run_stacked_debert.ipynb

{
 "nbformat": 4,
 "nbformat_minor": 0,
 "metadata": {
  "colab": {
   "name": "run_stacked_debert.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "kernelspec": {
   "name": "python3",
   "language": "python",
   "display_name": "Python 3"
  },
  "accelerator": "GPU",
  "pycharm": {
   "stem_cell": {
    "cell_type": "raw",
    "source": [],
    "metadata": {
     "collapsed": false
    }
   }
  }
 },
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KVlTCvEgXWR-",
    "colab_type": "text"
   },
   "source": [
    "## Requirements"
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "Xu4TfR-3DwRr",
    "colab_type": "code",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "outputId": "11f443e2-6696-415d-f2ee-8c13d6821d93",
    "pycharm": {
     "is_executing": false
    }
   },
   "source": [
    "from __future__ import absolute_import\n",
    "from __future__ import division\n",
    "from __future__ import print_function\n",
    "\n",
    "import csv\n",
    "import os\n",
    "import logging\n",
    "import argparse\n",
    "import random\n",
    "from tqdm import tqdm, trange\n",
    "import json\n",
    "\n",
    "import numpy as np\n",
    "import torch\n",
    "from torch.utils.data import TensorDataset, DataLoader, RandomSampler, SequentialSampler\n",
    "from torch.utils.data.distributed import DistributedSampler\n",
    "from sklearn.metrics import confusion_matrix, precision_recall_fscore_support, f1_score\n",
    "from plot_confusion_matrix import plot_confusion_matrix\n",
    "\n",
    "from models.stacked_debert_dae.tokenization import BertTokenizer\n",
    "from models.stacked_debert_dae.modeling import BertForSequenceClassification\n",
    "from models.stacked_debert_dae import BertAdam\n",
    "from models.stacked_debert_dae import PYTORCH_PRETRAINED_BERT_CACHE\n",
    "from models.stacked_debert_dae.denoising_mlp import AutoEncoder\n",
    "\n",
    "\n",
    "logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s -   %(message)s',\n",
    "                    datefmt = '%m/%d/%Y %H:%M:%S',\n",
    "                    level = logging.INFO)\n",
    "logger = logging.getLogger(__name__)"
   ],
   "execution_count": 107,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rRkvtJnYXZ3Y",
    "colab_type": "text"
   },
   "source": [
    "### Data processing functions"
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "5k-4ooYiXf4z",
    "colab_type": "code",
    "colab": {},
    "pycharm": {
     "is_executing": false
    }
   },
   "source": [
    "class InputExample(object):\n",
    "    \"\"\"A single training/test example for simple sequence classification.\"\"\"\n",
    "\n",
    "    def __init__(self, guid, text_a, text_b=None, label=None):\n",
    "        \"\"\"Constructs a InputExample.\n",
    "\n",
    "        Args:\n",
    "            guid: Unique id for the example.\n",
    "            text_a: string. The untokenized text of the first sequence. For single\n",
    "            sequence tasks, only this sequence must be specified.\n",
    "            text_b: (Optional) string. The untokenized text of the second sequence.\n",
    "            Only must be specified for sequence pair tasks.\n",
    "            label: (Optional) string. The label of the example. This should be\n",
    "            specified for train and dev examples, but not for test examples.\n",
    "        \"\"\"\n",
    "        self.guid = guid\n",
    "        self.text_a = text_a\n",
    "        self.text_b = text_b\n",
    "        self.label = label\n",
    "\n",
    "\n",
    "class InputFeatures(object):\n",
    "    \"\"\"A single set of features of data.\"\"\"\n",
    "\n",
    "    def __init__(self, input_ids, input_mask, segment_ids, label_id):\n",
    "        self.input_ids = input_ids\n",
    "        self.input_mask = input_mask\n",
    "        self.segment_ids = segment_ids\n",
    "        self.label_id = label_id\n",
    "\n",
    "\n",
    "class DataProcessor(object):\n",
    "    \"\"\"Base class for data converters for sequence classification data sets.\"\"\"\n",
    "\n",
    "    def get_train_examples(self, data_dir):\n",
    "        \"\"\"Gets a collection of `InputExample`s for the train set.\"\"\"\n",
    "        raise NotImplementedError()\n",
    "\n",
    "    def get_dev_examples(self, data_dir):\n",
    "        \"\"\"Gets a collection of `InputExample`s for the dev set.\"\"\"\n",
    "        raise NotImplementedError()\n",
    "\n",
    "    def get_test_examples(self, data_dir):\n",
    "        \"\"\"Gets a collection of `InputExample`s for the test set.\"\"\"\n",
    "        raise NotImplementedError()\n",
    "\n",
    "    def get_labels(self):\n",
    "        \"\"\"Gets the list of labels for this data set.\"\"\"\n",
    "        raise NotImplementedError()\n",
    "\n",
    "    @classmethod\n",
    "    def _read_tsv(cls, input_file, quotechar=None):\n",
    "        \"\"\"Reads a tab separated value file.\"\"\"\n",
    "        with open(input_file, \"r\", encoding='utf-8') as f:\n",
    "            reader = csv.reader(f, delimiter=\"\\t\", quotechar=quotechar)\n",
    "            lines = []\n",
    "            for line in reader:\n",
    "                lines.append(line)\n",
    "            return lines\n",
    "\n",
    "\n",
    "class SentenceClassificationProcessor(DataProcessor):\n",
    "    \"\"\"Processor for Intent classification dataset.\"\"\"\n",
    "\n",
    "    def __init__(self, labels_array):\n",
    "        self.labels_array = labels_array\n",
    "\n",
    "    def get_train_examples(self, data_dir):\n",
    "        \"\"\"See base class.\"\"\"\n",
    "        logger.info(\"LOOKING AT {}\".format(os.path.join(data_dir, \"train.tsv\")))\n",
    "        return self._create_examples(\n",
    "            self._read_tsv(os.path.join(data_dir, \"train.tsv\")), \"train\")\n",
    "\n",
    "    def get_dev_examples(self, data_dir):\n",
    "        \"\"\"See base class.\"\"\"\n",
    "        return self._create_examples(\n",
    "            self._read_tsv(os.path.join(data_dir, \"test.tsv\")), \"dev\")\n",
    "\n",
    "    def get_test_examples(self, data_dir):\n",
    "        \"\"\"See base class.\"\"\"\n",
    "        return self._create_examples(\n",
    "            self._read_tsv(os.path.join(data_dir, \"test.tsv\")), \"test\")\n",
    "\n",
    "    def get_labels(self):\n",
    "        \"\"\"See base class.\"\"\"\n",
    "        return self.labels_array\n",
    "\n",
    "    def _create_examples(self, lines, set_type):\n",
    "        \"\"\"\n",
    "        Creates examples for the training and dev sets.\n",
    "        Add target sentence for AutoEncoder.\n",
    "        \"\"\"\n",
    "        examples, examples_target = [], []\n",
    "        for (i, line) in enumerate(lines):\n",
    "            if i == 0:\n",
    "                continue\n",
    "            guid = \"%s-%s\" % (set_type, i)\n",
    "            text_a = line[0]\n",
    "            label = line[1]\n",
    "            examples.append(\n",
    "                InputExample(guid=guid, text_a=text_a, text_b=None, label=label))\n",
    "            if len(line) == 4:\n",
    "                target_a = line[3]\n",
    "                examples_target.append(\n",
    "                    InputExample(guid=guid, text_a=target_a, text_b=None, label=label))\n",
    "        return examples, examples_target\n",
    "\n",
    "\n",
    "def convert_examples_to_features(examples, label_list, max_seq_length, tokenizer):\n",
    "    \"\"\"Loads a data file into a list of `InputBatch`s.\"\"\"\n",
    "\n",
    "    label_map = {label : i for i, label in enumerate(label_list)}\n",
    "\n",
    "    features = []\n",
    "    for (ex_index, example) in enumerate(examples):\n",
    "        tokens_a = tokenizer.tokenize(example.text_a)\n",
    "\n",
    "        tokens_b = None\n",
    "        if example.text_b:\n",
    "            tokens_b = tokenizer.tokenize(example.text_b)\n",
    "            # Modifies `tokens_a` and `tokens_b` in place so that the total\n",
    "            # length is less than the specified length.\n",
    "            # Account for [CLS], [SEP], [SEP] with \"- 3\"\n",
    "            _truncate_seq_pair(tokens_a, tokens_b, max_seq_length - 3)\n",
    "        else:\n",
    "            # Account for [CLS] and [SEP] with \"- 2\"\n",
    "            if len(tokens_a) > max_seq_length - 2:\n",
    "                tokens_a = tokens_a[:(max_seq_length - 2)]\n",
    "\n",
    "        # The convention in BERT is:\n",
    "        # (a) For sequence pairs:\n",
    "        #  tokens:   [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]\n",
    "        #  type_ids: 0   0  0    0    0     0       0 0    1  1  1  1   1 1\n",
    "        # (b) For single sequences:\n",
    "        #  tokens:   [CLS] the dog is hairy . [SEP]\n",
    "        #  type_ids: 0   0   0   0  0     0 0\n",
    "        #\n",
    "        # Where \"type_ids\" are used to indicate whether this is the first\n",
    "        # sequence or the second sequence. The embedding vectors for `type=0` and\n",
    "        # `type=1` were learned during pre-training and are added to the wordpiece\n",
    "        # embedding vector (and position vector). This is not *strictly* necessary\n",
    "        # since the [SEP] token unambigiously separates the sequences, but it makes\n",
    "        # it easier for the model to learn the concept of sequences.\n",
    "        #\n",
    "        # For classification tasks, the first vector (corresponding to [CLS]) is\n",
    "        # used as as the \"sentence vector\". Note that this only makes sense because\n",
    "        # the entire model is fine-tuned.\n",
    "        tokens = [\"[CLS]\"] + tokens_a + [\"[SEP]\"]\n",
    "        segment_ids = [0] * len(tokens)\n",
    "\n",
    "        if tokens_b:\n",
    "            tokens += tokens_b + [\"[SEP]\"]\n",
    "            segment_ids += [1] * (len(tokens_b) + 1)\n",
    "\n",
    "        input_ids = tokenizer.convert_tokens_to_ids(tokens)\n",
    "\n",
    "        # The mask has 1 for real tokens and 0 for padding tokens. Only real\n",
    "        # tokens are attended to.\n",
    "        input_mask = [1] * len(input_ids)\n",
    "\n",
    "        # Zero-pad up to the sequence length.\n",
    "        padding = [0] * (max_seq_length - len(input_ids))\n",
    "        input_ids += padding\n",
    "        input_mask += padding\n",
    "        segment_ids += padding\n",
    "\n",
    "        assert len(input_ids) == max_seq_length\n",
    "        assert len(input_mask) == max_seq_length\n",
    "        assert len(segment_ids) == max_seq_length\n",
    "\n",
    "        label_id = label_map[example.label]\n",
    "        if ex_index < 5:\n",
    "            logger.info(\"*** Example ***\")\n",
    "            logger.info(\"guid: %s\" % (example.guid))\n",
    "            logger.info(\"tokens: %s\" % \" \".join(\n",
    "                    [str(x) for x in tokens]))\n",
    "            logger.info(\"input_ids: %s\" % \" \".join([str(x) for x in input_ids]))\n",
    "            logger.info(\"input_mask: %s\" % \" \".join([str(x) for x in input_mask]))\n",
    "            logger.info(\n",
    "                    \"segment_ids: %s\" % \" \".join([str(x) for x in segment_ids]))\n",
    "            logger.info(\"label: %s (id = %d)\" % (example.label, label_id))\n",
    "\n",
    "        features.append(\n",
    "                InputFeatures(input_ids=input_ids,\n",
    "                              input_mask=input_mask,\n",
    "                              segment_ids=segment_ids,\n",
    "                              label_id=label_id))\n",
    "    return features\n",
    "\n",
    "\n",
    "def _truncate_seq_pair(tokens_a, tokens_b, max_length):\n",
    "    \"\"\"Truncates a sequence pair in place to the maximum length.\"\"\"\n",
    "\n",
    "    # This is a simple heuristic which will always truncate the longer sequence\n",
    "    # one token at a time. This makes more sense than truncating an equal percent\n",
    "    # of tokens from each, since if one sequence is very short then each token\n",
    "    # that's truncated likely contains more information than a longer sequence.\n",
    "    while True:\n",
    "        total_length = len(tokens_a) + len(tokens_b)\n",
    "        if total_length <= max_length:\n",
    "            break\n",
    "        if len(tokens_a) > len(tokens_b):\n",
    "            tokens_a.pop()\n",
    "        else:\n",
    "            tokens_b.pop()"
   ],
   "execution_count": 108,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "y_uDWLkPXj46",
    "colab_type": "text"
   },
   "source": [
    "### Other functions"
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "bhdQG7SrXmVM",
    "colab_type": "code",
    "colab": {},
    "pycharm": {
     "is_executing": false
    }
   },
   "source": [
    "def accuracy(out, labels):\n",
    "    outputs = np.argmax(out, axis=1)\n",
    "    return np.sum(outputs == labels)\n",
    "\n",
    "\n",
    "def warmup_linear(x, warmup=0.002):\n",
    "    if x < warmup:\n",
    "        return x/warmup\n",
    "    return 1.0 - x\n"
   ],
   "execution_count": 109,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BmdWU0vwXqnc",
    "colab_type": "text"
   },
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BCSj1AfaXyBv",
    "colab_type": "text"
   },
   "source": [
    "### 1. Evaluation functions"
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "HXBzf9n2XwTF",
    "colab_type": "code",
    "colab": {},
    "pycharm": {
     "is_executing": false
    }
   },
   "source": [
    "# ======== EARLY STOPPING ========\n",
    "def evaluate_original_bert_model(args, model, device, eval_dataloader, eval_examples_dataloader, tr_loss, nb_tr_steps, global_step, batches=None):\n",
    "    # Batches: None if eval and small number if eval is being used in train\n",
    "    eval_loss, eval_accuracy = 0, 0\n",
    "    nb_eval_steps, nb_eval_examples = 0, 0\n",
    "\n",
    "    predicted = []\n",
    "    target = []\n",
    "    predicted_mismatch, target_mismatch, text_mismatch = [], [], []\n",
    "    counter = 0\n",
    "    for (input_ids, input_mask, segment_ids, label_ids), ex in zip(eval_dataloader, eval_examples_dataloader):\n",
    "        input_ids = input_ids.to(device)\n",
    "        input_mask = input_mask.to(device)\n",
    "        segment_ids = segment_ids.to(device)\n",
    "        label_ids = label_ids.to(device)\n",
    "\n",
    "        with torch.no_grad():\n",
    "            _, tmp_eval_loss = model(input_ids, segment_ids, input_mask, label_ids)\n",
    "            _, logits = model(input_ids, segment_ids, input_mask)\n",
    "\n",
    "        logits = logits.detach().cpu().numpy()\n",
    "        label_ids = label_ids.to('cpu').numpy()\n",
    "        tmp_eval_accuracy = accuracy(logits, label_ids)\n",
    "\n",
    "        logits_max = np.argmax(logits, axis=1)\n",
    "        for pred, lab, e in zip(logits_max, label_ids, ex):\n",
    "            predicted.append(pred)\n",
    "            target.append(lab)\n",
    "            if pred != lab:\n",
    "                predicted_mismatch.append(pred)\n",
    "                target_mismatch.append(lab)\n",
    "                text_mismatch.append(e)\n",
    "\n",
    "        eval_loss += tmp_eval_loss.mean().item()\n",
    "        eval_accuracy += tmp_eval_accuracy\n",
    "\n",
    "        nb_eval_examples += input_ids.size(0)\n",
    "        nb_eval_steps += 1\n",
    "        if batches is not None and counter >= batches:\n",
    "            break\n",
    "        counter += 1\n",
    "\n",
    "    eval_loss = eval_loss / nb_eval_steps\n",
    "    eval_accuracy = eval_accuracy / nb_eval_examples\n",
    "    loss = tr_loss / nb_tr_steps if args.do_train else None\n",
    "    result = {'eval_loss': eval_loss,\n",
    "              'eval_accuracy': eval_accuracy,\n",
    "              'global_step': global_step,\n",
    "              'loss': loss}\n",
    "    return eval_loss, result, target, predicted, text_mismatch, target_mismatch, predicted_mismatch\n",
    "\n",
    "\n",
    "def evaluate_model(args, model, device, eval_dataloader, eval_examples_dataloader, tr_loss, nb_tr_steps, global_step,\n",
    "                   autoencoder, model_second_stack, batches=None):\n",
    "    # Batches: None if eval and small number if eval is being used in train\n",
    "    eval_loss, eval_accuracy = 0, 0\n",
    "    nb_eval_steps, nb_eval_examples = 0, 0\n",
    "\n",
    "    predicted = []\n",
    "    target = []\n",
    "    predicted_mismatch, target_mismatch, text_mismatch = [], [], []\n",
    "    counter = 0\n",
    "    for (input_ids, input_mask, segment_ids, label_ids), ex in zip(eval_dataloader, eval_examples_dataloader):\n",
    "        # tqdm(zip(eval_dataloader, eval_examples_dataloader), desc=\"Evaluating\"):\n",
    "        input_ids = input_ids.to(device)\n",
    "        input_mask = input_mask.to(device)\n",
    "        segment_ids = segment_ids.to(device)\n",
    "        label_ids = label_ids.to(device)\n",
    "\n",
    "        with torch.no_grad():\n",
    "            encoded_layers, tmp_eval_loss = model(input_ids, segment_ids, input_mask, label_ids)\n",
    "            # _, out_decoder = autoencoder(encoded_layers)\n",
    "            # encoded_layers = out_decoder\n",
    "            encoded_layers = encoded_layers.view(-1, 768)\n",
    "            _, out_decoder = autoencoder(encoded_layers)\n",
    "            encoded_layers = out_decoder.view(args.eval_batch_size, -1, 768)\n",
    "            _, tmp_eval_loss = model_second_stack(input_ids, segment_ids, input_mask, label_ids,\n",
    "                                                  words_embeddings=encoded_layers)\n",
    "            _, logits = model_second_stack(input_ids, segment_ids, input_mask, words_embeddings=encoded_layers)\n",
    "\n",
    "        logits = logits.detach().cpu().numpy()\n",
    "        label_ids = label_ids.to('cpu').numpy()\n",
    "        tmp_eval_accuracy = accuracy(logits, label_ids)\n",
    "\n",
    "        logits_max = np.argmax(logits, axis=1)\n",
    "        for pred, lab, e in zip(logits_max, label_ids, ex):\n",
    "            predicted.append(pred)\n",
    "            target.append(lab)\n",
    "            if pred != lab:\n",
    "                predicted_mismatch.append(pred)\n",
    "                target_mismatch.append(lab)\n",
    "                text_mismatch.append(e)\n",
    "\n",
    "        eval_loss += tmp_eval_loss.mean().item()\n",
    "        eval_accuracy += tmp_eval_accuracy\n",
    "\n",
    "        nb_eval_examples += input_ids.size(0)\n",
    "        nb_eval_steps += 1\n",
    "\n",
    "        if batches is not None and counter >= batches:\n",
    "            break\n",
    "        counter += 1\n",
    "\n",
    "    eval_loss = eval_loss / nb_eval_steps\n",
    "    eval_accuracy = eval_accuracy / nb_eval_examples\n",
    "    loss = tr_loss / nb_tr_steps if args.do_train or args.do_train_second_layer else None\n",
    "    result = {'eval_loss': eval_loss,\n",
    "              'eval_accuracy': eval_accuracy,\n",
    "              'global_step': global_step,\n",
    "              'loss': loss}\n",
    "    return eval_loss, result, target, predicted, text_mismatch, target_mismatch, predicted_mismatch\n",
    "# ================================"
   ],
   "execution_count": 110,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "1zVZ3ORnXtEB",
    "colab_type": "text"
   },
   "source": [
    "### 2. Training functions: BERT, denoising transformer layer"
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "Go1dPwdnX6HE",
    "colab_type": "code",
    "colab": {},
    "pycharm": {
     "is_executing": false
    }
   },
   "source": [
    "def first_bert_layer(train_examples, num_train_steps, args, tokenizer, label_list, num_labels, device, n_gpu,\n",
    "                     eval_dataloader, eval_examples_dataloader):\n",
    "\n",
    "    # Prepare model\n",
    "    model = BertForSequenceClassification.from_pretrained(args.bert_model,\n",
    "                                                          cache_dir=PYTORCH_PRETRAINED_BERT_CACHE / 'distributed_{}'.format(\n",
    "                                                              args.local_rank),\n",
    "                                                          num_labels=num_labels)\n",
    "    if args.fp16:\n",
    "        model.half()\n",
    "    model.to(device)\n",
    "    if args.local_rank != -1:\n",
    "        try:\n",
    "            from apex.parallel import DistributedDataParallel as DDP\n",
    "        except ImportError:\n",
    "            raise ImportError(\n",
    "                \"Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.\")\n",
    "\n",
    "        model = DDP(model)\n",
    "    elif n_gpu > 1:\n",
    "        model = torch.nn.DataParallel(model)\n",
    "\n",
    "    # Prepare optimizer\n",
    "    param_optimizer = list(model.named_parameters())\n",
    "    no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']\n",
    "    optimizer_grouped_parameters = [\n",
    "        {'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},\n",
    "        {'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}\n",
    "    ]\n",
    "    t_total = num_train_steps\n",
    "    if args.local_rank != -1:\n",
    "        t_total = t_total // torch.distributed.get_world_size()\n",
    "    if args.fp16:\n",
    "        try:\n",
    "            from apex.optimizers import FP16_Optimizer\n",
    "            from apex.optimizers import FusedAdam\n",
    "        except ImportError:\n",
    "            raise ImportError(\n",
    "                \"Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.\")\n",
    "\n",
    "        optimizer = FusedAdam(optimizer_grouped_parameters,\n",
    "                              lr=args.learning_rate,\n",
    "                              bias_correction=False,\n",
    "                              max_grad_norm=1.0)\n",
    "        if args.loss_scale == 0:\n",
    "            optimizer = FP16_Optimizer(optimizer, dynamic_loss_scale=True)\n",
    "        else:\n",
    "            optimizer = FP16_Optimizer(optimizer, static_loss_scale=args.loss_scale)\n",
    "\n",
    "    else:\n",
    "        optimizer = BertAdam(optimizer_grouped_parameters,\n",
    "                             lr=args.learning_rate,\n",
    "                             warmup=args.warmup_proportion,\n",
    "                             t_total=t_total)\n",
    "\n",
    "    global_step = 0\n",
    "    nb_tr_steps = 0\n",
    "    tr_loss = 0\n",
    "    # EARLY STOPPING\n",
    "    last_eval_loss = float(\"inf\")\n",
    "    if args.do_train:\n",
    "        train_features = convert_examples_to_features(\n",
    "            train_examples, label_list, args.max_seq_length, tokenizer)\n",
    "        logger.info(\"***** Running training *****\")\n",
    "        logger.info(\"  Num examples = %d\", len(train_examples))\n",
    "        logger.info(\"  Batch size = %d\", args.train_batch_size)\n",
    "        logger.info(\"  Num steps = %d\", num_train_steps)\n",
    "        all_input_ids = torch.tensor([f.input_ids for f in train_features], dtype=torch.long)\n",
    "        all_input_mask = torch.tensor([f.input_mask for f in train_features], dtype=torch.long)\n",
    "        all_segment_ids = torch.tensor([f.segment_ids for f in train_features], dtype=torch.long)\n",
    "        all_label_ids = torch.tensor([f.label_id for f in train_features], dtype=torch.long)\n",
    "        train_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)\n",
    "        if args.local_rank == -1:\n",
    "            train_sampler = RandomSampler(train_data)\n",
    "        else:\n",
    "            train_sampler = DistributedSampler(train_data)\n",
    "        train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=args.train_batch_size)\n",
    "\n",
    "        model.train()\n",
    "        for _ in trange(int(args.num_train_epochs), desc=\"Epoch\"):\n",
    "            tr_loss = 0\n",
    "            nb_tr_examples, nb_tr_steps = 0, 0\n",
    "            for step, batch in enumerate(tqdm(train_dataloader, desc=\"Iteration\")):\n",
    "                batch = tuple(t.to(device) for t in batch)\n",
    "                input_ids, input_mask, segment_ids, label_ids = batch\n",
    "                _, loss = model(input_ids, segment_ids, input_mask, label_ids)\n",
    "                if n_gpu > 1:\n",
    "                    loss = loss.mean()  # mean() to average on multi-gpu.\n",
    "                if args.gradient_accumulation_steps > 1:\n",
    "                    loss = loss / args.gradient_accumulation_steps\n",
    "\n",
    "                if args.fp16:\n",
    "                    optimizer.backward(loss)\n",
    "                else:\n",
    "                    loss.backward()\n",
    "\n",
    "                tr_loss += loss.item()\n",
    "                nb_tr_examples += input_ids.size(0)\n",
    "                nb_tr_steps += 1\n",
    "                if (step + 1) % args.gradient_accumulation_steps == 0:\n",
    "                    # modify learning rate with special warm up BERT uses\n",
    "                    lr_this_step = args.learning_rate * warmup_linear(global_step / t_total, args.warmup_proportion)\n",
    "                    for param_group in optimizer.param_groups:\n",
    "                        param_group['lr'] = lr_this_step\n",
    "                    optimizer.step()\n",
    "                    optimizer.zero_grad()\n",
    "                    global_step += 1\n",
    "\n",
    "                    # ======== EARLY STOPPING ========\n",
    "                    eval_loss, _, _, _, _, _, _ = evaluate_original_bert_model(args, model, device, eval_dataloader,\n",
    "                                                                               eval_examples_dataloader, tr_loss,\n",
    "                                                                               nb_tr_steps, global_step, batches=4)\n",
    "                    if args.save_best_model:\n",
    "                        # Save a trained model - EARLY STOPPING\n",
    "                        if eval_loss <= last_eval_loss:\n",
    "                            model_to_save = model.module if hasattr(model,\n",
    "                                                                    'module') else model  # Only save the model it-self\n",
    "                            output_model_file = os.path.join(args.output_dir_first_layer, \"pytorch_model.bin\")\n",
    "                            torch.save(model_to_save.state_dict(), output_model_file)\n",
    "                            last_eval_loss = eval_loss\n",
    "                    # ================================\n",
    "\n",
    "    # Save a trained model\n",
    "    if not args.save_best_model:\n",
    "        model_to_save = model.module if hasattr(model, 'module') else model  # Only save the model it-self\n",
    "        output_model_file = os.path.join(args.output_dir_first_layer, \"pytorch_model.bin\")\n",
    "        torch.save(model_to_save.state_dict(), output_model_file)\n",
    "\n",
    "    return model, global_step, nb_tr_steps, tr_loss\n",
    "\n",
    "def autoencoder_layer(train_examples, train_examples_target, model_first_layer, args, tokenizer, label_list, device, n_gpu):\n",
    "\n",
    "    # Prepare model\n",
    "    autoencoder = AutoEncoder()\n",
    "\n",
    "    if args.fp16:\n",
    "        autoencoder.half()\n",
    "    autoencoder.to(device)\n",
    "    if args.local_rank != -1:\n",
    "        try:\n",
    "            from apex.parallel import DistributedDataParallel as DDP\n",
    "        except ImportError:\n",
    "            raise ImportError(\n",
    "                \"Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.\")\n",
    "\n",
    "        autoencoder = DDP(autoencoder)\n",
    "    elif n_gpu > 1:\n",
    "        autoencoder = torch.nn.DataParallel(autoencoder)\n",
    "\n",
    "    # Prepare optimizer\n",
    "    optimizer = torch.optim.Adam(autoencoder.parameters(), lr=0.001, weight_decay=1e-5)\n",
    "    criterion = torch.nn.MSELoss()\n",
    "\n",
    "    if args.do_train_second_layer:\n",
    "        train_features = convert_examples_to_features(\n",
    "            train_examples, label_list, args.max_seq_length, tokenizer)\n",
    "        logger.info(\"***** Running training *****\")\n",
    "        logger.info(\"  Num examples = %d\", len(train_examples))\n",
    "        logger.info(\"  Batch size = %d\", args.train_batch_size)\n",
    "        all_input_ids = torch.tensor([f.input_ids for f in train_features], dtype=torch.long)\n",
    "        all_input_mask = torch.tensor([f.input_mask for f in train_features], dtype=torch.long)\n",
    "        all_segment_ids = torch.tensor([f.segment_ids for f in train_features], dtype=torch.long)\n",
    "        all_label_ids = torch.tensor([f.label_id for f in train_features], dtype=torch.long)\n",
    "\n",
    "        train_features = convert_examples_to_features(\n",
    "            train_examples_target, label_list, args.max_seq_length, tokenizer)\n",
    "        logger.info(\"***** Running training *****\")\n",
    "        logger.info(\"  Num examples = %d\", len(train_examples_target))\n",
    "        logger.info(\"  Batch size = %d\", args.train_batch_size)\n",
    "        all_input_ids_target = torch.tensor([f.input_ids for f in train_features], dtype=torch.long)\n",
    "        all_input_mask_target = torch.tensor([f.input_mask for f in train_features], dtype=torch.long)\n",
    "        all_segment_ids_target = torch.tensor([f.segment_ids for f in train_features], dtype=torch.long)\n",
    "        all_label_ids_target = torch.tensor([f.label_id for f in train_features], dtype=torch.long)\n",
    "\n",
    "        train_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids,\n",
    "                                   all_input_ids_target, all_input_mask_target, all_segment_ids_target, all_label_ids_target)\n",
    "        if args.local_rank == -1:\n",
    "            train_sampler = RandomSampler(train_data)\n",
    "        else:\n",
    "            train_sampler = DistributedSampler(train_data)\n",
    "        train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=args.train_batch_size)\n",
    "\n",
    "        autoencoder.train()\n",
    "        for _ in trange(int(args.num_train_epochs_autoencoder), desc=\"Epoch\"):\n",
    "            tr_loss = 0\n",
    "            nb_tr_examples, nb_tr_steps = 0, 0\n",
    "            for step, batch in enumerate(tqdm(train_dataloader, desc=\"Iteration\")):\n",
    "                batch = tuple(t.to(device) for t in batch)\n",
    "                input_ids, input_mask, segment_ids, label_ids, input_ids_target, input_mask_target, segment_ids_target, label_ids_target = batch\n",
    "                # BERT model on INcomplete data\n",
    "                encoded_layers, _ = model_first_layer(input_ids, segment_ids, input_mask, label_ids)\n",
    "                encoded_layers = encoded_layers.view(-1, 768)\n",
    "                # BERT model on complete data\n",
    "                encoded_layers_complete, _ = model_first_layer(input_ids_target, segment_ids_target, input_mask_target, label_ids_target)\n",
    "                # Denoising Autoencoder trained on noisy input and clean output_cnn_hter_cr\n",
    "                encoded_layers_complete = encoded_layers_complete.view(-1, 768)\n",
    "                _, out_dec = autoencoder(encoded_layers)\n",
    "                loss = criterion(out_dec, encoded_layers_complete)\n",
    "                if n_gpu > 1:\n",
    "                    loss = loss.mean()  # mean() to average on multi-gpu.\n",
    "                if args.gradient_accumulation_steps > 1:\n",
    "                    loss = loss / args.gradient_accumulation_steps\n",
    "\n",
    "                optimizer.zero_grad()\n",
    "                loss.backward()\n",
    "                optimizer.step()\n",
    "\n",
    "                tr_loss += loss.item()\n",
    "                nb_tr_examples += input_ids.size(0)\n",
    "                nb_tr_steps += 1\n",
    "\n",
    "    return autoencoder\n",
    "\n",
    "\n",
    "def second_bert_layer(train_examples, num_train_steps, model_first_layer, autoencoder, args, tokenizer, label_list,\n",
    "                      num_labels, device, n_gpu, eval_dataloader, eval_examples_dataloader):\n",
    "\n",
    "    # Prepare model\n",
    "    model = BertForSequenceClassification.from_pretrained(args.bert_model,\n",
    "                                                          cache_dir=PYTORCH_PRETRAINED_BERT_CACHE / 'distributed_{}'.format(\n",
    "                                                              args.local_rank),\n",
    "                                                          num_labels=num_labels)\n",
    "    if args.fp16:\n",
    "        model.half()\n",
    "    model.to(device)\n",
    "    if args.local_rank != -1:\n",
    "        try:\n",
    "            from apex.parallel import DistributedDataParallel as DDP\n",
    "        except ImportError:\n",
    "            raise ImportError(\n",
    "                \"Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.\")\n",
    "\n",
    "        model = DDP(model)\n",
    "    elif n_gpu > 1:\n",
    "        model = torch.nn.DataParallel(model)\n",
    "\n",
    "    # Prepare optimizer\n",
    "    param_optimizer = list(model.named_parameters())\n",
    "    no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']\n",
    "    optimizer_grouped_parameters = [\n",
    "        {'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},\n",
    "        {'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}\n",
    "    ]\n",
    "    t_total = num_train_steps\n",
    "    if args.local_rank != -1:\n",
    "        t_total = t_total // torch.distributed.get_world_size()\n",
    "    if args.fp16:\n",
    "        try:\n",
    "            from apex.optimizers import FP16_Optimizer\n",
    "            from apex.optimizers import FusedAdam\n",
    "        except ImportError:\n",
    "            raise ImportError(\n",
    "                \"Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.\")\n",
    "\n",
    "        optimizer = FusedAdam(optimizer_grouped_parameters,\n",
    "                              lr=args.learning_rate,\n",
    "                              bias_correction=False,\n",
    "                              max_grad_norm=1.0)\n",
    "        if args.loss_scale == 0:\n",
    "            optimizer = FP16_Optimizer(optimizer, dynamic_loss_scale=True)\n",
    "        else:\n",
    "            optimizer = FP16_Optimizer(optimizer, static_loss_scale=args.loss_scale)\n",
    "\n",
    "    else:\n",
    "        optimizer = BertAdam(optimizer_grouped_parameters,\n",
    "                             lr=args.learning_rate,\n",
    "                             warmup=args.warmup_proportion,\n",
    "                             t_total=t_total)\n",
    "\n",
    "    global_step = 0\n",
    "    nb_tr_steps = 0\n",
    "    tr_loss = 0\n",
    "    # EARLY STOPPING\n",
    "    last_eval_loss = float(\"inf\")\n",
    "    if args.do_train_second_layer:\n",
    "        train_features = convert_examples_to_features(\n",
    "            train_examples, label_list, args.max_seq_length, tokenizer)\n",
    "        logger.info(\"***** Running training *****\")\n",
    "        logger.info(\"  Num examples = %d\", len(train_examples))\n",
    "        logger.info(\"  Batch size = %d\", args.train_batch_size)\n",
    "        logger.info(\"  Num steps = %d\", num_train_steps)\n",
    "        all_input_ids = torch.tensor([f.input_ids for f in train_features], dtype=torch.long)\n",
    "        all_input_mask = torch.tensor([f.input_mask for f in train_features], dtype=torch.long)\n",
    "        all_segment_ids = torch.tensor([f.segment_ids for f in train_features], dtype=torch.long)\n",
    "        all_label_ids = torch.tensor([f.label_id for f in train_features], dtype=torch.long)\n",
    "        train_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)\n",
    "        if args.local_rank == -1:\n",
    "            train_sampler = RandomSampler(train_data)\n",
    "        else:\n",
    "            train_sampler = DistributedSampler(train_data)\n",
    "        train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=args.train_batch_size)\n",
    "\n",
    "        model.train()\n",
    "        for _ in trange(int(args.num_train_epochs), desc=\"Epoch\"):\n",
    "            tr_loss = 0\n",
    "            nb_tr_examples, nb_tr_steps = 0, 0\n",
    "            for step, batch in enumerate(tqdm(train_dataloader, desc=\"Iteration\")):\n",
    "                batch = tuple(t.to(device) for t in batch)\n",
    "                input_ids, input_mask, segment_ids, label_ids = batch\n",
    "                encoded_layers, loss = model_first_layer(input_ids, segment_ids, input_mask, label_ids)\n",
    "                # AutoEncoder\n",
    "                encoded_layers = encoded_layers.view(-1, 768)\n",
    "                _, out_decoder = autoencoder(encoded_layers)\n",
    "                encoded_layers = out_decoder.view(args.train_batch_size, -1, 768)\n",
    "                _, loss = model(input_ids, segment_ids, input_mask, label_ids, words_embeddings=encoded_layers)\n",
    "                if n_gpu > 1:\n",
    "                    loss = loss.mean()  # mean() to average on multi-gpu.\n",
    "                if args.gradient_accumulation_steps > 1:\n",
    "                    loss = loss / args.gradient_accumulation_steps\n",
    "\n",
    "                if args.fp16:\n",
    "                    optimizer.backward(loss)\n",
    "                else:\n",
    "                    loss.backward()\n",
    "\n",
    "                tr_loss += loss.item()\n",
    "                nb_tr_examples += input_ids.size(0)\n",
    "                nb_tr_steps += 1\n",
    "                if (step + 1) % args.gradient_accumulation_steps == 0:\n",
    "                    # modify learning rate with special warm up BERT uses\n",
    "                    lr_this_step = args.learning_rate * warmup_linear(global_step / t_total, args.warmup_proportion)\n",
    "                    for param_group in optimizer.param_groups:\n",
    "                        param_group['lr'] = lr_this_step\n",
    "                    optimizer.step()\n",
    "                    optimizer.zero_grad()\n",
    "                    global_step += 1\n",
    "\n",
    "                    # ======== EARLY STOPPING ========\n",
    "                    eval_loss, _, _, _, _, _, _ = evaluate_model(args, model_first_layer, device, eval_dataloader,\n",
    "                                                                 eval_examples_dataloader, tr_loss, nb_tr_steps,\n",
    "                                                                 global_step, autoencoder, model, batches=4)\n",
    "                    if args.save_best_model:\n",
    "                        # Save a trained model - EARLY STOPPING\n",
    "                        if eval_loss <= last_eval_loss:\n",
    "                            model_to_save = model.module if hasattr(model,\n",
    "                                                                    'module') else model  # Only save the model it-self\n",
    "                            output_model_file = os.path.join(args.output_dir, \"pytorch_model.bin\")\n",
    "                            torch.save(model_to_save.state_dict(), output_model_file)\n",
    "                            last_eval_loss = eval_loss\n",
    "                    # ================================\n",
    "\n",
    "    # Save a trained model\n",
    "    model_to_save = model.module if hasattr(model, 'module') else model  # Only save the model it-self\n",
    "    output_model_file = os.path.join(args.output_dir, \"pytorch_model.bin\")\n",
    "    torch.save(model_to_save.state_dict(), output_model_file)\n",
    "\n",
    "    return model, global_step, nb_tr_steps, tr_loss"
   ],
   "execution_count": 111,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KqqiGSkyYFjx",
    "colab_type": "text"
   },
   "source": [
    "### 3. Main function\n",
    "Example from gtts-witai (TTS-STT) Chatbot dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Vi--ZGEYly6L",
    "colab_type": "text"
   },
   "source": [
    "#### 3.1 - Parser"
   ]
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "B3dOUx3Zl1IS",
    "colab_type": "code",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 469
    },
    "outputId": "31407277-9ced-4285-9f33-bcd0a2074ae7",
    "pycharm": {
     "is_executing": false
    }
   },
   "source": [
    "    # ipykernel_launcher.py: the following arguments are required: --data_dir, --bert_model, --task_name, --output_dir, --output_dir_first_layer\n",
    "    DATA_DIR = \"./data/intent_data/stterror_data/chatbot/gtts_witai/\"\n",
    "    BERT_MODEL = \"bert-base-uncased\"\n",
    "    TASK_NAME = \"chatbot_intent\"\n",
    "    OUTPUT_DIR = \"./results/results_stacked_debert_STTerror\"\n",
    "    SEED, BS_TRAIN, BS_EVAL, EPOCH, EPOCH_AE = [1, 8, 1, 3, 100]\n",
    "    EPOCH_1st = EPOCH\n",
    "    OUTPUT_DIR_1st_LAYER = OUTPUT_DIR + \"/chatbot/gtts_witai/bs\" + str(BS_TRAIN) +\"_epae\" + str(EPOCH_AE) + \"/chatbot_ep\" + str(EPOCH_1st) + \"_bs\" + str(BS_TRAIN) + \"_seed\" + str(SEED) + \"_first_layer_epae\" + str(EPOCH_AE) + \"/\"\n",
    "    OUTPUT_DIR_2nd_LAYER = OUTPUT_DIR + \"/chatbot/gtts_witai/bs\" + str(BS_TRAIN) +\"_epae\" + str(EPOCH_AE) + \"/chatbot_ep\" + str(EPOCH_1st) + \"_bs\" + str(BS_TRAIN) + \"_seed\" + str(SEED) + \"_second_layer_epae\" + str(EPOCH_AE) + \"/\"\n",
    "    \n",
    "    import easydict\n",
    "    args = easydict.EasyDict({\n",
    "        ## Required parameters\n",
    "        \"data_dir\": DATA_DIR, # The input data dir. Should contain the .tsv files (or other data files) for the task.\n",
    "        \"bert_model\": BERT_MODEL, # Bert pre-trained model selected in the list: bert-base-uncased, bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, bert-base-multilingual-cased, bert-base-chinese.\n",
    "        \"task_name\": TASK_NAME, # The name of the task to train\n",
    "        \"output_dir\": OUTPUT_DIR_2nd_LAYER, # The output_cnn_hter_cr directory where the model predictions and checkpoints will be written.\n",
    "        \"output_dir_first_layer\": OUTPUT_DIR_1st_LAYER, # The output_cnn_hter_cr directory where the model predictions and checkpoints will be written.\n",
    "        \"max_noise\": 0.1,\n",
    "        ## Other parameters\n",
    "        \"max_seq_length\": 128, # The maximum total input sequence length after WordPiece tokenization. Sequences longer than this will be truncated, and sequences shorter than this will be padded.\n",
    "        \"do_train\": True, # Whether to train first BERT layer\n",
    "        \"do_train_second_layer\": True, # Whether to train 2nd BERT (transformers)\n",
    "        \"do_train_autoencoder\": True, # Whether to train the denoinsing layer\n",
    "        \"do_eval\": True, # Whether to run eval on the dev set\n",
    "        \"do_lower_case\": True, # Set this flag if you are using an uncased model\n",
    "        \"train_batch_size\": BS_TRAIN, # Total batch size for training\n",
    "        \"eval_batch_size\": BS_EVAL, # Total batch size for eval\n",
    "        \"learning_rate\": 5e-5, # The initial learning rate for Adam\n",
    "        \"num_train_epochs\": EPOCH, # Total number of training epochs to perform\n",
    "        \"num_train_epochs_autoencoder\": EPOCH_AE, # Total number of training epochs to perform\n",
    "        \"warmup_proportion\": 0.1, # Proportion of training to perform linear learning rate warmup for. E.g., 0.1 = 10%% of training\n",
    "        \"no_cuda\": True, # Whether not to use CUDA when available\n",
    "        \"save_best_model\": True, # Whether to do early stopping or not\n",
    "        \"local_rank\": -1, # local_rank for distributed training on gpus\n",
    "        \"seed\": SEED, # random seed for initialization\n",
    "        \"gradient_accumulation_steps\": 1, # Number of updates steps to accumulate before performing a backward/update pass\n",
    "        \"fp16\": True, # Whether to use 16-bit float precision instead of 32-bit\n",
    "        \"loss_scale\": 0, # Loss scaling to improve fp16 numeric stability. Only used when fp16 set to True. 0 (default value): dynamic loss scaling. Positive power of 2: static loss scaling value.\n",
    "        })"
   ],
   "execution_count": 112,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "LGKlh0h-l5MP",
    "colab_type": "text"
   },
   "source": [
    "### 3.2 - Variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "outputs": [
    {
     "name": "stderr",
     "text": [
      "08/21/2020 11:58:32 - INFO - __main__ -   device: cpu n_gpu: 1, distributed training: False, 16-bits training: True\n",
      "08/21/2020 11:58:32 - INFO - models.stacked_debert_dae.tokenization -   loading vocabulary file https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-uncased-vocab.txt from cache at /home/ceslea/.pytorch_pretrained_bert/26bc1ad6c0ac742e9b52263248f6d0f00068293b33709fae12320c0e35ccfbbb.542ce4285a40d23a559526243235df47c5f75c197f04f37d1a0c124c32c9a084\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   LOOKING AT ./data/intent_data/stterror_data/chatbot/gtts_witai/train.tsv\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   *** Example ***\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   guid: dev-1\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   tokens: [CLS] when is the next bus from ga ##rch ##ing for ##sch ##ung ##sz ##ent ##rum . [SEP]\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   input_ids: 101 2043 2003 1996 2279 3902 2013 11721 11140 2075 2005 11624 5575 17112 4765 6824 1012 102 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 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 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 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 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   input_mask: 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 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 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 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   segment_ids: 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 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 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 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 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 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   label: 0 (id = 0)\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   *** Example ***\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   guid: dev-2\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   tokens: [CLS] how do i get from a easy to h ##bs . [SEP]\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   input_ids: 101 2129 2079 1045 2131 2013 1037 3733 2000 1044 5910 1012 102 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 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 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 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   input_mask: 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 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 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 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   segment_ids: 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 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 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 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 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 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   label: 1 (id = 1)\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   *** Example ***\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   guid: dev-3\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   tokens: [CLS] how i can get from marie ##np ##lat ##z to ga ##rch ##ing . [SEP]\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   input_ids: 101 2129 1045 2064 2131 2013 5032 16275 20051 2480 2000 11721 11140 2075 1012 102 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 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 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 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 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   input_mask: 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 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 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 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   segment_ids: 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 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 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 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 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 0 0 0\n",
      "08/21/2020 11:58:32 - INFO - __main__ -   label: 1 (id = 1)\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   *** Example ***\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   guid: dev-4\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   tokens: [CLS] how can i get from ga ##rch ##ing to ga ##rch ##ing for ##sch ##ung ##sz ##ent ##rum . [SEP]\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   input_ids: 101 2129 2064 1045 2131 2013 11721 11140 2075 2000 11721 11140 2075 2005 11624 5575 17112 4765 6824 1012 102 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 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 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 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 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   input_mask: 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 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 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 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   segment_ids: 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 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 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 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 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 0 0 0\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   label: 1 (id = 1)\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   *** Example ***\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   guid: dev-5\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   tokens: [CLS] i wanna go un ##pl ##ug . [SEP]\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   input_ids: 101 1045 10587 2175 4895 24759 15916 1012 102 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 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 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 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   input_mask: 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 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 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 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 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   segment_ids: 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 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 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 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 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 0 0 0\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   label: 1 (id = 1)\n",
      "08/21/2020 11:58:33 - INFO - __main__ -   ***** Running evaluation *****\n",
      "08/21/2020 11:58:33 - INFO - __main__ -     Num examples = 106\n",
      "08/21/2020 11:58:33 - INFO - __main__ -     Batch size = 1\n"
     ],
     "output_type": "stream"
    }
   ],
   "source": [
    "    processors = {\n",
    "        \"chatbot_intent\": SentenceClassificationProcessor,\n",
    "        \"sentiment140_sentiment\": SentenceClassificationProcessor,\n",
    "    }\n",
    "\n",
    "    num_labels_task = {\n",
    "        \"chatbot_intent\": 2,\n",
    "        \"sentiment140_sentiment\": 2,\n",
    "    }\n",
    "\n",
    "    labels_array = {\n",
    "        \"chatbot_intent\": [\"0\", \"1\"],\n",
    "        \"sentiment140_sentiment\": [\"0\", \"1\"],\n",
    "    }\n",
    "\n",
    "    labels_array_int = {\n",
    "        \"chatbot_intent\": [0, 1],\n",
    "        \"sentiment140_sentiment\": [0, 1],\n",
    "    }\n",
    "\n",
    "    if args.local_rank == -1 or args.no_cuda:\n",
    "        device = torch.device(\"cuda\" if torch.cuda.is_available() and not args.no_cuda else \"cpu\")\n",
    "        n_gpu = torch.cuda.device_count()\n",
    "    else:\n",
    "        torch.cuda.set_device(args.local_rank)\n",
    "        device = torch.device(\"cuda\", args.local_rank)\n",
    "        n_gpu = 1\n",
    "        # Initializes the distributed backend which will take care of sychronizing nodes/GPUs\n",
    "        torch.distributed.init_process_group(backend='nccl')\n",
    "    logger.info(\"device: {} n_gpu: {}, distributed training: {}, 16-bits training: {}\".format(\n",
    "        device, n_gpu, bool(args.local_rank != -1), args.fp16))\n",
    "\n",
    "    if args.gradient_accumulation_steps < 1:\n",
    "        raise ValueError(\"Invalid gradient_accumulation_steps parameter: {}, should be >= 1\".format(\n",
    "                            args.gradient_accumulation_steps))\n",
    "\n",
    "    args.train_batch_size = int(args.train_batch_size / args.gradient_accumulation_steps)\n",
    "\n",
    "    random.seed(args.seed)\n",
    "    np.random.seed(args.seed)\n",
    "    torch.manual_seed(args.seed)\n",
    "    if n_gpu > 0:\n",
    "        torch.cuda.manual_seed_all(args.seed)\n",
    "\n",
    "    if not args.do_train and not args.do_eval and not args.do_train_second_layer:\n",
    "        raise ValueError(\"At least one of `do_train` or `do_eval` must be True.\")\n",
    "\n",
    "    if os.path.exists(args.output_dir) and os.listdir(args.output_dir) and args.do_train_second_layer:\n",
    "        raise ValueError(\"Output directory ({}) already exists and is not empty.\".format(args.output_dir))\n",
    "    os.makedirs(args.output_dir, exist_ok=True)\n",
    "\n",
    "    if os.path.exists(args.output_dir_first_layer) and os.listdir(args.output_dir_first_layer) and args.do_train:\n",
    "        raise ValueError(\"Output directory ({}) already exists and is not empty.\".format(args.output_dir_first_layer))\n",
    "    os.makedirs(args.output_dir_first_layer, exist_ok=True)\n",
    "\n",
    "    task_name = args.task_name.lower()\n",
    "\n",
    "    if task_name not in processors:\n",
    "        raise ValueError(\"Task not found: %s\" % (task_name))\n",
    "\n",
    "    processor = processors[task_name](labels_array[task_name])\n",
    "    num_labels = num_labels_task[task_name]\n",
    "    label_list = processor.get_labels()\n",
    "\n",
    "    tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)\n",
    "\n",
    "    train_examples = None\n",
    "    train_examples_target = None\n",
    "    num_train_steps = None\n",
    "    if args.do_train or args.do_train_second_layer or args.do_train_autoencoder:\n",
    "        train_examples, train_examples_target = processor.get_train_examples(args.data_dir)\n",
    "        num_train_steps = int(\n",
    "            len(train_examples) / args.train_batch_size / args.gradient_accumulation_steps * args.num_train_epochs)\n",
    "\n",
    "    # ======== EARLY STOPPING ========\n",
    "    # Load evaluation dataset for use in early stopping\n",
    "    # Eval dataloader\n",
    "    eval_examples, _ = processor.get_dev_examples(args.data_dir)\n",
    "    eval_features = convert_examples_to_features(\n",
    "        eval_examples, label_list, args.max_seq_length, tokenizer)\n",
    "    logger.info(\"***** Running evaluation *****\")\n",
    "    logger.info(\"  Num examples = %d\", len(eval_examples))\n",
    "    logger.info(\"  Batch size = %d\", args.eval_batch_size)\n",
    "    all_input_ids = torch.tensor([f.input_ids for f in eval_features], dtype=torch.long)\n",
    "    all_input_mask = torch.tensor([f.input_mask for f in eval_features], dtype=torch.long)\n",
    "    all_segment_ids = torch.tensor([f.segment_ids for f in eval_features], dtype=torch.long)\n",
    "    all_label_ids = torch.tensor([f.label_id for f in eval_features], dtype=torch.long)\n",
    "    eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)\n",
    "    # Run prediction for full data\n",
    "    eval_sampler = SequentialSampler(eval_data)\n",
    "    eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)\n",
    "\n",
    "    # Get texts\n",
    "    text = []\n",
    "    for e in eval_examples:\n",
    "        text.append(e.text_a)\n",
    "    eval_examples_dataloader = DataLoader(text, shuffle=False, batch_size=args.eval_batch_size)\n",
    "    # ================================"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    }
   }
  },
  {
   "cell_type": "code",
   "metadata": {
    "id": "sJ2SVp6uYUm0",
    "colab_type": "code",
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 469
    },
    "outputId": "8d0298e6-505b-4fc2-aeb4-6144ac37d73e",
    "pycharm": {
     "is_executing": false
    }
   },
   "source": [
    "    # Train 1st BERT\n",
    "    global_step, nb_tr_steps, tr_loss = 0, 0, 0\n",
    "    if args.do_train:\n",
    "        model, global_step, nb_tr_steps, tr_loss = first_bert_layer(train_examples, num_train_steps, args, tokenizer,\n",
    "                                                                    label_list, num_labels, device, n_gpu,\n",
    "                                                                    eval_dataloader, eval_examples_dataloader)\n",
    "\n",
    "    # Load a trained model that you have fine-tuned\n",
    "    output_model_file = os.path.join(args.output_dir_first_layer, \"pytorch_model.bin\")\n",
    "    model_state_dict = torch.load(output_model_file)\n",
    "    model_first_layer = BertForSequenceClassification.from_pretrained(args.bert_model, state_dict=model_state_dict,\n",
    "                                                                      num_labels=num_labels)\n",
    "    model_first_layer.to(device)"
   ],
   "execution_count": 114,
   "outputs": [
    {
     "name": "stderr",
     "text": [
      "08/21/2020 11:58:33 - INFO - models.stacked_debert_dae.modeling -   loading archive file https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-uncased.tar.gz from cache at /home/ceslea/.pytorch_pretrained_bert/distributed_-1/9c41111e2de84547a463fd39217199738d1e3deb72d4fec4399e6e241983c6f0.ae3cef932725ca7a30cdcb93fc6e09150a55e2a130ec7af63975a16c153ae2ba\n",
      "08/21/2020 11:58:33 - INFO - models.stacked_debert_dae.modeling -   extracting archive file /home/ceslea/.pytorch_pretrained_bert/distributed_-1/9c41111e2de84547a463fd39217199738d1e3deb72d4fec4399e6e241983c6f0.ae3cef932725ca7a30cdcb93fc6e09150a55e2a130ec7af63975a16c153ae2ba to temp dir /tmp/tmp6hbb6xjm\n",
      "08/21/2020 11:58:36 - INFO - models.stacked_debert_dae.modeling -   Model config {\n  \"attention_probs_dropout_prob\": 0.1,\n  \"hidden_act\": \"gelu\",\n  \"hidden_dropout_prob\": 0.1,\n  \"hidden_size\": 768,\n  \"initializer_range\": 0.02,\n  \"intermediate_size\": 3072,\n  \"max_position_embeddings\": 512,\n  \"num_attention_heads\": 12,\n  \"num_hidden_layers\": 12,\n  \"type_vocab_size\": 2,\n  \"vocab_size\": 30522\n}\n\n",
      "08/21/2020 11:58:39 - INFO - models.stacked_debert_dae.modeling -   Weights of BertForSequenceClassification not initialized from pretrained model: ['classifier.weight', 'classifier.bias']\n",
      "08/21/2020 11:58:39 - INFO - models.stacked_debert_dae.modeling -   Weights from pretrained model not used in BertForSequenceClassification: ['cls.predictions.bias', 'cls.predictions.transform.dense.weight', 'cls.predictions.transform.dense.bias', 'cls.predictions.decoder.weight', 'cls.seq_relationship.weight', 'cls.seq_relationship.bias', 'cls.predictions.transform.LayerNorm.weight', 'cls.predictions.transform.LayerNorm.bias']\n"
     ],
     "output_type": "stream"
    },
    {
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mModuleNotFoundError\u001b[0m                       Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-111-a2095e9ef69c>\u001b[0m in \u001b[0;36mfirst_bert_layer\u001b[0;34m(train_examples, num_train_steps, args, tokenizer, label_list, num_labels, device, n_gpu, eval_dataloader, eval_examples_dataloader)\u001b[0m\n\u001b[1;32m     34\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 35\u001b[0;31m             \u001b[0;32mfrom\u001b[0m \u001b[0mapex\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moptimizers\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mFP16_Optimizer\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     36\u001b[0m             \u001b[0;32mfrom\u001b[0m \u001b[0mapex\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moptimizers\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mFusedAdam\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mModuleNotFoundError\u001b[0m: No module named 'apex'",
      "\nDuring handling of the above exception, another exception occurred:\n",
      "\u001b[0;31mImportError\u001b[0m                               Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-114-8ea78b831292>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      4\u001b[0m     model, global_step, nb_tr_steps, tr_loss = first_bert_layer(train_examples, num_train_steps, args, tokenizer,\n\u001b[1;32m      5\u001b[0m                                                                 \u001b[0mlabel_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnum_labels\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdevice\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_gpu\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 6\u001b[0;31m                                                                 eval_dataloader, eval_examples_dataloader)\n\u001b[0m\u001b[1;32m      7\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0;31m# Load a trained model that you have fine-tuned\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-111-a2095e9ef69c>\u001b[0m in \u001b[0;36mfirst_bert_layer\u001b[0;34m(train_examples, num_train_steps, args, tokenizer, label_list, num_labels, device, n_gpu, eval_dataloader, eval_examples_dataloader)\u001b[0m\n\u001b[1;32m     37\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mImportError\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     38\u001b[0m             raise ImportError(\n\u001b[0;32m---> 39\u001b[0;31m                 \"Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.\")\n\u001b[0m\u001b[1;32m     40\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     41\u001b[0m         optimizer = FusedAdam(optimizer_grouped_parameters,\n",
      "\u001b[0;31mImportError\u001b[0m: Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training."
     ],
     "ename": "ImportError",
     "evalue": "Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.",
     "output_type": "error"
    }
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "    # Train autoencoder\n",
    "    # TODO: Early Stopping for AutoEncoder\n",
    "    if args.do_train_autoencoder:\n",
    "        autoencoder = autoencoder_layer(train_examples, train_examples_target, model_first_layer, args, tokenizer, label_list, device, n_gpu)\n",
    "\n",
    "        # Save a trained model\n",
    "        model_to_save = autoencoder.module if hasattr(autoencoder, 'module') else autoencoder  # Only save the model it-self\n",
    "        output_model_file = os.path.join(args.output_dir, \"autoencoder_pytorch_model.bin\")\n",
    "        torch.save(model_to_save.state_dict(), output_model_file)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "    # Train 2nd BERT layer\n",
    "    if args.do_train_second_layer:\n",
    "        # Load trained autoencoder\n",
    "        output_model_file_ae = os.path.join(args.output_dir, \"autoencoder_pytorch_model.bin\")\n",
    "        model_state_dict_ae = torch.load(output_model_file_ae)\n",
    "        autoencoder = AutoEncoder()\n",
    "        autoencoder.load_state_dict(model_state_dict_ae)\n",
    "        autoencoder.to(device)\n",
    "\n",
    "        model_second_stack, global_step, nb_tr_steps, tr_loss = second_bert_layer(train_examples, num_train_steps,\n",
    "                                                                                  model_first_layer, autoencoder, args,\n",
    "                                                                                  tokenizer, label_list, num_labels,\n",
    "                                                                                  device, n_gpu, eval_dataloader,\n",
    "                                                                                  eval_examples_dataloader)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "    # ======== LOAD TEST DATA ========\n",
    "    # Test dataloader\n",
    "    test_examples, _ = processor.get_test_examples(args.data_dir)\n",
    "    test_features = convert_examples_to_features(test_examples, label_list, args.max_seq_length, tokenizer)\n",
    "    logger.info(\"***** Loading test data *****\")\n",
    "    logger.info(\"  Num examples = %d\", len(test_examples))\n",
    "    logger.info(\"  Batch size = %d\", args.eval_batch_size)\n",
    "    all_input_ids = torch.tensor([f.input_ids for f in test_features], dtype=torch.long)\n",
    "    all_input_mask = torch.tensor([f.input_mask for f in test_features], dtype=torch.long)\n",
    "    all_segment_ids = torch.tensor([f.segment_ids for f in test_features], dtype=torch.long)\n",
    "    all_label_ids = torch.tensor([f.label_id for f in test_features], dtype=torch.long)\n",
    "    test_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)\n",
    "    # Run prediction for full data\n",
    "    test_sampler = SequentialSampler(test_data)\n",
    "    test_dataloader = DataLoader(test_data, sampler=test_sampler, batch_size=args.eval_batch_size)\n",
    "\n",
    "    # Get texts\n",
    "    text = []\n",
    "    for e in test_examples:\n",
    "        text.append(e.text_a)\n",
    "    test_examples_dataloader = DataLoader(text, shuffle=False, batch_size=args.eval_batch_size)\n",
    "\n",
    "    # ======== TEST with TEST DATA ========\n",
    "    # Load a trained model that you have fine-tuned\n",
    "    model_state_dict = torch.load(os.path.join(args.output_dir_first_layer, \"pytorch_model.bin\"))\n",
    "    model = BertForSequenceClassification.from_pretrained(args.bert_model, state_dict=model_state_dict, num_labels=num_labels)\n",
    "    model.to(device)\n",
    "\n",
    "    # Load trained autoencoder\n",
    "    output_model_file_ae = os.path.join(args.output_dir, \"autoencoder_pytorch_model.bin\")\n",
    "    model_state_dict_ae = torch.load(output_model_file_ae)\n",
    "    autoencoder = AutoEncoder()\n",
    "    autoencoder.load_state_dict(model_state_dict_ae)\n",
    "    autoencoder.to(device)\n",
    "\n",
    "    # Load second layer\n",
    "    model_state_dict_second_stack = torch.load(os.path.join(args.output_dir, \"pytorch_model.bin\"))\n",
    "    model_second_stack = BertForSequenceClassification.from_pretrained(args.bert_model,\n",
    "                                                                       state_dict=model_state_dict_second_stack,\n",
    "                                                                       num_labels=num_labels)\n",
    "    model_second_stack.to(device)\n",
    "\n",
    "    if args.do_eval and (args.local_rank == -1 or torch.distributed.get_rank() == 0):\n",
    "        model.eval()\n",
    "        eval_loss, result, target, predicted, text_mismatch, target_mismatch, predicted_mismatch = \\\n",
    "            evaluate_model(args, model, device, test_dataloader, test_examples_dataloader, tr_loss, nb_tr_steps,\n",
    "                           global_step, autoencoder, model_second_stack, batches=None)\n",
    "\n",
    "        # Save model information\n",
    "        # CHECK IF target and predicted need to be array\n",
    "        labels = labels_array_int[task_name]\n",
    "        result['precision_macro'], result['recall_macro'], result['f1_macro'], support =\\\n",
    "            precision_recall_fscore_support(target, predicted, average='macro', labels=labels)\n",
    "        result['precision_micro'], result['recall_micro'], result['f1_micro'], support =\\\n",
    "            precision_recall_fscore_support(target, predicted, average='micro', labels=labels)\n",
    "        result['precision_weighted'], result['recall_weighted'], result['f1_weighted'], support =\\\n",
    "            precision_recall_fscore_support(target, predicted, average='weighted', labels=labels)\n",
    "        result['confusion_matrix'] = confusion_matrix(target, predicted, labels=labels).tolist()\n",
    "\n",
    "        output_eval_filename = \"eval_results\"\n",
    "        if not args.do_train:\n",
    "            output_eval_filename = \"eval_results_test\"\n",
    "\n",
    "        # target_asarray = np.asarray(target)\n",
    "        # predicted_asarray = np.asarray(predicted)\n",
    "        classes = np.asarray(labels_array[task_name])\n",
    "        classes_idx = None\n",
    "        if 'webapplications' in task_name:\n",
    "            classes = np.asarray(['0', '1', '3', '4', '5', '6', '7'])  # there is no class 2 in test\n",
    "            classes_idx = np.asarray([0, 1, 2, 3, 4, 5, 6])\n",
    "        target_asarray = np.asarray(target)\n",
    "        predicted_asarray = np.asarray(predicted)\n",
    "        ax, fig = plot_confusion_matrix(target_asarray, predicted_asarray, classes=classes,\n",
    "                                        normalize=True, title='Normalized confusion matrix', rotate=False,\n",
    "                                        classes_idx=classes_idx)\n",
    "        fig.savefig(os.path.join(args.output_dir, output_eval_filename + \"_confusion.png\"))\n",
    "\n",
    "        output_eval_file = os.path.join(args.output_dir, output_eval_filename + \".json\")\n",
    "        with open(output_eval_file, \"w\") as writer:\n",
    "            json.dump(result, writer, indent=2)\n",
    "\n",
    "        output_eval_file = os.path.join(args.output_dir, output_eval_filename + \".txt\")\n",
    "        with open(output_eval_file, \"w\") as writer:\n",
    "            logger.info(\"***** Eval results *****\")\n",
    "            for key in sorted(result.keys()):\n",
    "                logger.info(\"  %s = %s\", key, str(result[key]))\n",
    "                writer.write(\"%s = %s\\n\" % (key, str(result[key])))\n",
    "\n",
    "        # Example sentences and original and predicted labels\n",
    "        output_eval_filename = \"eval_examples\"\n",
    "        if not args.do_train:\n",
    "            output_eval_filename = \"eval_examples_test\"\n",
    "\n",
    "        dev_dict = {'sentence': text, 'label_target': target, 'label_prediction': predicted}\n",
    "        keys = ['sentence', 'label_target', 'label_prediction']\n",
    "        output_examples_file = os.path.join(args.output_dir, output_eval_filename + \".tsv\")\n",
    "        file_test = open(output_examples_file, 'wt')\n",
    "        dict_writer = csv.writer(file_test, delimiter='\\t')\n",
    "        dict_writer.writerow(keys)\n",
    "        r = zip(*dev_dict.values())\n",
    "        for d in r:\n",
    "            dict_writer.writerow(d)\n",
    "\n",
    "        # Mismatched example sentences and original and predicted labels\n",
    "        dev_dict = {'sentence': text_mismatch, 'label_target': target_mismatch, 'label_prediction': predicted_mismatch}\n",
    "        keys = ['sentence', 'label_target', 'label_prediction']\n",
    "        output_examples_file = os.path.join(args.output_dir, output_eval_filename + \"_mismatch.tsv\")\n",
    "        file_test = open(output_examples_file, 'wt')\n",
    "        dict_writer = csv.writer(file_test, delimiter='\\t')\n",
    "        dict_writer.writerow(keys)\n",
    "        r = zip(*dev_dict.values())\n",
    "        for d in r:\n",
    "            dict_writer.writerow(d)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n",
     "is_executing": false
    }
   }
  }
 ]
}