BERT_ner_embedding.py

#! usr/bin/env python3
# -*- coding:utf-8 -*-
"""
Copyright 2018 The Google AI Language Team Authors.
BASED ON Google_BERT.
@Author:zhoukaiyin
Adjust code for chinese ner
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import collections
import os
import warnings
with warnings.catch_warnings():  
    warnings.filterwarnings("ignore", category=FutureWarning)
    from bert import modeling_ner_embedding as modeling
    # from bert import modeling as modeling
    from bert import optimization
    from bert import tokenization
    import tensorflow as tf
    from sklearn.metrics import f1_score,precision_score,recall_score
    from tensorflow.python.ops import math_ops
    #import tf_metrics
import pickle
import time
import sys
import json
import requests
from tensorflow.contrib import predictor
import pathlib
import collections
import random
import numpy as np
import spacy
import en_core_web_sm
nlp = en_core_web_sm.load()
from nltk.sentiment import SentimentIntensityAnalyzer
vader_analyzer = SentimentIntensityAnalyzer()
#tf.logging.logger.basicConfig(filename='output_nlu_computer_test.log', level=logging.DEBUG)

#sys.stdout = open("output_nlu_computer_test.log", "w")

flags = tf.flags

FLAGS = flags.FLAGS

flags.DEFINE_string(
    "data_dir", None,
    "The input datadir.",
)

flags.DEFINE_string(
    "bert_config_file", None,
    "The config json file corresponding to the pre-trained BERT model."
)

flags.DEFINE_string(
    "task_name", "Intent", "The name of the task to train."
)

flags.DEFINE_string(
    "output_dir", None,
    "The output directory where the model checkpoints will be written."
)

## Other parameters
flags.DEFINE_string(
    "init_checkpoint", None,
    "Initial checkpoint (usually from a pre-trained BERT model)."
)

flags.DEFINE_string(
    "saved_model_dir", None,
    "Saved model directory (usually contain a .pb file)."
)

flags.DEFINE_bool(
    "do_lower_case", True,
    "Whether to lower case the input text."
)

flags.DEFINE_integer(
    "max_seq_length", 128,
    "The maximum total input sequence length after WordPiece tokenization."
)

flags.DEFINE_bool(
    "do_train", False,
    "Whether to run training."
)
flags.DEFINE_bool("use_tpu", False, "Whether to use TPU or GPU/CPU.")

flags.DEFINE_bool("use_crf", False, "Whether to add one crf layer on top of BERT")

flags.DEFINE_bool("do_eval", False, "Whether to run eval on the dev set.")

flags.DEFINE_bool("do_predict", False,"Whether to run the model in inference mode on the test set.")

flags.DEFINE_bool("do_realtime_inference", False,"Whether to run the model in inference mode on the test set.")

flags.DEFINE_bool("do_realtime_inference_v2", False,"Whether to run the model in inference mode on the test set.")

flags.DEFINE_bool("do_export", False, "Whether to run eval on the dev set.")

flags.DEFINE_integer("train_batch_size", 32, "Total batch size for training.")

flags.DEFINE_integer("eval_batch_size", 32, "Total batch size for eval.")

flags.DEFINE_integer("predict_batch_size", 32, "Total batch size for predict.")

flags.DEFINE_float("learning_rate", 2e-5, "The initial learning rate for Adam.")

flags.DEFINE_float("num_train_epochs", 20, "Total number of training epochs to perform.")

flags.DEFINE_string("export_dir", 'output', "directory to save exported model")

flags.DEFINE_string("data_delimiter", ' ', "delimiter to split data and labels")

flags.DEFINE_string("train_txt", "train.txt", "Name of train txt")
flags.DEFINE_string("dev_txt", "dev.txt", "Name of dev txt")
flags.DEFINE_string("test_txt", "test.txt", "Name of test txt")

flags.DEFINE_string("post_matching", "mean", "mean/concat. (default: mean)")
flags.DEFINE_bool("use_polarity", True, "Whether to use sentiment olarity features. (default: True)")

flags.DEFINE_float(
    "warmup_proportion", 0.1,
    "Proportion of training to perform linear learning rate warmup for. "
    "E.g., 0.1 = 10% of training.")

flags.DEFINE_integer("save_checkpoints_steps", 1000,
                     "How often to save the model checkpoint.")

flags.DEFINE_integer("iterations_per_loop", 1000,
                     "How many steps to make in each estimator call.")

flags.DEFINE_string("vocab_file", None,
                    "The vocabulary file that the BERT model was trained on.")
tf.flags.DEFINE_string("master", None, "[Optional] TensorFlow master URL.")
flags.DEFINE_integer(
    "num_tpu_cores", 8,
    "Only used if `use_tpu` is True. Total number of TPU cores to use.")

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

    def __init__(self, guid, label, text):
        """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.
          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 = text
        self.label = label


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

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


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):
        """Gets the list of labels for this data set."""
        raise NotImplementedError()

    @classmethod
    def _read_data(cls, input_file):
        """Reads a BIO data."""
        with open(input_file) as f:
            lines = []
            words = []
            word = ''
            label = 'NA'
            for line in f:
                contends = line.strip()
                if label == 'NA' and len(contends) > 0: # start of one example, label
                    label = line.strip().split()[1]
                elif len(contends) > 0:
                    word = line.strip()

                if len(contends) == 0 and label != 'NA':
                    w = ' '.join([word for word in words if len(word) > 0])
                    lines.append([w, label])

                    words = []
                    label = 'NA'
                    continue
                if word:
                    words.append(word)
                    word = ''
            return lines


class IntentProcessor(DataProcessor):
    def get_train_examples(self, data_dir):
        return self._create_example(
            self._read_data(os.path.join(data_dir, FLAGS.train_txt)), "train"
        )

    def get_dev_examples(self, data_dir):
        return self._create_example(
            self._read_data(os.path.join(data_dir, FLAGS.dev_txt)), "dev"
        )

    def get_test_examples(self,data_dir):
        return self._create_example(
            self._read_data(os.path.join(data_dir, FLAGS.test_txt)), "test")

    def get_labels(self,data_dir=None):
        labels = set()
        with open(data_dir, 'r') as f:
            lines = f.readlines()
        for l in lines:
            l = l.strip().split()
            
            if len(l) == 2 and len(l[1])>0:  
                labels.add(l[1])
        return sorted(list(labels))

    def _create_example(self, lines, set_type):
        examples = []
        for (i, line) in enumerate(lines):
            guid = "%s-%s" % (set_type, i)
            label = tokenization.convert_to_unicode(line[1])
            text = tokenization.convert_to_unicode(line[0])
            examples.append(InputExample(guid=guid, label=label, text=text))
        return examples


def write_tokens(tokens,mode):
    if mode=="test":
        path = os.path.join(FLAGS.output_dir, "token_"+mode+".txt")
        wf = open(path,'a')
        for token in tokens:
            if token!="**NULL**":
                wf.write(token+'\n')
        wf.close()


def convert_single_example(ex_index, example, label_list, max_seq_length, tokenizer):

    def _get_tokens(textlist, max_seq_length):
        tokens = []
        for i, word in enumerate(textlist):
            token = tokenizer.tokenize(word)
            tokens.extend(token)

        if len(tokens) >= max_seq_length - 1:
            tokens = tokens[0:(max_seq_length - 2)]

        return tokens
    # end def

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

  
    ntokens = []
    segment_ids = []
    ntokens.append("[CLS]")
    segment_ids.append(0)
    textlist = example.text.split(' ')
    tokens = _get_tokens(textlist, max_seq_length)
    for token in tokens:
        ntokens.append(token)
        segment_ids.append(0)
    ntokens.append("[SEP]")
    segment_ids.append(0)

  
    input_ids = tokenizer.convert_tokens_to_ids(ntokens)

  # 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.
    while len(input_ids) < max_seq_length:
        input_ids.append(0)
        input_mask.append(0)
        segment_ids.append(0)

    ner_feature = []
    doc = nlp(example.text)
    doc = [str(x) for x in doc.ents]
    for i, word in enumerate(textlist):
        token = tokenizer.tokenize(word)
        if word in doc:
            ner_feature.extend([1]*len(token))
        else:
            ner_feature.extend([0]*len(token))
    if len(ner_feature) >= max_seq_length - 1:
        ner_feature = ner_feature[0:(max_seq_length - 2)]
    assert len(ner_feature)==len(tokens)
    while len(ner_feature) < max_seq_length:
        ner_feature.append(0)

    polarity_score = vader_analyzer.polarity_scores(example.text)
    polarity = [score for (name, score) in polarity_score.items()]

    assert len(input_ids) == max_seq_length
    assert len(input_mask) == max_seq_length
    assert len(segment_ids) == max_seq_length
    assert len(ner_feature) == max_seq_length
    assert len(polarity) == 4


    label_id = label_map[example.label]
    if ex_index < 5:
        tf.logging.info("*** Example ***")
        tf.logging.info("guid: %s" % (example.guid))
        tf.logging.info("tokens: %s" % " ".join(
            [tokenization.printable_text(x) for x in tokens]))
        tf.logging.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
        tf.logging.info("input_mask: %s" % " ".join([str(x) for x in input_mask]))
        tf.logging.info("segment_ids: %s" % " ".join([str(x) for x in segment_ids]))
        tf.logging.info("label: %s (id = %d)" % (example.label, label_id))
        tf.logging.info("ner_feature: %s"% " ".join([str(x) for x in ner_feature]))
        tf.logging.info("polarity: %s"% " ".join([str(x) for x in polarity]))
        

    feature = InputFeatures(
          input_ids=input_ids,
          input_mask=input_mask,
          segment_ids=segment_ids,
          label_id=label_id,
          ner_feature=ner_feature,
          polarity=polarity)
    return feature


def file_based_convert_examples_to_features(
    examples, label_list, max_seq_length, tokenizer, output_file):
  """Convert a set of `InputExample`s to a TFRecord file."""

  writer = tf.python_io.TFRecordWriter(output_file)

  for (ex_index, example) in enumerate(examples):
    if ex_index % 5000 == 0:
      tf.logging.info("Writing example %d of %d" % (ex_index, len(examples)))

    feature = convert_single_example(ex_index, example, label_list,
                                     max_seq_length, tokenizer)

    def create_int_feature(values):
      f = tf.train.Feature(int64_list=tf.train.Int64List(value=list(values)))
      return f
    def create_float_feature(values):
      f = tf.train.Feature(float_list=tf.train.FloatList(value=list(values)))
      return f

    features = collections.OrderedDict()
    features["input_ids"] = create_int_feature(feature.input_ids)
    features["input_mask"] = create_int_feature(feature.input_mask)
    features["segment_ids"] = create_int_feature(feature.segment_ids)
    features["label_ids"] = create_int_feature([feature.label_id])
    features["ner_feature"] = create_int_feature(feature.ner_feature)
    features["polarity"] = create_float_feature(feature.polarity)

    tf_example = tf.train.Example(features=tf.train.Features(feature=features))
    writer.write(tf_example.SerializeToString())
  writer.close()


def file_based_input_fn_builder(input_file, seq_length, is_training,
                                drop_remainder):
  """Creates an `input_fn` closure to be passed to TPUEstimator."""

  name_to_features = {
      "input_ids": tf.FixedLenFeature([seq_length], tf.int64),
      "input_mask": tf.FixedLenFeature([seq_length], tf.int64),
      "segment_ids": tf.FixedLenFeature([seq_length], tf.int64),
      "label_ids": tf.FixedLenFeature([], tf.int64),
      "ner_feature": tf.FixedLenFeature([seq_length], tf.int64),
      "polarity": tf.FixedLenFeature([4], tf.float32),
  }

  def _decode_record(record, name_to_features):
    """Decodes a record to a TensorFlow example."""
    example = tf.parse_single_example(record, name_to_features)

    # tf.Example only supports tf.int64, but the TPU only supports tf.int32.
    # So cast all int64 to int32.
    for name in list(example.keys()):
      t = example[name]
      if t.dtype == tf.int64:
        t = tf.to_int32(t)
      example[name] = t

    return example

  def input_fn(params):
    """The actual input function."""
    batch_size = params["batch_size"]

    # For training, we want a lot of parallel reading and shuffling.
    # For eval, we want no shuffling and parallel reading doesn't matter.
    d = tf.data.TFRecordDataset(input_file)
    if is_training:
      d = d.repeat()
      d = d.shuffle(buffer_size=100)

    d = d.apply(
        tf.contrib.data.map_and_batch(
            lambda record: _decode_record(record, name_to_features),
            batch_size=batch_size,
            drop_remainder=drop_remainder))

    return d

  return input_fn





def create_model(bert_config, is_training, input_ids, input_mask, segment_ids,
                 labels, num_labels, use_one_hot_embeddings, ner_feature, polarity):
  """Creates a classification model."""
  model = modeling.BertModel(
      config=bert_config,
      is_training=is_training,
      input_ids=input_ids,
      input_mask=input_mask,
      token_type_ids=segment_ids,
      ner_embedding=ner_feature,
      use_one_hot_embeddings=use_one_hot_embeddings)

  # In the demo, we are doing a simple classification task on the entire
  # segment.
  #
  # If you want to use the token-level output, use model.get_sequence_output()
  # instead.
  output_layer = model.get_pooled_output()

  hidden_size = output_layer.shape[-1].value


  with tf.variable_scope("loss"):
    output_weights = tf.get_variable(
        "output_weights", [num_labels, hidden_size],
        initializer=tf.truncated_normal_initializer(stddev=0.02))
    
    if FLAGS.post_matching=='concat' and FLAGS.use_polarity:
        output_weights = tf.get_variable(
            "output_weights", [num_labels, hidden_size*2],
            initializer=tf.truncated_normal_initializer(stddev=0.02))
    
    output_bias = tf.get_variable(
        "output_bias", [num_labels], initializer=tf.zeros_initializer())
    
    if FLAGS.use_polarity:
        polarity_weights = tf.get_variable(
            "polarity_weights", [4, hidden_size],
            initializer=tf.truncated_normal_initializer(stddev=0.02))

        polarity_bias = tf.get_variable(
            "polarity_bias", [hidden_size], initializer=tf.zeros_initializer())


    if is_training:
      output_layer = tf.nn.dropout(output_layer, keep_prob=0.9)
    if FLAGS.use_polarity:
        polarity_logits = tf.matmul(polarity, polarity_weights)
        polarity_logits = tf.nn.bias_add(polarity_logits, polarity_bias)

    if FLAGS.post_matching=='mean' and FLAGS.use_polarity:
        output_layer = (output_layer + polarity_logits) / 2.0
    elif FLAGS.post_matching=='concat' and FLAGS.use_polarity:
        output_layer = tf.concat([output_layer, polarity_logits], axis=-1)

    logits = tf.matmul(output_layer, output_weights, transpose_b=True)
    logits = tf.nn.bias_add(logits, output_bias)
    probabilities = tf.nn.softmax(logits, axis=-1)
    log_probs = tf.nn.log_softmax(logits, axis=-1)

    one_hot_labels = tf.one_hot(labels, depth=num_labels, dtype=tf.float32)

    per_example_loss = -tf.reduce_sum(one_hot_labels * log_probs, axis=-1)
    loss = tf.reduce_mean(per_example_loss)
    predict = tf.argmax(probabilities, axis=-1)

    return (loss, per_example_loss, logits, probabilities, predict)


def model_fn_builder(bert_config, num_labels, init_checkpoint, learning_rate,
                     num_train_steps, num_warmup_steps, use_tpu,
                     use_one_hot_embeddings):
  """Returns `model_fn` closure for TPUEstimator."""

  def model_fn(features, labels, mode, params):  # pylint: disable=unused-argument
    """The `model_fn` for TPUEstimator."""

    tf.logging.info("*** Features ***")
    for name in sorted(features.keys()):
      tf.logging.info("  name = %s, shape = %s" % (name, features[name].shape))

    input_ids = features["input_ids"]
    input_mask = features["input_mask"]
    segment_ids = features["segment_ids"]
    label_ids = features["label_ids"]
    ner_feature = features["ner_feature"]
    polarity = features["polarity"]
    is_real_example = None
    if "is_real_example" in features:
      is_real_example = tf.cast(features["is_real_example"], dtype=tf.float32)
    else:
      is_real_example = tf.ones(tf.shape(label_ids), dtype=tf.float32)

    is_training = (mode == tf.estimator.ModeKeys.TRAIN)

    (total_loss, per_example_loss, logits, probabilities, predict) = create_model(
        bert_config, is_training, input_ids, input_mask, segment_ids, label_ids,
        num_labels, use_one_hot_embeddings, ner_feature, polarity)

    tvars = tf.trainable_variables()
    initialized_variable_names = {}
    scaffold_fn = None
    if init_checkpoint:
      (assignment_map, initialized_variable_names
      ) = modeling.get_assignment_map_from_checkpoint(tvars, init_checkpoint)
      if use_tpu:

        def tpu_scaffold():
          tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
          return tf.train.Scaffold()

        scaffold_fn = tpu_scaffold
      else:
        tf.train.init_from_checkpoint(init_checkpoint, assignment_map)

    tf.logging.info("**** Trainable Variables ****")
    for var in tvars:
      init_string = ""
      if var.name in initialized_variable_names:
        init_string = ", *INIT_FROM_CKPT*"
      tf.logging.info("  name = %s, shape = %s%s", var.name, var.shape,
                      init_string)

    output_spec = None
    accuracy = tf.metrics.accuracy(
                    labels=label_ids, predictions=predict, weights=tf.ones(tf.shape(label_ids)))
    #grads = tf.gradients(total_loss, tvars)[0]
    logging_hook = tf.train.LoggingTensorHook({"loss": total_loss, "accuracy": accuracy[1]}, every_n_iter=10)
    if mode == tf.estimator.ModeKeys.TRAIN:

      train_op = optimization.create_optimizer(
          total_loss, learning_rate, num_train_steps, num_warmup_steps, use_tpu)

      output_spec = tf.contrib.tpu.TPUEstimatorSpec(
          mode=mode,
          loss=total_loss,
          train_op=train_op,
          training_hooks=[logging_hook],
          scaffold_fn=scaffold_fn)
    elif mode == tf.estimator.ModeKeys.EVAL:

      def metric_fn(per_example_loss, label_ids, logits, is_real_example):
        predictions = tf.argmax(logits, axis=-1, output_type=tf.int32)
        accuracy = tf.metrics.accuracy(
            labels=label_ids, predictions=predictions, weights=is_real_example)
        loss = tf.metrics.mean(values=per_example_loss, weights=is_real_example)
        return {
            "eval_accuracy": accuracy,
            "eval_loss": loss,
        }

      eval_metrics = (metric_fn,
                      [per_example_loss, label_ids, logits, is_real_example])
      output_spec = tf.contrib.tpu.TPUEstimatorSpec(
          mode=mode,
          loss=total_loss,
          eval_metrics=eval_metrics,
          evaluation_hooks=[logging_hook],
          scaffold_fn=scaffold_fn)
    else:
      output_spec = tf.contrib.tpu.TPUEstimatorSpec(
          mode=mode,
          predictions={"predict": predict},
          prediction_hooks=[logging_hook],
          scaffold_fn=scaffold_fn)
    return output_spec

  return model_fn


# This function is not used by this file but is still used by the Colab and
# people who depend on it.
def input_fn_builder(features, seq_length, is_training, drop_remainder):
  """Creates an `input_fn` closure to be passed to TPUEstimator."""

  all_input_ids = []
  all_input_mask = []
  all_segment_ids = []
  all_label_ids = []

  for feature in features:
    all_input_ids.append(feature.input_ids)
    all_input_mask.append(feature.input_mask)
    all_segment_ids.append(feature.segment_ids)
    all_label_ids.append(feature.label_id)

  def input_fn(params):
    """The actual input function."""
    batch_size = params["batch_size"]

    num_examples = len(features)

    # This is for demo purposes and does NOT scale to large data sets. We do
    # not use Dataset.from_generator() because that uses tf.py_func which is
    # not TPU compatible. The right way to load data is with TFRecordReader.
    d = tf.data.Dataset.from_tensor_slices({
        "input_ids":
            tf.constant(
                all_input_ids, shape=[num_examples, seq_length],
                dtype=tf.int32),
        "input_mask":
            tf.constant(
                all_input_mask,
                shape=[num_examples, seq_length],
                dtype=tf.int32),
        "segment_ids":
            tf.constant(
                all_segment_ids,
                shape=[num_examples, seq_length],
                dtype=tf.int32),
        "label_ids":
            tf.constant(all_label_ids, shape=[num_examples], dtype=tf.int32),
    })

    if is_training:
      d = d.repeat()
      d = d.shuffle(buffer_size=100)

    d = d.batch(batch_size=batch_size, drop_remainder=drop_remainder)
    return d

  return input_fn


# This function is not used by this file but is still used by the Colab and
# people who depend on it.
def convert_examples_to_features(examples, label_list, max_seq_length,
                                 tokenizer):
  """Convert a set of `InputExample`s to a list of `InputFeatures`."""

  features = []
  for (ex_index, example) in enumerate(examples):
    if ex_index % 10000 == 0:
      tf.logging.info("Writing example %d of %d" % (ex_index, len(examples)))

    feature = convert_single_example(ex_index, example, label_list,
                                     max_seq_length, tokenizer)

    features.append(feature)
  return features

def write_to_label_test(predict_examples, label_predicts=None, label_list=None, file_name="label_test.txt"):
    output_predict_file = os.path.join(FLAGS.output_dir, file_name)
    id2label = {}
    for (i, label) in enumerate(label_list):
        id2label[i] = label
    with open(output_predict_file,'w') as writer:
        #writer.write('Time used: {:.1f} seconds.\n\n'.format(end_time - start_time))
        #writer.write('intent_predicts: {}'.format(intent_predicts))
        #writer.write('intent_ids_lists: {}'.format(intent_ids_lists))
        for i, label in enumerate(label_predicts):
            text = predict_examples[i].text.split(' ')

            correct_label = predict_examples[i].label
            pred_label = id2label[label_predicts[i]]
            writer.write(correct_label + ' ' + pred_label + '\n')


def main(_):
  tf.logging.set_verbosity(tf.logging.INFO)

  processors = {
        "intent": IntentProcessor
    }

  tokenization.validate_case_matches_checkpoint(FLAGS.do_lower_case,
                                                FLAGS.init_checkpoint)

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

  bert_config = modeling.BertConfig.from_json_file(FLAGS.bert_config_file)

  if FLAGS.max_seq_length > bert_config.max_position_embeddings:
    raise ValueError(
        "Cannot use sequence length %d because the BERT model "
        "was only trained up to sequence length %d" %
        (FLAGS.max_seq_length, bert_config.max_position_embeddings))

  tf.gfile.MakeDirs(FLAGS.output_dir)

  task_name = FLAGS.task_name.lower()

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

  processor = processors[task_name]()

  label_list = processor.get_labels(os.path.join(FLAGS.data_dir, "train.txt"))

  tokenizer = tokenization.FullTokenizer(
      vocab_file=FLAGS.vocab_file, do_lower_case=FLAGS.do_lower_case)

  tpu_cluster_resolver = None
  if FLAGS.use_tpu and FLAGS.tpu_name:
    tpu_cluster_resolver = tf.contrib.cluster_resolver.TPUClusterResolver(
        FLAGS.tpu_name, zone=FLAGS.tpu_zone, project=FLAGS.gcp_project)

  is_per_host = tf.contrib.tpu.InputPipelineConfig.PER_HOST_V2
  run_config = tf.contrib.tpu.RunConfig(
      cluster=tpu_cluster_resolver,
      master=FLAGS.master,
      model_dir=FLAGS.output_dir,
      save_checkpoints_steps=FLAGS.save_checkpoints_steps,
      tpu_config=tf.contrib.tpu.TPUConfig(
          iterations_per_loop=FLAGS.iterations_per_loop,
          num_shards=FLAGS.num_tpu_cores,
          per_host_input_for_training=is_per_host))

  train_examples = None
  num_train_steps = None
  num_warmup_steps = None
  if FLAGS.do_train:
    train_examples = processor.get_train_examples(FLAGS.data_dir)
    num_train_steps = int(
        len(train_examples) / FLAGS.train_batch_size * FLAGS.num_train_epochs)
    num_warmup_steps = int(num_train_steps * FLAGS.warmup_proportion)

  model_fn = model_fn_builder(
      bert_config=bert_config,
      num_labels=len(label_list),
      init_checkpoint=FLAGS.init_checkpoint,
      learning_rate=FLAGS.learning_rate,
      num_train_steps=num_train_steps,
      num_warmup_steps=num_warmup_steps,
      use_tpu=FLAGS.use_tpu,
      use_one_hot_embeddings=FLAGS.use_tpu)

  # If TPU is not available, this will fall back to normal Estimator on CPU
  # or GPU.
  estimator = tf.contrib.tpu.TPUEstimator(
      use_tpu=FLAGS.use_tpu,
      model_fn=model_fn,
      config=run_config,
      train_batch_size=FLAGS.train_batch_size,
      eval_batch_size=FLAGS.eval_batch_size,
      predict_batch_size=FLAGS.predict_batch_size)

  if FLAGS.do_train:
    train_file = os.path.join(FLAGS.output_dir, "train.tf_record")
    file_based_convert_examples_to_features(
        train_examples, label_list, FLAGS.max_seq_length, tokenizer, train_file)
    tf.logging.info("***** Running training *****")
    tf.logging.info("  Num examples = %d", len(train_examples))
    tf.logging.info("  Batch size = %d", FLAGS.train_batch_size)
    tf.logging.info("  Num steps = %d", num_train_steps)
    train_input_fn = file_based_input_fn_builder(
        input_file=train_file,
        seq_length=FLAGS.max_seq_length,
        is_training=True,
        drop_remainder=True)
    estimator.train(input_fn=train_input_fn, max_steps=num_train_steps)

  if FLAGS.do_eval:
    eval_examples = processor.get_dev_examples(FLAGS.data_dir)
    num_actual_eval_examples = len(eval_examples)
    if FLAGS.use_tpu:
      # TPU requires a fixed batch size for all batches, therefore the number
      # of examples must be a multiple of the batch size, or else examples
      # will get dropped. So we pad with fake examples which are ignored
      # later on. These do NOT count towards the metric (all tf.metrics
      # support a per-instance weight, and these get a weight of 0.0).
      while len(eval_examples) % FLAGS.eval_batch_size != 0:
        eval_examples.append(PaddingInputExample())

    eval_file = os.path.join(FLAGS.output_dir, "eval.tf_record")
    file_based_convert_examples_to_features(
        eval_examples, label_list, FLAGS.max_seq_length, tokenizer, eval_file)

    tf.logging.info("***** Running evaluation *****")
    tf.logging.info("  Num examples = %d (%d actual, %d padding)",
                    len(eval_examples), num_actual_eval_examples,
                    len(eval_examples) - num_actual_eval_examples)
    tf.logging.info("  Batch size = %d", FLAGS.eval_batch_size)

    # This tells the estimator to run through the entire set.
    eval_steps = None
    # However, if running eval on the TPU, you will need to specify the
    # number of steps.
    if FLAGS.use_tpu:
      assert len(eval_examples) % FLAGS.eval_batch_size == 0
      eval_steps = int(len(eval_examples) // FLAGS.eval_batch_size)

    eval_drop_remainder = True if FLAGS.use_tpu else False
    eval_input_fn = file_based_input_fn_builder(
        input_file=eval_file,
        seq_length=FLAGS.max_seq_length,
        is_training=False,
        drop_remainder=eval_drop_remainder)

    result = estimator.evaluate(input_fn=eval_input_fn, steps=eval_steps)

    output_eval_file = os.path.join(FLAGS.output_dir, "eval_results.txt")
    with tf.gfile.GFile(output_eval_file, "w") as writer:
      tf.logging.info("***** Eval results *****")
      for key in sorted(result.keys()):
        tf.logging.info("  %s = %s", key, str(result[key]))
        writer.write("%s = %s\n" % (key, str(result[key])))

  if FLAGS.do_predict:
    
    predict_examples = processor.get_test_examples(FLAGS.data_dir)
    num_actual_predict_examples = len(predict_examples)
    if FLAGS.use_tpu:
      # TPU requires a fixed batch size for all batches, therefore the number
      # of examples must be a multiple of the batch size, or else examples
      # will get dropped. So we pad with fake examples which are ignored
      # later on.
      while len(predict_examples) % FLAGS.predict_batch_size != 0:
        predict_examples.append(PaddingInputExample())

    predict_file = os.path.join(FLAGS.output_dir, "predict.tf_record")
    file_based_convert_examples_to_features(predict_examples, label_list,
                                            FLAGS.max_seq_length, tokenizer,
                                            predict_file)

    tf.logging.info("***** Running prediction*****")
    tf.logging.info("  Num examples = %d (%d actual, %d padding)",
                    len(predict_examples), num_actual_predict_examples,
                    len(predict_examples) - num_actual_predict_examples)
    tf.logging.info("  Batch size = %d", FLAGS.predict_batch_size)

    predict_drop_remainder = True if FLAGS.use_tpu else False
    predict_input_fn = file_based_input_fn_builder(
        input_file=predict_file,
        seq_length=FLAGS.max_seq_length,
        is_training=False,
        drop_remainder=predict_drop_remainder)

    #result = estimator.predict(input_fn=predict_input_fn)
    predictions = estimator.predict(input_fn=predict_input_fn)
    label_predicts = []
    for prediction in predictions:
      label_predicts.append(prediction['predict'])
    label_predicts = list(label_predicts)

    
    write_to_label_test(
      predict_examples, label_predicts=label_predicts, label_list=label_list,
      file_name="label_test.txt"
    )
    """
    output_predict_file = os.path.join(FLAGS.output_dir, "test_results.txt")
    with tf.gfile.GFile(output_predict_file, "w") as writer:
      num_written_lines = 0
      tf.logging.info("***** Predict results *****")
      for (i, prediction) in enumerate(result):
        predict = prediction["predict"]
        if i >= num_actual_predict_examples:
          break
        output_line = "\t".join(
            str(class_probability)
            for class_probability in probabilities) + "\n"
        writer.write(output_line)
        num_written_lines += 1
    assert num_written_lines == num_actual_predict_examples
    """


if __name__ == "__main__":
  flags.mark_flag_as_required("data_dir")
  flags.mark_flag_as_required("task_name")
  flags.mark_flag_as_required("vocab_file")
  flags.mark_flag_as_required("bert_config_file")
  flags.mark_flag_as_required("output_dir")
  tf.app.run()