notebooks/specialization.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import json\n",
    "import numpy as np\n",
    "from tqdm import tqdm\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import seaborn as sns\n",
    "palette = sns.color_palette(\"colorblind\")\n",
    "\n",
    "from utils import load_responses, responses_to_acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 260/260 [00:01<00:00, 201.99it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "llama-3-8b has 260 tasks\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 260/260 [00:01<00:00, 228.89it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "llama-3-8b-instruct has 260 tasks\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 260/260 [00:01<00:00, 210.56it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lawma-8b has 260 tasks\n"
     ]
    }
   ],
   "source": [
    "responses = load_responses({}, ['llama-3-8b'], 'results/scaling-experiments/')\n",
    "responses['llama-3-8b-instruct-ft-1e'] = responses.pop('llama-3-8b')  # rename\n",
    "models = ['llama-3-8b-instruct', 'lawma-8b']\n",
    "responses = load_responses(responses, ['llama-3-8b-instruct', 'lawma-8b'], 'results/model_responses/')\n",
    "base_accs = {model: {k: responses_to_acc(v) for k, v in responses[model].items()} for model in responses}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10/10 [00:00<00:00, 39.16it/s]\n",
      "100%|██████████| 10/10 [00:00<00:00, 100.79it/s]\n",
      "100%|██████████| 10/10 [00:00<00:00, 100.98it/s]\n",
      "100%|██████████| 9/9 [00:00<00:00, 703.57it/s]\n",
      "100%|██████████| 12/12 [00:00<00:00, 30.01it/s]\n",
      "100%|██████████| 9/9 [00:00<00:00, 464.81it/s]\n"
     ]
    }
   ],
   "source": [
    "def load_folder_accs(base_dir):\n",
    "    accs = {}\n",
    "    for task in tqdm(os.listdir(base_dir)):\n",
    "        task_files = os.listdir(os.path.join(base_dir, task))\n",
    "\n",
    "        accs[task] = {}\n",
    "        for file in task_files:\n",
    "            with open(os.path.join(base_dir, task, file), 'r') as f:\n",
    "                responses = json.load(f)\n",
    "            accs[task] = responses_to_acc(responses)\n",
    "            \n",
    "    return accs\n",
    "\n",
    "base_dir = 'results/specialization/'\n",
    "outer_dirs = {'warmup': '', 'no-warmup': '+'}\n",
    "base_models = {'llama-3-8b-instruct': '0e', \n",
    "               'llama-3-8b-instruct-ft-1e': '1e',\n",
    "               'lawma-8b': '3e'}\n",
    "\n",
    "ft_trials = {}\n",
    "for outer_dir, outer_dir_suffix in outer_dirs.items():\n",
    "    for model, model_preffix in base_models.items():\n",
    "        ft_trials[model_preffix + outer_dir_suffix] = load_folder_accs(base_dir + outer_dir + '/' + model + '/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "task_names = {\n",
    "        'sc_issuearea': 'SC Issue Area',\n",
    "        'sc_decisiondirection': 'SC Direction',\n",
    "        'sc_casedisposition': 'SC Disposition',\n",
    "        'sc_casesource': 'SC Case Source',\n",
    "        'sc_lcdisposition': 'SC LC Disposition',\n",
    "        'songer_geniss': 'Songer Gen. Issue',\n",
    "        'songer_direct1': 'Songer Direction',\n",
    "        'songer_treat': 'Songer Treatment',\n",
    "        'songer_origin': 'Songer Case Origin',\n",
    "        'sc_lcdispositiondirection': 'SC LC Direction',\n",
    "}\n",
    "\n",
    "tasks = task_names.keys()\n",
    "tasks = sorted(tasks, key=lambda x: base_accs['lawma-8b'][x])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def responses_to_acc_maj(responses, majority_class):\n",
    "    y_true = [r['ground_truth'] for r in responses]\n",
    "    y_pred = [r['model_response'] for r in responses]\n",
    "\n",
    "    if type(majority_class) == int:\n",
    "        majority_class = ' ' + str(majority_class)\n",
    "\n",
    "    ids_majority = [i for i, true in enumerate(y_true) if true == majority_class]\n",
    "    y_true = [y_true[i] for i in ids_majority]\n",
    "    y_pred = [y_pred[i] for i in ids_majority]\n",
    "\n",
    "    if len(y_true) == 0:\n",
    "        print(majority_class, responses)\n",
    "\n",
    "    is_digit = y_true[0]\n",
    "    if type(is_digit) == list:\n",
    "        is_digit = is_digit[0]\n",
    "    is_digit = is_digit.strip().isdigit()\n",
    "    if is_digit:\n",
    "        y_pred = [find_first_integer(pred) for pred in y_pred]\n",
    "    \n",
    "    return accuracy(y_true, y_pred), len(y_true)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def take_max(accs, task, key):\n",
    "    acc1 = accs[key][task]\n",
    "    if key+'+' in accs and task in accs[key+'+']:\n",
    "        return max(acc1, accs[key+'+'][task])\n",
    "    return acc1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "accs = {}\n",
    "for m in ['llama-3-8b-instruct', 'llama-3-8b-instruct-ft-1e', 'lawma-8b']:\n",
    "    accs[m] = [base_accs[m][task] for task in tasks]\n",
    "for key in ['0e', '1e', '3e']:\n",
    "    accs[key] = [take_max(ft_trials, task, key) for task in tasks]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1100x210 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "to_plot = ['0e', 'llama-3-8b-instruct-ft-1e', '1e', 'lawma-8b', '3e']\n",
    "bars_to_plot = [accs[k] for k in to_plot]\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(11, 2.1))\n",
    "index = np.arange(len(tasks))\n",
    "\n",
    "ax.grid()\n",
    "\n",
    "suffix = '+ specialized for a single task'\n",
    "labels = [\n",
    "    'Llama 3 8B Inst ' + suffix,\n",
    "    'Llama 3 8B Inst + 1 epoch on everything',\n",
    "    suffix,\n",
    "    'Lawma 8B (3 epochs on everything)',\n",
    "    suffix,\n",
    "]\n",
    "\n",
    "bar_width = 1. / (len(bars_to_plot)+0.5)\n",
    "bar_colors = list(palette[:5]) + [palette[6]]\n",
    "for i in range(len(bars_to_plot)):\n",
    "    positions = index + i * bar_width - bar_width * 2\n",
    "    ax.bar(positions, bars_to_plot[i], bar_width, color=bar_colors[i])\n",
    "              \n",
    "ax.set_ylabel('Accuracy', fontsize=12)\n",
    "ax.set_ylim(0.7, 0.96)\n",
    "ax.set_xlim(-0.5, len(tasks) - 0.4)\n",
    "\n",
    "ax.tick_params(axis='y', labelsize=11)\n",
    "\n",
    "ax.set_xticks(index + bar_width / 2)\n",
    "task_labels = [task_names[task] for task in tasks]\n",
    "task_labels = [l.split(' ')[0] + '\\n' + ' '.join(l.split(' ')[1:]) for l in task_labels]\n",
    "ax.set_xticklabels(task_labels, rotation=25, ha='right', fontsize=12, va='top')\n",
    "\n",
    "legend_patches = []\n",
    "from matplotlib.patches import Patch\n",
    "for i, label in enumerate(labels):\n",
    "    legend_patches.append(Patch(color=bar_colors[i], label=label))\n",
    "legend_patches = [legend_patches[0]] + [Patch(facecolor='none', edgecolor='none', label='')] + legend_patches[1:]\n",
    "\n",
    "re_order = [0, 2, 4, 1, 3, 5]\n",
    "legend_patches = [legend_patches[i] for i in re_order]\n",
    "\n",
    "# place legend below the plot, outside the plot area\n",
    "ax.legend(handles=legend_patches,loc='upper center', bbox_to_anchor=(0.5, -0.55), ncol=2, frameon=False, fontsize=12)\n",
    "\n",
    "# save as pdf\n",
    "plt.savefig('plots/lawma-specialized.pdf', bbox_inches='tight')"
   ]
  }
 ],
 "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": "3.10.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}