run_bert_link_prediction.py

# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION.  All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""BERT finetuning runner."""

from __future__ import absolute_import, division, print_function
import shutil
import argparse
import csv
import logging
import os
import random
import sys

import numpy as np
import torch
from torch.utils.data import (DataLoader, RandomSampler, SequentialSampler,
                              TensorDataset)
from torch.utils.data.distributed import DistributedSampler
from tqdm import tqdm, trange

from torch.nn import CrossEntropyLoss, MSELoss
from scipy.stats import pearsonr, spearmanr
from sklearn.metrics import matthews_corrcoef, f1_score
from sklearn import metrics

from pytorch_pretrained_bert.file_utils import PYTORCH_PRETRAINED_BERT_CACHE, WEIGHTS_NAME, CONFIG_NAME
from pytorch_pretrained_bert.modeling import BertForSequenceClassification, BertConfig
from pytorch_pretrained_bert.tokenization import BertTokenizer
from pytorch_pretrained_bert.optimization import BertAdam, WarmupLinearSchedule

# os.environ['CUDA_VISIBLE_DEVICES']= '1'
#torch.backends.cudnn.deterministic = True

logger = logging.getLogger(__name__)


class InputExample(object):
    """A single training/test example for simple sequence classification."""

    def __init__(self, guid, text_a, text_b=None, text_c=None, label=None):
        """Constructs a InputExample.

        Args:
            guid: Unique id for the example.
            text_a: string. The untokenized text of the first sequence. For single
            sequence tasks, only this sequence must be specified.
            text_b: (Optional) string. The untokenized text of the second sequence.
            Only must be specified for sequence pair tasks.
            text_c: (Optional) string. The untokenized text of the third sequence.
            Only must be specified for sequence triple tasks.
            label: (Optional) string. The label of the example. This should be
            specified for train and dev examples, but not for test examples.
        """
        self.guid = guid
        self.text_a = text_a
        self.text_b = text_b
        self.text_c = text_c
        self.label = label


class InputFeatures(object):
    """A single set of features of data."""

    def __init__(self, input_ids, input_mask, segment_ids, label_id):
        self.input_ids = input_ids
        self.input_mask = input_mask
        self.segment_ids = segment_ids
        self.label_id = label_id


class DataProcessor(object):
    """Base class for data converters for sequence classification data sets."""

    def get_train_examples(self, data_dir):
        """Gets a collection of `InputExample`s for the train set."""
        raise NotImplementedError()

    def get_dev_examples(self, data_dir):
        """Gets a collection of `InputExample`s for the dev set."""
        raise NotImplementedError()

    def get_labels(self, data_dir):
        """Gets the list of labels for this data set."""
        raise NotImplementedError()

    @classmethod
    def _read_tsv(cls, input_file, quotechar=None):
        """Reads a tab separated value file."""
        with open(input_file, "r", encoding="utf-8") as f:
            reader = csv.reader(f, delimiter="\t", quotechar=quotechar)
            lines = []
            for line in reader:
                if sys.version_info[0] == 2:
                    line = list(unicode(cell, 'utf-8') for cell in line)
                lines.append(line)
            return lines


class KGProcessor(DataProcessor):
    """Processor for knowledge graph data set."""
    def __init__(self):
        self.labels = set()

    def get_train_examples(self, data_dir):
        """See base class."""
        return self._create_examples(
            self._read_tsv(os.path.join(data_dir, "train.tsv")), "train", data_dir)

    def get_dev_examples(self, data_dir):
        """See base class."""
        return self._create_examples(
            self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev", data_dir)

    def get_test_examples(self, data_dir):
      """See base class."""
      return self._create_examples(
          self._read_tsv(os.path.join(data_dir, "test.tsv")), "test", data_dir)

    def get_relations(self, data_dir):
        """Gets all labels (relations) in the knowledge graph."""
        # return list(self.labels)
        with open(os.path.join(data_dir, "relations.txt"), 'r') as f:
            lines = f.readlines()
            relations = []
            for line in lines:
                relations.append(line.strip())
        return relations

    def get_labels(self, data_dir):
        """Gets all labels (0, 1) for triples in the knowledge graph."""
        return ["0", "1"]

    def get_entities(self, data_dir):
        """Gets all entities in the knowledge graph."""
        # return list(self.labels)
        with open(os.path.join(data_dir, "entities.txt"), 'r') as f:
            lines = f.readlines()
            entities = []
            for line in lines:
                entities.append(line.strip())
        return entities

    def get_train_triples(self, data_dir):
        """Gets training triples."""
        return self._read_tsv(os.path.join(data_dir, "train.tsv"))

    def get_dev_triples(self, data_dir):
        """Gets validation triples."""
        return self._read_tsv(os.path.join(data_dir, "dev.tsv"))

    def get_test_triples(self, data_dir):
        """Gets test triples."""
        return self._read_tsv(os.path.join(data_dir, "test.tsv"))
    # lines are a list of tuples from line to line: We use this to match entities/relations with the text
    def _create_examples(self, lines, set_type, data_dir):
        """Creates examples for the training and dev sets."""
        # entity to text
        ent2text = {}
        with open(os.path.join(data_dir, "entity2text.txt"), 'r') as f: # reading the data from text, and adding entity2text pairs
            ent_lines = f.readlines()
            for line in ent_lines:
                temp = line.strip().split('\t')
                if len(temp) == 2:
                    end = temp[1]#.find(',')
                    ent2text[temp[0]] = temp[1]#[:end]
  
        if data_dir.find("FB15") != -1:
            with open(os.path.join(data_dir, "entity2textlong.txt"), 'r') as f:
                ent_lines = f.readlines()
                for line in ent_lines:
                    temp = line.strip().split('\t')
                    #first_sent_end_position = temp[1].find(".")
                    ent2text[temp[0]] = temp[1]#[:first_sent_end_position + 1] 

        entities = list(ent2text.keys()) # gets only entities

        rel2text = {}
        with open(os.path.join(data_dir, "relation2text.txt"), 'r') as f:
            rel_lines = f.readlines()
            for line in rel_lines:
                temp = line.strip().split('\t')
                rel2text[temp[0]] = temp[1]      
    # Gets set of tuples
        lines_str_set = set(['\t'.join(line) for line in lines])
        examples = []
        # Loops each tuple and can create negative examples (50% chance)
        count = 0
        for (i, line) in enumerate(lines): # i=index, line= tuple
            # if count == 1000: # only get 10,00
            #     break
            count+=1
            head_ent_text = ent2text[line[0]]#finds match from entity2text to train head entities
            tail_ent_text = ent2text[line[2]]#finds match from entity2text to train tail entities
            relation_text = rel2text[line[1]]#finds match from rel2text to train relations

            if set_type == "dev" or set_type == "test":

                label = "1"

                guid = "%s-%s" % (set_type, i)
                text_a = head_ent_text
                text_b = relation_text
                text_c = tail_ent_text 
                self.labels.add(label)
                examples.append(
                    InputExample(guid=guid, text_a=text_a, text_b=text_b, text_c = text_c, label=label))
                
            elif set_type == "train":
                guid = "%s-%s" % (set_type, i)
                # print("IN TRAIN" + guid)
                text_a = head_ent_text
                text_b = relation_text
                text_c = tail_ent_text 
                examples.append(
                    InputExample(guid=guid, text_a=text_a, text_b=text_b, text_c = text_c, label="1"))

                rnd = random.random()
                guid = "%s-%s" % (set_type + "_corrupt", i)
                if rnd <= 0.5:# 50/50 chance of corrupting head OR tail: changing head/tail entity to negative exampe
                    # corrupting head
                    for j in range(5): # corrupt 5 heads
                        tmp_head = ''
                        while True:
                            tmp_ent_list = set(entities)
                            tmp_ent_list.remove(line[0]) # Remove the first head entity out of all entities
                            tmp_ent_list = list(tmp_ent_list) # convert to list
                            tmp_head = random.choice(tmp_ent_list) # get random head from set of entities
                            tmp_triple_str = tmp_head + '\t' + line[1] + '\t' + line[2] # create new negative example with corrupt head
                            if tmp_triple_str not in lines_str_set:
                                break                    
                        tmp_head_text = ent2text[tmp_head]
                        examples.append( # Add NEW NEGATIVE EXAMPLE
                            InputExample(guid=guid, text_a=tmp_head_text, text_b=text_b, text_c = text_c, label="0"))       
                else:
                    # corrupting tail
                    tmp_tail = ''
                    for j in range(5):
                        while True:
                            tmp_ent_list = set(entities)
                            tmp_ent_list.remove(line[2])
                            tmp_ent_list = list(tmp_ent_list)
                            tmp_tail = random.choice(tmp_ent_list)
                            tmp_triple_str = line[0] + '\t' + line[1] + '\t' + tmp_tail
                            if tmp_triple_str not in lines_str_set:
                                break
                        tmp_tail_text = ent2text[tmp_tail]
                        examples.append(
                            InputExample(guid=guid, text_a=text_a, text_b=text_b, text_c = tmp_tail_text, label="0"))                                                  
        return examples

def convert_examples_to_features(examples, label_list, max_seq_length, tokenizer, print_info = True):
    """Loads a data file into a list of `InputBatch`s."""

    label_map = {label : i for i, label in enumerate(label_list)}

    features = []
    # go through each real example (1), or negative (0)
    for (ex_index, example) in enumerate(examples):
        # print("Hello")
        if ex_index % 10000 == 0 and print_info:
            logger.info("Writing example %d of %d" % (ex_index, len(examples)))
        # if ex_index == 1000: # TEMPORARY
        #     break;
        tokens_a = tokenizer.tokenize(example.text_a) # tokenizes the head

        tokens_b = None
        tokens_c = None

        if example.text_b and example.text_c: # if the example has a relation and tail
            tokens_b = tokenizer.tokenize(example.text_b)
            tokens_c = tokenizer.tokenize(example.text_c)
            # Modifies `tokens_a`, `tokens_b` and `tokens_c`in place so that the total
            # length is less than the specified length.
            # Account for [CLS], [SEP], [SEP], [SEP] with "- 4"
            #_truncate_seq_pair(tokens_a, tokens_b, max_seq_length - 3)
            _truncate_seq_triple(tokens_a, tokens_b, tokens_c, max_seq_length - 4) # truncate the sequence
        else:
            # Account for [CLS] and [SEP] with "- 2"
            if len(tokens_a) > max_seq_length - 2:
                tokens_a = tokens_a[:(max_seq_length - 2)]

        # The convention in BERT is:
        # (a) For sequence pairs:
        #  tokens:   [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
        #  type_ids: 0   0  0    0    0     0       0 0    1  1  1  1   1 1
        # (b) For single sequences:
        #  tokens:   [CLS] the dog is hairy . [SEP]
        #  type_ids: 0   0   0   0  0     0 0
        # (c) for sequence triples:
        #  tokens: [CLS] Steve Jobs [SEP] founded [SEP] Apple Inc .[SEP]
        #  type_ids: 0 0 0 0 1 1 0 0 0 0
        #
        # Where "type_ids" are used to indicate whether this is the first
        # sequence or the second sequence or the third sequence. The embedding vectors for `type=0` and
        # `type=1` were learned during pre-training and are added to the wordpiece
        # embedding vector (and position vector). This is not *strictly* necessary
        # since the [SEP] token unambiguously separates the sequences, but it makes
        # it easier for the model to learn the concept of sequences.
        #
        # For classification tasks, the first vector (corresponding to [CLS]) is
        # used as as the "sentence vector". Note that this only makes sense because
        # the entire model is fine-tuned.
        tokens = ["[CLS]"] + tokens_a + ["[SEP]"]
        segment_ids = [0] * len(tokens)

        if tokens_b: # since we are doing link prediciton, we need entity and tail
            tokens += tokens_b + ["[SEP]"]
            segment_ids += [1] * (len(tokens_b) + 1)
        if tokens_c:
            tokens += tokens_c + ["[SEP]"]
            segment_ids += [0] * (len(tokens_c) + 1)        

        input_ids = tokenizer.convert_tokens_to_ids(tokens) # converts words -> ids (001, 002) from the pretrained bert

        # The mask has 1 for real tokens and 0 for padding tokens. Only real
        # tokens are attended to.
        input_mask = [1] * len(input_ids)

        # Zero-pad up to the sequence length.
        padding = [0] * (max_seq_length - len(input_ids)) # for input seq that are shorter, pad them w 0s
        input_ids += padding
        input_mask += padding
        segment_ids += padding

        assert len(input_ids) == max_seq_length
        assert len(input_mask) == max_seq_length
        assert len(segment_ids) == max_seq_length

        label_id = label_map[example.label]

        if ex_index < 5 and print_info:
            logger.info("*** Example ***")
            logger.info("guid: %s" % (example.guid))
            logger.info("tokens: %s" % " ".join(
                    [str(x) for x in tokens]))
            logger.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
            logger.info("input_mask: %s" % " ".join([str(x) for x in input_mask]))
            logger.info(
                    "segment_ids: %s" % " ".join([str(x) for x in segment_ids]))
            logger.info("label: %s (id = %d)" % (example.label, label_id))

        features.append(
                InputFeatures(input_ids=input_ids,
                              input_mask=input_mask,
                              segment_ids=segment_ids,
                              label_id=label_id))
    return features # return list of all feature 1. input_ids (goat: 120) 2. input_mask(1-real, 0-pad)
                    # 3. segment_ids(00011000 , differentiates the sequence into three parts: head, rel, tail)


def _truncate_seq_pair(tokens_a, tokens_b, max_length):
    """Truncates a sequence pair in place to the maximum length."""

    # This is a simple heuristic which will always truncate the longer sequence
    # one token at a time. This makes more sense than truncating an equal percent
    # of tokens from each, since if one sequence is very short then each token
    # that's truncated likely contains more information than a longer sequence.
    while True:
        total_length = len(tokens_a) + len(tokens_b)
        if total_length <= max_length:
            break
        if len(tokens_a) > len(tokens_b):
            tokens_a.pop()
        else:
            tokens_b.pop()

def _truncate_seq_triple(tokens_a, tokens_b, tokens_c, max_length):
    """Truncates a sequence triple in place to the maximum length."""

    # This is a simple heuristic which will always truncate the longer sequence
    # one token at a time. This makes more sense than truncating an equal percent
    # of tokens from each, since if one sequence is very short then each token
    # that's truncated likely contains more information than a longer sequence.
    while True:
        total_length = len(tokens_a) + len(tokens_b) + len(tokens_c)
        if total_length <= max_length:
            break
        if len(tokens_a) > len(tokens_b) and len(tokens_a) > len(tokens_c):
            tokens_a.pop()
        elif len(tokens_b) > len(tokens_a) and len(tokens_b) > len(tokens_c):
            tokens_b.pop()
        elif len(tokens_c) > len(tokens_a) and len(tokens_c) > len(tokens_b):
            tokens_c.pop()
        else:
            tokens_c.pop()

def simple_accuracy(preds, labels):
    return (preds == labels).mean()

def compute_metrics(task_name, preds, labels):
    assert len(preds) == len(labels)
    if task_name == "kg":
        return {"acc": simple_accuracy(preds, labels)}
    else:
        raise KeyError(task_name)


def main():
    print("HERE")
    parser = argparse.ArgumentParser()

    ## Required parameters
    parser.add_argument("--data_dir",
                        default=None,
                        type=str,
                        required=True,
                        help="The input data dir. Should contain the .tsv files (or other data files) for the task.")
    parser.add_argument("--bert_model", default=None, type=str, required=True,
                        help="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.")
    parser.add_argument("--task_name",
                        default=None,
                        type=str,
                        required=True,
                        help="The name of the task to train.")
    parser.add_argument("--output_dir",
                        default=None,
                        type=str,
                        required=True,
                        help="The output directory where the model predictions and checkpoints will be written.")

    ## Other parameters
    parser.add_argument("--cache_dir",
                        default="",
                        type=str,
                        help="Where do you want to store the pre-trained models downloaded from s3")
    parser.add_argument("--max_seq_length",
                        default=128,
                        type=int,
                        help="The maximum total input sequence length after WordPiece tokenization. \n"
                             "Sequences longer than this will be truncated, and sequences shorter \n"
                             "than this will be padded.")
    parser.add_argument("--do_train",
                        action='store_true',
                        help="Whether to run training.")
    parser.add_argument("--do_eval",
                        action='store_true',
                        help="Whether to run eval on the dev set.")
    parser.add_argument("--do_predict",
                        action='store_true',
                        help="Whether to run eval on the test set.")
    parser.add_argument("--do_lower_case",
                        action='store_true',
                        help="Set this flag if you are using an uncased model.")
    parser.add_argument("--train_batch_size",
                        default=32,
                        type=int,
                        help="Total batch size for training.")
    parser.add_argument("--eval_batch_size",
                        default=8,
                        type=int,
                        help="Total batch size for eval.")
    parser.add_argument("--learning_rate",
                        default=5e-5,
                        type=float,
                        help="The initial learning rate for Adam.")
    parser.add_argument("--num_train_epochs",
                        default=3.0,
                        type=float,
                        help="Total number of training epochs to perform.")
    parser.add_argument("--warmup_proportion",
                        default=0.1,
                        type=float,
                        help="Proportion of training to perform linear learning rate warmup for. "
                             "E.g., 0.1 = 10%% of training.")
    parser.add_argument("--no_cuda",
                        action='store_true',
                        help="Whether not to use CUDA when available")
    parser.add_argument("--local_rank",
                        type=int,
                        default=-1,
                        help="local_rank for distributed training on gpus")
    parser.add_argument('--seed',
                        type=int,
                        default=42,
                        help="random seed for initialization")
    parser.add_argument('--gradient_accumulation_steps',
                        type=int,
                        default=1,
                        help="Number of updates steps to accumulate before performing a backward/update pass.")
    parser.add_argument('--fp16',
                        action='store_true',
                        help="Whether to use 16-bit float precision instead of 32-bit")
    parser.add_argument('--loss_scale',
                        type=float, default=0,
                        help="Loss scaling to improve fp16 numeric stability. Only used when fp16 set to True.\n"
                             "0 (default value): dynamic loss scaling.\n"
                             "Positive power of 2: static loss scaling value.\n")
    parser.add_argument('--server_ip', type=str, default='', help="Can be used for distant debugging.")
    parser.add_argument('--server_port', type=str, default='', help="Can be used for distant debugging.")
    args = parser.parse_args()

    if args.server_ip and args.server_port:
        # Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
        import ptvsd
        print("Waiting for debugger attach")
        ptvsd.enable_attach(address=(args.server_ip, args.server_port), redirect_output=True)
        ptvsd.wait_for_attach()

    processors = {
        "kg": KGProcessor,
    }
# Since we are just running on remote sheet server, our device is an object on which a torch Tensor will be allocated
    if args.local_rank == -1 or args.no_cuda:
        device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
        n_gpu = torch.cuda.device_count()
    else:
        torch.cuda.set_device(args.local_rank)
        device = torch.device("cuda", args.local_rank)
        n_gpu = 1
        # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
        torch.distributed.init_process_group(backend='nccl')

    logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s -   %(message)s',
                        datefmt = '%m/%d/%Y %H:%M:%S',
                        level = logging.INFO if args.local_rank in [-1, 0] else logging.WARN)

    logger.info("device: {} n_gpu: {}, distributed training: {}, 16-bits training: {}".format(
        device, n_gpu, bool(args.local_rank != -1), args.fp16))

    if args.gradient_accumulation_steps < 1:
        raise ValueError("Invalid gradient_accumulation_steps parameter: {}, should be >= 1".format(
                            args.gradient_accumulation_steps))

    args.train_batch_size = args.train_batch_size // args.gradient_accumulation_steps
    args.seed = random.randint(1, 200)
    random.seed(args.seed)
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
# To get determinism, set all gpus to same seed
    if n_gpu > 0:
        torch.cuda.manual_seed_all(args.seed)

    if not args.do_train and not args.do_eval:
        raise ValueError("At least one of `do_train` or `do_eval` must be True.")

    if os.path.exists(args.output_dir) and os.listdir(args.output_dir) and args.do_train:
        shutil.rmtree(args.output_dir)
        # raise ValueError("Output directory ({}) already exists and is not empty.".format(args.output_dir))
    if not os.path.exists(args.output_dir):
        os.makedirs(args.output_dir)

    task_name = args.task_name.lower()

    if task_name not in processors:
        raise ValueError("Task not found: %s" % (task_name))

    processor = processors[task_name]()
    # Gets the list of all tuples
    label_list = processor.get_labels(args.data_dir)
    num_labels = len(label_list)

    entity_list = processor.get_entities(args.data_dir)
    # print(entity_list)

    tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)

    train_examples = None
    num_train_optimization_steps = 0
    if args.do_train:
        train_examples = processor.get_train_examples(args.data_dir)
        # get examples in each batch size and multiply by epochs to get total optimization steps
        num_train_optimization_steps = int(
            len(train_examples) / args.train_batch_size / args.gradient_accumulation_steps) * args.num_train_epochs
        if args.local_rank != -1:
            num_train_optimization_steps = num_train_optimization_steps // torch.distributed.get_world_size() # returns # optimization steps for each process in current group

    # Prepare model
    cache_dir = args.cache_dir if args.cache_dir else os.path.join(str(PYTORCH_PRETRAINED_BERT_CACHE), 'distributed_{}'.format(args.local_rank))
    model = BertForSequenceClassification.from_pretrained(args.bert_model, # creates bert classification model
              cache_dir=cache_dir,
              num_labels=num_labels) # Labels are binary: 1 or 0
    if args.fp16:
        model.half()
    model.to(device) # we allocate all model tensors to our sheet server device (cuda)
    if args.local_rank != -1:
        try:
            from apex.parallel import DistributedDataParallel as DDP
        except ImportError:
            raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")

        model = DDP(model)
    elif n_gpu > 1: # if theres more than one gpu on server, can run parallel
        model = torch.nn.DataParallel(model)
        #model = torch.nn.parallel.data_parallel(model)
    # Prepare optimizer
    param_optimizer = list(model.named_parameters()) # named_parameters are
    no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']# apply weight decay to all parameters other than bias and layer norm
    optimizer_grouped_parameters = [
        {'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
        {'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
        ]
    if args.fp16:
        try:
            from apex.optimizers import FP16_Optimizer
            from apex.optimizers import FusedAdam
        except ImportError:
            raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")

        optimizer = FusedAdam(optimizer_grouped_parameters,
                              lr=args.learning_rate,
                              bias_correction=False,
                              max_grad_norm=1.0)
        if args.loss_scale == 0:
            optimizer = FP16_Optimizer(optimizer, dynamic_loss_scale=True)
        else:
            optimizer = FP16_Optimizer(optimizer, static_loss_scale=args.loss_scale)
        warmup_linear = WarmupLinearSchedule(warmup=args.warmup_proportion,
                                             t_total=num_train_optimization_steps)        

    else:
        optimizer = BertAdam(optimizer_grouped_parameters, # optimization model
                             lr=args.learning_rate,
                             warmup=args.warmup_proportion, #
                             t_total=num_train_optimization_steps) # gets the number of steps to optimize (total epoch optimizatinos)

    global_step = 0
    nb_tr_steps = 0
    tr_loss = 0
    if args.do_train:
        # passes in the train_examples, so a list of InputExample objects (text_a = head, text_b = relation, text_c = tail)
        train_features = convert_examples_to_features(
            train_examples, label_list, args.max_seq_length, tokenizer)
        logger.info("***** Running training *****")
        logger.info("  Num examples = %d", len(train_examples))
        logger.info("  Batch size = %d", args.train_batch_size)
        logger.info("  Num steps = %d", num_train_optimization_steps)
        #creates a tensor of type long int
        all_input_ids = torch.tensor([f.input_ids for f in train_features], dtype=torch.long)
        all_input_mask = torch.tensor([f.input_mask for f in train_features], dtype=torch.long)
        all_segment_ids = torch.tensor([f.segment_ids for f in train_features], dtype=torch.long)

        all_label_ids = torch.tensor([f.label_id for f in train_features], dtype=torch.long)

        train_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)
        if args.local_rank == -1:
            train_sampler = RandomSampler(train_data) #randomly permutes a batch of indices and yields next index/key to fetch
        else:
            train_sampler = DistributedSampler(train_data)
            # Combines a dataset and a sampler, and provides an iterable over the given dataset. https://pytorch.org/docs/stable/data.html
        train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=args.train_batch_size)

        model.train() #Puts in train mode. https://huggingface.co/transformers/training.html
        print(model)
        for _ in trange(int(args.num_train_epochs), desc="Epoch"): # every epoch
            tr_loss = 0
            nb_tr_examples, nb_tr_steps = 0, 0
            for step, batch in enumerate(tqdm(train_dataloader, desc="Iteration")):
                batch = tuple(t.to(device) for t in batch) # gets the batch of samples from dataloader
                input_ids, input_mask, segment_ids, label_ids = batch # input_mask is used so padded ones (0) wont have any attention to them in encoder

                # define a new function to compute loss values for both output_modes
                logits = model(input_ids, segment_ids, input_mask, labels=None) # the vector of raw (non-normalized) predictions that a classification model generates
                # print("The logits are: ", logits, logits.shape) # 32 x 2
                # print(label_ids.view(-1))
                loss_fct = CrossEntropyLoss()
                loss = loss_fct(logits.view(-1, num_labels), label_ids.view(-1)) # unsure of # rows
                # logits batch size: (32 x 2) and labels batch #: (32 x 1)
                if n_gpu > 1:
                    loss = loss.mean() # mean() to average on multi-gpu.
                if args.gradient_accumulation_steps > 1:
                    loss = loss / args.gradient_accumulation_steps

                if args.fp16:
                    optimizer.backward(loss)
                else:
                    loss.backward()

                tr_loss += loss.item() # adds total loss
                nb_tr_examples += input_ids.size(0) # adds 32 examples to count total examples
                nb_tr_steps += 1 #
                if (step + 1) % args.gradient_accumulation_steps == 0: # when the # steps reaches time to optimize gradient
                    if args.fp16:
                        # modify learning rate with special warm up BERT uses
                        # if args.fp16 is False, BertAdam is used that handles this automatically
                        lr_this_step = args.learning_rate * warmup_linear.get_lr(global_step/num_train_optimization_steps,
                                                                                 args.warmup_proportion)
                        for param_group in optimizer.param_groups:
                            param_group['lr'] = lr_this_step
                    optimizer.step() # adjust the parameters by the gradients collected in the backward pass
                    optimizer.zero_grad() # reset the gradients of model parameters after each pass
                    global_step += 1
            print("Training loss: ", tr_loss, nb_tr_examples)
# Loss 1 epoch: 16.94.
    if args.do_train and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
        # Save a trained model, configuration and tokenizer
        model_to_save = model.module if hasattr(model, 'module') else model  # Only save the model it-self

        # If we save using the predefined names, we can load using `from_pretrained`
        output_model_file = os.path.join(args.output_dir, WEIGHTS_NAME)
        output_config_file = os.path.join(args.output_dir, CONFIG_NAME)

        torch.save(model_to_save.state_dict(), output_model_file) # saves our trained model to pytorch_model.bin
        model_to_save.config.to_json_file(output_config_file)
        tokenizer.save_vocabulary(args.output_dir)

        # Load a trained model and vocabulary that you have fine-tuned
        model = BertForSequenceClassification.from_pretrained(args.output_dir, num_labels=num_labels)
        tokenizer = BertTokenizer.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
    else:
        model = BertForSequenceClassification.from_pretrained(args.bert_model, num_labels=num_labels)
    model.to(device)

    if args.do_eval and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
        
        eval_examples = processor.get_dev_examples(args.data_dir)
        eval_features = convert_examples_to_features(
            eval_examples, label_list, args.max_seq_length, tokenizer)
        logger.info("***** Running evaluation *****")
        logger.info("  Num examples = %d", len(eval_examples))
        logger.info("  Batch size = %d", args.eval_batch_size)
        all_input_ids = torch.tensor([f.input_ids for f in eval_features], dtype=torch.long)
        all_input_mask = torch.tensor([f.input_mask for f in eval_features], dtype=torch.long)
        all_segment_ids = torch.tensor([f.segment_ids for f in eval_features], dtype=torch.long)

        all_label_ids = torch.tensor([f.label_id for f in eval_features], dtype=torch.long)

        eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)
        # Run prediction for full data
        eval_sampler = SequentialSampler(eval_data)
        eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)
        
        # Load a trained model and vocabulary that you have fine-tuned
        model = BertForSequenceClassification.from_pretrained(args.output_dir, num_labels=num_labels)
        tokenizer = BertTokenizer.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
        model.to(device)

        model.eval() # sets the model layers for evaluation
        eval_loss = 0
        nb_eval_steps = 0
        preds = []

        for input_ids, input_mask, segment_ids, label_ids in tqdm(eval_dataloader, desc="Evaluating"):
            input_ids = input_ids.to(device)
            input_mask = input_mask.to(device)
            segment_ids = segment_ids.to(device)
            label_ids = label_ids.to(device)

            with torch.no_grad(): #reduce memory usage and speed up computations but you won’t be able to backprop
                logits = model(input_ids, segment_ids, input_mask, labels=None)

            # create eval loss and other metric required by the task
            loss_fct = CrossEntropyLoss()
            tmp_eval_loss = loss_fct(logits.view(-1, num_labels), label_ids.view(-1))
            print(label_ids.view(-1))
            
            eval_loss += tmp_eval_loss.mean().item()
            nb_eval_steps += 1
            if len(preds) == 0:
                preds.append(logits.detach().cpu().numpy()) # converting to regular array lists to store loss
            else:
                preds[0] = np.append(
                    preds[0], logits.detach().cpu().numpy(), axis=0)

        eval_loss = eval_loss / nb_eval_steps
        preds = preds[0]

        preds = np.argmax(preds, axis=1) # finds the max element in each row
        result = compute_metrics(task_name, preds, all_label_ids.numpy())
        loss = tr_loss/nb_tr_steps if args.do_train else None

        result['eval_loss'] = eval_loss
        result['global_step'] = global_step
        result['loss'] = loss
        # We are documenting the loss for that evaluation
        output_eval_file = os.path.join(args.output_dir, "eval_results.txt")
        with open(output_eval_file, "w") as writer:
            logger.info("***** Eval results *****")
            for key in sorted(result.keys()):
                logger.info("  %s = %s", key, str(result[key]))
                writer.write("%s = %s\n" % (key, str(result[key])))

    if args.do_predict and (args.local_rank == -1 or torch.distributed.get_rank() == 0):

        train_triples = processor.get_train_triples(args.data_dir)
        dev_triples = processor.get_dev_triples(args.data_dir)
        test_triples = processor.get_test_triples(args.data_dir)
        all_triples = train_triples + dev_triples + test_triples

        all_triples_str_set = set()
        for triple in all_triples:
            triple_str = '\t'.join(triple)
            all_triples_str_set.add(triple_str)

        eval_examples = processor.get_test_examples(args.data_dir)
        eval_features = convert_examples_to_features(
            eval_examples, label_list, args.max_seq_length, tokenizer)
        logger.info("***** Running Prediction *****")
        logger.info("  Num examples = %d", len(eval_examples))
        logger.info("  Batch size = %d", args.eval_batch_size)
        all_input_ids = torch.tensor([f.input_ids for f in eval_features], dtype=torch.long)
        all_input_mask = torch.tensor([f.input_mask for f in eval_features], dtype=torch.long)
        all_segment_ids = torch.tensor([f.segment_ids for f in eval_features], dtype=torch.long)

        all_label_ids = torch.tensor([f.label_id for f in eval_features], dtype=torch.long)

        eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)
        # Run prediction for full data
        eval_sampler = SequentialSampler(eval_data)
        eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)
        # Load a trained model and vocabulary that you have fine-tuned
        model = BertForSequenceClassification.from_pretrained(args.output_dir, num_labels=num_labels)
        tokenizer = BertTokenizer.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
        model.to(device)
        model.eval()
        eval_loss = 0
        nb_eval_steps = 0
        preds = []

        for input_ids, input_mask, segment_ids, label_ids in tqdm(eval_dataloader, desc="Testing"):
            input_ids = input_ids.to(device)
            input_mask = input_mask.to(device)
            segment_ids = segment_ids.to(device)
            label_ids = label_ids.to(device)

            with torch.no_grad():
                logits = model(input_ids, segment_ids, input_mask, labels=None)

            loss_fct = CrossEntropyLoss()
            tmp_eval_loss = loss_fct(logits.view(-1, num_labels), label_ids.view(-1))
            
            eval_loss += tmp_eval_loss.mean().item()
            nb_eval_steps += 1
            if len(preds) == 0:
                preds.append(logits.detach().cpu().numpy())
            else:
                preds[0] = np.append(
                    preds[0], logits.detach().cpu().numpy(), axis=0)

        eval_loss = eval_loss / nb_eval_steps
        preds = preds[0]
        print(preds, preds.shape)
        
        all_label_ids = all_label_ids.numpy()

        preds = np.argmax(preds, axis=1)

        result = compute_metrics(task_name, preds, all_label_ids) # USes simple accuracy: Finds proportion of correct preds
        loss = tr_loss/nb_tr_steps if args.do_train else None

        result['eval_loss'] = eval_loss
        result['global_step'] = global_step
        result['loss'] = loss

        output_eval_file = os.path.join(args.output_dir, "test_results.txt")
        with open(output_eval_file, "w") as writer:
            logger.info("***** Test results *****")
            for key in sorted(result.keys()):
                logger.info("  %s = %s", key, str(result[key]))
                writer.write("%s = %s\n" % (key, str(result[key])))
        print("Triple classification acc is : ")
        print(metrics.accuracy_score(all_label_ids, preds))

        # run link prediction
        ranks = []
        ranks_left = []
        ranks_right = []

        hits_left = []
        hits_right = []
        hits = []

        top_ten_hit_count = 0

        for i in range(10):
            hits_left.append([])
            hits_right.append([])
            hits.append([])
        '''
        file_prefix = str(args.data_dir[7:])
        f = open(file_prefix + '_ranks.txt','r')
        lines = f.readlines()
        for line in lines:
            temp = line.strip().split()
            rank1 = int(temp[0])
            ranks_left.append(rank1+1)
            print('left: ', rank1)
            ranks.append(rank1+1)
            if rank1 < 10:
                top_ten_hit_count += 1
            rank2 = int(temp[1])
            ranks.append(rank2+1)
            ranks_right.append(rank2+1)
            print('right: ', rank2)
            print('mean rank until now: ', np.mean(ranks))
            if rank2 < 10:
                top_ten_hit_count += 1
            print("hit@10 until now: ", top_ten_hit_count * 1.0 / len(ranks))                
            for hits_level in range(10):
                if rank1 <= hits_level:
                    hits[hits_level].append(1.0)
                    hits_left[hits_level].append(1.0)
                else:
                    hits[hits_level].append(0.0)
                    hits_left[hits_level].append(0.0)

                if rank2 <= hits_level:
                    hits[hits_level].append(1.0)
                    hits_right[hits_level].append(1.0)
                else:
                    hits[hits_level].append(0.0)
                    hits_right[hits_level].append(0.0)
    
        '''
        for test_triple in test_triples: # run through each tuple in test set
            head = test_triple[0]
            relation = test_triple[1]
            tail = test_triple[2]
            #print(test_triple, head, relation, tail)

            head_corrupt_list = [test_triple]
            for corrupt_ent in entity_list:
                if corrupt_ent != head:
                    tmp_triple = [corrupt_ent, relation, tail] # creating tuples that replace the truth head
                    tmp_triple_str = '\t'.join(tmp_triple)
                    if tmp_triple_str not in all_triples_str_set:
                        # may be slow
                        head_corrupt_list.append(tmp_triple)
            # since we are in test, we wont corrupt examples, we did that above!
            tmp_examples = processor._create_examples(head_corrupt_list, "test", args.data_dir)
            print(len(tmp_examples))
            tmp_features = convert_examples_to_features(tmp_examples, label_list, args.max_seq_length, tokenizer, print_info = False)
            all_input_ids = torch.tensor([f.input_ids for f in tmp_features], dtype=torch.long)
            all_input_mask = torch.tensor([f.input_mask for f in tmp_features], dtype=torch.long)
            all_segment_ids = torch.tensor([f.segment_ids for f in tmp_features], dtype=torch.long)
            all_label_ids = torch.tensor([f.label_id for f in tmp_features], dtype=torch.long)

            eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)
            # Run prediction for temp data, each tuple is corrupted with head,
            eval_sampler = SequentialSampler(eval_data)
            eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)
            model.eval()

            preds = []
            
            for input_ids, input_mask, segment_ids, label_ids in tqdm(eval_dataloader, desc="Testing"):
                # loop through and update loss for each batch
                input_ids = input_ids.to(device)
                input_mask = input_mask.to(device)
                segment_ids = segment_ids.to(device)
                label_ids = label_ids.to(device)
                
                with torch.no_grad():
                    logits = model(input_ids, segment_ids, input_mask, labels=None)
                if len(preds) == 0:
                    batch_logits = logits.detach().cpu().numpy()
                    preds.append(batch_logits)

                else:
                    batch_logits = logits.detach().cpu().numpy()
                    preds[0] = np.append(preds[0], batch_logits, axis=0) # adds 32(batch_size) new rows each time

            preds = preds[0]
            # get the dimension corresponding to current label 1
            # print("Preds: " + preds, preds.shape)
            rel_values = preds[:, all_label_ids[0]] # get all in column 1: corrupt logits
            rel_values = torch.tensor(rel_values)
            # print("rel_values: " + rel_values, rel_values.shape)
            _, argsort1 = torch.sort(rel_values, descending=True) # sort from largest to smallest
            #print(max_values)
            #print(argsort1)
            argsort1 = argsort1.cpu().numpy()
            rank1 = np.where(argsort1 == 0)[0][0] # get the highest
            print('left: ', rank1)
            ranks.append(rank1+1)
            ranks_left.append(rank1+1)
            if rank1 < 10:
                top_ten_hit_count += 1

            #Now do same thing for tail corruption
            tail_corrupt_list = [test_triple]
            for corrupt_ent in entity_list:
                if corrupt_ent != tail:
                    tmp_triple = [head, relation, corrupt_ent]
                    tmp_triple_str = '\t'.join(tmp_triple)
                    if tmp_triple_str not in all_triples_str_set:
                        # may be slow
                        tail_corrupt_list.append(tmp_triple)

            tmp_examples = processor._create_examples(tail_corrupt_list, "test", args.data_dir)
            #print(len(tmp_examples))
            tmp_features = convert_examples_to_features(tmp_examples, label_list, args.max_seq_length, tokenizer, print_info = False)
            all_input_ids = torch.tensor([f.input_ids for f in tmp_features], dtype=torch.long)
            all_input_mask = torch.tensor([f.input_mask for f in tmp_features], dtype=torch.long)
            all_segment_ids = torch.tensor([f.segment_ids for f in tmp_features], dtype=torch.long)
            all_label_ids = torch.tensor([f.label_id for f in tmp_features], dtype=torch.long)

            eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)
            # Run prediction for temp data
            eval_sampler = SequentialSampler(eval_data)
            eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)
            model.eval()
            preds = []        

            for input_ids, input_mask, segment_ids, label_ids in tqdm(eval_dataloader, desc="Testing"):
            
                input_ids = input_ids.to(device)
                input_mask = input_mask.to(device)
                segment_ids = segment_ids.to(device)
                label_ids = label_ids.to(device)
                
                with torch.no_grad():
                    logits = model(input_ids, segment_ids, input_mask, labels=None)
                if len(preds) == 0:
                    batch_logits = logits.detach().cpu().numpy()
                    preds.append(batch_logits)

                else:
                    batch_logits = logits.detach().cpu().numpy()
                    preds[0] = np.append(preds[0], batch_logits, axis=0) 

            preds = preds[0]
            # get the dimension corresponding to current label 1
            rel_values = preds[:, all_label_ids[0]]
            rel_values = torch.tensor(rel_values)
            _, argsort1 = torch.sort(rel_values, descending=True)
            argsort1 = argsort1.cpu().numpy()
            rank2 = np.where(argsort1 == 0)[0][0]
            ranks.append(rank2+1)
            ranks_right.append(rank2+1)
            print('right: ', rank2)
            print('mean rank until now: ', np.mean(ranks))
            if rank2 < 10:
                top_ten_hit_count += 1
            print("hit@10 until now: ", top_ten_hit_count * 1.0 / len(ranks))

            file_prefix = str(args.data_dir[7:]) + "_" + str(args.train_batch_size) + "_" + str(args.learning_rate) + "_" + str(args.max_seq_length) + "_" + str(args.num_train_epochs)
            #file_prefix = str(args.data_dir[7:])
            f = open(file_prefix + '_ranks.txt','a')
            f.write(str(rank1) + '\t' + str(rank2) + '\n')
            f.close()
            # this could be done more elegantly, but here you go
            for hits_level in range(10):
                if rank1 <= hits_level:
                    hits[hits_level].append(1.0)
                    hits_left[hits_level].append(1.0)
                else:
                    hits[hits_level].append(0.0)
                    hits_left[hits_level].append(0.0)

                if rank2 <= hits_level:
                    hits[hits_level].append(1.0)
                    hits_right[hits_level].append(1.0)
                else:
                    hits[hits_level].append(0.0)
                    hits_right[hits_level].append(0.0)
    

        for i in [0,2,9]:
            logger.info('Hits left @{0}: {1}'.format(i+1, np.mean(hits_left[i])))
            logger.info('Hits right @{0}: {1}'.format(i+1, np.mean(hits_right[i])))
            logger.info('Hits @{0}: {1}'.format(i+1, np.mean(hits[i])))
        logger.info('Mean rank left: {0}'.format(np.mean(ranks_left)))
        logger.info('Mean rank right: {0}'.format(np.mean(ranks_right)))
        logger.info('Mean rank: {0}'.format(np.mean(ranks)))
        logger.info('Mean reciprocal rank left: {0}'.format(np.mean(1./np.array(ranks_left))))
        logger.info('Mean reciprocal rank right: {0}'.format(np.mean(1./np.array(ranks_right))))
        logger.info('Mean reciprocal rank: {0}'.format(np.mean(1./np.array(ranks))))            
              
if __name__ == "__main__":
    main()