# -*- coding: utf-8 -*- # MegEngine is Licensed under the Apache License, Version 2.0 (the "License") # # Copyright (c) 2014-2020 Megvii Inc. All rights reserved. # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. import csv import os import megengine as mge import numpy as np from megengine.data import DataLoader from megengine.data.dataset import ArrayDataset from megengine.data.sampler import RandomSampler, SequentialSampler from tokenization import BertTokenizer logger = mge.get_logger(__name__) class DataProcessor: """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_tsv(cls, input_file, quotechar=None): """Reads a tab separated value file.""" with open(input_file, "r") as f: reader = csv.reader(f, delimiter="\t", quotechar=quotechar) lines = [] for line in reader: lines.append(line) return lines class InputFeatures: """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 InputExample: """A single training/test example for simple sequence classification.""" def __init__(self, guid, text_a, text_b=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. 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.label = label class MrpcProcessor(DataProcessor): """Processor for the MRPC data set (GLUE version).""" def get_train_examples(self, data_dir): """See base class.""" logger.info("LOOKING AT {}".format(os.path.join(data_dir, "train.tsv"))) return self._create_examples( self._read_tsv(os.path.join(data_dir, "train.tsv")), "train" ) 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" ) def get_labels(self): """See base class.""" return ["0", "1"] def _create_examples(self, lines, set_type): """Creates examples for the training and dev sets.""" examples = [] for (i, line) in enumerate(lines): if i == 0: continue guid = "%s-%s" % (set_type, i) text_a = line[3] text_b = line[4] label = line[0] examples.append( InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label) ) return examples 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 convert_examples_to_features(examples, label_list, max_seq_length, tokenizer): """Loads a data file into a list of `InputBatch`s.""" label_map = {label: i for i, label in enumerate(label_list)} features = [] for (ex_index, example) in enumerate(examples): tokens_a = tokenizer.tokenize(example.text_a) tokens_b = None if example.text_b: tokens_b = tokenizer.tokenize(example.text_b) # Modifies `tokens_a` and `tokens_b` in place so that the total # length is less than the specified length. # Account for [CLS], [SEP], [SEP] with "- 3" _truncate_seq_pair(tokens_a, tokens_b, max_seq_length - 3) 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 # # Where "type_ids" are used to indicate whether this is the first # sequence or the second 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 unambigiously 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: tokens += tokens_b + ["[SEP]"] segment_ids += [1] * (len(tokens_b) + 1) input_ids = tokenizer.convert_tokens_to_ids(tokens) # 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)) 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 < 0: logger.info("*** Example ***") logger.info("guid: {}".format(example.guid)) logger.info("tokens: {}".format(" ".join([str(x) for x in tokens]))) logger.info("input_ids: {}".format(" ".join([str(x) for x in input_ids]))) logger.info("input_mask: {}".format(" ".join([str(x) for x in input_mask]))) logger.info("segment_ids: {}".format(" ".join([str(x) for x in segment_ids]))) logger.info("label: {} (id = {})".format(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 class MRPCDataset: def __init__(self, args): self.args = args self.processor = MrpcProcessor() self.label_list = self.processor.get_labels() self.tokenizer = BertTokenizer( args.vocab_file, do_lower_case=args.do_lower_case ) def to_inputs(self, inp): return ( np.array([f.input_ids for f in inp]).astype(np.int32), np.array([f.input_mask for f in inp]).astype(np.float32), np.array([f.segment_ids for f in inp]).astype(np.int32), np.array([f.label_id for f in inp]).astype(np.int32), ) def get_dataloader(self, examples, batch_size, is_random=False): features = convert_examples_to_features( examples, self.label_list, self.args.max_seq_length, self.tokenizer ) all_input_ids, all_input_mask, all_segment_ids, all_label_ids = self.to_inputs( features ) dataset = ArrayDataset( all_input_ids, all_input_mask, all_segment_ids, all_label_ids ) if is_random: sampler = RandomSampler( dataset=dataset, batch_size=batch_size, drop_last=True ) else: sampler = SequentialSampler( dataset=dataset, batch_size=batch_size, drop_last=True ) dataloader = DataLoader(dataset=dataset, sampler=sampler,) return dataloader, len(features) def get_train_dataloader(self): examples = self.processor.get_train_examples(self.args.data_dir) return self.get_dataloader(examples, self.args.train_batch_size, is_random=True) def get_eval_dataloader(self): examples = self.processor.get_dev_examples(self.args.data_dir) return self.get_dataloader(examples, self.args.eval_batch_size, is_random=False)