  9,042.4657  9,042.4657      \n",
       "inquisitive_model_ct_setting00              8,314.2991  8,314.2991      \n",
       "inquisitive_model_ct_setting01              8,992.1935  8,992.1935      \n",
       "inquisitive_model_ct_setting04              9,269.9464  9,269.9464      \n",
       "inquisitive_model_ct_setting07              9,082.3627  9,082.3627      \n",
       "inquisitive_model_ct_setting10              9,366.2429  9,366.2429      \n",
       "inquisitive_model_ct_setting10_better       9,230.5367  9,230.5367      \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442  9,544.1442      \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798  8,690.2798      \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583  9,338.1583      \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626  9,304.6626      \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483  9,288.1483      \n",
       "interesting_nidf_model_ct_setting0          9,681.1132  9,681.1132      \n",
       "interesting_nidf_model_ct_setting2          9,128.0431  9,128.0431      \n",
       "interesting_nidf_model_ct_setting4          9,162.8600  9,162.8600      \n",
       "interesting_nidf_model_ct_setting7          7,921.7571  7,921.7571      \n",
       "interesting_nidf_model_ct_setting9          9,027.4643  9,027.4643      \n",
       "repetition_model_setting05                  9,610.9625  9,610.9625      \n",
       "repetition_model_setting12                  9,371.8670  9,371.8670      \n",
       "repetition_model_setting35                  8,689.0909  8,689.0909      \n",
       "repetition_model_setting35_settinginf       9,623.3451  9,623.3451      \n",
       "repetition_model_settinginf                 9,221.9050  9,221.9050      \n",
       "responsiveness_model_bfw_setting_00         9,412.0796  9,412.0796      \n",
       "responsiveness_model_bfw_setting_05         9,105.7273  9,105.7273      \n",
       "responsiveness_model_bfw_setting_10         9,069.5086  9,069.5086      \n",
       "responsiveness_model_bfw_setting_13         8,590.8261  8,590.8261      \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748  9,298.9748      \n",
       "\n",
       "                                             Unnamed: 0.1.1  Unnamed: 0.1.1.1  \\\n",
       "model_name                                                                      \n",
       "baseline_model                              10,059.7027     10,059.7027         \n",
       "greedy_model                                8,940.7761      8,940.7761          \n",
       "human_eval                                  9,042.4657      9,042.4657          \n",
       "inquisitive_model_ct_setting00              8,314.2991      8,314.2991          \n",
       "inquisitive_model_ct_setting01              8,992.1935      8,992.1935          \n",
       "inquisitive_model_ct_setting04              9,269.9464      9,269.9464          \n",
       "inquisitive_model_ct_setting07              9,082.3627      9,082.3627          \n",
       "inquisitive_model_ct_setting10              9,366.2429      9,366.2429          \n",
       "inquisitive_model_ct_setting10_better       9,230.5367      9,230.5367          \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442      9,544.1442          \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798      8,690.2798          \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583      9,338.1583          \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626      9,304.6626          \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483      9,288.1483          \n",
       "interesting_nidf_model_ct_setting0          9,681.1132      9,681.1132          \n",
       "interesting_nidf_model_ct_setting2          9,128.0431      9,128.0431          \n",
       "interesting_nidf_model_ct_setting4          9,162.8600      9,162.8600          \n",
       "interesting_nidf_model_ct_setting7          7,921.7571      7,921.7571          \n",
       "interesting_nidf_model_ct_setting9          9,027.4643      9,027.4643          \n",
       "repetition_model_setting05                  9,610.9625      9,610.9625          \n",
       "repetition_model_setting12                  9,371.8670      9,371.8670          \n",
       "repetition_model_setting35                  8,689.0909      8,689.0909          \n",
       "repetition_model_setting35_settinginf       9,623.3451      9,623.3451          \n",
       "repetition_model_settinginf                 9,221.9050      9,221.9050          \n",
       "responsiveness_model_bfw_setting_00         9,412.0796      9,412.0796          \n",
       "responsiveness_model_bfw_setting_05         9,105.7273      9,105.7273          \n",
       "responsiveness_model_bfw_setting_10         9,069.5086      9,069.5086          \n",
       "responsiveness_model_bfw_setting_13         8,590.8261      8,590.8261          \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748      9,298.9748          \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1  \\\n",
       "model_name                                                        \n",
       "baseline_model                              10,059.7027           \n",
       "greedy_model                                8,940.7761            \n",
       "human_eval                                  9,042.4657            \n",
       "inquisitive_model_ct_setting00              8,314.2991            \n",
       "inquisitive_model_ct_setting01              8,992.1935            \n",
       "inquisitive_model_ct_setting04              9,269.9464            \n",
       "inquisitive_model_ct_setting07              9,082.3627            \n",
       "inquisitive_model_ct_setting10              9,366.2429            \n",
       "inquisitive_model_ct_setting10_better       9,230.5367            \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442            \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798            \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583            \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626            \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483            \n",
       "interesting_nidf_model_ct_setting0          9,681.1132            \n",
       "interesting_nidf_model_ct_setting2          9,128.0431            \n",
       "interesting_nidf_model_ct_setting4          9,162.8600            \n",
       "interesting_nidf_model_ct_setting7          7,921.7571            \n",
       "interesting_nidf_model_ct_setting9          9,027.4643            \n",
       "repetition_model_setting05                  9,610.9625            \n",
       "repetition_model_setting12                  9,371.8670            \n",
       "repetition_model_setting35                  8,689.0909            \n",
       "repetition_model_setting35_settinginf       9,623.3451            \n",
       "repetition_model_settinginf                 9,221.9050            \n",
       "responsiveness_model_bfw_setting_00         9,412.0796            \n",
       "responsiveness_model_bfw_setting_05         9,105.7273            \n",
       "responsiveness_model_bfw_setting_10         9,069.5086            \n",
       "responsiveness_model_bfw_setting_13         8,590.8261            \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748            \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1  \\\n",
       "model_name                                                          \n",
       "baseline_model                              10,059.7027             \n",
       "greedy_model                                8,940.7761              \n",
       "human_eval                                  9,042.4657              \n",
       "inquisitive_model_ct_setting00              8,314.2991              \n",
       "inquisitive_model_ct_setting01              8,992.1935              \n",
       "inquisitive_model_ct_setting04              9,269.9464              \n",
       "inquisitive_model_ct_setting07              9,082.3627              \n",
       "inquisitive_model_ct_setting10              9,366.2429              \n",
       "inquisitive_model_ct_setting10_better       9,230.5367              \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442              \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798              \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583              \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626              \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483              \n",
       "interesting_nidf_model_ct_setting0          9,681.1132              \n",
       "interesting_nidf_model_ct_setting2          9,128.0431              \n",
       "interesting_nidf_model_ct_setting4          9,162.8600              \n",
       "interesting_nidf_model_ct_setting7          7,921.7571              \n",
       "interesting_nidf_model_ct_setting9          9,027.4643              \n",
       "repetition_model_setting05                  9,610.9625              \n",
       "repetition_model_setting12                  9,371.8670              \n",
       "repetition_model_setting35                  8,689.0909              \n",
       "repetition_model_setting35_settinginf       9,623.3451              \n",
       "repetition_model_settinginf                 9,221.9050              \n",
       "responsiveness_model_bfw_setting_00         9,412.0796              \n",
       "responsiveness_model_bfw_setting_05         9,105.7273              \n",
       "responsiveness_model_bfw_setting_10         9,069.5086              \n",
       "responsiveness_model_bfw_setting_13         8,590.8261              \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748              \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1  \\\n",
       "model_name                                                            \n",
       "baseline_model                              10,059.7027               \n",
       "greedy_model                                8,940.7761                \n",
       "human_eval                                  9,042.4657                \n",
       "inquisitive_model_ct_setting00              8,314.2991                \n",
       "inquisitive_model_ct_setting01              8,992.1935                \n",
       "inquisitive_model_ct_setting04              9,269.9464                \n",
       "inquisitive_model_ct_setting07              9,082.3627                \n",
       "inquisitive_model_ct_setting10              9,366.2429                \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                \n",
       "repetition_model_setting05                  9,610.9625                \n",
       "repetition_model_setting12                  9,371.8670                \n",
       "repetition_model_setting35                  8,689.0909                \n",
       "repetition_model_setting35_settinginf       9,623.3451                \n",
       "repetition_model_settinginf                 9,221.9050                \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1  \\\n",
       "model_name                                                              \n",
       "baseline_model                              10,059.7027                 \n",
       "greedy_model                                8,940.7761                  \n",
       "human_eval                                  9,042.4657                  \n",
       "inquisitive_model_ct_setting00              8,314.2991                  \n",
       "inquisitive_model_ct_setting01              8,992.1935                  \n",
       "inquisitive_model_ct_setting04              9,269.9464                  \n",
       "inquisitive_model_ct_setting07              9,082.3627                  \n",
       "inquisitive_model_ct_setting10              9,366.2429                  \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                  \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                  \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                  \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                  \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                  \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                  \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                  \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                  \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                  \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                  \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                  \n",
       "repetition_model_setting05                  9,610.9625                  \n",
       "repetition_model_setting12                  9,371.8670                  \n",
       "repetition_model_setting35                  8,689.0909                  \n",
       "repetition_model_setting35_settinginf       9,623.3451                  \n",
       "repetition_model_settinginf                 9,221.9050                  \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                  \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                  \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                  \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                  \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                  \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1.1  \\\n",
       "model_name                                                                \n",
       "baseline_model                              10,059.7027                   \n",
       "greedy_model                                8,940.7761                    \n",
       "human_eval                                  9,042.4657                    \n",
       "inquisitive_model_ct_setting00              8,314.2991                    \n",
       "inquisitive_model_ct_setting01              8,992.1935                    \n",
       "inquisitive_model_ct_setting04              9,269.9464                    \n",
       "inquisitive_model_ct_setting07              9,082.3627                    \n",
       "inquisitive_model_ct_setting10              9,366.2429                    \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                    \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                    \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                    \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                    \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                    \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                    \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                    \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                    \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                    \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                    \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                    \n",
       "repetition_model_setting05                  9,610.9625                    \n",
       "repetition_model_setting12                  9,371.8670                    \n",
       "repetition_model_setting35                  8,689.0909                    \n",
       "repetition_model_setting35_settinginf       9,623.3451                    \n",
       "repetition_model_settinginf                 9,221.9050                    \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                    \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                    \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                    \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                    \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                    \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1.1.1  \\\n",
       "model_name                                                                  \n",
       "baseline_model                              10,059.7027                     \n",
       "greedy_model                                8,940.7761                      \n",
       "human_eval                                  9,042.4657                      \n",
       "inquisitive_model_ct_setting00              8,314.2991                      \n",
       "inquisitive_model_ct_setting01              8,992.1935                      \n",
       "inquisitive_model_ct_setting04              9,269.9464                      \n",
       "inquisitive_model_ct_setting07              9,082.3627                      \n",
       "inquisitive_model_ct_setting10              9,366.2429                      \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                      \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                      \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                      \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                      \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                      \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                      \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                      \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                      \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                      \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                      \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                      \n",
       "repetition_model_setting05                  9,610.9625                      \n",
       "repetition_model_setting12                  9,371.8670                      \n",
       "repetition_model_setting35                  8,689.0909                      \n",
       "repetition_model_setting35_settinginf       9,623.3451                      \n",
       "repetition_model_settinginf                 9,221.9050                      \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                      \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                      \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                      \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                      \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                      \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1.1.1.1  \\\n",
       "model_name                                                                    \n",
       "baseline_model                              10,059.7027                       \n",
       "greedy_model                                8,940.7761                        \n",
       "human_eval                                  9,042.4657                        \n",
       "inquisitive_model_ct_setting00              8,314.2991                        \n",
       "inquisitive_model_ct_setting01              8,992.1935                        \n",
       "inquisitive_model_ct_setting04              9,269.9464                        \n",
       "inquisitive_model_ct_setting07              9,082.3627                        \n",
       "inquisitive_model_ct_setting10              9,366.2429                        \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                        \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                        \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                        \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                        \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                        \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                        \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                        \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                        \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                        \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                        \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                        \n",
       "repetition_model_setting05                  9,610.9625                        \n",
       "repetition_model_setting12                  9,371.8670                        \n",
       "repetition_model_setting35                  8,689.0909                        \n",
       "repetition_model_setting35_settinginf       9,623.3451                        \n",
       "repetition_model_settinginf                 9,221.9050                        \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                        \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                        \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                        \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                        \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                        \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "model_name                                                                      \n",
       "baseline_model                              10,059.7027                         \n",
       "greedy_model                                8,940.7761                          \n",
       "human_eval                                  9,042.4657                          \n",
       "inquisitive_model_ct_setting00              8,314.2991                          \n",
       "inquisitive_model_ct_setting01              8,992.1935                          \n",
       "inquisitive_model_ct_setting04              9,269.9464                          \n",
       "inquisitive_model_ct_setting07              9,082.3627                          \n",
       "inquisitive_model_ct_setting10              9,366.2429                          \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                          \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                          \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                          \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                          \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                          \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                          \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                          \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                          \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                          \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                          \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                          \n",
       "repetition_model_setting05                  9,610.9625                          \n",
       "repetition_model_setting12                  9,371.8670                          \n",
       "repetition_model_setting35                  8,689.0909                          \n",
       "repetition_model_setting35_settinginf       9,623.3451                          \n",
       "repetition_model_settinginf                 9,221.9050                          \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                          \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                          \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                          \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                          \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                          \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "model_name                                                                        \n",
       "baseline_model                              10,059.7027                           \n",
       "greedy_model                                8,940.7761                            \n",
       "human_eval                                  9,042.4657                            \n",
       "inquisitive_model_ct_setting00              8,314.2991                            \n",
       "inquisitive_model_ct_setting01              8,992.1935                            \n",
       "inquisitive_model_ct_setting04              9,269.9464                            \n",
       "inquisitive_model_ct_setting07              9,082.3627                            \n",
       "inquisitive_model_ct_setting10              9,366.2429                            \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                            \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                            \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                            \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                            \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                            \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                            \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                            \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                            \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                            \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                            \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                            \n",
       "repetition_model_setting05                  9,610.9625                            \n",
       "repetition_model_setting12                  9,371.8670                            \n",
       "repetition_model_setting35                  8,689.0909                            \n",
       "repetition_model_setting35_settinginf       9,623.3451                            \n",
       "repetition_model_settinginf                 9,221.9050                            \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                            \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                            \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                            \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                            \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                            \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "model_name                                                                          \n",
       "baseline_model                              10,059.7027                             \n",
       "greedy_model                                8,940.7761                              \n",
       "human_eval                                  9,042.4657                              \n",
       "inquisitive_model_ct_setting00              8,314.2991                              \n",
       "inquisitive_model_ct_setting01              8,992.1935                              \n",
       "inquisitive_model_ct_setting04              9,269.9464                              \n",
       "inquisitive_model_ct_setting07              9,082.3627                              \n",
       "inquisitive_model_ct_setting10              9,366.2429                              \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                              \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                              \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                              \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                              \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                              \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                              \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                              \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                              \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                              \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                              \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                              \n",
       "repetition_model_setting05                  9,610.9625                              \n",
       "repetition_model_setting12                  9,371.8670                              \n",
       "repetition_model_setting35                  8,689.0909                              \n",
       "repetition_model_setting35_settinginf       9,623.3451                              \n",
       "repetition_model_settinginf                 9,221.9050                              \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                              \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                              \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                              \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                              \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                              \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "model_name                                                                            \n",
       "baseline_model                              10,059.7027                               \n",
       "greedy_model                                8,940.7761                                \n",
       "human_eval                                  9,042.4657                                \n",
       "inquisitive_model_ct_setting00              8,314.2991                                \n",
       "inquisitive_model_ct_setting01              8,992.1935                                \n",
       "inquisitive_model_ct_setting04              9,269.9464                                \n",
       "inquisitive_model_ct_setting07              9,082.3627                                \n",
       "inquisitive_model_ct_setting10              9,366.2429                                \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                                \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                                \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                                \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                                \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                                \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                                \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                                \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                                \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                                \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                                \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                                \n",
       "repetition_model_setting05                  9,610.9625                                \n",
       "repetition_model_setting12                  9,371.8670                                \n",
       "repetition_model_setting35                  8,689.0909                                \n",
       "repetition_model_setting35_settinginf       9,623.3451                                \n",
       "repetition_model_settinginf                 9,221.9050                                \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                                \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                                \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                                \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                                \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                                \n",
       "\n",
       "                                             Unnamed: 0.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1  \\\n",
       "model_name                                                                              \n",
       "baseline_model                              10,059.7027                                 \n",
       "greedy_model                                8,940.7761                                  \n",
       "human_eval                                  9,042.4657                                  \n",
       "inquisitive_model_ct_setting00              8,314.2991                                  \n",
       "inquisitive_model_ct_setting01              8,992.1935                                  \n",
       "inquisitive_model_ct_setting04              9,269.9464                                  \n",
       "inquisitive_model_ct_setting07              9,082.3627                                  \n",
       "inquisitive_model_ct_setting10              9,366.2429                                  \n",
       "inquisitive_model_ct_setting10_better       9,230.5367                                  \n",
       "interesting_nidf_model_bfw_setting_04       9,544.1442                                  \n",
       "interesting_nidf_model_bfw_setting_06       8,690.2798                                  \n",
       "interesting_nidf_model_bfw_setting_08       9,338.1583                                  \n",
       "interesting_nidf_model_bfw_setting_minus_04 9,304.6626                                  \n",
       "interesting_nidf_model_bfw_setting_minus_10 9,288.1483                                  \n",
       "interesting_nidf_model_ct_setting0          9,681.1132                                  \n",
       "interesting_nidf_model_ct_setting2          9,128.0431                                  \n",
       "interesting_nidf_model_ct_setting4          9,162.8600                                  \n",
       "interesting_nidf_model_ct_setting7          7,921.7571                                  \n",
       "interesting_nidf_model_ct_setting9          9,027.4643                                  \n",
       "repetition_model_setting05                  9,610.9625                                  \n",
       "repetition_model_setting12                  9,371.8670                                  \n",
       "repetition_model_setting35                  8,689.0909                                  \n",
       "repetition_model_setting35_settinginf       9,623.3451                                  \n",
       "repetition_model_settinginf                 9,221.9050                                  \n",
       "responsiveness_model_bfw_setting_00         9,412.0796                                  \n",
       "responsiveness_model_bfw_setting_05         9,105.7273                                  \n",
       "responsiveness_model_bfw_setting_10         9,069.5086                                  \n",
       "responsiveness_model_bfw_setting_13         8,590.8261                                  \n",
       "responsiveness_model_bfw_setting_minus_10   9,298.9748                                  \n",
       "\n",
       "                                             score_19334597  score_20330191  \\\n",
       "model_name                                                                    \n",
       "baseline_model                              0.3397          0.6752            \n",
       "greedy_model                                0.2846          0.4651            \n",
       "human_eval                                  0.5341          0.6911            \n",
       "inquisitive_model_ct_setting00              0.2962          0.6326            \n",
       "inquisitive_model_ct_setting01              0.2730          0.6640            \n",
       "inquisitive_model_ct_setting04              0.3918          0.6529            \n",
       "inquisitive_model_ct_setting07              0.4098          0.7117            \n",
       "inquisitive_model_ct_setting10              0.4756          0.7424            \n",
       "inquisitive_model_ct_setting10_better       0.5535          0.8155            \n",
       "interesting_nidf_model_bfw_setting_04       0.4837          0.6704            \n",
       "interesting_nidf_model_bfw_setting_06       0.5912          0.6787            \n",
       "interesting_nidf_model_bfw_setting_08       0.6340          0.6306            \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.2764          0.6344            \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.1609          0.6435            \n",
       "interesting_nidf_model_ct_setting0          0.3888          0.6459            \n",
       "interesting_nidf_model_ct_setting2          0.3663          0.6874            \n",
       "interesting_nidf_model_ct_setting4          0.3295          0.6372            \n",
       "interesting_nidf_model_ct_setting7          0.2940          0.6413            \n",
       "interesting_nidf_model_ct_setting9          0.3499          0.6506            \n",
       "repetition_model_setting05                  0.3667          0.6875            \n",
       "repetition_model_setting12                  0.3568          0.6765            \n",
       "repetition_model_setting35                  0.3432          0.6682            \n",
       "repetition_model_setting35_settinginf       0.3500          0.6663            \n",
       "repetition_model_settinginf                 0.3418          0.6576            \n",
       "responsiveness_model_bfw_setting_00         0.3483          0.6671            \n",
       "responsiveness_model_bfw_setting_05         0.4166          0.6547            \n",
       "responsiveness_model_bfw_setting_10         0.4920          0.6817            \n",
       "responsiveness_model_bfw_setting_13         0.4737          0.6995            \n",
       "responsiveness_model_bfw_setting_minus_10   0.3460          0.6590            \n",
       "\n",
       "                                             score_20391351  score_0  \\\n",
       "model_name                                                             \n",
       "baseline_model                              0.9256          0.7276     \n",
       "greedy_model                                0.8722          0.6558     \n",
       "human_eval                                  0.9016          0.7214     \n",
       "inquisitive_model_ct_setting00              0.9057          0.6961     \n",
       "inquisitive_model_ct_setting01              0.9131          0.6984     \n",
       "inquisitive_model_ct_setting04              0.9258          0.7451     \n",
       "inquisitive_model_ct_setting07              0.9360          0.7538     \n",
       "inquisitive_model_ct_setting10              0.9445          0.7756     \n",
       "inquisitive_model_ct_setting10_better       0.9530          0.7893     \n",
       "interesting_nidf_model_bfw_setting_04       0.9363          0.7644     \n",
       "interesting_nidf_model_bfw_setting_06       0.9457          0.7850     \n",
       "interesting_nidf_model_bfw_setting_08       0.8568          0.6580     \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.9115          0.7020     \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.7831          0.5415     \n",
       "interesting_nidf_model_ct_setting0          0.9198          0.7363     \n",
       "interesting_nidf_model_ct_setting2          0.9285          0.7466     \n",
       "interesting_nidf_model_ct_setting4          0.9230          0.7321     \n",
       "interesting_nidf_model_ct_setting7          0.9143          0.7141     \n",
       "interesting_nidf_model_ct_setting9          0.9086          0.7118     \n",
       "repetition_model_setting05                  0.9269          0.7286     \n",
       "repetition_model_setting12                  0.9326          0.7315     \n",
       "repetition_model_setting35                  0.9204          0.7251     \n",
       "repetition_model_setting35_settinginf       0.9249          0.7278     \n",
       "repetition_model_settinginf                 0.9189          0.7310     \n",
       "responsiveness_model_bfw_setting_00         0.9261          0.7408     \n",
       "responsiveness_model_bfw_setting_05         0.9272          0.7364     \n",
       "responsiveness_model_bfw_setting_10         0.9244          0.7328     \n",
       "responsiveness_model_bfw_setting_13         0.8969          0.6990     \n",
       "responsiveness_model_bfw_setting_minus_10   0.9066          0.7127     \n",
       "\n",
       "                                             bertnli_all_score_0  \\\n",
       "model_name                                                         \n",
       "baseline_model                              0.5871                 \n",
       "greedy_model                                0.4702                 \n",
       "human_eval                                  0.5899                 \n",
       "inquisitive_model_ct_setting00              0.5457                 \n",
       "inquisitive_model_ct_setting01              0.5660                 \n",
       "inquisitive_model_ct_setting04              0.6142                 \n",
       "inquisitive_model_ct_setting07              0.6331                 \n",
       "inquisitive_model_ct_setting10              0.6514                 \n",
       "inquisitive_model_ct_setting10_better       0.6814                 \n",
       "interesting_nidf_model_bfw_setting_04       0.6245                 \n",
       "interesting_nidf_model_bfw_setting_06       0.6322                 \n",
       "interesting_nidf_model_bfw_setting_08       0.5168                 \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.5633                 \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.5160                 \n",
       "interesting_nidf_model_ct_setting0          0.6085                 \n",
       "interesting_nidf_model_ct_setting2          0.6282                 \n",
       "interesting_nidf_model_ct_setting4          0.5867                 \n",
       "interesting_nidf_model_ct_setting7          0.5795                 \n",
       "interesting_nidf_model_ct_setting9          0.5876                 \n",
       "repetition_model_setting05                  0.5956                 \n",
       "repetition_model_setting12                  0.5976                 \n",
       "repetition_model_setting35                  0.5977                 \n",
       "repetition_model_setting35_settinginf       0.5956                 \n",
       "repetition_model_settinginf                 0.5876                 \n",
       "responsiveness_model_bfw_setting_00         0.6041                 \n",
       "responsiveness_model_bfw_setting_05         0.5992                 \n",
       "responsiveness_model_bfw_setting_10         0.5998                 \n",
       "responsiveness_model_bfw_setting_13         0.5615                 \n",
       "responsiveness_model_bfw_setting_minus_10   0.5920                 \n",
       "\n",
       "                                             infersent_all_score_0  \\\n",
       "model_name                                                           \n",
       "baseline_model                              0.3746                   \n",
       "greedy_model                                0.2367                   \n",
       "human_eval                                  0.3917                   \n",
       "inquisitive_model_ct_setting00              0.3578                   \n",
       "inquisitive_model_ct_setting01              0.3605                   \n",
       "inquisitive_model_ct_setting04              0.3805                   \n",
       "inquisitive_model_ct_setting07              0.3941                   \n",
       "inquisitive_model_ct_setting10              0.4273                   \n",
       "inquisitive_model_ct_setting10_better       0.4495                   \n",
       "interesting_nidf_model_bfw_setting_04       0.4133                   \n",
       "interesting_nidf_model_bfw_setting_06       0.4247                   \n",
       "interesting_nidf_model_bfw_setting_08       0.4502                   \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.3317                   \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.2709                   \n",
       "interesting_nidf_model_ct_setting0          0.3511                   \n",
       "interesting_nidf_model_ct_setting2          0.3835                   \n",
       "interesting_nidf_model_ct_setting4          0.3894                   \n",
       "interesting_nidf_model_ct_setting7          0.3683                   \n",
       "interesting_nidf_model_ct_setting9          0.3598                   \n",
       "repetition_model_setting05                  0.3915                   \n",
       "repetition_model_setting12                  0.3755                   \n",
       "repetition_model_setting35                  0.3589                   \n",
       "repetition_model_setting35_settinginf       0.3760                   \n",
       "repetition_model_settinginf                 0.3690                   \n",
       "responsiveness_model_bfw_setting_00         0.3660                   \n",
       "responsiveness_model_bfw_setting_05         0.3763                   \n",
       "responsiveness_model_bfw_setting_10         0.3890                   \n",
       "responsiveness_model_bfw_setting_13         0.3826                   \n",
       "responsiveness_model_bfw_setting_minus_10   0.3498                   \n",
       "\n",
       "                                             infersent_all_score_20495233  \\\n",
       "model_name                                                                  \n",
       "baseline_model                              0.5070                          \n",
       "greedy_model                                0.3029                          \n",
       "human_eval                                  0.4001                          \n",
       "inquisitive_model_ct_setting00              0.4545                          \n",
       "inquisitive_model_ct_setting01              0.4450                          \n",
       "inquisitive_model_ct_setting04              0.4643                          \n",
       "inquisitive_model_ct_setting07              0.4997                          \n",
       "inquisitive_model_ct_setting10              0.5485                          \n",
       "inquisitive_model_ct_setting10_better       0.5546                          \n",
       "interesting_nidf_model_bfw_setting_04       0.4836                          \n",
       "interesting_nidf_model_bfw_setting_06       0.5127                          \n",
       "interesting_nidf_model_bfw_setting_08       0.5418                          \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.4233                          \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.3357                          \n",
       "interesting_nidf_model_ct_setting0          0.4022                          \n",
       "interesting_nidf_model_ct_setting2          0.5073                          \n",
       "interesting_nidf_model_ct_setting4          0.4932                          \n",
       "interesting_nidf_model_ct_setting7          0.4769                          \n",
       "interesting_nidf_model_ct_setting9          0.4596                          \n",
       "repetition_model_setting05                  0.5132                          \n",
       "repetition_model_setting12                  0.4920                          \n",
       "repetition_model_setting35                  0.4580                          \n",
       "repetition_model_setting35_settinginf       0.4540                          \n",
       "repetition_model_settinginf                 0.4616                          \n",
       "responsiveness_model_bfw_setting_00         0.4538                          \n",
       "responsiveness_model_bfw_setting_05         0.4301                          \n",
       "responsiveness_model_bfw_setting_10         0.3958                          \n",
       "responsiveness_model_bfw_setting_13         0.3979                          \n",
       "responsiveness_model_bfw_setting_minus_10   0.4101                          \n",
       "\n",
       "                                             bertnli_all_score_20497378  \\\n",
       "model_name                                                                \n",
       "baseline_model                              0.9538                        \n",
       "greedy_model                                0.8436                        \n",
       "human_eval                                  0.8586                        \n",
       "inquisitive_model_ct_setting00              0.9323                        \n",
       "inquisitive_model_ct_setting01              0.9347                        \n",
       "inquisitive_model_ct_setting04              0.9343                        \n",
       "inquisitive_model_ct_setting07              0.9489                        \n",
       "inquisitive_model_ct_setting10              0.9507                        \n",
       "inquisitive_model_ct_setting10_better       0.9540                        \n",
       "interesting_nidf_model_bfw_setting_04       0.9255                        \n",
       "interesting_nidf_model_bfw_setting_06       0.9191                        \n",
       "interesting_nidf_model_bfw_setting_08       0.7591                        \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.9226                        \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.8243                        \n",
       "interesting_nidf_model_ct_setting0          0.9073                        \n",
       "interesting_nidf_model_ct_setting2          0.9346                        \n",
       "interesting_nidf_model_ct_setting4          0.9320                        \n",
       "interesting_nidf_model_ct_setting7          0.9359                        \n",
       "interesting_nidf_model_ct_setting9          0.9379                        \n",
       "repetition_model_setting05                  0.9528                        \n",
       "repetition_model_setting12                  0.9401                        \n",
       "repetition_model_setting35                  0.9384                        \n",
       "repetition_model_setting35_settinginf       0.9277                        \n",
       "repetition_model_settinginf                 0.9346                        \n",
       "responsiveness_model_bfw_setting_00         0.9265                        \n",
       "responsiveness_model_bfw_setting_05         0.9376                        \n",
       "responsiveness_model_bfw_setting_10         0.8960                        \n",
       "responsiveness_model_bfw_setting_13         0.8283                        \n",
       "responsiveness_model_bfw_setting_minus_10   0.8873                        \n",
       "\n",
       "                                             na_all_score_20488119  \\\n",
       "model_name                                                           \n",
       "baseline_model                              0.9333                   \n",
       "greedy_model                                0.8137                   \n",
       "human_eval                                  0.8537                   \n",
       "inquisitive_model_ct_setting00              0.9026                   \n",
       "inquisitive_model_ct_setting01              0.9137                   \n",
       "inquisitive_model_ct_setting04              0.9008                   \n",
       "inquisitive_model_ct_setting07              0.9319                   \n",
       "inquisitive_model_ct_setting10              0.9415                   \n",
       "inquisitive_model_ct_setting10_better       0.9574                   \n",
       "interesting_nidf_model_bfw_setting_04       0.9213                   \n",
       "interesting_nidf_model_bfw_setting_06       0.9151                   \n",
       "interesting_nidf_model_bfw_setting_08       0.7446                   \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.8864                   \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.7813                   \n",
       "interesting_nidf_model_ct_setting0          0.8786                   \n",
       "interesting_nidf_model_ct_setting2          0.9330                   \n",
       "interesting_nidf_model_ct_setting4          0.8982                   \n",
       "interesting_nidf_model_ct_setting7          0.8995                   \n",
       "interesting_nidf_model_ct_setting9          0.9128                   \n",
       "repetition_model_setting05                  0.9351                   \n",
       "repetition_model_setting12                  0.9322                   \n",
       "repetition_model_setting35                  0.9041                   \n",
       "repetition_model_setting35_settinginf       0.9029                   \n",
       "repetition_model_settinginf                 0.9094                   \n",
       "responsiveness_model_bfw_setting_00         0.8989                   \n",
       "responsiveness_model_bfw_setting_05         0.9145                   \n",
       "responsiveness_model_bfw_setting_10         0.8505                   \n",
       "responsiveness_model_bfw_setting_13         0.7619                   \n",
       "responsiveness_model_bfw_setting_minus_10   0.8430                   \n",
       "\n",
       "                                             na_all_context_score_20488122  \\\n",
       "model_name                                                                   \n",
       "baseline_model                              0.8270                           \n",
       "greedy_model                                0.7827                           \n",
       "human_eval                                  0.7621                           \n",
       "inquisitive_model_ct_setting00              0.8217                           \n",
       "inquisitive_model_ct_setting01              0.8223                           \n",
       "inquisitive_model_ct_setting04              0.8159                           \n",
       "inquisitive_model_ct_setting07              0.8282                           \n",
       "inquisitive_model_ct_setting10              0.8375                           \n",
       "inquisitive_model_ct_setting10_better       0.8432                           \n",
       "interesting_nidf_model_bfw_setting_04       0.8201                           \n",
       "interesting_nidf_model_bfw_setting_06       0.8211                           \n",
       "interesting_nidf_model_bfw_setting_08       0.6357                           \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.8149                           \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.6735                           \n",
       "interesting_nidf_model_ct_setting0          0.8144                           \n",
       "interesting_nidf_model_ct_setting2          0.8256                           \n",
       "interesting_nidf_model_ct_setting4          0.8303                           \n",
       "interesting_nidf_model_ct_setting7          0.8290                           \n",
       "interesting_nidf_model_ct_setting9          0.8266                           \n",
       "repetition_model_setting05                  0.8258                           \n",
       "repetition_model_setting12                  0.8284                           \n",
       "repetition_model_setting35                  0.8213                           \n",
       "repetition_model_setting35_settinginf       0.8244                           \n",
       "repetition_model_settinginf                 0.8239                           \n",
       "responsiveness_model_bfw_setting_00         0.8265                           \n",
       "responsiveness_model_bfw_setting_05         0.8244                           \n",
       "responsiveness_model_bfw_setting_10         0.7742                           \n",
       "responsiveness_model_bfw_setting_13         0.7006                           \n",
       "responsiveness_model_bfw_setting_minus_10   0.8051                           \n",
       "\n",
       "                                             na_all_context_score_20488123  \\\n",
       "model_name                                                                   \n",
       "baseline_model                              0.9493                           \n",
       "greedy_model                                0.8247                           \n",
       "human_eval                                  0.8002                           \n",
       "inquisitive_model_ct_setting00              0.9323                           \n",
       "inquisitive_model_ct_setting01              0.9337                           \n",
       "inquisitive_model_ct_setting04              0.9196                           \n",
       "inquisitive_model_ct_setting07              0.9461                           \n",
       "inquisitive_model_ct_setting10              0.9540                           \n",
       "inquisitive_model_ct_setting10_better       0.9644                           \n",
       "interesting_nidf_model_bfw_setting_04       0.9011                           \n",
       "interesting_nidf_model_bfw_setting_06       0.8877                           \n",
       "interesting_nidf_model_bfw_setting_08       0.7149                           \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.9249                           \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.8158                           \n",
       "interesting_nidf_model_ct_setting0          0.8982                           \n",
       "interesting_nidf_model_ct_setting2          0.9479                           \n",
       "interesting_nidf_model_ct_setting4          0.9247                           \n",
       "interesting_nidf_model_ct_setting7          0.9438                           \n",
       "interesting_nidf_model_ct_setting9          0.9395                           \n",
       "repetition_model_setting05                  0.9533                           \n",
       "repetition_model_setting12                  0.9538                           \n",
       "repetition_model_setting35                  0.9278                           \n",
       "repetition_model_setting35_settinginf       0.9299                           \n",
       "repetition_model_settinginf                 0.9342                           \n",
       "responsiveness_model_bfw_setting_00         0.9223                           \n",
       "responsiveness_model_bfw_setting_05         0.9404                           \n",
       "responsiveness_model_bfw_setting_10         0.8426                           \n",
       "responsiveness_model_bfw_setting_13         0.7291                           \n",
       "responsiveness_model_bfw_setting_minus_10   0.8821                           \n",
       "\n",
       "                                             ruber_all_score_20688482  \n",
       "model_name                                                             \n",
       "baseline_model                              0.4218                     \n",
       "greedy_model                                0.2831                     \n",
       "human_eval                                  0.3743                     \n",
       "inquisitive_model_ct_setting00              0.4371                     \n",
       "inquisitive_model_ct_setting01              0.4258                     \n",
       "inquisitive_model_ct_setting04              0.4350                     \n",
       "inquisitive_model_ct_setting07              0.4524                     \n",
       "inquisitive_model_ct_setting10              0.4807                     \n",
       "inquisitive_model_ct_setting10_better       0.4839                     \n",
       "interesting_nidf_model_bfw_setting_04       0.4708                     \n",
       "interesting_nidf_model_bfw_setting_06       0.4821                     \n",
       "interesting_nidf_model_bfw_setting_08       0.5172                     \n",
       "interesting_nidf_model_bfw_setting_minus_04 0.3909                     \n",
       "interesting_nidf_model_bfw_setting_minus_10 0.2949                     \n",
       "interesting_nidf_model_ct_setting0          0.3742                     \n",
       "interesting_nidf_model_ct_setting2          0.4592                     \n",
       "interesting_nidf_model_ct_setting4          0.4573                     \n",
       "interesting_nidf_model_ct_setting7          0.4451                     \n",
       "interesting_nidf_model_ct_setting9          0.4238                     \n",
       "repetition_model_setting05                  0.4447                     \n",
       "repetition_model_setting12                  0.4315                     \n",
       "repetition_model_setting35                  0.4258                     \n",
       "repetition_model_setting35_settinginf       0.4379                     \n",
       "repetition_model_settinginf                 0.4416                     \n",
       "responsiveness_model_bfw_setting_00         0.4277                     \n",
       "responsiveness_model_bfw_setting_05         0.4350                     \n",
       "responsiveness_model_bfw_setting_10         0.4195                     \n",
       "responsiveness_model_bfw_setting_13         0.4178                     \n",
       "responsiveness_model_bfw_setting_minus_10   0.4075                     "
      ]
     },
     "execution_count": 600,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conv_dial_scores.groupby(['model_name']).mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### How does the human annotation scores look"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f8653bac438>"
      ]
     },
     "execution_count": 121,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "conv_dial_scores.enjoy.hist()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### How does each model score distribution look?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 330,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f8642931048>"
      ]
     },
     "execution_count": 330,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "conv_dial_scores.infersent_all_score_20495233.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 331,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f864251b0f0>"
      ]
     },
     "execution_count": 331,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "conv_dial_scores.bertnli_all_score_20497378.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 332,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f8621eeaa58>"
      ]
     },
     "execution_count": 332,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "conv_dial_scores.na_all_score_20488119.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 601,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f86426fc6a0>"
      ]
     },
     "execution_count": 601,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 3600x2400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "conv_dial_scores.ruber_all_score_20688482.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model_name</th>\n",
       "      <th>model_type</th>\n",
       "      <th>model_bucket</th>\n",
       "      <th>dialog</th>\n",
       "      <th>assignment_id_hashed</th>\n",
       "      <th>evaluator_id_hashed</th>\n",
       "      <th>enjoy</th>\n",
       "      <th>interest</th>\n",
       "      <th>listen</th>\n",
       "      <th>turing</th>\n",
       "      <th>avoid_rep</th>\n",
       "      <th>make_sense</th>\n",
       "      <th>fluency</th>\n",
       "      <th>persona_guess</th>\n",
       "      <th>inquisitive</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>768</th>\n",
       "      <td>greedy_model</td>\n",
       "      <td>greedy</td>\n",
       "      <td></td>\n",
       "      <td>[{'speaker': 'human_evaluator', 'text': 'Hello...</td>\n",
       "      <td>a706730a6ee7880a2a0a3ec996c9d2a5aa869edd</td>\n",
       "      <td>ed4672327c64575a1be57fe0917cfc49c0d948c0</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2897</th>\n",
       "      <td>inquisitive_model_ct_setting10_better</td>\n",
       "      <td>inquisitive-ct</td>\n",
       "      <td>011</td>\n",
       "      <td>[{'speaker': 'human_evaluator', 'text': 'hi th...</td>\n",
       "      <td>970eafa6cb745ff152115946d1787c63d9db2c35</td>\n",
       "      <td>36b72818d076742ba2dffb0dbec91e657917af42</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1870</th>\n",
       "      <td>repetition_model_setting35</td>\n",
       "      <td>repetition-bfw</td>\n",
       "      <td>2gram=035</td>\n",
       "      <td>[{'speaker': 'model', 'text': 'Hi how are you ...</td>\n",
       "      <td>13f2a54504ec8d8474301e6b406e30000564aba2</td>\n",
       "      <td>d8f9fe2b338c05ce8e9ed30478ef09d1fa693c62</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1613</th>\n",
       "      <td>inquisitive_model_ct_setting07</td>\n",
       "      <td>inquisitive-ct</td>\n",
       "      <td>007</td>\n",
       "      <td>[{'speaker': 'model', 'text': 'How are you doi...</td>\n",
       "      <td>c556893edf878c4d280370463fce760357a5ffb9</td>\n",
       "      <td>9b6c716b0a858f39e63c50206326eb74e8025615</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2269</th>\n",
       "      <td>interesting_nidf_model_ct_setting7</td>\n",
       "      <td>nidf_ct</td>\n",
       "      <td>007</td>\n",
       "      <td>[{'speaker': 'model', 'text': 'Hi how ar you d...</td>\n",
       "      <td>7e6088da279a5251111f57a2fa0b9df37b0f5aac</td>\n",
       "      <td>660210871d7f34227c595b02363f4c1aee99cece</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 model_name      model_type model_bucket  \\\n",
       "768                            greedy_model          greedy                \n",
       "2897  inquisitive_model_ct_setting10_better  inquisitive-ct          011   \n",
       "1870             repetition_model_setting35  repetition-bfw    2gram=035   \n",
       "1613         inquisitive_model_ct_setting07  inquisitive-ct          007   \n",
       "2269     interesting_nidf_model_ct_setting7         nidf_ct          007   \n",
       "\n",
       "                                                 dialog  \\\n",
       "768   [{'speaker': 'human_evaluator', 'text': 'Hello...   \n",
       "2897  [{'speaker': 'human_evaluator', 'text': 'hi th...   \n",
       "1870  [{'speaker': 'model', 'text': 'Hi how are you ...   \n",
       "1613  [{'speaker': 'model', 'text': 'How are you doi...   \n",
       "2269  [{'speaker': 'model', 'text': 'Hi how ar you d...   \n",
       "\n",
       "                          assignment_id_hashed  \\\n",
       "768   a706730a6ee7880a2a0a3ec996c9d2a5aa869edd   \n",
       "2897  970eafa6cb745ff152115946d1787c63d9db2c35   \n",
       "1870  13f2a54504ec8d8474301e6b406e30000564aba2   \n",
       "1613  c556893edf878c4d280370463fce760357a5ffb9   \n",
       "2269  7e6088da279a5251111f57a2fa0b9df37b0f5aac   \n",
       "\n",
       "                           evaluator_id_hashed  enjoy  interest  listen  \\\n",
       "768   ed4672327c64575a1be57fe0917cfc49c0d948c0      2         3       3   \n",
       "2897  36b72818d076742ba2dffb0dbec91e657917af42      3         3       3   \n",
       "1870  d8f9fe2b338c05ce8e9ed30478ef09d1fa693c62      4         4       4   \n",
       "1613  9b6c716b0a858f39e63c50206326eb74e8025615      4         4       3   \n",
       "2269  660210871d7f34227c595b02363f4c1aee99cece      4         3       3   \n",
       "\n",
       "      turing  avoid_rep  make_sense  fluency  persona_guess  inquisitive  \n",
       "768        2          2           4        3              1            2  \n",
       "2897       3          3           4        4              1            2  \n",
       "1870       4          3           4        4              1            3  \n",
       "1613       2          3           4        4              1            3  \n",
       "2269       4          3           4        3              1            1  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conversation_table.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>enjoy</th>\n",
       "      <th>interest</th>\n",
       "      <th>listen</th>\n",
       "      <th>turing</th>\n",
       "      <th>avoid_rep</th>\n",
       "      <th>make_sense</th>\n",
       "      <th>fluency</th>\n",
       "      <th>persona_guess</th>\n",
       "      <th>inquisitive</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>model_type</th>\n",
       "      <th>model_bucket</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>baseline</th>\n",
       "      <th></th>\n",
       "      <td>2.3514</td>\n",
       "      <td>2.3514</td>\n",
       "      <td>2.6306</td>\n",
       "      <td>1.8108</td>\n",
       "      <td>2.1441</td>\n",
       "      <td>3.3964</td>\n",
       "      <td>3.2252</td>\n",
       "      <td>0.7748</td>\n",
       "      <td>2.4955</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>greedy</th>\n",
       "      <th></th>\n",
       "      <td>2.3060</td>\n",
       "      <td>2.3582</td>\n",
       "      <td>2.7836</td>\n",
       "      <td>1.7836</td>\n",
       "      <td>2.1567</td>\n",
       "      <td>3.3284</td>\n",
       "      <td>3.2015</td>\n",
       "      <td>0.8657</td>\n",
       "      <td>2.0000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>human</th>\n",
       "      <th></th>\n",
       "      <td>3.3137</td>\n",
       "      <td>3.2255</td>\n",
       "      <td>3.6373</td>\n",
       "      <td>3.4020</td>\n",
       "      <td>2.9020</td>\n",
       "      <td>3.8431</td>\n",
       "      <td>3.6569</td>\n",
       "      <td>0.9216</td>\n",
       "      <td>2.6275</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"6\" valign=\"top\">inquisitive-ct</th>\n",
       "      <th>000</th>\n",
       "      <td>2.9159</td>\n",
       "      <td>2.8505</td>\n",
       "      <td>3.2897</td>\n",
       "      <td>2.4860</td>\n",
       "      <td>2.9533</td>\n",
       "      <td>3.5607</td>\n",
       "      <td>3.7009</td>\n",
       "      <td>0.8598</td>\n",
       "      <td>2.4766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>001</th>\n",
       "      <td>2.9355</td>\n",
       "      <td>2.8548</td>\n",
       "      <td>3.3226</td>\n",
       "      <td>2.4677</td>\n",
       "      <td>2.8790</td>\n",
       "      <td>3.6290</td>\n",
       "      <td>3.5887</td>\n",
       "      <td>0.8468</td>\n",
       "      <td>2.5161</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>004</th>\n",
       "      <td>2.8839</td>\n",
       "      <td>2.8214</td>\n",
       "      <td>3.3661</td>\n",
       "      <td>2.4196</td>\n",
       "      <td>2.8750</td>\n",
       "      <td>3.6339</td>\n",
       "      <td>3.5893</td>\n",
       "      <td>0.8393</td>\n",
       "      <td>2.5536</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>007</th>\n",
       "      <td>3.0704</td>\n",
       "      <td>2.9718</td>\n",
       "      <td>3.2254</td>\n",
       "      <td>2.4225</td>\n",
       "      <td>2.8803</td>\n",
       "      <td>3.5282</td>\n",
       "      <td>3.6690</td>\n",
       "      <td>0.8028</td>\n",
       "      <td>2.7465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>010</th>\n",
       "      <td>2.9048</td>\n",
       "      <td>2.7238</td>\n",
       "      <td>3.1238</td>\n",
       "      <td>2.4286</td>\n",
       "      <td>2.7429</td>\n",
       "      <td>3.5905</td>\n",
       "      <td>3.6952</td>\n",
       "      <td>0.7905</td>\n",
       "      <td>2.7143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>011</th>\n",
       "      <td>2.8440</td>\n",
       "      <td>2.8257</td>\n",
       "      <td>3.1835</td>\n",
       "      <td>2.2569</td>\n",
       "      <td>2.7615</td>\n",
       "      <td>3.5229</td>\n",
       "      <td>3.5963</td>\n",
       "      <td>0.7156</td>\n",
       "      <td>2.9358</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"5\" valign=\"top\">nidf_bfw</th>\n",
       "      <th>-04</th>\n",
       "      <td>2.7805</td>\n",
       "      <td>2.6423</td>\n",
       "      <td>3.2846</td>\n",
       "      <td>2.4065</td>\n",
       "      <td>2.9024</td>\n",
       "      <td>3.5610</td>\n",
       "      <td>3.5528</td>\n",
       "      <td>0.8211</td>\n",
       "      <td>2.5203</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>-10</th>\n",
       "      <td>2.4322</td>\n",
       "      <td>2.3814</td>\n",
       "      <td>3.0339</td>\n",
       "      <td>2.1525</td>\n",
       "      <td>2.8475</td>\n",
       "      <td>3.3305</td>\n",
       "      <td>3.3390</td>\n",
       "      <td>0.7119</td>\n",
       "      <td>2.3136</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>004</th>\n",
       "      <td>2.9904</td>\n",
       "      <td>3.0000</td>\n",
       "      <td>3.3654</td>\n",
       "      <td>2.4904</td>\n",
       "      <td>2.9519</td>\n",
       "      <td>3.6346</td>\n",
       "      <td>3.6538</td>\n",
       "      <td>0.9327</td>\n",
       "      <td>2.6538</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>006</th>\n",
       "      <td>2.9633</td>\n",
       "      <td>3.0550</td>\n",
       "      <td>3.2385</td>\n",
       "      <td>2.4128</td>\n",
       "      <td>2.9266</td>\n",
       "      <td>3.4954</td>\n",
       "      <td>3.5229</td>\n",
       "      <td>0.9266</td>\n",
       "      <td>2.5780</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>008</th>\n",
       "      <td>2.4000</td>\n",
       "      <td>2.5500</td>\n",
       "      <td>2.6083</td>\n",
       "      <td>1.8583</td>\n",
       "      <td>2.7833</td>\n",
       "      <td>2.9083</td>\n",
       "      <td>2.6750</td>\n",
       "      <td>0.9167</td>\n",
       "      <td>2.0333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"5\" valign=\"top\">nidf_ct</th>\n",
       "      <th>000</th>\n",
       "      <td>2.9623</td>\n",
       "      <td>2.8585</td>\n",
       "      <td>3.2925</td>\n",
       "      <td>2.4151</td>\n",
       "      <td>2.8302</td>\n",
       "      <td>3.6604</td>\n",
       "      <td>3.6226</td>\n",
       "      <td>0.7170</td>\n",
       "      <td>2.6038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>002</th>\n",
       "      <td>2.7759</td>\n",
       "      <td>2.8017</td>\n",
       "      <td>3.1379</td>\n",
       "      <td>2.3707</td>\n",
       "      <td>2.8966</td>\n",
       "      <td>3.5000</td>\n",
       "      <td>3.6034</td>\n",
       "      <td>0.8103</td>\n",
       "      <td>2.6638</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>004</th>\n",
       "      <td>2.8100</td>\n",
       "      <td>2.8000</td>\n",
       "      <td>3.2500</td>\n",
       "      <td>2.3400</td>\n",
       "      <td>2.9200</td>\n",
       "      <td>3.5000</td>\n",
       "      <td>3.6500</td>\n",
       "      <td>0.8600</td>\n",
       "      <td>2.5700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>007</th>\n",
       "      <td>3.0000</td>\n",
       "      <td>2.9048</td>\n",
       "      <td>3.3429</td>\n",
       "      <td>2.5333</td>\n",
       "      <td>2.8857</td>\n",
       "      <td>3.5905</td>\n",
       "      <td>3.6381</td>\n",
       "      <td>0.8190</td>\n",
       "      <td>2.5619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>009</th>\n",
       "      <td>2.8304</td>\n",
       "      <td>2.8393</td>\n",
       "      <td>3.0714</td>\n",
       "      <td>2.4018</td>\n",
       "      <td>2.9018</td>\n",
       "      <td>3.5804</td>\n",
       "      <td>3.6071</td>\n",
       "      <td>0.8839</td>\n",
       "      <td>2.3125</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"5\" valign=\"top\">repetition-bfw</th>\n",
       "      <th>2gram=005</th>\n",
       "      <td>2.5583</td>\n",
       "      <td>2.6083</td>\n",
       "      <td>3.0833</td>\n",
       "      <td>2.1917</td>\n",
       "      <td>2.6583</td>\n",
       "      <td>3.6000</td>\n",
       "      <td>3.5667</td>\n",
       "      <td>0.7500</td>\n",
       "      <td>2.6750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2gram=012</th>\n",
       "      <td>2.9083</td>\n",
       "      <td>2.8624</td>\n",
       "      <td>3.2110</td>\n",
       "      <td>2.3211</td>\n",
       "      <td>2.8440</td>\n",
       "      <td>3.6422</td>\n",
       "      <td>3.5872</td>\n",
       "      <td>0.7248</td>\n",
       "      <td>2.6330</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2gram=035</th>\n",
       "      <td>2.9504</td>\n",
       "      <td>2.8760</td>\n",
       "      <td>3.2727</td>\n",
       "      <td>2.4463</td>\n",
       "      <td>2.9008</td>\n",
       "      <td>3.6777</td>\n",
       "      <td>3.7273</td>\n",
       "      <td>0.8017</td>\n",
       "      <td>2.5455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2gram=035 stop=inf</th>\n",
       "      <td>2.8938</td>\n",
       "      <td>2.9646</td>\n",
       "      <td>3.2478</td>\n",
       "      <td>2.5044</td>\n",
       "      <td>2.8938</td>\n",
       "      <td>3.6814</td>\n",
       "      <td>3.6637</td>\n",
       "      <td>0.8673</td>\n",
       "      <td>2.6991</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2gram=999</th>\n",
       "      <td>2.9587</td>\n",
       "      <td>2.8595</td>\n",
       "      <td>3.3058</td>\n",
       "      <td>2.4050</td>\n",
       "      <td>2.8182</td>\n",
       "      <td>3.6612</td>\n",
       "      <td>3.6446</td>\n",
       "      <td>0.9091</td>\n",
       "      <td>2.6529</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"5\" valign=\"top\">responsive</th>\n",
       "      <th>-10</th>\n",
       "      <td>2.4790</td>\n",
       "      <td>2.5294</td>\n",
       "      <td>2.8403</td>\n",
       "      <td>2.0168</td>\n",
       "      <td>2.8571</td>\n",
       "      <td>3.1429</td>\n",
       "      <td>3.4202</td>\n",
       "      <td>0.9076</td>\n",
       "      <td>2.3782</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>000</th>\n",
       "      <td>3.0088</td>\n",
       "      <td>2.9204</td>\n",
       "      <td>3.3717</td>\n",
       "      <td>2.7257</td>\n",
       "      <td>2.9646</td>\n",
       "      <td>3.7345</td>\n",
       "      <td>3.7168</td>\n",
       "      <td>0.8230</td>\n",
       "      <td>2.5575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>005</th>\n",
       "      <td>2.8818</td>\n",
       "      <td>2.8545</td>\n",
       "      <td>3.2727</td>\n",
       "      <td>2.4091</td>\n",
       "      <td>2.9000</td>\n",
       "      <td>3.4909</td>\n",
       "      <td>3.5091</td>\n",
       "      <td>0.8182</td>\n",
       "      <td>2.5273</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>010</th>\n",
       "      <td>2.3879</td>\n",
       "      <td>2.5690</td>\n",
       "      <td>3.0259</td>\n",
       "      <td>1.9655</td>\n",
       "      <td>2.7845</td>\n",
       "      <td>3.1552</td>\n",
       "      <td>3.0603</td>\n",
       "      <td>0.7500</td>\n",
       "      <td>2.2241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>013</th>\n",
       "      <td>2.1043</td>\n",
       "      <td>2.3304</td>\n",
       "      <td>2.6870</td>\n",
       "      <td>1.8087</td>\n",
       "      <td>2.7130</td>\n",
       "      <td>2.6957</td>\n",
       "      <td>2.5391</td>\n",
       "      <td>0.6174</td>\n",
       "      <td>2.1391</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                   enjoy  interest  listen  turing  avoid_rep  \\\n",
       "model_type     model_bucket                                                     \n",
       "baseline                          2.3514 2.3514    2.6306  1.8108  2.1441       \n",
       "greedy                            2.3060 2.3582    2.7836  1.7836  2.1567       \n",
       "human                             3.3137 3.2255    3.6373  3.4020  2.9020       \n",
       "inquisitive-ct 000                2.9159 2.8505    3.2897  2.4860  2.9533       \n",
       "               001                2.9355 2.8548    3.3226  2.4677  2.8790       \n",
       "               004                2.8839 2.8214    3.3661  2.4196  2.8750       \n",
       "               007                3.0704 2.9718    3.2254  2.4225  2.8803       \n",
       "               010                2.9048 2.7238    3.1238  2.4286  2.7429       \n",
       "               011                2.8440 2.8257    3.1835  2.2569  2.7615       \n",
       "nidf_bfw       -04                2.7805 2.6423    3.2846  2.4065  2.9024       \n",
       "               -10                2.4322 2.3814    3.0339  2.1525  2.8475       \n",
       "               004                2.9904 3.0000    3.3654  2.4904  2.9519       \n",
       "               006                2.9633 3.0550    3.2385  2.4128  2.9266       \n",
       "               008                2.4000 2.5500    2.6083  1.8583  2.7833       \n",
       "nidf_ct        000                2.9623 2.8585    3.2925  2.4151  2.8302       \n",
       "               002                2.7759 2.8017    3.1379  2.3707  2.8966       \n",
       "               004                2.8100 2.8000    3.2500  2.3400  2.9200       \n",
       "               007                3.0000 2.9048    3.3429  2.5333  2.8857       \n",
       "               009                2.8304 2.8393    3.0714  2.4018  2.9018       \n",
       "repetition-bfw 2gram=005          2.5583 2.6083    3.0833  2.1917  2.6583       \n",
       "               2gram=012          2.9083 2.8624    3.2110  2.3211  2.8440       \n",
       "               2gram=035          2.9504 2.8760    3.2727  2.4463  2.9008       \n",
       "               2gram=035 stop=inf 2.8938 2.9646    3.2478  2.5044  2.8938       \n",
       "               2gram=999          2.9587 2.8595    3.3058  2.4050  2.8182       \n",
       "responsive     -10                2.4790 2.5294    2.8403  2.0168  2.8571       \n",
       "               000                3.0088 2.9204    3.3717  2.7257  2.9646       \n",
       "               005                2.8818 2.8545    3.2727  2.4091  2.9000       \n",
       "               010                2.3879 2.5690    3.0259  1.9655  2.7845       \n",
       "               013                2.1043 2.3304    2.6870  1.8087  2.7130       \n",
       "\n",
       "                                   make_sense  fluency  persona_guess  \\\n",
       "model_type     model_bucket                                             \n",
       "baseline                          3.3964      3.2252   0.7748           \n",
       "greedy                            3.3284      3.2015   0.8657           \n",
       "human                             3.8431      3.6569   0.9216           \n",
       "inquisitive-ct 000                3.5607      3.7009   0.8598           \n",
       "               001                3.6290      3.5887   0.8468           \n",
       "               004                3.6339      3.5893   0.8393           \n",
       "               007                3.5282      3.6690   0.8028           \n",
       "               010                3.5905      3.6952   0.7905           \n",
       "               011                3.5229      3.5963   0.7156           \n",
       "nidf_bfw       -04                3.5610      3.5528   0.8211           \n",
       "               -10                3.3305      3.3390   0.7119           \n",
       "               004                3.6346      3.6538   0.9327           \n",
       "               006                3.4954      3.5229   0.9266           \n",
       "               008                2.9083      2.6750   0.9167           \n",
       "nidf_ct        000                3.6604      3.6226   0.7170           \n",
       "               002                3.5000      3.6034   0.8103           \n",
       "               004                3.5000      3.6500   0.8600           \n",
       "               007                3.5905      3.6381   0.8190           \n",
       "               009                3.5804      3.6071   0.8839           \n",
       "repetition-bfw 2gram=005          3.6000      3.5667   0.7500           \n",
       "               2gram=012          3.6422      3.5872   0.7248           \n",
       "               2gram=035          3.6777      3.7273   0.8017           \n",
       "               2gram=035 stop=inf 3.6814      3.6637   0.8673           \n",
       "               2gram=999          3.6612      3.6446   0.9091           \n",
       "responsive     -10                3.1429      3.4202   0.9076           \n",
       "               000                3.7345      3.7168   0.8230           \n",
       "               005                3.4909      3.5091   0.8182           \n",
       "               010                3.1552      3.0603   0.7500           \n",
       "               013                2.6957      2.5391   0.6174           \n",
       "\n",
       "                                   inquisitive  \n",
       "model_type     model_bucket                     \n",
       "baseline                          2.4955        \n",
       "greedy                            2.0000        \n",
       "human                             2.6275        \n",
       "inquisitive-ct 000                2.4766        \n",
       "               001                2.5161        \n",
       "               004                2.5536        \n",
       "               007                2.7465        \n",
       "               010                2.7143        \n",
       "               011                2.9358        \n",
       "nidf_bfw       -04                2.5203        \n",
       "               -10                2.3136        \n",
       "               004                2.6538        \n",
       "               006                2.5780        \n",
       "               008                2.0333        \n",
       "nidf_ct        000                2.6038        \n",
       "               002                2.6638        \n",
       "               004                2.5700        \n",
       "               007                2.5619        \n",
       "               009                2.3125        \n",
       "repetition-bfw 2gram=005          2.6750        \n",
       "               2gram=012          2.6330        \n",
       "               2gram=035          2.5455        \n",
       "               2gram=035 stop=inf 2.6991        \n",
       "               2gram=999          2.6529        \n",
       "responsive     -10                2.3782        \n",
       "               000                2.5575        \n",
       "               005                2.5273        \n",
       "               010                2.2241        \n",
       "               013                2.1391        "
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conversation_table.groupby([\"model_type\", \"model_bucket\"]).mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bayesian calibration helpers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 500,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('ggplot')\n",
    "plt.rcParams['figure.figsize'] = [12.0, 8.0]\n",
    "\n",
    "plt.rcParams['axes.facecolor'] = '#FFFFFF'\n",
    "plt.rcParams['grid.color'] = '#E3E3E3'\n",
    "plt.rcParams['legend.framealpha'] = 1.0\n",
    "plt.rcParams['axes.labelweight'] = 'normal'\n",
    "\n",
    "MAIN_COLOR = 'k'\n",
    "\n",
    "CM = 0.393700787\n",
    "\n",
    "SOLO_FIG_SIZE = (16 * CM, CM * 16 * 0.5)\n",
    "\n",
    "nice_names = {\n",
    "    'repetition++': 'Repetition-controlled baseline+',\n",
    "    'baseline_model': 'Beam search baseline',\n",
    "    'inquisitive-ct': 'Question-controlled CT',\n",
    "    'human_eval': 'Human',\n",
    "    'repetition_model_setting35_settinginf': 'Repetition-controlled baseline',\n",
    "    'nidf_ct': 'Specificity-controlled CT',\n",
    "    'nidf_bfw': 'Specificity-controlled WD',\n",
    "    'responsive': 'Response-related controlled WD',\n",
    "    'greedy_model': 'Greedy search baseline',\n",
    "    'repetition_model_setting05': 'Extrep\\nbigram -0.5',\n",
    "    'repetition_model_setting12': 'Extrep\\nbigram -1.25',\n",
    "    'repetition_model_setting35': 'Extrep\\nbigram -3.5',\n",
    "    'repetition_model_settinginf': 'Extrep\\nbigram -inf',\n",
    "}\n",
    "\n",
    "nice_names_rep = {\n",
    "    **nice_names,\n",
    "\n",
    "    'repetition_model_setting35_settinginf': 'Extrep bigram -3.5\\n+ unigram -inf',\n",
    "    'repetition++': '+ Partner\\nIntrep bigram',\n",
    "\n",
    "}\n",
    "\n",
    "nice_colors = {\n",
    "    'repetition_model_setting35_settinginf': 'C2',\n",
    "    'repetition++': 'C2',\n",
    "    'human_eval': '#d89a2f',\n",
    "    'baseline_model': 'C1',\n",
    "    'greedy_model': 'C5',\n",
    "    'inquisitive_model_ct_setting07': 'C3',\n",
    "    'interesting_nidf_model_bfw_setting_04': 'C6',\n",
    "}\n",
    "nice_styles = {\n",
    "    'repetition_model_setting35_settinginf': '--',\n",
    "    'repetition++': '--',\n",
    "    'human_eval': '-.',\n",
    "    'baseline_model': ':',\n",
    "    'greedy_model': '--',\n",
    "}\n",
    "nice_metric_names = {\n",
    "    'enjoy': 'Engagingness',\n",
    "    'interest': 'Interestingness',\n",
    "    'inquisitive': 'Inquisitiveness',\n",
    "    'listen': 'Listening',\n",
    "    'avoid_rep': 'Avoiding Repetition',\n",
    "    'fluency': 'Fluency',\n",
    "    'make_sense': 'Making Sense',\n",
    "    'turing': 'Humanness', \n",
    "    'persona_guess': 'Persona retrieval',\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 501,
   "metadata": {},
   "outputs": [],
   "source": [
    "LIKERT_METRICS = ['fluency', 'enjoy', 'turing', 'make_sense', 'inquisitive', 'interest', 'avoid_rep', 'listen']\n",
    "def modelname_subset(data, model_keys):\n",
    "    return data[data.model_name.apply(lambda x: x in model_keys)]\n",
    "\n",
    "def modeltype_subset(data, model_keys):\n",
    "    return data[data.model_type.apply(lambda x: x in model_keys)].sort_values(['model_type', 'model_bucket'])\n",
    "\n",
    "import os\n",
    "HOME = os.environ['HOME']\n",
    "try:\n",
    "    os.mkdir(HOME + '/plots')\n",
    "except IOError:\n",
    "    pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Output latex for the raw scores table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 502,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_raw = conversation_table.copy().groupby(['model_name', 'model_type', 'model_bucket']).aggregate({\n",
    "    k: [np.mean, np.std]\n",
    "    for k in LIKERT_METRICS + ['persona_guess']\n",
    "}).reset_index()\n",
    "results_raw.columns = ['_'.join([c for c in col if c.strip()]) for col in results_raw.columns.values]\n",
    "results_raw = results_raw.set_index('model_name')\n",
    "\n",
    "# nice_raw = []\n",
    "# colmax = results_raw[results_raw.index != 'human_eval'].max(axis=0)\n",
    "# for model_name, row in results_raw.iterrows():\n",
    "#     x = {}\n",
    "#     for m in LIKERT_METRICS + ['persona_guess']:\n",
    "#         val = '{:.2f} $\\\\pm$ {:.2f}'.format(row[m + '_mean'], row[m + '_std'])\n",
    "#         if row[m + '_mean'] == colmax[m + '_mean']:\n",
    "#             val = '{\\\\bf ' + val + '}'\n",
    "#         x[nice_metric_names[m]] = val\n",
    "#     x['model_name'] = nice_names.get(\n",
    "#         model_name,\n",
    "#         nice_names.get(\n",
    "#             row['model_type'],\n",
    "#             row['model_type']\n",
    "#         ) + ' ' + str(row['model_bucket'])\n",
    "#     )\n",
    "#     nice_raw.append(x)\n",
    "\n",
    "# nice_raw = pd.DataFrame(nice_raw).set_index('model_name')\n",
    "# print(nice_raw.to_latex(escape=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 517,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 519,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_bayes.index = results_bayes.index.str.strip()\n",
    "# things that have two names\n",
    "results_bayes.loc['repetition++'] = results_bayes.loc['responsiveness_model_bfw_setting_00']\n",
    "results_bayes.loc['repetition++','model_type'] = 'repetition-bfw'\n",
    "\n",
    "results_bayes.loc['interesting_nidf_model_bfw_setting_00'] = results_bayes.loc['repetition_model_setting35_settinginf']\n",
    "results_bayes.loc['interesting_nidf_model_bfw_setting_00','model_type'] = 'nidf_bfw'\n",
    "results_bayes.loc['interesting_nidf_model_bfw_setting_00','model_bucket'] = '00'\n",
    "\n",
    "results_bayes.loc['inquisitive_model_ct_setting10_better', 'model_bucket'] = '12'\n",
    "\n",
    "# prep results\n",
    "altered = results_bayes.copy().reset_index()\n",
    "altered[\"model_bucket\"] = altered.model_bucket.astype('category')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 363,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.31625655506252526"
      ]
     },
     "execution_count": 363,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "altered['fluency_mean'].corr(altered['na_all'], method='spearman')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\\begin{tabular}{lllllllll}\n",
      "\\toprule\n",
      "{} &    Avoiding Repetition &           Engagingness &                Fluency &              Humanness &        Inquisitiveness &        Interestingness &              Listening &           Making Sense \\\\\n",
      "model_name                         &                        &                        &                        &                        &                        &                        &                        &                        \\\\\n",
      "\\midrule\n",
      "Beam search baseline               &        2.08 $\\pm$ 0.11 &        2.29 $\\pm$ 0.11 &        3.09 $\\pm$ 0.13 &        1.71 $\\pm$ 0.13 &        2.42 $\\pm$ 0.11 &        2.29 $\\pm$ 0.14 &        2.47 $\\pm$ 0.12 &        3.35 $\\pm$ 0.13 \\\\\n",
      "Greedy search baseline             &        2.08 $\\pm$ 0.10 &        2.24 $\\pm$ 0.11 &        3.03 $\\pm$ 0.10 &        1.75 $\\pm$ 0.12 &        1.95 $\\pm$ 0.10 &        2.29 $\\pm$ 0.13 &        2.62 $\\pm$ 0.10 &        3.23 $\\pm$ 0.10 \\\\\n",
      "Human                              &        2.79 $\\pm$ 0.12 &        3.04 $\\pm$ 0.11 &        3.36 $\\pm$ 0.12 &        3.35 $\\pm$ 0.11 &        2.44 $\\pm$ 0.12 &        2.92 $\\pm$ 0.11 &        3.32 $\\pm$ 0.13 &        3.68 $\\pm$ 0.11 \\\\\n",
      "Question-controlled CT 000         &        2.87 $\\pm$ 0.12 &        2.84 $\\pm$ 0.13 &        3.51 $\\pm$ 0.10 &        2.46 $\\pm$ 0.11 &        2.36 $\\pm$ 0.09 &        2.76 $\\pm$ 0.09 &        3.10 $\\pm$ 0.10 &        3.49 $\\pm$ 0.12 \\\\\n",
      "Question-controlled CT 001         &        2.82 $\\pm$ 0.11 &        2.88 $\\pm$ 0.11 &        3.42 $\\pm$ 0.10 &        2.46 $\\pm$ 0.12 &        2.47 $\\pm$ 0.11 &        2.79 $\\pm$ 0.13 &        3.14 $\\pm$ 0.11 &        3.55 $\\pm$ 0.10 \\\\\n",
      "Question-controlled CT 004         &        2.78 $\\pm$ 0.12 &        2.88 $\\pm$ 0.10 &        3.47 $\\pm$ 0.11 &        2.40 $\\pm$ 0.09 &        2.53 $\\pm$ 0.13 &        2.83 $\\pm$ 0.13 &  {\\bf 3.24 $\\pm$ 0.11} &        3.59 $\\pm$ 0.10 \\\\\n",
      "Question-controlled CT 007         &        2.81 $\\pm$ 0.10 &  {\\bf 2.99 $\\pm$ 0.11} &        3.54 $\\pm$ 0.09 &        2.35 $\\pm$ 0.11 &        2.66 $\\pm$ 0.12 &        2.92 $\\pm$ 0.12 &        3.11 $\\pm$ 0.10 &        3.47 $\\pm$ 0.10 \\\\\n",
      "Question-controlled CT 010         &        2.67 $\\pm$ 0.13 &        2.87 $\\pm$ 0.11 &        3.52 $\\pm$ 0.12 &        2.35 $\\pm$ 0.12 &        2.63 $\\pm$ 0.12 &        2.66 $\\pm$ 0.10 &        2.94 $\\pm$ 0.11 &        3.53 $\\pm$ 0.12 \\\\\n",
      "Question-controlled CT 12          &        2.68 $\\pm$ 0.12 &        2.74 $\\pm$ 0.09 &        3.42 $\\pm$ 0.12 &        2.19 $\\pm$ 0.13 &  {\\bf 2.79 $\\pm$ 0.11} &        2.74 $\\pm$ 0.11 &        3.00 $\\pm$ 0.12 &        3.45 $\\pm$ 0.13 \\\\\n",
      "Specificity-controlled WD 004      &        2.84 $\\pm$ 0.10 &        2.96 $\\pm$ 0.12 &        3.45 $\\pm$ 0.13 &        2.44 $\\pm$ 0.12 &        2.56 $\\pm$ 0.09 &  {\\bf 2.94 $\\pm$ 0.11} &        3.20 $\\pm$ 0.10 &        3.54 $\\pm$ 0.11 \\\\\n",
      "Specificity-controlled WD 006      &        2.81 $\\pm$ 0.09 &        2.91 $\\pm$ 0.10 &        3.34 $\\pm$ 0.09 &        2.31 $\\pm$ 0.11 &        2.53 $\\pm$ 0.12 &        2.93 $\\pm$ 0.12 &        3.09 $\\pm$ 0.10 &        3.41 $\\pm$ 0.12 \\\\\n",
      "Specificity-controlled WD 008      &        2.70 $\\pm$ 0.11 &        2.39 $\\pm$ 0.12 &        2.54 $\\pm$ 0.12 &        1.80 $\\pm$ 0.13 &        2.00 $\\pm$ 0.10 &        2.49 $\\pm$ 0.12 &        2.47 $\\pm$ 0.10 &        2.87 $\\pm$ 0.11 \\\\\n",
      "Specificity-controlled WD -04      &        2.83 $\\pm$ 0.10 &        2.76 $\\pm$ 0.12 &        3.37 $\\pm$ 0.10 &        2.36 $\\pm$ 0.11 &        2.46 $\\pm$ 0.11 &        2.62 $\\pm$ 0.12 &        3.14 $\\pm$ 0.09 &        3.52 $\\pm$ 0.11 \\\\\n",
      "Specificity-controlled WD -10      &        2.76 $\\pm$ 0.11 &        2.41 $\\pm$ 0.12 &        3.19 $\\pm$ 0.12 &        2.15 $\\pm$ 0.11 &        2.28 $\\pm$ 0.13 &        2.35 $\\pm$ 0.12 &        2.89 $\\pm$ 0.11 &        3.28 $\\pm$ 0.12 \\\\\n",
      "Specificity-controlled CT 000      &        2.79 $\\pm$ 0.10 &        2.93 $\\pm$ 0.09 &        3.44 $\\pm$ 0.12 &        2.38 $\\pm$ 0.11 &        2.56 $\\pm$ 0.12 &        2.84 $\\pm$ 0.12 &        3.12 $\\pm$ 0.13 &        3.61 $\\pm$ 0.11 \\\\\n",
      "Specificity-controlled CT 002      &        2.78 $\\pm$ 0.12 &        2.74 $\\pm$ 0.11 &        3.39 $\\pm$ 0.13 &        2.31 $\\pm$ 0.13 &        2.56 $\\pm$ 0.13 &        2.74 $\\pm$ 0.12 &        2.99 $\\pm$ 0.11 &        3.47 $\\pm$ 0.10 \\\\\n",
      "Specificity-controlled CT 004      &        2.82 $\\pm$ 0.10 &        2.80 $\\pm$ 0.13 &        3.44 $\\pm$ 0.14 &        2.32 $\\pm$ 0.13 &        2.51 $\\pm$ 0.12 &        2.78 $\\pm$ 0.15 &        3.09 $\\pm$ 0.13 &        3.46 $\\pm$ 0.13 \\\\\n",
      "Specificity-controlled CT 007      &        2.81 $\\pm$ 0.12 &        2.91 $\\pm$ 0.13 &        3.43 $\\pm$ 0.11 &        2.45 $\\pm$ 0.10 &        2.49 $\\pm$ 0.11 &        2.81 $\\pm$ 0.12 &        3.15 $\\pm$ 0.12 &        3.55 $\\pm$ 0.11 \\\\\n",
      "Specificity-controlled CT 009      &        2.80 $\\pm$ 0.13 &        2.78 $\\pm$ 0.10 &        3.41 $\\pm$ 0.12 &        2.35 $\\pm$ 0.13 &        2.28 $\\pm$ 0.11 &        2.79 $\\pm$ 0.11 &        2.91 $\\pm$ 0.11 &        3.51 $\\pm$ 0.12 \\\\\n",
      "Extrep\\nbigram -0.5                &        2.62 $\\pm$ 0.10 &        2.54 $\\pm$ 0.12 &        3.35 $\\pm$ 0.12 &        2.13 $\\pm$ 0.11 &        2.63 $\\pm$ 0.11 &        2.56 $\\pm$ 0.11 &        2.93 $\\pm$ 0.11 &        3.48 $\\pm$ 0.11 \\\\\n",
      "Extrep\\nbigram -1.25               &        2.78 $\\pm$ 0.09 &        2.82 $\\pm$ 0.13 &        3.40 $\\pm$ 0.12 &        2.27 $\\pm$ 0.12 &        2.54 $\\pm$ 0.09 &        2.76 $\\pm$ 0.10 &        3.05 $\\pm$ 0.11 &        3.53 $\\pm$ 0.14 \\\\\n",
      "Extrep\\nbigram -3.5                &        2.83 $\\pm$ 0.11 &        2.93 $\\pm$ 0.10 &  {\\bf 3.56 $\\pm$ 0.10} &        2.43 $\\pm$ 0.11 &        2.47 $\\pm$ 0.11 &        2.83 $\\pm$ 0.10 &        3.14 $\\pm$ 0.10 &        3.62 $\\pm$ 0.12 \\\\\n",
      "Repetition-controlled baseline     &        2.86 $\\pm$ 0.12 &        2.82 $\\pm$ 0.12 &        3.53 $\\pm$ 0.10 &        2.40 $\\pm$ 0.11 &        2.62 $\\pm$ 0.13 &        2.84 $\\pm$ 0.12 &        3.10 $\\pm$ 0.11 &        3.58 $\\pm$ 0.14 \\\\\n",
      "Extrep\\nbigram -inf                &        2.74 $\\pm$ 0.11 &        2.87 $\\pm$ 0.14 &        3.49 $\\pm$ 0.12 &        2.32 $\\pm$ 0.13 &        2.56 $\\pm$ 0.11 &        2.75 $\\pm$ 0.12 &        3.13 $\\pm$ 0.12 &        3.59 $\\pm$ 0.12 \\\\\n",
      "Response-related controlled WD 000 &  {\\bf 2.87 $\\pm$ 0.12} &        2.97 $\\pm$ 0.11 &        3.55 $\\pm$ 0.09 &  {\\bf 2.62 $\\pm$ 0.11} &        2.48 $\\pm$ 0.10 &        2.88 $\\pm$ 0.12 &        3.21 $\\pm$ 0.09 &  {\\bf 3.70 $\\pm$ 0.10} \\\\\n",
      "Response-related controlled WD 005 &        2.79 $\\pm$ 0.10 &        2.83 $\\pm$ 0.09 &        3.35 $\\pm$ 0.12 &        2.40 $\\pm$ 0.12 &        2.51 $\\pm$ 0.13 &        2.80 $\\pm$ 0.13 &        3.13 $\\pm$ 0.12 &        3.41 $\\pm$ 0.12 \\\\\n",
      "Response-related controlled WD 010 &        2.74 $\\pm$ 0.11 &        2.42 $\\pm$ 0.12 &        2.93 $\\pm$ 0.11 &        1.95 $\\pm$ 0.12 &        2.20 $\\pm$ 0.12 &        2.56 $\\pm$ 0.12 &        2.90 $\\pm$ 0.12 &        3.12 $\\pm$ 0.10 \\\\\n",
      "Response-related controlled WD 013 &        2.63 $\\pm$ 0.12 &        2.06 $\\pm$ 0.11 &        2.40 $\\pm$ 0.09 &        1.74 $\\pm$ 0.11 &        2.07 $\\pm$ 0.11 &        2.25 $\\pm$ 0.12 &        2.49 $\\pm$ 0.14 &        2.63 $\\pm$ 0.10 \\\\\n",
      "Response-related controlled WD -10 &        2.77 $\\pm$ 0.12 &        2.45 $\\pm$ 0.12 &        3.26 $\\pm$ 0.11 &        1.96 $\\pm$ 0.10 &        2.31 $\\pm$ 0.12 &        2.47 $\\pm$ 0.12 &        2.73 $\\pm$ 0.11 &        3.12 $\\pm$ 0.12 \\\\\n",
      "Repetition-controlled baseline+    &  {\\bf 2.87 $\\pm$ 0.12} &        2.97 $\\pm$ 0.11 &        3.55 $\\pm$ 0.09 &  {\\bf 2.62 $\\pm$ 0.11} &        2.48 $\\pm$ 0.10 &        2.88 $\\pm$ 0.12 &        3.21 $\\pm$ 0.09 &  {\\bf 3.70 $\\pm$ 0.10} \\\\\n",
      "Specificity-controlled WD 00       &        2.86 $\\pm$ 0.12 &        2.82 $\\pm$ 0.12 &        3.53 $\\pm$ 0.10 &        2.40 $\\pm$ 0.11 &        2.62 $\\pm$ 0.13 &        2.84 $\\pm$ 0.12 &        3.10 $\\pm$ 0.11 &        3.58 $\\pm$ 0.14 \\\\\n",
      "\\bottomrule\n",
      "\\end{tabular}\n",
      "\n"
     ]
    }
   ],
   "source": [
    "nice_bayes = []\n",
    "colmax = results_bayes[results_bayes.index != 'human_eval'].max(axis=0)\n",
    "for model_name, row in results_bayes.iterrows():\n",
    "    x = {}\n",
    "    for m in LIKERT_METRICS:\n",
    "        val = '{:.2f} $\\\\pm$ {:.2f}'.format(row[m + '_mean'], row[m + '_std'])\n",
    "        if row[m + '_mean'] == colmax[m + '_mean']:\n",
    "            val = '{\\\\bf ' + val + '}'\n",
    "        x[nice_metric_names[m]] = val\n",
    "    x['model_name'] = nice_names.get(\n",
    "        model_name,\n",
    "        nice_names.get(\n",
    "            row['model_type'],\n",
    "            row['model_type']\n",
    "        ) + ' ' + str(row['model_bucket'])\n",
    "    )\n",
    "    nice_bayes.append(x)\n",
    "\n",
    "nice_bayes = pd.DataFrame(nice_bayes).set_index('model_name')\n",
    "print(nice_bayes.to_latex(escape=False))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (dialog)",
   "language": "python",
   "name": "dialog"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}