commonsense.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Commonsense (AI2 Reasoning Challenge)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from datasets import load_dataset\n",
    "import tqdm\n",
    "from torch.utils.data import DataLoader\n",
    "import sys\n",
    "import pickle as pkl\n",
    "sys.path.append(\"../\")\n",
    "\n",
    "from train import train_ensemble\n",
    "\n",
    "\n",
    "hf_token = \"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "ds_names = [\"arc_challenge\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_model_data = {\n",
    "    \"meta-llama__Llama-2-70b-chat-hf\": {},\n",
    "    \"mistralai__Mixtral-8x7B-Instruct-v0.1\" : {},\n",
    "    \"google__gemma-7b\": {},\n",
    "    \"microsoft__phi-2\": {},\n",
    "    \"mistralai__Mistral-7B-v0.1\": {},\n",
    "}\n",
    "\n",
    "model_names = list(all_model_data.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "meta-llama__Llama-2-70b-chat-hf\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [01:05<00:00, 65.59s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mistralai__Mixtral-8x7B-Instruct-v0.1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:58<00:00, 58.19s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "google__gemma-7b\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:55<00:00, 55.62s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "microsoft__phi-2\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:55<00:00, 55.71s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mistralai__Mistral-7B-v0.1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:56<00:00, 56.30s/it]\n"
     ]
    }
   ],
   "source": [
    "def load_data(model_name, ds_name):\n",
    "    data = load_dataset(\n",
    "    f\"HuggingFaceEvalInternal/{model_name}-details-private\",\n",
    "    name=f\"{model_name}__leaderboard_{ds_name}\",\n",
    "    split=\"latest\",\n",
    "    token=hf_token)\n",
    "\n",
    "    \n",
    "    data_len = len(data)\n",
    "    remove_idx = [idx for idx in range(len(data)) if np.array(data[\"resps\"][idx]).shape[0] != 4 or data[\"doc\"][idx][\"answerKey\"] not in [\"A\", \"B\", \"C\", \"D\"]]\n",
    "\n",
    "    doc_ids = [data[\"doc_id\"][idx] for idx in range(data_len) if idx not in remove_idx]\n",
    "    resps = [data[\"resps\"][idx] for idx in range(data_len) if idx not in remove_idx]\n",
    "    labels = [data[\"doc\"][idx][\"answerKey\"] for idx in range(data_len) if idx not in remove_idx]\n",
    "    idx = np.argsort(doc_ids)\n",
    "\n",
    "    probs = np.array(resps).squeeze()[:, :, 0].astype(float)\n",
    "    labels = np.array(labels)\n",
    "    choices = sorted(list(set(labels)))\n",
    "    labels_list = [choices.index(lbl) for lbl in labels]\n",
    "    label_arr = np.array(labels_list).astype(int)\n",
    "\n",
    "    return probs[idx], label_arr[idx]\n",
    "\n",
    "for mn in list(all_model_data.keys()):\n",
    "    print(mn)\n",
    "    for ds_name in tqdm.tqdm(ds_names):\n",
    "        x, y = load_data(mn, ds_name)\n",
    "        all_model_data[mn][ds_name] = {\n",
    "            \"data\" : x,\n",
    "            \"label\": y\n",
    "        }\n",
    "\n",
    "with open(\"commonsense_data.pkl\", \"wb\") as f:\n",
    "    pkl.dump(all_model_data, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"commonsense_data.pkl\", \"rb\") as f:\n",
    "    all_model_data = pkl.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_names = list(all_model_data.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "meta-llama__Llama-2-70b-chat-hf: Acc: 0.4895\n",
      "mistralai__Mixtral-8x7B-Instruct-v0.1: Acc: 0.6329\n",
      "google__gemma-7b: Acc: 0.5664\n",
      "microsoft__phi-2: Acc: 0.5629\n",
      "mistralai__Mistral-7B-v0.1: Acc: 0.5455\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/stekin6/sft/inference_llm/other_results/../train.py:77: UserWarning: nn.init.xavier_uniform is now deprecated in favor of nn.init.xavier_uniform_.\n",
      "  torch.nn.init.xavier_uniform(m.weight)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training ended, total time: 4.74\n",
      "Novel Acc = 64.1875 +- 5.19\n",
      "64.1875\n"
     ]
    }
   ],
   "source": [
    "ds_name = ds_names[0]\n",
    "results = {}\n",
    "\n",
    "results[ds_name] = {}\n",
    "all_x, all_y = [], []\n",
    "for mn in all_model_data.keys():\n",
    "    x, y = all_model_data[mn][ds_name].values()\n",
    "    all_x.append(x)\n",
    "    all_y.append(y)\n",
    "all_x = np.concatenate(all_x, axis=1)\n",
    "\n",
    "try:\n",
    "    for y_idx in all_y:\n",
    "        for i in range(len(y_idx)):\n",
    "            assert(y_idx[i] == y[i])\n",
    "except AssertionError as e:\n",
    "    print(e)\n",
    "    print(f\"skipping ... {ds_name}\")\n",
    "\n",
    "concat_data = np.concatenate([all_x, y[:, None]], axis=1)\n",
    "rand_id = np.random.permutation(len(concat_data))\n",
    "train_data = concat_data[rand_id]\n",
    "\n",
    "train_size = int(len(x) * 0.75)\n",
    "train_data, test_data = concat_data[:train_size], concat_data[train_size:]\n",
    "\n",
    "val_size = int(len(train_data) * 0.25)\n",
    "train_size = len(train_data) - val_size\n",
    "\n",
    "novel_data = test_data[:, :-1]\n",
    "novel_label = test_data[:, -1]\n",
    "for i, novel_arr in enumerate(np.split(novel_data, len(model_names), axis=1)):\n",
    "    pred = novel_arr.argmax(axis=1)\n",
    "    acc = np.mean(novel_label == pred)\n",
    "    print(f\"{model_names[i]}: Acc: {acc:.4f}\")\n",
    "    results[ds_name][model_names[i]] = acc * 100\n",
    "\n",
    "\n",
    "batch_size = 64\n",
    "train_loader = DataLoader(train_data, batch_size=batch_size, shuffle=True)\n",
    "val_loader = DataLoader(train_data[train_size:], batch_size=batch_size, shuffle=True)\n",
    "novel_loader = DataLoader(test_data, batch_size=batch_size, shuffle=True)\n",
    "\n",
    "space_size = all_x[0].shape[0] // len(model_names)\n",
    "\n",
    "exp_result = train_ensemble(model_names, train_loader, val_loader, novel_loader,\n",
    "                n_epochs=500, save_dir=\".\",\n",
    "                space_size=space_size, verbose=False)\n",
    "results[ds_name][\"TOPLA\"] = exp_result[\"test_acc\"]\n",
    "print(exp_result[\"test_acc\"])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "avg_scores = {mn: 0 for mn in model_names}\n",
    "avg_scores[\"TOPLA\"] = 0\n",
    "for ds_name, res in results.items():\n",
    "    for mn, acc in res.items():\n",
    "        avg_scores[mn] += acc\n",
    "\n",
    "for mn in model_names:\n",
    "    avg_scores[mn] = avg_scores[mn] / len(results.keys())\n",
    "avg_scores[\"TOPLA\"] /= len(results.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'mistralai__Mixtral-8x7B-Instruct-v0.1': 63.286713286713294,\n",
       " 'google__gemma-7b': 56.64335664335665,\n",
       " 'microsoft__phi-2': 56.29370629370629,\n",
       " 'mistralai__Mistral-7B-v0.1': 54.54545454545454,\n",
       " 'TOPLA': 63.604166666666664}"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "avg_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "width = 0.5  # the width of the bars\n",
    "multiplier = 0\n",
    "mstrip_names = []\n",
    "x = np.arange(len(avg_scores.keys()))\n",
    "for i, (mn, acc) in enumerate(avg_scores.items()):\n",
    "    strip_name = mn.split(\"__\")\n",
    "    strip_name = strip_name[1] if len(strip_name) == 2 else strip_name[0]\n",
    "    offset = width * multiplier\n",
    "    rects = ax.bar(x[i] + offset, acc, width, label=strip_name)\n",
    "    # mstrip_names.append(strip_name)\n",
    "\n",
    "ax.set_ylabel('Performance Acc (%)')\n",
    "ax.set_title('Commonsense ARC, Base Models and TOPLA Performance')\n",
    "ax.set_xticks(x, mstrip_names, rotation=10)\n",
    "ax.legend(loc='upper right')\n",
    "ax.set_ylim(0, 100)\n",
    "ax.yaxis.grid(zorder=0)\n",
    "plt.show()\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "mistral",
   "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": "3.9.19"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}