Heatmap_Generation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "4aa36bb9",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import random\n",
    "import requests\n",
    "import os\n",
    "import copy\n",
    "import csv\n",
    "import statistics\n",
    "\n",
    "from sklearn.manifold import TSNE\n",
    "from scipy.stats import pearsonr\n",
    "from scipy.stats import scoreatpercentile\n",
    "from scipy.stats import kendalltau\n",
    "from scipy.stats import spearmanr\n",
    "from scipy.stats import weightedtau\n",
    "from sklearn.cluster import KMeans\n",
    "\n",
    "from tabulate import tabulate\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from string import ascii_letters\n",
    "from tqdm.notebook import tqdm, trange\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1841debe",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"heatmap_raw_data.json\",\"r\",encoding=\"utf-8\") as f:\n",
    "    raw_data = json.load(f)\n",
    "    \n",
    "output_collection = raw_data[\"output_collection\"]\n",
    "majority_collection = raw_data[\"majority_collection\"]\n",
    "testsize_collection = raw_data[\"testsize_collection\"]\n",
    "all_dist_collections = raw_data[\"all_dist_collections\"]\n",
    "output_collection_electra = raw_data[\"output_collection_electra\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "494a6138",
   "metadata": {},
   "source": [
    "# Intrinsic Few-Shot Hardness (Fig.2 )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "4dfbf299",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 684x648 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 684x648 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def heatmap_generation(aspects=[\"cola\",\"sst2\",\"mrpc\",\"qqp\",\"mnli\",\"qnli\",\"rte\",\"boolq\",\"cb\",\"copa\",\"wic\"] \\\n",
    "                       , scope=\"clusters\", mode = \"corr\"):\n",
    "\n",
    "    method_cluster = {}\n",
    "    method_cluster[\"prompt-fs\"] = [\"lmbff\",\"nullp\",\"bitfit\",\"adapet\"]\n",
    "    method_cluster[\"lw-fs\"] = [\"ptuning\",\"compacter\"]\n",
    "    method_cluster[\"fewshot\"] = [\"lmbff\",\"nullp\",\"bitfit\",\"adapet\",\"ptuning\",\"compacter\"]\n",
    "    # method_cluster[\"fewshot-avg\"] = [\"few-avg\"] \n",
    "#     method_cluster[\"lw-ft\"] = [\"p-tuning(f)\",\"adapter(f)\"]\n",
    "#     method_cluster[\"lm-ft\"] = [\"roberta-b(f)\",\"roberta-l(f)\",\"bert-b(f)\",\"bert-l(f)\",\"t5-b(f)\",\"t5-l(f)\",\"t5-11b(f)\"]\n",
    "#     method_cluster[\"human\"] = [\"human\"]\n",
    "\n",
    "\n",
    "\n",
    "    included_tasks = [\"cola\",\"sst2\",\"mrpc\",\"qqp\",\"mnli\",\"qnli\",\"rte\",\"boolq\",\"cb\",\"copa\",\"wic\"]\n",
    "\n",
    "\n",
    "    included_methods = []\n",
    "    for k,v in method_cluster.items():\n",
    "        for item in list(v):\n",
    "            if item not in included_methods:      \n",
    "                included_methods += [item]\n",
    "    \n",
    "    # included_tasks = aspects\n",
    "    \n",
    "    head = [\"methods\"]\n",
    "    \n",
    "    def generate_score(mode,lst1,lst2):\n",
    "        if mode == \"corr\":\n",
    "            return spearmanr(lst1, lst2)[0]\n",
    "            # return pearsonr(lst1, lst2)[0]\n",
    "        elif mode == \"kendall_tau\":\n",
    "            return kendalltau(np.argsort(lst1), np.argsort(lst2))[0]\n",
    "        elif mode == \"weighted_tau\":\n",
    "            return weightedtau(np.argsort(lst1), np.argsort(lst2))[0]\n",
    "        \n",
    "    def cluster_score_avg(cluster1, cluster2, task_collector):\n",
    "        all_score = 0\n",
    "        for v1 in cluster1:\n",
    "            for v2 in cluster2:\n",
    "                all_score += task_collector[v1][v2]\n",
    "                \n",
    "        return all_score/(len(cluster1)*len(cluster2))\n",
    "    \n",
    "    def extract_score(method,task):\n",
    "        temp_score = 0\n",
    "        temp_collection = {}\n",
    "        if method in output_collection.keys():\n",
    "            temp_collection = output_collection\n",
    "        else:\n",
    "            temp_collection = output_dic\n",
    "            \n",
    "        if method in [\"sensitivity\",\"diversity\"]:\n",
    "            temp_score -= temp_collection[method][task]*100\n",
    "        elif method in [\"rda\",\"coverage\"]:\n",
    "            # rda is computed as a sum over the examples\n",
    "            temp_score -= temp_collection[method][task]\n",
    "        else:\n",
    "            temp_score += temp_collection[method][task]-majority_collection[task]\n",
    "                    \n",
    "        return temp_score\n",
    "\n",
    "    # capture the correlation for each two tasks \n",
    "    df = pd.DataFrame()\n",
    "    task_collector = {}\n",
    "    \n",
    "    if scope == \"methods\":\n",
    "        for i,method in enumerate(included_methods):\n",
    "            head.append(method)    \n",
    "        df[\"methods\"] = included_methods\n",
    "    \n",
    "    for method1 in included_methods:\n",
    "        temp = []\n",
    "        task_collector[method1] = {}\n",
    "        \n",
    "        for method2 in included_methods:\n",
    "            lst1 = []\n",
    "            lst2 = []\n",
    "\n",
    "            for task in included_tasks:\n",
    "                lst1.append(extract_score(method1,task))\n",
    "                lst2.append(extract_score(method2,task))\n",
    "\n",
    "            temp_score = generate_score(mode,lst1,lst2)\n",
    "            temp.append(round(temp_score,4))\n",
    "            task_collector[method1][method2] = temp_score \n",
    "            \n",
    "        if scope == \"methods\":\n",
    "            df[method1] = temp\n",
    "            \n",
    "            \n",
    "    if scope == \"methods\":\n",
    "        df.set_index('methods', drop=True, inplace=True)\n",
    "    \n",
    "        \n",
    "    if scope == \"clusters\":\n",
    "            for key in list(method_cluster.keys()):\n",
    "                head.append(key)\n",
    "                \n",
    "            df[\"clusters\"] = list(method_cluster.keys())\n",
    "            \n",
    "            for cluster1 in df[\"clusters\"]:\n",
    "                temp = []\n",
    "                for cluster2 in df[\"clusters\"]:\n",
    "                    temp_score = cluster_score_avg(method_cluster[cluster1], method_cluster[cluster2], task_collector)\n",
    "                    temp.append(round(temp_score,4))\n",
    "            \n",
    "                df[cluster1] = temp\n",
    "\n",
    "            df.set_index('clusters', drop=True, inplace=True)\n",
    "        \n",
    "    mask = np.triu(np.ones_like(df, dtype=bool))\n",
    "    f, ax = plt.subplots(figsize=(9.5, 9))\n",
    "\n",
    "    mask[0][0] = False\n",
    "    mask[1][1] = False\n",
    "    mask[2][2] = False\n",
    "    cmap = sns.diverging_palette(230, 20, as_cmap=True)\n",
    "    \n",
    "    if scope == \"clusters\":\n",
    "        sns.heatmap(df,annot=True,mask=mask, cmap='gist_gray_r', vmax=1, center=0.78,cbar=False,\n",
    "                    square=False, linewidths=.6, cbar_kws={\"shrink\": .5}, annot_kws={\"fontsize\":50})\n",
    "        \n",
    "        ax.set(ylabel=None)\n",
    "        ax.tick_params(axis='x', labelsize= 30)\n",
    "        ax.tick_params(axis='y', labelsize= 30)\n",
    "    \n",
    "    \n",
    "    if scope == \"methods\":\n",
    "        \n",
    "        mask[3][3] = False\n",
    "        mask[4][4] = False\n",
    "        mask[5][5] = False\n",
    "        sns.heatmap(df,annot=True,mask=mask, cmap='gist_gray_r', vmax=1, center=0.6,cbar=False,\n",
    "            square=False, linewidths=.6, cbar_kws={\"shrink\": .5}, annot_kws={\"fontsize\":30})\n",
    "        \n",
    "        ax.set(ylabel=None)\n",
    "        ax.tick_params(axis='x', labelsize= 15)\n",
    "        ax.tick_params(axis='y', labelsize= 15)\n",
    "\n",
    "    \n",
    "    # image_format = 'svg' # e.g .png, .svg, etc.\n",
    "    # image_name = 'glue_intrinsic_'+scope[:-1]+'.svg'\n",
    "    # f.savefig(image_name, format=image_format, dpi=1200)\n",
    "    \n",
    "    image_name = 'glue_intrinsic_'+scope[:-1]+'.pdf'\n",
    "    f.savefig(image_name, format=\"pdf\", bbox_inches=\"tight\", dpi=1200)\n",
    "\n",
    "    plt.show()\n",
    "    \n",
    "heatmap_generation(scope=\"methods\")\n",
    "heatmap_generation(scope=\"clusters\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd1da126",
   "metadata": {},
   "source": [
    "# Comparing Automatic Metrics For Intrinsic Few-Shot Hardness (Tab. 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "912652e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Table 1 and Fig 4 in the paper\n",
    "# note that the correlation scores for Spread/RDA can be slightly different from those reported in the paper \n",
    "#     due to the choice of method clusters\n",
    "# our conclusion still holds. Correlation scores for the involved hardness metrics, \n",
    "#     in the format of here (reported in the paper):\n",
    "#    Spread: 0.52 (0.47), RDA 0.41 (0.36), Sensitivity 0.02 (0.06)\n",
    "\n",
    "def percentile_filter(dists, percentile):\n",
    "    threshold = np.percentile(dists, percentile)\n",
    "    new_dists = []\n",
    "\n",
    "    for dist in dists:\n",
    "        if dist >= threshold:\n",
    "            new_dists.append(10**dist)\n",
    "            \n",
    "    # print(threshold)\n",
    "    return(sum(new_dists)/len(new_dists))\n",
    "\n",
    "\n",
    "def heatmap_generation(aspects, scope=\"clusters\", mode = \"corr\"):\n",
    "    # scope: methods,clusters\n",
    "    method_cluster = {}\n",
    "    method_cluster[\"prompt-fs\"] = [\"lmbff\",\"nullp\",\"bitfit\",\"adapet\"] # \"manual\",\n",
    "    method_cluster[\"lw-fs\"] = [\"ptuning\",\"compacter\"] #,\"adapter\"\n",
    "#     method_cluster[\"fewshot-avg\"] = [\"few-avg\"] \n",
    "    method_cluster[\"fewshot\"] = [\"lmbff\",\"nullp\",\"bitfit\",\"adapet\",\"ptuning\",\"compacter\"] #\"manual\",,\"few-avg\"\n",
    "\n",
    "    method_cluster[\"Spread\"] = [\"Spread\"]\n",
    "    method_cluster[\"sensitivity\"] = [\"sensitivity\"]\n",
    "    method_cluster[\"rda\"] = [\"rda\"]\n",
    "#     method_cluster[\"hardness-avg\"] = [\"sensitivity\",\"rda\"]\n",
    "    \n",
    "\n",
    "    included_methods = []\n",
    "    for k,v in method_cluster.items():\n",
    "        for item in list(v):\n",
    "            if item not in included_methods:      \n",
    "                included_methods += [item]\n",
    "    \n",
    "    included_tasks = aspects\n",
    "    \n",
    "    head = [\"methods\"]\n",
    "    \n",
    "    def generate_score(mode,lst1,lst2):\n",
    "        if mode == \"corr\":\n",
    "            return spearmanr(lst1, lst2)[0]\n",
    "            # return pearsonr(lst1, lst2)[0]\n",
    "        elif mode == \"kendall_tau\":\n",
    "            return kendalltau(np.argsort(lst1), np.argsort(lst2))[0]\n",
    "        elif mode == \"weighted_tau\":\n",
    "            return weightedtau(np.argsort(lst1), np.argsort(lst2))[0]\n",
    "        \n",
    "    def cluster_score_avg(cluster1, cluster2, task_collector):\n",
    "        all_score = 0\n",
    "        for v1 in cluster1:\n",
    "            for v2 in cluster2:\n",
    "                all_score += task_collector[v1][v2]\n",
    "                \n",
    "        return all_score/(len(cluster1)*len(cluster2))\n",
    "    \n",
    "    def extract_score(method,task):\n",
    "        temp_score = 0\n",
    "        \n",
    "        if method in [\"sensitivity\",\"diversity\"]:\n",
    "            temp_score -= output_collection[method][task]*100\n",
    "        elif method in [\"rda\",\"coverage\"]:\n",
    "            # rda is computed as a sum over the examples\n",
    "            temp_score -= output_collection[method][task]# /(testsize_collection[task]/100) # 0\n",
    "        elif method == \"Spread\":\n",
    "            temp_score = -1*percentile_filter(all_dist_collections[task], 99.9)\n",
    "        else:\n",
    "            temp_score += output_collection[method][task]-majority_collection[task]*100\n",
    "        \n",
    "        return temp_score\n",
    "\n",
    "    # capture the correlation for each two tasks \n",
    "    df = pd.DataFrame()\n",
    "    task_collector = {}\n",
    "    \n",
    "    if scope == \"methods\":\n",
    "        for i,method in enumerate(included_methods):\n",
    "            head.append(method)    \n",
    "        df[\"methods\"] = included_methods\n",
    "    \n",
    "    for method1 in included_methods:\n",
    "        temp = []\n",
    "        task_collector[method1] = {}\n",
    "        \n",
    "        for method2 in included_methods:\n",
    "            lst1 = []\n",
    "            lst2 = []\n",
    "\n",
    "            for task in included_tasks:\n",
    "                lst1.append(extract_score(method1,task))\n",
    "                lst2.append(extract_score(method2,task))\n",
    "\n",
    "            temp_score = generate_score(mode,lst1,lst2)\n",
    "            temp.append(round(temp_score,2))\n",
    "            task_collector[method1][method2] = temp_score \n",
    "            \n",
    "        if scope == \"methods\":\n",
    "            df[method1] = temp\n",
    "            \n",
    "            \n",
    "    if scope == \"methods\":\n",
    "        df.set_index('methods', drop=True, inplace=True)\n",
    "    \n",
    "        \n",
    "    if scope == \"clusters\":\n",
    "            for key in list(method_cluster.keys()):\n",
    "                head.append(key)\n",
    "                \n",
    "            df[\"clusters\"] = list(method_cluster.keys())\n",
    "            \n",
    "            for cluster1 in df[\"clusters\"]:\n",
    "                temp = []\n",
    "                for cluster2 in df[\"clusters\"]:\n",
    "                    temp_score = cluster_score_avg(method_cluster[cluster1], method_cluster[cluster2], task_collector)\n",
    "                    temp.append(round(temp_score,2))\n",
    "            \n",
    "                df[cluster1] = temp\n",
    "\n",
    "            df.set_index('clusters', drop=True, inplace=True)\n",
    "    \n",
    "    # print the averaged fewshot correlation\n",
    "    if scope == \"methods\":\n",
    "        for col in df:\n",
    "            temp = 0\n",
    "            dup = 0\n",
    "            for method in method_cluster[\"fewshot\"]:\n",
    "                if method != col:\n",
    "                    temp += df[col][method]\n",
    "                else:\n",
    "                    dup = 1\n",
    "            print(col,\":\", round(temp/(len(method_cluster[\"fewshot\"])-dup),2))\n",
    "        \n",
    "    mask = np.triu(np.ones_like(df, dtype=bool))\n",
    "    f, ax = plt.subplots(figsize=(9, 9))\n",
    "    cmap = sns.diverging_palette(230, 20, as_cmap=True)\n",
    "    \n",
    "    for i in range(len(mask)):\n",
    "        mask[i][i] = False\n",
    "    \n",
    "    if scope == \"methods\":\n",
    "        sns.heatmap(df,annot=True,mask=mask, cmap='gist_gray_r', vmax=1, center=0.3,cbar=False,\n",
    "                square=False, linewidths=.5, cbar_kws={\"shrink\": .5}, annot_kws={\"fontsize\":20})\n",
    "        ax.tick_params(axis='x', labelsize= 10)\n",
    "        ax.tick_params(axis='y', labelsize= 10)\n",
    "    else:\n",
    "        sns.heatmap(df,annot=True,mask=mask, cmap='gist_gray_r', vmax=1, center=0.6,cbar=False,\n",
    "                square=False, linewidths=.5, cbar_kws={\"shrink\": .5}, annot_kws={\"fontsize\":50})\n",
    "        ax.tick_params(axis='x', labelsize= 30)\n",
    "        ax.tick_params(axis='y', labelsize= 30)\n",
    "        \n",
    "    ax.set(ylabel=None)\n",
    "\n",
    "#     image_format = 'svg' # e.g .png, .svg, etc.\n",
    "#     image_name = 'nli_intrinsic_'+scope[:-1]+'.svg'\n",
    "#     f.savefig(image_name, format=image_format, dpi=1200)\n",
    "    \n",
    "    image_name = 'nli_intrinsic_'+scope[:-1]+'.pdf'\n",
    "    f.savefig(image_name, format=\"pdf\", bbox_inches=\"tight\", dpi=1200)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "62e0e41c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lmbff : 0.72\n",
      "nullp : 0.78\n",
      "bitfit : 0.8\n",
      "adapet : 0.63\n",
      "ptuning : 0.72\n",
      "compacter : 0.71\n",
      "adapter : 0.79\n",
      "Spread : 0.52\n",
      "sensitivity : 0.02\n",
      "rda : 0.41\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x648 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "temp_aspects = ['anaphora', 'boolean', 'comparative', 'conditional', 'counting', \\\n",
    "                'event', 'gender', 'lexico_syntactic', 'mnli', 'monotonicity', \\\n",
    "                'monotonicity_hard', 'monotonicity_simple', 'negation', 'ner','ner_merged', \\\n",
    "                'ner_entity', 'ner_event', 'ner_location', \\\n",
    "                'ner_organization', 'ner_person', 'ner_time',  \\\n",
    "                'paraphrase', 'puns','quantifier', 'relation_extraction', 'rte', 'semantic_role', 'sentiment']\n",
    "\n",
    "heatmap_generation(temp_aspects, scope=\"methods\", mode = \"corr\")\n",
    "# heatmap_generation(temp_aspects, scope=\"clusters\", mode = \"corr\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "af47c086",
   "metadata": {},
   "source": [
    "# Measuring IFH across pre-trained models (Sec. 4.4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "cd95ceec",
   "metadata": {},
   "outputs": [],
   "source": [
    "def electra_heatmap_generation(aspects, scope=\"clusters\", mode = \"corr\"):\n",
    "    # scope: methods,clusters\n",
    "    method_cluster = {}\n",
    "    method_cluster[\"roberta_fs\"]=[\"lmbff\",\"nullp\",\"bitfit\",\"adapet\"]\n",
    "    \n",
    "    method_cluster[\"lmbff\"] = [\"el_lmbff\"]\n",
    "    method_cluster[\"nullp\"] = [\"el_nullp\"]\n",
    "    method_cluster[\"bitfit\"] = [\"el_bitfit\"]\n",
    "    method_cluster[\"adapet\"] = [\"el_adapet\"]\n",
    "    \n",
    "    # method_cluster[\"electra_fs\"] = [\"el_lmbff\",\"el_nullp\",\"el_bitfit\",\"el_adapet\"]\n",
    "    # method_cluster[\"hardness-avg\"] = [\"sensitivity\",\"rda\"]\n",
    "    # method_cluster[\"ours\"] = [\"diversity\",\"coverage\"] #,\"mixed\"\n",
    "\n",
    "    included_methods = []\n",
    "    for k,v in method_cluster.items():\n",
    "        for item in list(v):\n",
    "            if item not in included_methods:      \n",
    "                included_methods += [item]\n",
    "    \n",
    "    included_tasks = aspects\n",
    "    \n",
    "    head = [\"methods\"]\n",
    "    \n",
    "    def generate_score(mode,lst1,lst2):\n",
    "        if mode == \"corr\":\n",
    "            return spearmanr(lst1, lst2)[0]\n",
    "            # return pearsonr(lst1, lst2)[0]\n",
    "        elif mode == \"kendall_tau\":\n",
    "            return kendalltau(np.argsort(lst1), np.argsort(lst2))[0]\n",
    "        elif mode == \"weighted_tau\":\n",
    "            return weightedtau(np.argsort(lst1), np.argsort(lst2))[0]\n",
    "        \n",
    "    def cluster_score_avg(cluster1, cluster2, task_collector):\n",
    "        all_score = 0\n",
    "        for v1 in cluster1:\n",
    "            for v2 in cluster2:\n",
    "                all_score += task_collector[v1][v2]\n",
    "                \n",
    "        return all_score/(len(cluster1)*len(cluster2))\n",
    "    \n",
    "    def extract_score(method,task):\n",
    "        temp_score = 0\n",
    "        \n",
    "        if method in [\"sensitivity\",\"diversity\"]:\n",
    "            temp_score -= output_collection[method][task]*100\n",
    "        elif method in [\"rda\",\"coverage\"]:\n",
    "            # rda is computed as a sum over the examples\n",
    "            temp_score -= output_collection[method][task]# /(testsize_collection[task]/100) # 0\n",
    "        elif \"el_\" in method:\n",
    "            temp_score += output_collection_electra[method.replace(\"el_\",\"\")][task]-majority_collection[task]*100\n",
    "        else:\n",
    "            temp_score += output_collection[method][task]-majority_collection[task]*100\n",
    "        \n",
    "        return temp_score\n",
    "\n",
    "    # capture the correlation for each two tasks \n",
    "    df = pd.DataFrame()\n",
    "    task_collector = {}\n",
    "    \n",
    "    if scope == \"methods\":\n",
    "        for i,method in enumerate(included_methods):\n",
    "            head.append(method)    \n",
    "        df[\"methods\"] = included_methods\n",
    "    \n",
    "    for method1 in included_methods:\n",
    "        temp = []\n",
    "        task_collector[method1] = {}\n",
    "        \n",
    "        for method2 in included_methods:\n",
    "            lst1 = []\n",
    "            lst2 = []\n",
    "\n",
    "            for task in included_tasks:\n",
    "                lst1.append(extract_score(method1,task))\n",
    "                lst2.append(extract_score(method2,task))\n",
    "\n",
    "            temp_score = generate_score(mode,lst1,lst2)\n",
    "            temp.append(round(temp_score,2))\n",
    "            task_collector[method1][method2] = temp_score \n",
    "            \n",
    "        if scope == \"methods\":\n",
    "            df[method1] = temp\n",
    "            \n",
    "            \n",
    "    if scope == \"methods\":\n",
    "        df.set_index('methods', drop=True, inplace=True)\n",
    "    \n",
    "        \n",
    "    if scope == \"clusters\":\n",
    "            for key in list(method_cluster.keys()):\n",
    "                head.append(key)\n",
    "                \n",
    "            df[\"clusters\"] = list(method_cluster.keys())\n",
    "            \n",
    "            for cluster1 in df[\"clusters\"]:\n",
    "                temp = []\n",
    "                for cluster2 in df[\"clusters\"]:\n",
    "                    temp_score = cluster_score_avg(method_cluster[cluster1], method_cluster[cluster2], task_collector)\n",
    "                    temp.append(round(temp_score,2))\n",
    "            \n",
    "                df[cluster1] = temp\n",
    "\n",
    "            df.set_index('clusters', drop=True, inplace=True)\n",
    "            \n",
    "    if scope == \"methods\":\n",
    "        for col in df:\n",
    "            temp = 0\n",
    "            dup = 0\n",
    "            for method in method_cluster[\"roberta_fewshot\"]:\n",
    "                if method != col:\n",
    "                    temp += df[col][method]\n",
    "                else:\n",
    "                    dup = 1\n",
    "            print(col,\":\", round(temp/(len(method_cluster[\"roberta_fewshot\"])-dup),2))\n",
    "            \n",
    "    mask = np.triu(np.ones_like(df, dtype=bool))\n",
    "    f, ax = plt.subplots(figsize=(9, 9))\n",
    "    cmap = sns.diverging_palette(230, 20, as_cmap=True)\n",
    "    \n",
    "    for i in range(len(mask)):\n",
    "        mask[i][i] = False\n",
    "    \n",
    "    if scope == \"methods\":\n",
    "        sns.heatmap(df,annot=True,mask=mask, cmap='RdYlGn', vmax=1, center=0.6,cbar=False,\n",
    "                square=False, linewidths=.5, cbar_kws={\"shrink\": .5}, annot_kws={\"fontsize\":30})\n",
    "    else:\n",
    "        sns.heatmap(df,annot=True,mask=mask, cmap='RdYlGn', vmax=1, center=0.4,cbar=False,\n",
    "                square=False, linewidths=.5, cbar_kws={\"shrink\": .5}, annot_kws={\"fontsize\":30})\n",
    "        \n",
    "    ax.set(ylabel=None)\n",
    "    ax.tick_params(axis='x', labelsize= 20)\n",
    "    ax.tick_params(axis='y', labelsize= 20)\n",
    "    \n",
    "    image_format = 'svg' # e.g .png, .svg, etc.\n",
    "    image_name = 'heatmap3.svg'\n",
    "\n",
    "    f.savefig(image_name, format=image_format, dpi=1200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "9ac8150a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# electra_heatmap_generation(temp_aspects, scope=\"clusters\", mode = \"corr\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "4d39ceb0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Detailed Heatmap for Fig 2.\n",
    "\n",
    "def electra_heatmap_generation_v2(aspects, scope=\"clusters\", mode = \"corr\"):\n",
    "    # scope: methods,clusters\n",
    "    method_cluster = {}\n",
    "    method_cluster[\"roberta_fs\"]=[\"rb_lmbff\",\"rb_nullp\",\"rb_bitfit\",\"rb_adapet\"]#,\"manual\",\"few-avg\"\n",
    "    el_methods = [\"el_lmbff\",\"el_nullp\",\"el_bitfit\",\"el_adapet\"]\n",
    "    # method_cluster[\"hardness-avg\"] = [\"sensitivity\",\"rda\"]\n",
    "\n",
    "    included_methods = []\n",
    "    for k,v in method_cluster.items():\n",
    "        for item in list(v):\n",
    "            if item not in included_methods:      \n",
    "                included_methods += [item]\n",
    "    \n",
    "    included_tasks = aspects\n",
    "    \n",
    "    head = [\"methods\"]\n",
    "    \n",
    "    def generate_score(mode,lst1,lst2):\n",
    "        if mode == \"corr\":\n",
    "            return spearmanr(lst1, lst2)[0]\n",
    "            # return pearsonr(lst1, lst2)[0]\n",
    "        elif mode == \"kendall_tau\":\n",
    "            return kendalltau(np.argsort(lst1), np.argsort(lst2))[0]\n",
    "        elif mode == \"weighted_tau\":\n",
    "            return weightedtau(np.argsort(lst1), np.argsort(lst2))[0]\n",
    "        \n",
    "    def cluster_score_avg(cluster1, cluster2, task_collector):\n",
    "        all_score = 0\n",
    "        for v1 in cluster1:\n",
    "            for v2 in cluster2:\n",
    "                all_score += task_collector[v1][v2]\n",
    "                \n",
    "        return all_score/(len(cluster1)*len(cluster2))\n",
    "    \n",
    "    def extract_score(method,task):\n",
    "        temp_score = 0\n",
    "        \n",
    "        if method in [\"sensitivity\",\"diversity\"]:\n",
    "            temp_score -= output_collection[method][task]*100\n",
    "        elif method in [\"rda\",\"coverage\"]:\n",
    "            # rda is computed as a sum over the examples\n",
    "            temp_score -= output_collection[method][task]# /(testsize_collection[task]/100) # 0\n",
    "        elif \"el_\" in method:\n",
    "            temp_score += output_collection_electra[method.replace(\"el_\",\"\")][task]-majority_collection[task]*100\n",
    "        elif \"rb_\" in method:\n",
    "            temp_score += output_collection[method.replace(\"rb_\",\"\")][task]-majority_collection[task]*100\n",
    "        else:\n",
    "            temp_score += output_collection[method][task]-majority_collection[task]*100\n",
    "        \n",
    "        return temp_score\n",
    "\n",
    "    # capture the correlation for each two tasks \n",
    "    df = pd.DataFrame()\n",
    "    task_collector = {}\n",
    "    \n",
    "    if scope == \"methods\":\n",
    "        for i,method in enumerate(included_methods):\n",
    "            head.append(method)    \n",
    "        df[\"methods\"] = included_methods\n",
    "    \n",
    "    for method1 in el_methods: #included_methods:\n",
    "        temp = []\n",
    "        task_collector[method1] = {}\n",
    "        \n",
    "        for method2 in included_methods: # el_methods:\n",
    "            lst1 = []\n",
    "            lst2 = []\n",
    "\n",
    "            for task in included_tasks:\n",
    "                lst1.append(extract_score(method1,task))\n",
    "                lst2.append(extract_score(method2,task))\n",
    "\n",
    "            temp_score = generate_score(mode,lst1,lst2)\n",
    "            temp.append(round(temp_score,2))\n",
    "            task_collector[method1][method2] = temp_score \n",
    "            \n",
    "        if scope == \"methods\":\n",
    "            df[method1] = temp\n",
    "            \n",
    "            \n",
    "    if scope == \"methods\":\n",
    "        df.set_index('methods', drop=True, inplace=True)\n",
    "    \n",
    "        \n",
    "    if scope == \"clusters\":\n",
    "            for key in list(method_cluster.keys()):\n",
    "                head.append(key)\n",
    "                \n",
    "            df[\"clusters\"] = list(method_cluster.keys())\n",
    "            \n",
    "            for cluster1 in df[\"clusters\"]:\n",
    "                temp = []\n",
    "                for cluster2 in df[\"clusters\"]:\n",
    "                    temp_score = cluster_score_avg(method_cluster[cluster1], method_cluster[cluster2], task_collector)\n",
    "                    temp.append(round(temp_score,2))\n",
    "            \n",
    "                df[cluster1] = temp\n",
    "\n",
    "            df.set_index('clusters', drop=True, inplace=True)\n",
    "\n",
    "            \n",
    "    mask = np.triu(np.ones_like(df, dtype=bool))\n",
    "    f, ax = plt.subplots(figsize=(9, 9))\n",
    "    cmap = sns.diverging_palette(230, 20, as_cmap=True)\n",
    "    \n",
    "    for i in range(len(mask)):\n",
    "        mask[i][i] = False\n",
    "        \n",
    "    sns.set_theme(style=\"ticks\")\n",
    "    if scope == \"methods\":\n",
    "        # cmap='RdYlGn'\n",
    "        sns.heatmap(df,annot=True,mask=mask, cmap='gist_gray_r', vmax=1, center=0.3,cbar=False,\n",
    "                square=False, linewidths=.5, cbar_kws={\"shrink\": .5}, annot_kws={\"fontsize\":30})\n",
    "    else:\n",
    "        sns.heatmap(df,annot=True,mask=mask, cmap='BuGn', vmax=1, center=0.3,cbar=False,\n",
    "                square=False, linewidths=.5, cbar_kws={\"shrink\": .5}, annot_kws={\"fontsize\":30})\n",
    "        \n",
    "    ax.set(ylabel=None)\n",
    "    ax.tick_params(axis='x', labelsize= 20)\n",
    "    ax.tick_params(axis='y', labelsize= 20)\n",
    "    \n",
    "    image_format = 'pdf' # e.g .png, .svg, etc.\n",
    "    image_name = 'backbone_ablation.pdf'\n",
    "\n",
    "    f.savefig(image_name, format=\"pdf\", bbox_inches=\"tight\", dpi=1200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "65556ebb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x648 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "electra_heatmap_generation_v2(temp_aspects, scope=\"methods\", mode = \"corr\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9d8444ef",
   "metadata": {},
   "source": [
    "# Decreasing Few-Shot hardness via dataset decomposition (Sec. 5 & Tab. 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "02e19374",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Steps:\n",
    "\n",
    "# 1. download the decomp-NER dataset with the link provided in readme\n",
    "# 2. download the corresponding repos:\n",
    "#         LMBFF: https://github.com/princeton-nlp/LM-BFF\n",
    "#         AdaPET: https://github.com/rrmenon10/ADAPET\n",
    "#         Null-prompt & Prompt-Bitfit: https://github.com/ucinlp/null-prompts\n",
    "# 3. convert the provided dataset into the required format of the repos and put them under ./data/\n",
    "# 4. run the commands in ./remote_commands in this repository\n",
    "# 5*. feel free to contact me if you meet any problems with running the models.\n",
    "\n",
    "# for reference, here is original code for cluster generation with instance representation\n",
    "\n",
    "import torch\n",
    "from transformers import AutoModel, AutoTokenizer\n",
    "\n",
    "def create_simcse_embeddings(tokenizer, model, texts):\n",
    "    device = torch.device('cuda')\n",
    "    # create tokenized inputs\n",
    "    batch_size = 65\n",
    "\n",
    "    # naive batching\n",
    "    if len(texts) < batch_size:\n",
    "        inputs = tokenizer(texts, padding=True, truncation=True, return_tensors=\"pt\")\n",
    "        inputs = inputs.to(device)\n",
    "        with torch.no_grad():\n",
    "            batch_embeddings = model(**inputs, output_hidden_states=True, return_dict=True).pooler_output\n",
    "            embeddings = batch_embeddings.tolist()\n",
    "            del batch_embeddings\n",
    "            torch.cuda.empty_cache()\n",
    "    else:\n",
    "        embeddings = []\n",
    "        num_batch = len(texts)//batch_size\n",
    "\n",
    "        for i in trange(num_batch+1):\n",
    "            batch_start = i*batch_size\n",
    "            batch_end = min(len(texts), (i+1)*batch_size)\n",
    "            batch_texts = texts[batch_start:batch_end]\n",
    "\n",
    "            inputs = tokenizer(batch_texts, padding=True, truncation=True, return_tensors=\"pt\")\n",
    "            inputs = inputs.to(device)\n",
    "\n",
    "            with torch.no_grad():\n",
    "                batch_embeddings = model(**inputs, output_hidden_states=True, return_dict=True).pooler_output\n",
    "                embeddings.extend(batch_embeddings.detach().cpu().tolist())\n",
    "\n",
    "                # save cuda memory\n",
    "                del batch_embeddings\n",
    "                del inputs\n",
    "                torch.cuda.empty_cache()\n",
    "\n",
    "    return embeddings\n",
    "\n",
    "\n",
    "def extract_texts(dataset):\n",
    "    # nli specific way to extract texts\n",
    "    texts = []\n",
    "    for item in dataset:\n",
    "        mapping = {'not-entailed':'Clearly','entailed':'Similarly'}\n",
    "        temp = item[\"context\"].replace(\".\",\" \")+mapping[item[\"label\"]]+\", I believe \"+item[\"hypothesis\"]\n",
    "        texts.append(temp)\n",
    "\n",
    "    return texts\n",
    "\n",
    "\n",
    "def model_clustering(ner_fewdatasets):\n",
    "    # update the code with GPU use\n",
    "    device = torch.device('cuda')\n",
    "\n",
    "    tokenizer = AutoTokenizer.from_pretrained(\"princeton-nlp/sup-simcse-bert-base-uncased\")\n",
    "    model = AutoModel.from_pretrained(\"princeton-nlp/sup-simcse-bert-base-uncased\").to(device)    \n",
    "    \n",
    "    # clustering with bert-based embeddings\n",
    "    \n",
    "    dataset={\"train\":[],\"dev\":[],\"test\":[]}\n",
    "    for split in [\"train\",\"dev\",\"test\"]:\n",
    "        temp_filename = \"./nli/DNC-master/\"+split+\"/recast_ner_data.json\"\n",
    "        dataset[split].extend(nli_dnc_loading(temp_filename, \"dnc\"))\n",
    "        \n",
    "    all_examples = []\n",
    "    reverse_split_list = []\n",
    "    for split in [\"train\",\"dev\",\"test\"]:\n",
    "        all_examples.extend(dataset[split])\n",
    "        for i in range(len(dataset[split])):\n",
    "            reverse_split_list.append(split)\n",
    "    \n",
    "    all_texts = extract_texts(all_examples)\n",
    "    embeddings = create_simcse_embeddings(tokenizer, model, all_texts)\n",
    "    \n",
    "    with open(\"./embeddings/ner_embedding.json\",\"w\",encoding=\"utf-8\") as f:\n",
    "        json.dump(embeddings,f)\n",
    "        \n",
    "    with open(\"./embeddings/ner_embedding.json\",\"r\",encoding=\"utf-8\") as f:\n",
    "        embeddings = json.load(f)\n",
    "    \n",
    "    # clustering the embeddings and redistribute to 6 clusters\n",
    "    kmeans = KMeans(n_clusters=6, random_state=42).fit(embeddings)\n",
    "    \n",
    "    bert_clusters = {}\n",
    "    for i in range(6):\n",
    "        bert_clusters[i] = {\"train\":[],\"dev\":[],\"test\":[]}\n",
    "        \n",
    "    for i, label in enumerate(kmeans.labels_):\n",
    "        bert_clusters[int(label)][reverse_split_list[i]].append(all_examples[i])\n",
    "    \n",
    "    return bert_clusters\n",
    "\n",
    "\n",
    "def cluster_similarity_analysis():\n",
    "    head = [\"clusters\"]\n",
    "    row_tasks = [\"ner_person\",\"ner_entity\",\"ner_location\",\"ner_event\",\"ner_organization\",\"ner_time\"] #\n",
    "    head.extend(row_tasks)\n",
    "    column_tasks = [\"ner_bert_0\",\"ner_bert_1\",\"ner_bert_2\",\"ner_bert_3\",\"ner_bert_4\",\"ner_bert_5\"] #\n",
    "    mydata= []\n",
    "    \n",
    "    def dataset_sim(item_list1, item_list2):\n",
    "        # A and B / A or B\n",
    "        overlap=0\n",
    "\n",
    "        for item1 in item_list1:\n",
    "            for item2 in item_list2:\n",
    "                identical = 1\n",
    "                for k,v in item1.items():\n",
    "                    if item2[k] != v:\n",
    "                        identical=0\n",
    "                if identical:\n",
    "                    overlap+=1\n",
    "        return overlap/(len(item_list1)+len(item_list2)-overlap)\n",
    "        \n",
    "    for k1 in column_tasks:\n",
    "        temp = [k1]\n",
    "        \n",
    "        for k2 in row_tasks:\n",
    "              temp.append(dataset_sim(ner_fewdatasets[k1][\"test\"],ner_fewdatasets[k2][\"test\"]))\n",
    "        mydata.append(temp)\n",
    "    \n",
    "    \n",
    "    print(tabulate(mydata, headers=head, tablefmt=\"grid\", floatfmt=\".3f\"))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}