len_statisitcs.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yuxiaoming/zhuyunchang/Programs/anaconda3/envs/py36/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:523: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint8 = np.dtype([(\"qint8\", np.int8, 1)])\n",
      "/home/yuxiaoming/zhuyunchang/Programs/anaconda3/envs/py36/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:524: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint8 = np.dtype([(\"quint8\", np.uint8, 1)])\n",
      "/home/yuxiaoming/zhuyunchang/Programs/anaconda3/envs/py36/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:525: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint16 = np.dtype([(\"qint16\", np.int16, 1)])\n",
      "/home/yuxiaoming/zhuyunchang/Programs/anaconda3/envs/py36/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_quint16 = np.dtype([(\"quint16\", np.uint16, 1)])\n",
      "/home/yuxiaoming/zhuyunchang/Programs/anaconda3/envs/py36/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:527: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  _np_qint32 = np.dtype([(\"qint32\", np.int32, 1)])\n",
      "/home/yuxiaoming/zhuyunchang/Programs/anaconda3/envs/py36/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:532: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.\n",
      "  np_resource = np.dtype([(\"resource\", np.ubyte, 1)])\n",
      "I0116 15:13:26.303243 139865309423424 file_utils.py:39] PyTorch version 1.0.1.post2 available.\n",
      "I0116 15:13:26.762387 139865309423424 modeling_xlnet.py:194] Better speed can be achieved with apex installed from https://www.github.com/nvidia/apex .\n",
      "I0116 15:13:28.816495 139865309423424 tokenization_utils.py:373] loading file https://s3.amazonaws.com/models.huggingface.co/bert/roberta-large-vocab.json from cache at /home/yuxiaoming/.cache/torch/transformers/1ae1f5b6e2b22b25ccc04c000bb79ca847aa226d0761536b011cf7e5868f0655.ef00af9e673c7160b4d41cfda1f48c5f4cba57d5142754525572a846a1ab1b9b\n",
      "I0116 15:13:28.817844 139865309423424 tokenization_utils.py:373] loading file https://s3.amazonaws.com/models.huggingface.co/bert/roberta-large-merges.txt from cache at /home/yuxiaoming/.cache/torch/transformers/f8f83199a6270d582d6245dc100e99c4155de81c9745c6248077018fe01abcfb.70bec105b4158ed9a1747fea67a43f5dee97855c64d62b6ec3742f4cfdb5feda\n",
      "I0116 15:13:28.936852 139865309423424 data_process.py:201] ***** Loading train examples *****\n",
      "I0116 15:13:29.239394 139865309423424 data_process.py:201] ***** Loading dev examples *****\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Num train stories 17801\n",
      "Num dev stories 1532\n"
     ]
    }
   ],
   "source": [
    "import logging\n",
    "logging.basicConfig(format='%(asctime)s - %(levelname)s - %(name)s:%(lineno)s -   %(message)s',\n",
    "                    datefmt='%m/%d %H:%M:%S', level=logging.INFO)\n",
    "from transformers import RobertaTokenizer\n",
    "from data_process import AlphaNliProcessor, StoryFeatures\n",
    "\n",
    "tokenizer = RobertaTokenizer.from_pretrained('roberta-large', do_lower_case=False)\n",
    "processor = AlphaNliProcessor('./alphanli/', tokenizer)\n",
    "\n",
    "train_examples = processor.get_examples('train')\n",
    "dev_examples = processor.get_examples('dev')\n",
    "\n",
    "print('Num train stories', len(train_examples))\n",
    "print('Num dev stories', len(dev_examples))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "I0116 15:13:29.260528 139865309423424 data_process.py:139] *** No. 0 example ***\n",
      "I0116 15:13:29.261903 139865309423424 data_process.py:140] story_id: 00050cbb-049e-444f-a17a-04c882da4693-1\n",
      "I0116 15:13:29.262756 139865309423424 data_process.py:142] tokens:\n",
      "<s> Ch ad Ġwent Ġto Ġget Ġthe Ġwheel Ġalignment Ġmeasured Ġon Ġhis Ġcar . </s> Ch ad Ġwas Ġwaiting Ġfor Ġhis Ġcar Ġto Ġbe Ġwashed . </s> The Ġmechanic Ġprovided Ġa Ġworking Ġalignment Ġwith Ġnew Ġbody Ġwork . </s> <pad>\n",
      "<s> Ch ad Ġwent Ġto Ġget Ġthe Ġwheel Ġalignment Ġmeasured Ġon Ġhis Ġcar . </s> Ch ad Ġwas Ġwaiting Ġfor Ġhis Ġcar Ġto Ġbe Ġfinished . </s> The Ġmechanic Ġprovided Ġa Ġworking Ġalignment Ġwith Ġnew Ġbody Ġwork . </s> <pad>\n",
      "<s> Ch ad Ġwent Ġto Ġget Ġthe Ġwheel Ġalignment Ġmeasured Ġon Ġhis Ġcar . </s> Ch ad 's Ġcar Ġhad Ġall Ġsorts Ġof Ġother Ġproblems Ġbesides Ġalignment . </s> The Ġmechanic Ġprovided Ġa Ġworking Ġalignment Ġwith Ġnew Ġbody Ġwork .\n",
      "I0116 15:13:29.263556 139865309423424 data_process.py:144] token_ids:\n",
      "0 4771 625 439 7 120 5 5964 22432 9550 15 39 512 4 2 4771 625 21 2445 13 39 512 7 28 15158 4 2 133 25682 1286 10 447 22432 19 92 809 173 4 2 1\n",
      "0 4771 625 439 7 120 5 5964 22432 9550 15 39 512 4 2 4771 625 21 2445 13 39 512 7 28 1550 4 2 133 25682 1286 10 447 22432 19 92 809 173 4 2 1\n",
      "0 4771 625 439 7 120 5 5964 22432 9550 15 39 512 4 2 4771 625 18 512 56 70 9477 9 97 1272 12035 22432 4 2 133 25682 1286 10 447 22432 19 92 809 173 4\n",
      "I0116 15:13:29.264322 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:13:29.265244 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      "I0116 15:13:29.265966 139865309423424 data_process.py:150] labels: 0.0 1.0 1.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1\n",
      "I0116 15:13:29.272979 139865309423424 data_process.py:139] *** No. 1 example ***\n",
      "I0116 15:13:29.273740 139865309423424 data_process.py:140] story_id: 000731f7-c71d-49c8-b8cd-b6848933db99-1\n",
      "I0116 15:13:29.274556 139865309423424 data_process.py:142] tokens:\n",
      "<s> Ch ad Ġloves ĠBarry ĠBonds . </s> Ch ad Ġwent Ġto Ġmeet ĠBarry ĠBonds . </s> Ch ad Ġensured Ġthat Ġhe Ġtook Ġa Ġpicture Ġto Ġremember Ġthe Ġevent . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> Ch ad Ġloves ĠBarry ĠBonds . </s> Ch ad Ġfailed Ġto Ġmeet ĠBarry ĠBonds . </s> Ch ad Ġensured Ġthat Ġhe Ġtook Ġa Ġpicture Ġto Ġremember Ġthe Ġevent . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> Ch ad Ġloves ĠBarry ĠBonds . </s> Ch ad Ġfought ĠBarry ĠBonds . </s> Ch ad Ġensured Ġthat Ġhe Ġtook Ġa Ġpicture Ġto Ġremember Ġthe Ġevent . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "I0116 15:13:29.275304 139865309423424 data_process.py:144] token_ids:\n",
      "0 4771 625 6138 7600 23419 4 2 4771 625 439 7 972 7600 23419 4 2 4771 625 14953 14 37 362 10 2170 7 2145 5 515 4 2 1 1 1 1 1 1 1 1 1\n",
      "0 4771 625 6138 7600 23419 4 2 4771 625 1447 7 972 7600 23419 4 2 4771 625 14953 14 37 362 10 2170 7 2145 5 515 4 2 1 1 1 1 1 1 1 1 1\n",
      "0 4771 625 6138 7600 23419 4 2 4771 625 4951 7600 23419 4 2 4771 625 14953 14 37 362 10 2170 7 2145 5 515 4 2 1 1 1 1 1 1 1 1 1 1 1\n",
      "I0116 15:13:29.276007 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:13:29.276695 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:13:29.277364 139865309423424 data_process.py:150] labels: 1.0 0.0 0.0 1.0 0.0 1.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 0.0 -1 -1\n",
      "I0116 15:13:29.282725 139865309423424 data_process.py:139] *** No. 2 example ***\n",
      "I0116 15:13:29.283457 139865309423424 data_process.py:140] story_id: 00093549-05ba-4ae1-8216-64ef59a972a9-1\n",
      "I0116 15:13:29.284258 139865309423424 data_process.py:142] tokens:\n",
      "<s> H omer Ġbought Ġa Ġgas Ġgrill Ġfor Ġthe Ġsummer . </s> He Ġwanted Ġa Ġsteak . </s> They Ġgrilled Ġsteak Ġfor Ġthe Ġfirst Ġtime Ġon Ġthe Ġgrill . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> H omer Ġbought Ġa Ġgas Ġgrill Ġfor Ġthe Ġsummer . </s> He Ġdid Ġnot Ġlike Ġeating Ġste aks . </s> They Ġgrilled Ġsteak Ġfor Ġthe Ġfirst Ġtime Ġon Ġthe Ġgrill . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> H omer Ġbought Ġa Ġgas Ġgrill Ġfor Ġthe Ġsummer . </s> He Ġwas Ġreally Ġcraving Ġa Ġsteak Ġdinner . </s> They Ġgrilled Ġsteak Ġfor Ġthe Ġfirst Ġtime Ġon Ġthe Ġgrill . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "I0116 15:13:29.284973 139865309423424 data_process.py:144] token_ids:\n",
      "0 725 11032 2162 10 1123 20212 13 5 1035 4 2 894 770 10 19464 4 2 1213 20346 19464 13 5 78 86 15 5 20212 4 2 1 1 1 1 1 1 1 1 1 1\n",
      "0 725 11032 2162 10 1123 20212 13 5 1035 4 2 894 222 45 101 4441 11235 6629 4 2 1213 20346 19464 13 5 78 86 15 5 20212 4 2 1 1 1 1 1 1 1\n",
      "0 725 11032 2162 10 1123 20212 13 5 1035 4 2 894 21 269 32659 10 19464 3630 4 2 1213 20346 19464 13 5 78 86 15 5 20212 4 2 1 1 1 1 1 1 1\n",
      "I0116 15:13:29.285659 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:13:29.286344 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0\n",
      "I0116 15:13:29.286964 139865309423424 data_process.py:150] labels: 1.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1\n",
      "I0116 15:13:29.293604 139865309423424 data_process.py:139] *** No. 3 example ***\n",
      "I0116 15:13:29.294343 139865309423424 data_process.py:140] story_id: 000d05af-5db1-4d9d-b4e2-6c01b108b0b1-1\n",
      "I0116 15:13:29.295039 139865309423424 data_process.py:142] tokens:\n",
      "<s> B ren na Ġand ĠI Ġused Ġto Ġbe Ġbest Ġfriends . </s> B ren na Ġand ĠI Ġfought Ġover Ġa Ġboy . </s> We Ġnever Ġtalked Ġagain . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> B ren na Ġand ĠI Ġused Ġto Ġbe Ġbest Ġfriends . </s> Bre anna Ġand ĠI Ġwent Ġto Ġthe Ġmall Ġtogether . </s> We Ġnever Ġtalked Ġagain . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> B ren na Ġand ĠI Ġused Ġto Ġbe Ġbest Ġfriends . </s> B ren na Ġand ĠI Ġcompromised Ġover Ġa Ġboy . </s> We Ġnever Ġtalked Ġagain . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "I0116 15:13:29.295748 139865309423424 data_process.py:144] token_ids:\n",
      "0 387 2558 2133 8 38 341 7 28 275 964 4 2 387 2558 2133 8 38 4951 81 10 2143 4 2 170 393 3244 456 4 2 1 1 1 1 1 1 1 1 1 1\n",
      "0 387 2558 2133 8 38 341 7 28 275 964 4 2 31607 4057 8 38 439 7 5 9367 561 4 2 170 393 3244 456 4 2 1 1 1 1 1 1 1 1 1 1\n",
      "0 387 2558 2133 8 38 341 7 28 275 964 4 2 387 2558 2133 8 38 13969 81 10 2143 4 2 170 393 3244 456 4 2 1 1 1 1 1 1 1 1 1 1\n",
      "I0116 15:13:29.296391 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:13:29.297030 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "I0116 15:13:29.297654 139865309423424 data_process.py:150] labels: 1.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 -1 -1 -1 -1\n",
      "I0116 15:13:29.303876 139865309423424 data_process.py:139] *** No. 4 example ***\n",
      "I0116 15:13:29.304584 139865309423424 data_process.py:140] story_id: 000d9d91-9fec-4ae0-b2c0-79b12fb4f860-1\n",
      "I0116 15:13:29.305537 139865309423424 data_process.py:142] tokens:\n",
      "<s> Josh Ġbought Ġa Ġpar rot Ġas Ġa Ġpet . </s> Josh Ġrealized Ġthat Ġthe Ġpar rot Ġcould Ġtalk !. </s> Josh Ġwas Ġso Ġexcited Ġthat Ġhe Ġtaught Ġhis Ġpar rot Ġhow Ġto Ġsay Ġits Ġname ! </s> <pad> <pad>\n",
      "<s> Josh Ġbought Ġa Ġpar rot Ġas Ġa Ġpet . </s> Josh Ġrealized Ġthat Ġthe Ġdog Ġcould Ġtalk !. </s> Josh Ġwas Ġso Ġexcited Ġthat Ġhe Ġtaught Ġhis Ġpar rot Ġhow Ġto Ġsay Ġits Ġname ! </s> <pad> <pad> <pad>\n",
      "<s> Josh Ġbought Ġa Ġpar rot Ġas Ġa Ġpet . </s> Josh Ġrealized Ġthat Ġthe Ġpar rot Ġhad Ġno Ġvoice . </s> Josh Ġwas Ġso Ġexcited Ġthat Ġhe Ġtaught Ġhis Ġpar rot Ġhow Ġto Ġsay Ġits Ġname ! </s> <pad>\n",
      "I0116 15:13:29.306243 139865309423424 data_process.py:144] token_ids:\n",
      "0 32879 2162 10 2242 12179 25 10 4716 4 2 32879 5426 14 5 2242 12179 115 1067 42202 2 32879 21 98 2283 14 37 5850 39 2242 12179 141 7 224 63 766 328 2 1 1\n",
      "0 32879 2162 10 2242 12179 25 10 4716 4 2 32879 5426 14 5 2335 115 1067 42202 2 32879 21 98 2283 14 37 5850 39 2242 12179 141 7 224 63 766 328 2 1 1 1\n",
      "0 32879 2162 10 2242 12179 25 10 4716 4 2 32879 5426 14 5 2242 12179 56 117 2236 4 2 32879 21 98 2283 14 37 5850 39 2242 12179 141 7 224 63 766 328 2 1\n",
      "I0116 15:13:29.306874 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:13:29.307507 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0\n",
      "I0116 15:13:29.308099 139865309423424 data_process.py:150] labels: 1.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.0 0.0 -1 -1 -1 -1 -1 -1\n"
     ]
    }
   ],
   "source": [
    "train_features = StoryFeatures.convert_from_examples(train_examples, tokenizer, max_hyp_num=22, max_seq_len=72)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "I0116 15:14:10.007160 139865309423424 data_process.py:139] *** No. 0 example ***\n",
      "I0116 15:14:10.008227 139865309423424 data_process.py:140] story_id: 58090d3f-8a91-4c89-83ef-2b4994de9d241\n",
      "I0116 15:14:10.009174 139865309423424 data_process.py:142] tokens:\n",
      "<s> Ron Ġstarted Ġhis Ġnew Ġjob Ġas Ġa Ġlandsc aper Ġtoday . </s> Ron Ġignores Ġhis Ġbosses 's Ġorders Ġand Ġcalled Ġhim Ġan Ġidiot . </s> Ron Ġis Ġimmediately Ġfired Ġfor Ġins ub ord ination . </s> <pad> <pad> <pad>\n",
      "<s> Ron Ġstarted Ġhis Ġnew Ġjob Ġas Ġa Ġlandsc aper Ġtoday . </s> Ron 's Ġboss Ġcalled Ġhim Ġan Ġidiot . </s> Ron Ġis Ġimmediately Ġfired Ġfor Ġins ub ord ination . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "I0116 15:14:10.009883 139865309423424 data_process.py:144] token_ids:\n",
      "0 27674 554 39 92 633 25 10 19342 15888 452 4 2 27674 30462 39 14063 18 3365 8 373 123 41 30603 4 2 27674 16 1320 2277 13 7540 1792 3109 8111 4 2 1 1 1\n",
      "0 27674 554 39 92 633 25 10 19342 15888 452 4 2 27674 18 3504 373 123 41 30603 4 2 27674 16 1320 2277 13 7540 1792 3109 8111 4 2 1 1 1 1 1 1 1\n",
      "I0116 15:14:10.010824 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:14:10.011609 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0\n",
      "I0116 15:14:10.012171 139865309423424 data_process.py:150] labels: 1.0 0.0\n",
      "I0116 15:14:10.014473 139865309423424 data_process.py:139] *** No. 1 example ***\n",
      "I0116 15:14:10.015080 139865309423424 data_process.py:140] story_id: 91f9d1d9-934c-44f9-9677-3614def2874b2\n",
      "I0116 15:14:10.016184 139865309423424 data_process.py:142] tokens:\n",
      "<s> S andy Ġlived Ġin ĠNew ĠYork . </s> It Ġstormed Ġin ĠNew ĠYork . </s> S andy Ġwas Ġprepared . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> S andy Ġlived Ġin ĠNew ĠYork . </s> She Ġpart ied Ġall Ġnight . </s> S andy Ġwas Ġprepared . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "I0116 15:14:10.016715 139865309423424 data_process.py:144] token_ids:\n",
      "0 104 10708 3033 11 188 469 4 2 243 16965 11 188 469 4 2 104 10708 21 2460 4 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      "0 104 10708 3033 11 188 469 4 2 2515 233 2550 70 363 4 2 104 10708 21 2460 4 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      "I0116 15:14:10.017312 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:14:10.017842 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:14:10.018568 139865309423424 data_process.py:150] labels: 1.0 0.0\n",
      "I0116 15:14:10.021867 139865309423424 data_process.py:139] *** No. 2 example ***\n",
      "I0116 15:14:10.022521 139865309423424 data_process.py:140] story_id: dfedfdca-3c9d-451f-bf9a-abbaa8bbf54f-1\n",
      "I0116 15:14:10.023462 139865309423424 data_process.py:142] tokens:\n",
      "<s> Mary 's Ġmom Ġcame Ġhome Ġwith Ġmore Ġbananas Ġthan Ġthey Ġcould Ġpossibly Ġeat . </s> Mary Ġand Ġher Ġmom Ġdecided Ġto Ġmake Ġchocolate Ġcovered Ġfrozen Ġbananas Ġto Ġavoid Ġwaste . </s> That Ġwas Ġthe Ġbest Ġway Ġever Ġto Ġeat\n",
      "<s> Mary 's Ġmom Ġcame Ġhome Ġwith Ġmore Ġbananas Ġthan Ġthey Ġcould Ġpossibly Ġeat . </s> So ĠMary Ġmade Ġpineapple Ġsplits Ġfor Ġeveryone . </s> That Ġwas Ġthe Ġbest Ġway Ġever Ġto Ġeat Ġa Ġbanana ! </s> <pad> <pad> <pad>\n",
      "I0116 15:14:10.024199 139865309423424 data_process.py:144] token_ids:\n",
      "0 24877 18 3795 376 184 19 55 31130 87 51 115 3544 3529 4 2 24877 8 69 3795 1276 7 146 7548 2913 9214 31130 7 1877 3844 4 2 1711 21 5 275 169 655 7 3529\n",
      "0 24877 18 3795 376 184 19 55 31130 87 51 115 3544 3529 4 2 2847 2708 156 33336 26118 13 961 4 2 1711 21 5 275 169 655 7 3529 10 23611 328 2 1 1 1\n",
      "I0116 15:14:10.024877 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:14:10.025588 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0\n",
      "I0116 15:14:10.026302 139865309423424 data_process.py:150] labels: 1.0 0.0\n",
      "I0116 15:14:10.028644 139865309423424 data_process.py:139] *** No. 3 example ***\n",
      "I0116 15:14:10.029164 139865309423424 data_process.py:140] story_id: 08302614-b7fc-4fd5-ad07-8c264dc4dcb0-1\n",
      "I0116 15:14:10.029840 139865309423424 data_process.py:142] tokens:\n",
      "<s> Jim Ġwas Ġworking Ġon Ġa Ġproject . </s> Jim Ġfound Ġhe Ġwas Ġmissing Ġan Ġitem . </s> Luckily , Ġhe Ġfound Ġit Ġon Ġa Ġnearby Ġshelf . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> Jim Ġwas Ġworking Ġon Ġa Ġproject . </s> Jim Ġneeded Ġa Ġcertain Ġanimal Ġfor Ġit . </s> Luckily , Ġhe Ġfound Ġit Ġon Ġa Ġnearby Ġshelf . </s> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "I0116 15:14:10.030475 139865309423424 data_process.py:144] token_ids:\n",
      "0 24021 21 447 15 10 695 4 2 24021 303 37 21 1716 41 6880 4 2 42498 6 37 303 24 15 10 3027 15263 4 2 1 1 1 1 1 1 1 1 1 1 1\n",
      "0 24021 21 447 15 10 695 4 2 24021 956 10 1402 3477 13 24 4 2 42498 6 37 303 24 15 10 3027 15263 4 2 1 1 1 1 1 1 1 1 1 1 1\n",
      "I0116 15:14:10.031028 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:14:10.031630 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:14:10.032115 139865309423424 data_process.py:150] labels: 1.0 0.0\n",
      "I0116 15:14:10.034316 139865309423424 data_process.py:139] *** No. 4 example ***\n",
      "I0116 15:14:10.034779 139865309423424 data_process.py:140] story_id: 38b4e3a2-e815-427b-b7f1-64ffd40d6f84-1\n",
      "I0116 15:14:10.035419 139865309423424 data_process.py:142] tokens:\n",
      "<s> Sean Ġwas Ġsitting Ġat Ġhis Ġdesk . </s> He Ġnoticed Ġthe Ġchair Ġleg Ġwas Ġfalling Ġoff . </s> After Ġa Ġminute , Ġhe Ġwas Ġable Ġto Ġput Ġthe Ġchair Ġback Ġtogether . </s> <pad> <pad> <pad> <pad> <pad> <pad>\n",
      "<s> Sean Ġwas Ġsitting Ġat Ġhis Ġdesk . </s> He Ġleaned Ġtoo Ġfar Ġback Ġand Ġhis Ġchair Ġtipped Ġover . </s> After Ġa Ġminute , Ġhe Ġwas Ġable Ġto Ġput Ġthe Ġchair Ġback Ġtogether . </s> <pad> <pad> <pad> <pad>\n",
      "I0116 15:14:10.035994 139865309423424 data_process.py:144] token_ids:\n",
      "0 25285 21 2828 23 39 8429 4 2 894 5324 5 3428 2985 21 3064 160 4 2 4993 10 2289 6 37 21 441 7 342 5 3428 124 561 4 2 1 1 1 1 1 1\n",
      "0 25285 21 2828 23 39 8429 4 2 894 27627 350 444 124 8 39 3428 13402 81 4 2 4993 10 2289 6 37 21 441 7 342 5 3428 124 561 4 2 1 1 1 1\n",
      "I0116 15:14:10.036589 139865309423424 data_process.py:146] token_type_ids:\n",
      "0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "I0116 15:14:10.037146 139865309423424 data_process.py:148] input_mask:\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0\n",
      "1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0\n",
      "I0116 15:14:10.037662 139865309423424 data_process.py:150] labels: 1.0 0.0\n"
     ]
    }
   ],
   "source": [
    "dev_features = StoryFeatures.convert_from_examples(dev_examples, tokenizer, max_hyp_num=2, max_seq_len=72)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "seq_lens = []\n",
    "for f in train_features + dev_features:\n",
    "    seq_lens.extend(list(x for x in map(sum, f.input_mask) if x != 0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "max seq len: 67\n",
      "mean seq len: 34.858269246671625\n",
      "99% of seq len <= 50\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.title('seq_lens')\n",
    "counts, bins, patch = plt.hist(seq_lens, bins=10, log=False)\n",
    "width = bins[1] - bins[0]\n",
    "for a, b in zip(bins, counts):\n",
    "    plt.text(a + width / 2, b + 0.05, '%.0f' % b if b > 0 else '', ha='center', va='bottom',fontsize=7)\n",
    "plt.show()\n",
    "print('max seq len:', max(seq_lens))\n",
    "print('mean seq len:', sum(seq_lens) / len(seq_lens))\n",
    "print('99% of seq len <=', sorted(seq_lens)[int(0.99 * len(seq_lens))])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}