{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "c975c670-97cc-453e-bba6-3639cf8d5e89",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/jue@together.xyz/miniconda3/envs/nebula-fav2/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "import json\n",
    "from datasets import load_dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "e818dca8-bd10-4b89-9fcd-5cd9252b4e07",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset hellaswag (/home/jue@together.xyz/.cache/huggingface/datasets/hellaswag/default/0.1.0/512a66dd8b1b1643ab4a48aa4f150d04c91680da6a4096498a5e5f799623d5ae)\n",
      "100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 53.00it/s]\n"
     ]
    }
   ],
   "source": [
    "task_name = 'hellaswag'\n",
    "data = load_dataset(task_name)\n",
    "data.shuffle(seed=42)\n",
    "with open(f'../seed_data{task_name}.jsonl', 'w') as f:\n",
    "    for i_item, item in enumerate(data['train']):\n",
    "        text = item['ctx'] + item['endings'][int(item['label'])]\n",
    "        f.write(\n",
    "            json.dumps({'text': text, 'source': task_name}) + '\\n'\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c42aae00-be85-4b64-9325-f6a6139c6ee6",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Downloading builder script: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3.38k/3.38k [00:00<00:00, 32.1MB/s]\n",
      "Downloading metadata: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1.91k/1.91k [00:00<00:00, 20.0MB/s]\n",
      "Downloading readme: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6.60k/6.60k [00:00<00:00, 45.1MB/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading and preparing dataset boolq/default to /home/jue@together.xyz/.cache/huggingface/datasets/boolq/default/0.1.0/bf0dd57da941c50de94ae3ce3cef7fea48c08f337a4b7aac484e9dddc5aa24e5...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Downloading data files:   0%|                                                                                                                                                         | 0/2 [00:00<?, ?it/s]\n",
      "Downloading data:   0%|                                                                                                                                                         | 0.00/6.53M [00:00<?, ?B/s]\u001b[A\n",
      "Downloading data:   0%|▍                                                                                                                                                | 17.4k/6.53M [00:00<00:43, 150kB/s]\u001b[A\n",
      "Downloading data:   1%|█▏                                                                                                                                               | 54.3k/6.53M [00:00<00:26, 246kB/s]\u001b[A\n",
      "Downloading data:   2%|██▎                                                                                                                                               | 104k/6.53M [00:00<00:18, 346kB/s]\u001b[A\n",
      "Downloading data:   3%|████▍                                                                                                                                             | 197k/6.53M [00:00<00:11, 554kB/s]\u001b[A\n",
      "Downloading data:   5%|███████▋                                                                                                                                          | 343k/6.53M [00:00<00:07, 865kB/s]\u001b[A\n",
      "Downloading data:  10%|█████████████▊                                                                                                                                   | 623k/6.53M [00:00<00:03, 1.50MB/s]\u001b[A\n",
      "Downloading data:  17%|████████████████████████▋                                                                                                                       | 1.12M/6.53M [00:00<00:02, 2.58MB/s]\u001b[A\n",
      "Downloading data:  31%|█████████████████████████████████████████████▏                                                                                                  | 2.05M/6.53M [00:00<00:00, 4.67MB/s]\u001b[A\n",
      "Downloading data: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6.53M/6.53M [00:01<00:00, 6.28MB/s]\u001b[A\n",
      "Downloading data files:  50%|████████████████████████████████████████████████████████████████████████▌                                                                        | 1/2 [00:01<00:01,  1.95s/it]\n",
      "Downloading data:   0%|                                                                                                                                                         | 0.00/2.24M [00:00<?, ?B/s]\u001b[A\n",
      "Downloading data:   1%|█▏                                                                                                                                               | 17.4k/2.24M [00:00<00:14, 150kB/s]\u001b[A\n",
      "Downloading data:   2%|███▌                                                                                                                                             | 54.3k/2.24M [00:00<00:08, 246kB/s]\u001b[A\n",
      "Downloading data:   5%|██████▊                                                                                                                                           | 104k/2.24M [00:00<00:06, 347kB/s]\u001b[A\n",
      "Downloading data:   9%|████████████▋                                                                                                                                     | 195k/2.24M [00:00<00:03, 549kB/s]\u001b[A\n",
      "Downloading data:  15%|██████████████████████▏                                                                                                                           | 341k/2.24M [00:00<00:02, 863kB/s]\u001b[A\n",
      "Downloading data:  28%|███████████████████████████████████████▉                                                                                                         | 617k/2.24M [00:00<00:01, 1.49MB/s]\u001b[A\n",
      "Downloading data:  49%|██████████████████████████████████████████████████████████████████████▏                                                                         | 1.09M/2.24M [00:00<00:00, 2.53MB/s]\u001b[A\n",
      "Downloading data: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2.24M/2.24M [00:00<00:00, 2.61MB/s]\u001b[A\n",
      "Downloading data files: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:03<00:00,  1.85s/it]\n",
      "                                                                                                                                                                                                            \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset boolq downloaded and prepared to /home/jue@together.xyz/.cache/huggingface/datasets/boolq/default/0.1.0/bf0dd57da941c50de94ae3ce3cef7fea48c08f337a4b7aac484e9dddc5aa24e5. Subsequent calls will reuse this data.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 139.08it/s]\n"
     ]
    }
   ],
   "source": [
    "task_name = 'boolq'\n",
    "data = load_dataset(task_name)\n",
    "data.shuffle(seed=42)\n",
    "with open(f'../seed_data{task_name}.jsonl', 'w') as f:\n",
    "    for i_item, item in enumerate(data['train']):\n",
    "        text = f\"{item['passage']}\\nQuestion: {item['question']}?\\nAnswer: {item['answer']}\"\n",
    "        f.write(\n",
    "            json.dumps({'text': text, 'source': task_name}) + '\\n'\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "878216f4-74e4-46ba-bfcd-c95348c10415",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset ai2_arc (/home/jue@together.xyz/.cache/huggingface/datasets/ai2_arc/ARC-Challenge/1.0.0/1569c2591ea2683779581d9fb467203d9aa95543bb9b75dcfde5da92529fd7f6)\n",
      "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 624.09it/s]\n"
     ]
    }
   ],
   "source": [
    "task_name = 'arc_challenge'\n",
    "data = load_dataset('ai2_arc', 'ARC-Challenge')\n",
    "data.shuffle(seed=42)\n",
    "with open(f'../seed_data{task_name}.jsonl', 'w') as f:\n",
    "    for i_item, item in enumerate(data['train']):\n",
    "        i_a = item['choices']['label'].index(item['answerKey'])\n",
    "        q = item['question']\n",
    "        a = item['choices']['text'][i_a]\n",
    "        text = \"Question: \" + q + \"\\nAnswer:\" + a\n",
    "        f.write(\n",
    "            json.dumps({'text': text, 'source': task_name}) + '\\n'\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "7cb51285-d5ce-4ae2-bd7f-ac15e87c4fb7",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset ai2_arc (/home/jue@together.xyz/.cache/huggingface/datasets/ai2_arc/ARC-Easy/1.0.0/1569c2591ea2683779581d9fb467203d9aa95543bb9b75dcfde5da92529fd7f6)\n",
      "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 426.87it/s]\n"
     ]
    }
   ],
   "source": [
    "task_name = 'arc_easy'\n",
    "data = load_dataset('ai2_arc', 'ARC-Easy')\n",
    "data.shuffle(seed=42)\n",
    "with open(f'../seed_data{task_name}.jsonl', 'w') as f:\n",
    "    for i_item, item in enumerate(data['train']):\n",
    "        i_a = item['choices']['label'].index(item['answerKey'])\n",
    "        q = item['question']\n",
    "        a = item['choices']['text'][i_a]\n",
    "        text = \"Question: \" + q + \"\\nAnswer:\" + a\n",
    "        f.write(\n",
    "            json.dumps({'text': text, 'source': task_name}) + '\\n'\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9347c339-4559-4111-bf44-ce2b624edc9e",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset super_glue (/home/jue@together.xyz/.cache/huggingface/datasets/super_glue/copa/1.0.3/bb9675f958ebfee0d5d6dc5476fafe38c79123727a7258d515c450873dbdbbed)\n",
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 1055.52it/s]\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/super_glue/copa/1.0.3/bb9675f958ebfee0d5d6dc5476fafe38c79123727a7258d515c450873dbdbbed/cache-06e156489a3d17ef.arrow\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/super_glue/copa/1.0.3/bb9675f958ebfee0d5d6dc5476fafe38c79123727a7258d515c450873dbdbbed/cache-f4bd523109c343ed.arrow\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/super_glue/copa/1.0.3/bb9675f958ebfee0d5d6dc5476fafe38c79123727a7258d515c450873dbdbbed/cache-b8d0981688b46f46.arrow\n"
     ]
    }
   ],
   "source": [
    "task_name = 'copa'\n",
    "data = load_dataset('super_glue', task_name)\n",
    "data.shuffle(seed=42)\n",
    "with open(f'../seed_data{task_name}.jsonl', 'w') as f:\n",
    "    for i_item, item in enumerate(data['train']):\n",
    "        i_a = item['label']\n",
    "        premise = item['premise']\n",
    "        question = item['question']\n",
    "        choice = item['choice1'] if i_a == 0 else item['choice2']\n",
    "        choice = choice[0].lower() + choice[1:]\n",
    "\n",
    "        connector = {\n",
    "            \"cause\": \"because\",\n",
    "            \"effect\": \"therefore\",\n",
    "        }[question]\n",
    "\n",
    "        text = premise.strip()[:-1] + f\" {connector}\" + ' ' + choice\n",
    "        f.write(\n",
    "            json.dumps({'text': text, 'source': task_name}) + '\\n'\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "dc24ba3b-aa6a-4665-87aa-08471a87de9b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset piqa (/home/jue@together.xyz/.cache/huggingface/datasets/piqa/plain_text/1.1.0/6c611c1a9bf220943c4174e117d3b660859665baf1d43156230116185312d011)\n",
      "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 910.95it/s]\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/piqa/plain_text/1.1.0/6c611c1a9bf220943c4174e117d3b660859665baf1d43156230116185312d011/cache-64434cd628e6e9e0.arrow\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/piqa/plain_text/1.1.0/6c611c1a9bf220943c4174e117d3b660859665baf1d43156230116185312d011/cache-97a6e3c466ced62d.arrow\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/piqa/plain_text/1.1.0/6c611c1a9bf220943c4174e117d3b660859665baf1d43156230116185312d011/cache-6b1e5924150382c7.arrow\n"
     ]
    }
   ],
   "source": [
    "task_name = 'piqa'\n",
    "data = load_dataset(task_name)\n",
    "data.shuffle(seed=42)\n",
    "with open(f'../seed_data{task_name}.jsonl', 'w') as f:\n",
    "    for i_item, item in enumerate(data['train']):\n",
    "        i_a = item['label']\n",
    "        goal = item['goal']\n",
    "        sol = item['sol1'] if i_a == 0 else item['sol2']\n",
    "\n",
    "        text = \"Question: \" + goal + \"\\nAnswer: \" + sol\n",
    "        f.write(\n",
    "            json.dumps({'text': text, 'source': task_name}) + '\\n'\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "501744bd-0260-4a3c-89dc-df264ff3306e",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset winogrande (/home/jue@together.xyz/.cache/huggingface/datasets/winogrande/winogrande_xl/1.1.0/a826c3d3506aefe0e9e9390dcb53271070536586bab95849876b2c1743df56e2)\n",
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 1014.75it/s]\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/winogrande/winogrande_xl/1.1.0/a826c3d3506aefe0e9e9390dcb53271070536586bab95849876b2c1743df56e2/cache-ebcf1ebbadaf243c.arrow\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/winogrande/winogrande_xl/1.1.0/a826c3d3506aefe0e9e9390dcb53271070536586bab95849876b2c1743df56e2/cache-ea79c2b85e9d846f.arrow\n",
      "Loading cached shuffled indices for dataset at /home/jue@together.xyz/.cache/huggingface/datasets/winogrande/winogrande_xl/1.1.0/a826c3d3506aefe0e9e9390dcb53271070536586bab95849876b2c1743df56e2/cache-5aca6c830a1dfa33.arrow\n"
     ]
    }
   ],
   "source": [
    "task_name = 'winogrande'\n",
    "data = load_dataset(task_name, 'winogrande_xl')\n",
    "data.shuffle(seed=42)\n",
    "with open(f'../seed_data{task_name}.jsonl', 'w') as f:\n",
    "    for i_item, item in enumerate(data['train']):\n",
    "        i_a = item['answer']\n",
    "        sentence = item['sentence']\n",
    "        option = item['option1'] if i_a == 1 else item['option2']\n",
    "        \n",
    "        text = sentence.replace('_', option)\n",
    "        f.write(\n",
    "            json.dumps({'text': text, 'source': task_name}) + '\\n'\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "65d4b641-2b78-476f-89ac-6ecb48e7d044",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Found cached dataset json (/home/jue@together.xyz/.cache/huggingface/datasets/juewang___json/juewang--target-data-72e09ed206c54e99/0.0.0/8bb11242116d547c741b2e8a1f18598ffdd40a1d4f2a2872c7a28b697434bc96)\n"
     ]
    }
   ],
   "source": [
    "dataset = load_dataset(\"juewang/target-data\", data_files='seed_data/*.jsonl', split='train')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a0554b4f-c5c0-4621-8b91-3daf05e6fbdf",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "nebula-fav2",
   "language": "python",
   "name": "nebula-fav2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}