pretrain.py

# Copyright 2018 Dong-Hyun Lee, Kakao Brain.

""" Pretrain transformer with Masked LM and Sentence Classification """

import random
from random import randint, shuffle
from random import random as rand
import json
import fire

import torch
import torch.nn as nn
from tensorboardX import SummaryWriter

import tokenization
import models
import optim
import train

from utils import set_seeds, get_device, get_random_word, truncate_tokens_pair

def get_random_word(vocab_words):
    i = random.randint(0, len(vocab_words)-1)
    return vocab_words[i]

def seek_random_offset(f, back_margin=2000):
    """ seek random offset of file pointer """
    f.seek(0, 2)
    # we remain some amount of text to read
    max_offset = f.tell() - back_margin
    f.seek(randint(0, max_offset), 0)
    f.readline() # throw away an incomplete sentence

class SentPairDataLoader():
    """ Load sentence pair (sequential or random order) from corpus """
    def __init__(self, file, batch_size, tokenize, max_len, short_sampling_prob=0.1, pipeline=[]):
        super().__init__()
        self.f_pos = open(file, 'r') # file pointer for a positive sample
        self.f_neg = open(file, 'r') # file pointer for a negative (random) sample
        self.tokenize = tokenize # tokenize function
        self.max_len = max_len # maximum length of tokens
        self.short_sampling_prob = short_sampling_prob
        self.pipeline = pipeline
        self.batch_size = batch_size

    def read_tokens(self, f, length, discard_last_and_restart=True):
        """ Read tokens from file pointer with limited length """
        tokens = []
        while len(tokens) < length:
            line = f.readline()
            if not line: # end of file
                return None
            if not line.strip(): # blank line (delimiter of documents)
                if discard_last_and_restart:
                    tokens = [] # throw all and restart
                    continue
                else:
                    return tokens # return last tokens in the document
            tokens.extend(self.tokenize(line.strip()))
        return tokens

    def __iter__(self): # iterator to load data
        while True:
            batch = []
            for i in range(self.batch_size):
                # sampling length of each tokens_a and tokens_b
                # sometimes sample a short sentence to match between train and test sequences
                len_tokens = randint(1, int(self.max_len / 2)) \
                    if rand() < self.short_sampling_prob \
                    else int(self.max_len / 2)

                is_next = rand() < 0.5 # whether token_b is next to token_a or not

                tokens_a = self.read_tokens(self.f_pos, len_tokens, True)
                seek_random_offset(self.f_neg)
                f_next = self.f_pos if is_next else self.f_neg
                tokens_b = self.read_tokens(f_next, len_tokens, False)

                if tokens_a is None or tokens_b is None: # end of file
                    self.f_pos.seek(0, 0) # reset file pointer
                    return

                instance = (is_next, tokens_a, tokens_b)
                for proc in self.pipeline:
                    instance = proc(instance)

                batch.append(instance)

            # To Tensor
            batch_tensors = [torch.tensor(x, dtype=torch.long) for x in zip(*batch)]
            yield batch_tensors


class Pipeline():
    """ Pre-process Pipeline Class : callable """
    def __init__(self):
        super().__init__()

    def __call__(self, instance):
        raise NotImplementedError


class Preprocess4Pretrain(Pipeline):
    """ Pre-processing steps for pretraining transformer """
    def __init__(self, max_pred, mask_prob, vocab_words, indexer, max_len=512):
        super().__init__()
        self.max_len = max_len
        self.max_pred = max_pred # max tokens of prediction
        self.mask_prob = mask_prob # masking probability
        self.vocab_words = vocab_words # vocabulary (sub)words
        self.indexer = indexer # function from token to token index
        self.max_len = max_len

    def __call__(self, instance):
        is_next, tokens_a, tokens_b = instance

        # -3  for special tokens [CLS], [SEP], [SEP]
        truncate_tokens_pair(tokens_a, tokens_b, self.max_len - 3)

        # Add Special Tokens
        tokens = ['[CLS]'] + tokens_a + ['[SEP]'] + tokens_b + ['[SEP]']
        segment_ids = [0]*(len(tokens_a)+2) + [1]*(len(tokens_b)+1)
        input_mask = [1]*len(tokens)

        # For masked Language Models
        masked_tokens, masked_pos = [], []
        # the number of prediction is sometimes less than max_pred when sequence is short
        n_pred = min(self.max_pred, max(1, int(round(len(tokens)*self.mask_prob))))
        # candidate positions of masked tokens
        cand_pos = [i for i, token in enumerate(tokens)
                    if tokens != '[CLS]' and tokens != '[SEP]']
        shuffle(cand_pos)
        for pos in cand_pos[:n_pred]:
            masked_tokens.append(tokens[pos])
            masked_pos.append(pos)
            if rand() < 0.8: # 80%
                tokens[pos] = '[MASK]'
            elif rand() < 0.5: # 10%
                tokens[pos] = get_random_word(self.vocab_words)
        # when n_pred < max_pred, we only calculate loss within n_pred
        masked_weights = [1]*len(masked_tokens)

        # Token Indexing
        input_ids = self.indexer(tokens)
        masked_ids = self.indexer(masked_tokens)

        # Zero Padding
        n_pad = self.max_len - len(input_ids)
        input_ids.extend([0]*n_pad)
        segment_ids.extend([0]*n_pad)
        input_mask.extend([0]*n_pad)

        # Zero Padding for masked target
        if self.max_pred > n_pred:
            n_pad = self.max_pred - n_pred
            masked_ids.extend([0]*n_pad)
            masked_pos.extend([0]*n_pad)
            masked_weights.extend([0]*n_pad)

        return (input_ids, segment_ids, input_mask, masked_ids, masked_pos, masked_weights, is_next)


class BertModel4Pretrain(nn.Module):
    "Bert Model for Pretrain : Masked LM and next sentence classification"
    def __init__(self, cfg):
        super().__init__()
        self.transformer = models.Transformer(cfg)
        self.fc = nn.Linear(cfg.dim, cfg.dim)
        self.activ1 = nn.Tanh()
        self.linear = nn.Linear(cfg.dim, cfg.dim)
        self.activ2 = models.gelu
        self.norm = models.LayerNorm(cfg)
        self.classifier = nn.Linear(cfg.dim, 2)
        # decoder is shared with embedding layer
        embed_weight = self.transformer.embed.tok_embed.weight
        n_vocab, n_dim = embed_weight.size()
        self.decoder = nn.Linear(n_dim, n_vocab, bias=False)
        self.decoder.weight = embed_weight
        self.decoder_bias = nn.Parameter(torch.zeros(n_vocab))

    def forward(self, input_ids, segment_ids, input_mask, masked_pos):
        h = self.transformer(input_ids, segment_ids, input_mask)
        pooled_h = self.activ1(self.fc(h[:, 0]))
        masked_pos = masked_pos[:, :, None].expand(-1, -1, h.size(-1))
        h_masked = torch.gather(h, 1, masked_pos)
        h_masked = self.norm(self.activ2(self.linear(h_masked)))
        logits_lm = self.decoder(h_masked) + self.decoder_bias
        logits_clsf = self.classifier(pooled_h)

        return logits_lm, logits_clsf


def main(train_cfg='config/pretrain.json',
         model_cfg='config/bert_base.json',
         data_file='../tbc/books_large_all.txt',
         model_file=None,
         data_parallel=True,
         vocab='../uncased_L-12_H-768_A-12/vocab.txt',
         save_dir='../exp/bert/pretrain',
         log_dir='../exp/bert/pretrain/runs',
         max_len=512,
         max_pred=20,
         mask_prob=0.15):

    cfg = Config(**json.load(open(config, "r")))
    cfg_optim = train.Config(**json.load(open(cfg.cfg_optim, "r")))
    cfg_model = models.Config(**json.load(open(cfg.cfg_model, "r")))

    set_seeds(cfg.seed)

    tokenizer = tokenization.FullTokenizer(vocab_file=vocab, do_lower_case=True)
    tokenize = lambda x: tokenizer.tokenize(tokenizer.convert_to_unicode(x))

    pipeline = [Preprocess4Pretrain(max_pred,
                                    mask_prob,
                                    list(tokenizer.vocab.keys()),
                                    tokenizer.convert_tokens_to_ids,
                                    max_len)]
    data_iter = SentPairDataLoader(data_file,
                                   cfg.batch_size,
                                   tokenize,
                                   max_len,
                                   pipeline=pipeline)

    model = BertModel4Pretrain(model_cfg)
    criterion1 = nn.CrossEntropyLoss(reduction='none')
    criterion2 = nn.CrossEntropyLoss()

    optimizer = optim.optim4GPU(cfg, model)
    trainer = train.Trainer(cfg, model, data_iter, optimizer, save_dir, get_device())

    writer = SummaryWriter(log_dir=log_dir) # for tensorboardX

    def get_loss(model, batch, global_step): # make sure loss is tensor
        input_ids, segment_ids, input_mask, masked_ids, masked_pos, masked_weights, is_next = batch

        logits_lm, logits_clsf = model(input_ids, segment_ids, input_mask, masked_pos)
        loss_lm = criterion1(logits_lm.transpose(1, 2), masked_ids) # for masked LM
        loss_lm = (loss_lm*masked_weights.float()).mean()
        loss_clsf = criterion2(logits_clsf, is_next) # for sentence classification
        writer.add_scalars('data/scalar_group',
                           {'loss_lm': loss_lm.item(),
                            'loss_clsf': loss_clsf.item(),
                            'loss_total': (loss_lm + loss_clsf).item(),
                            'lr': optimizer.get_lr()[0],
                           },
                           global_step)
        return loss_lm + loss_clsf

    trainer.train(get_loss, model_file, None, data_parallel)


if __name__ == '__main__':
    fire.Fire(main)