classify_xnli_paper.py

# coding=utf-8
""" Finetuning multi-lingual models on XNLI/PAWSX (Bert, XLM, XLMRoberta)."""
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
import argparse
import glob
import logging
import random

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

from transformers import (
    AdamW,
    WEIGHTS_NAME,
    BertConfig,
    BertTokenizer,
    get_linear_schedule_with_warmup,
)

from third_party.modeling_bert import BertForSequenceClassification
from third_party.processors.constants import *

from third_party.processors.utils_classify_codeswitch import (
    convert_examples_to_features,
    SequencePairDataset,
    batchify,
    XnliProcessor,
    PawsxProcessor
)

try:
    from torch.utils.tensorboard import SummaryWriter
except ImportError:
    from tensorboardX import SummaryWriter

logger = logging.getLogger(__name__)


MODEL_CLASSES = {
    "bert": (BertConfig, BertForSequenceClassification, BertTokenizer),
}

PROCESSORS = {
    "xnli": XnliProcessor,
    "pawsx": PawsxProcessor,
}


def str2bool(v):
    return v.lower() in ('yes', 'true', 't', '1', 'y')


def compute_metrics(preds, labels):
    scores = {
        "acc": (preds == labels).mean(),
        "num": len(preds),
        "correct": (preds == labels).sum(),
    }
    return scores


def set_seed(args):
    random.seed(args.seed)
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    if args.n_gpu > 0:
        torch.cuda.manual_seed_all(args.seed)


def train(args, train_dataset, model, tokenizer, lang2id=None):
    """Train the model."""
    if args.local_rank in [-1, 0]:
        tb_writer = SummaryWriter()

    args.train_batch_size = args.per_gpu_train_batch_size * max(1, args.n_gpu)
    train_sampler = (
        RandomSampler(train_dataset)
        if args.local_rank == -1
        else DistributedSampler(train_dataset)
    )
    train_dataloader = DataLoader(
        train_dataset,
        sampler=train_sampler,
        batch_size=args.train_batch_size,
        num_workers=4,
        pin_memory=True,
        collate_fn=batchify,
    )

    if args.max_steps > 0:
        t_total = args.max_steps
        args.num_train_epochs = (
                args.max_steps
                // (len(train_dataloader) // args.gradient_accumulation_steps)
                + 1
        )
    else:
        t_total = (
                len(train_dataloader)
                // args.gradient_accumulation_steps
                * args.num_train_epochs
        )

    # Prepare optimizer and schedule (linear warmup and decay)
    no_decay = ["bias", "LayerNorm.weight"]
    optimizer_grouped_parameters = [
        {
            "params": [
                p
                for n, p in model.named_parameters()
                if not any(nd in n for nd in no_decay)
            ],
            "weight_decay": args.weight_decay,
        },
        {
            "params": [
                p
                for n, p in model.named_parameters()
                if any(nd in n for nd in no_decay)
            ],
            "weight_decay": 0.0,
        },
    ]
    optimizer = AdamW(
        optimizer_grouped_parameters, lr=args.learning_rate, eps=args.adam_epsilon
    )
    scheduler = get_linear_schedule_with_warmup(
        optimizer, num_warmup_steps=args.warmup_steps, num_training_steps=t_total
    )

    # Check if saved optimizer or scheduler states exist
    if os.path.isfile(
            os.path.join(args.model_name_or_path, "optimizer.pt")
    ) and os.path.isfile(os.path.join(args.model_name_or_path, "scheduler.pt")):
        # Load in optimizer and scheduler states
        optimizer.load_state_dict(
            torch.load(os.path.join(args.model_name_or_path, "optimizer.pt"))
        )
        scheduler.load_state_dict(
            torch.load(os.path.join(args.model_name_or_path, "scheduler.pt"))
        )

    if args.fp16:
        try:
            from apex import amp
        except ImportError:
            raise ImportError(
                "Please install apex from https://www.github.com/nvidia/apex to use fp16 training."
            )
        model, optimizer = amp.initialize(
            model, optimizer, opt_level=args.fp16_opt_level
        )

    # multi-gpu training (should be after apex fp16 initialization)
    if args.n_gpu > 1:
        model = torch.nn.DataParallel(model)

    # Distributed training (should be after apex fp16 initialization)
    if args.local_rank != -1:
        model = torch.nn.parallel.DistributedDataParallel(
            model,
            device_ids=[args.local_rank],
            output_device=args.local_rank,
            find_unused_parameters=True,
        )

    # Train!
    logger.info("***** Running training *****")
    logger.info("  Num examples = %d", len(train_dataset))
    logger.info("  Num Epochs = %d", args.num_train_epochs)
    logger.info(
        "  Instantaneous batch size per GPU = %d", args.per_gpu_train_batch_size
    )
    logger.info(
        "  Total train batch size (w. parallel, distributed & accumulation) = %d",
        args.train_batch_size
        * args.gradient_accumulation_steps
        * (torch.distributed.get_world_size() if args.local_rank != -1 else 1),
    )
    logger.info("  Gradient Accumulation steps = %d", args.gradient_accumulation_steps)
    logger.info("  Total optimization steps = %d", t_total)

    global_step = 0
    epochs_trained = 0
    steps_trained_in_current_epoch = 0
    # Check if continuing training from a checkpoint init_checkpoint
    # if os.path.exists(args.model_name_or_path):
    if args.init_checkpoint and os.path.exists(args.init_checkpoint):
        # set global_step to gobal_step of last saved checkpoint from model path
        # global_step = int(args.model_name_or_path.split("-")[-1].split("/")[0])
        global_step = int(args.init_checkpoint.split("-")[-1].split("/")[0])
        epochs_trained = global_step // (
                len(train_dataloader) // args.gradient_accumulation_steps
        )
        steps_trained_in_current_epoch = global_step % (
                len(train_dataloader) // args.gradient_accumulation_steps
        )

        logger.info(
            "  Continuing training from checkpoint, will skip to saved global_step"
        )
        logger.info("  Continuing training from epoch %d", epochs_trained)
        logger.info("  Continuing training from global step %d", global_step)
        logger.info(
            "  Will skip the first %d steps in the first epoch",
            steps_trained_in_current_epoch,
        )

    best_score = 0
    best_checkpoint = None
    tr_loss, logging_loss = 0.0, 0.0
    model.zero_grad()
    train_iterator = trange(
        epochs_trained,
        int(args.num_train_epochs),
        desc="Epoch",
        disable=args.local_rank not in [-1, 0],
    )
    set_seed(args)  # Added here for reproductibility
    accum_loss, loss_count = 0, 0
    for _ in train_iterator:
        epoch_iterator = tqdm(
            train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0]
        )
        for step, batch in enumerate(epoch_iterator):
            # Skip past any already trained steps if resuming training
            if steps_trained_in_current_epoch > 0:
                steps_trained_in_current_epoch -= 1
                continue

            model.train()
            batch = tuple(t.to(args.device) if t is not None else None for t in batch)

            inputs = dict()
            inputs['input_ids'] = batch[0]
            inputs['attention_mask'] = batch[1]
            inputs["token_type_ids"] = batch[2] if args.model_type in ["bert"] else None
            inputs['labels'] = batch[3]

            if args.use_syntax:
                inputs["dep_tag_ids"] = batch[4]
                inputs["pos_tag_ids"] = batch[5]
                inputs["dist_mat"] = batch[6]
                inputs["tree_depths"] = batch[7]

            outputs = model(**inputs)
            outputs = outputs[0]

            loss = outputs[0]

            if args.n_gpu > 1:
                loss = loss.mean()  # mean() to average on multi-gpu parallel training
            if args.gradient_accumulation_steps > 1:
                loss = loss / args.gradient_accumulation_steps

            if args.fp16:
                with amp.scale_loss(loss, optimizer) as scaled_loss:
                    scaled_loss.backward()
            else:
                loss.backward()

            tr_loss += loss.item()

            accum_loss += loss.item() * batch[0].size(0)
            loss_count += batch[0].size(0)
            log_info = 'Iteration [%.3f]' % (accum_loss / loss_count)
            epoch_iterator.set_description("%s" % log_info)

            if (step + 1) % args.gradient_accumulation_steps == 0:
                if args.fp16:
                    torch.nn.utils.clip_grad_norm_(
                        amp.master_params(optimizer), args.max_grad_norm
                    )
                else:
                    torch.nn.utils.clip_grad_norm_(
                        model.parameters(), args.max_grad_norm
                    )

                optimizer.step()
                scheduler.step()  # Update learning rate schedule
                model.zero_grad()
                global_step += 1

                if (
                        args.local_rank in [-1, 0]
                        and args.logging_steps > 0
                        and global_step % args.logging_steps == 0
                ):
                    # Log metrics
                    tb_writer.add_scalar("lr", scheduler.get_lr()[0], global_step)
                    tb_writer.add_scalar(
                        "loss",
                        (tr_loss - logging_loss) / args.logging_steps,
                        global_step,
                    )
                    logging_loss = tr_loss

                    # Only evaluate on single GPU otherwise metrics may not average well
                    if args.local_rank == -1 and args.evaluate_during_training:
                        results = evaluate(
                            args,
                            model,
                            tokenizer,
                            split=args.train_split,
                            language=args.train_language,
                            lang2id=lang2id,
                        )
                        for key, value in results.items():
                            tb_writer.add_scalar(
                                "eval_{}".format(key), value, global_step
                            )

                if (
                        args.local_rank in [-1, 0]
                        and args.save_steps > 0
                        and global_step % args.save_steps == 0
                ):
                    if args.eval_test_set:
                        output_predict_file = os.path.join(
                            args.output_dir, "eval_test_results.txt"
                        )
                        total = total_correct = 0.0
                        with open(output_predict_file, "a") as writer:
                            writer.write(
                                "\n======= Predict using the model from checkpoint-{}:\n".format(
                                    global_step
                                )
                            )
                            for language in args.predict_languages.split(","):
                                result = evaluate(
                                    args,
                                    model,
                                    tokenizer,
                                    split=args.test_split,
                                    language=language,
                                    lang2id=lang2id,
                                    prefix="checkpoint-" + str(global_step),
                                )
                                writer.write("{}={}\n".format(language, result["acc"]))
                                total += result["num"]
                                total_correct += result["correct"]
                            writer.write("total={}\n".format(total_correct / total))

                    if args.save_only_best_checkpoint:
                        if args.valid_languages:
                            total = total_correct = 0.0
                            for language in args.valid_languages.split(","):
                                result = evaluate(
                                    args,
                                    model,
                                    tokenizer,
                                    split="dev",
                                    language=language,
                                    lang2id=lang2id,
                                    prefix="checkpoint-" + str(global_step),
                                )
                                total += result["num"]
                                total_correct += result["correct"]
                            accuracy = total_correct / total
                        else:
                            result = evaluate(
                                args,
                                model,
                                tokenizer,
                                split="dev",
                                language=args.train_language,
                                lang2id=lang2id,
                                prefix=str(global_step),
                            )
                            accuracy = result["acc"]
                            logger.info(
                                " Dev accuracy {} = {}".format(
                                    args.train_language, result["acc"]
                                )
                            )
                        if accuracy > best_score:
                            logger.info(
                                " result['acc']={} > best_score={}".format(
                                    accuracy, best_score
                                )
                            )
                            output_dir = os.path.join(
                                args.output_dir, "checkpoint-best"
                            )
                            best_checkpoint = output_dir
                            best_score = accuracy
                            # Save model checkpoint
                            if not os.path.exists(output_dir):
                                os.makedirs(output_dir)
                            model_to_save = (
                                model.module if hasattr(model, "module") else model
                            )  # Take care of distributed/parallel training
                            model_to_save.save_pretrained(output_dir)
                            tokenizer.save_pretrained(output_dir)

                            torch.save(
                                args, os.path.join(output_dir, "training_args.bin")
                            )
                            logger.info("Saving model checkpoint to %s", output_dir)

                            torch.save(
                                optimizer.state_dict(),
                                os.path.join(output_dir, "optimizer.pt"),
                            )
                            torch.save(
                                scheduler.state_dict(),
                                os.path.join(output_dir, "scheduler.pt"),
                            )
                            logger.info(
                                "Saving optimizer and scheduler states to %s",
                                output_dir,
                            )
                    else:
                        # Save model checkpoint
                        output_dir = os.path.join(
                            args.output_dir, "checkpoint-{}".format(global_step)
                        )
                        if not os.path.exists(output_dir):
                            os.makedirs(output_dir)
                        model_to_save = (
                            model.module if hasattr(model, "module") else model
                        )  # Take care of distributed/parallel training
                        model_to_save.save_pretrained(output_dir)
                        tokenizer.save_pretrained(output_dir)

                        torch.save(args, os.path.join(output_dir, "training_args.bin"))
                        logger.info("Saving model checkpoint to %s", output_dir)

                        torch.save(
                            optimizer.state_dict(),
                            os.path.join(output_dir, "optimizer.pt"),
                        )
                        torch.save(
                            scheduler.state_dict(),
                            os.path.join(output_dir, "scheduler.pt"),
                        )
                        logger.info(
                            "Saving optimizer and scheduler states to %s", output_dir
                        )

            if args.max_steps > 0 and global_step > args.max_steps:
                epoch_iterator.close()
                break
        if args.max_steps > 0 and global_step > args.max_steps:
            train_iterator.close()
            break

    if args.local_rank in [-1, 0]:
        tb_writer.close()

    return global_step, tr_loss / global_step, best_score, best_checkpoint


def evaluate(
        args,
        model,
        tokenizer,
        split="train",
        language="en",
        lang2id=None,
        prefix="",
        output_file=None,
        label_list=None,
        output_only_prediction=True,
):
    """Evalute the model."""
    eval_task_names = (args.task_name,)
    eval_outputs_dirs = (args.output_dir,)

    results = {}
    for eval_task, eval_output_dir in zip(eval_task_names, eval_outputs_dirs):
        eval_dataset = load_and_cache_examples(
            args,
            eval_task,
            tokenizer,
            split=split,
            language=language,
            lang2id=lang2id,
            evaluate=True,
        )

        if not os.path.exists(eval_output_dir) and args.local_rank in [-1, 0]:
            os.makedirs(eval_output_dir)

        args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
        # Note that DistributedSampler samples randomly
        eval_sampler = SequentialSampler(eval_dataset)
        eval_dataloader = DataLoader(
            eval_dataset,
            sampler=eval_sampler,
            batch_size=args.eval_batch_size,
            num_workers=4,
            pin_memory=True,
            collate_fn=batchify,
        )

        # multi-gpu eval
        if args.n_gpu > 1 and not isinstance(model, torch.nn.DataParallel):
            model = torch.nn.DataParallel(model)

        # Eval!
        logger.info("***** Running evaluation {} {} *****".format(prefix, language))
        logger.info("  Num examples = %d", len(eval_dataset))
        logger.info("  Batch size = %d", args.eval_batch_size)
        eval_loss = 0.0
        nb_eval_steps = 0
        preds = None
        out_label_ids = None
        sentences = None
        for batch in tqdm(eval_dataloader, desc="Evaluating"):
            model.eval()
            batch = tuple(t.to(args.device) for t in batch if t is not None)

            with torch.no_grad():
                inputs = dict()
                inputs['input_ids'] = batch[0]
                inputs['attention_mask'] = batch[1]
                inputs["token_type_ids"] = batch[2] if args.model_type in ["bert"] else None
                inputs['labels'] = batch[3]

                if args.use_syntax:
                    inputs["dep_tag_ids"] = batch[4]
                    inputs["pos_tag_ids"] = batch[5]
                    inputs["dist_mat"] = batch[6]
                    inputs["tree_depths"] = batch[7]

                outputs = model(**inputs)
                tmp_eval_loss, logits = outputs[:2]

                eval_loss += tmp_eval_loss.mean().item()
            nb_eval_steps += 1

            if preds is None:
                preds = logits.detach().cpu().numpy()
                out_label_ids = inputs["labels"].detach().cpu().numpy()
                sentences = inputs["input_ids"].detach().cpu().numpy()
            else:
                preds = np.append(preds, logits.detach().cpu().numpy(), axis=0)
                out_label_ids = np.append(
                    out_label_ids, inputs["labels"].detach().cpu().numpy(), axis=0
                )
                sentences = np.append(
                    sentences, inputs["input_ids"].detach().cpu().numpy(), axis=0
                )

        eval_loss = eval_loss / nb_eval_steps
        if args.output_mode == "classification":
            pred_label_ids = np.argmax(preds, axis=1)
        else:
            raise ValueError("No other `output_mode` for XNLI.")
        result = compute_metrics(pred_label_ids, out_label_ids)
        results.update(result)

        if output_file:
            logger.info("***** Save prediction ******")
            with open(output_file, "w") as fout:
                pad_token_id = tokenizer.pad_token_id
                sentences = sentences.astype(int).tolist()
                sentences = [[w for w in s if w != pad_token_id] for s in sentences]
                sentences = [tokenizer.convert_ids_to_tokens(s) for s in sentences]
                for idx, (p, l, s) in enumerate(
                        zip(list(pred_label_ids), list(out_label_ids), sentences)
                ):
                    s = " ".join(s)
                    if label_list:
                        p = label_list[p]
                        l = label_list[l]
                    if output_only_prediction:
                        fout.write(str(p) + "\n")
                    else:
                        score = np.around(preds[idx], decimals=4)
                        score = ";".join([str(sc) for sc in score])
                        fout.write("{}\t{}\t{}\t{}\n".format(p, l, score, s))
        logger.info("***** Eval results {} {} *****".format(prefix, language))
        for key in sorted(result.keys()):
            logger.info("  %s = %s", key, str(result[key]))

    return results


def load_and_cache_examples(
        args, task, tokenizer, split="train", language="en",
        lang2id=None, evaluate=False, few_shot=-1
):
    # Make sure only the first process in distributed training process the
    # dataset, and the others will use the cache
    if args.local_rank not in [-1, 0] and not evaluate:
        torch.distributed.barrier()

    processor = PROCESSORS[task](args.covert_rate)
    output_mode = "classification"
    # Load data features from cache or dataset file
    lc = "_lc" if args.do_lower_case else ""
    cached_features_file = os.path.join(
        args.output_dir,
        "cached_{}_{}_{}_{}_{}{}".format(
            split,
            list(filter(None, args.model_name_or_path.split("/"))).pop(),
            str(args.max_seq_length),
            str(task),
            str('-'.join(language.split(','))),
            lc,
        ),
    )
    if os.path.exists(cached_features_file) and not args.overwrite_cache:
        logger.info("Loading features from cached file %s", cached_features_file)
        features = torch.load(cached_features_file)
    else:
        logger.info("Creating features from dataset file at %s", args.data_dir)
        label_list = processor.get_labels()
        if split == "train":
            examples = processor.get_train_examples(args.data_dir, language, args.swap_pairs)
        elif split == "translate-train":
            examples = processor.get_translate_train_examples(args.data_dir, language)
        elif split == "translate-test":
            examples = processor.get_translate_test_examples(args.data_dir, language)
        elif split == "dev":
            examples = processor.get_dev_examples(args.data_dir, language)
        elif split == "pseudo_test":
            examples = processor.get_pseudo_test_examples(args.data_dir, language)
        else:
            examples = processor.get_test_examples(args.data_dir, language, args.swap_pairs)

        if few_shot > 0 and split == 'train':
            logger.info("Original no. of examples = {}".format(len(examples)))
            examples = examples[: few_shot]
            logger.info('Using few-shot learning on {} examples'.format(len(examples)))

        features = convert_examples_to_features(
            examples,
            tokenizer,
            label_list=label_list,
            max_length=args.max_seq_length,
            output_mode=output_mode,
            sep_token_extra=bool(args.model_type in ["roberta", "xlm-roberta"]),
            pad_on_left=False,
            pad_token=tokenizer.convert_tokens_to_ids([tokenizer.pad_token])[0],
            pad_token_segment_id=0,
            lang2id=lang2id,
            use_syntax=args.use_syntax,
        )

        if args.local_rank in [-1, 0]:
            logger.info("Saving features into cached file %s", cached_features_file)
            torch.save(features, cached_features_file)

    # Make sure only the first process in distributed training process the
    # dataset, and the others will use the cache
    if args.local_rank == 0 and not evaluate:
        torch.distributed.barrier()

    dataset = SequencePairDataset(features)
    return dataset


def main():
    parser = argparse.ArgumentParser()
    parser.register('type', 'bool', str2bool)

    # Required parameters
    parser.add_argument(
        "--data_dir",
        default="./download/xnli_udpipe_processed_es",
        type=str,
        help="The input data dir. Should contain the .tsv files (or other data files) for the task.",
    )
    parser.add_argument(
        "--model_type",
        default="bert",
        type=str,
        help="Model type selected in the list: " + ", ".join(MODEL_CLASSES.keys()),
    )
    parser.add_argument(
        "--model_name_or_path",
        default="../bert-base-multilingual-cased",
        type=str,
        help="Path to pre-trained model or shortcut name selected in the list: "
    )
    parser.add_argument("--covert_rate", default=0.3, type=float, help="Covert rate")
    parser.add_argument(
        "--train_language",
        default="en",
        type=str,
        help="Train language if is different of the evaluation language.",
    )
    parser.add_argument(
        "--predict_languages",
        type=str,
        default="ar,bg,de,el,en,es,fr,hi,ru,tr,ur,vi,zh",
        help="prediction languages separated by ','.",
    )
    parser.add_argument(
        "--valid_languages",
        type=str,
        default=None,
        help="validation languages separated by ','.",
    )
    parser.add_argument(
        "--output_dir",
        default="./outputs/xnli_syn_codeswitch_0830",
        type=str,
        help="The output directory where the model predictions and checkpoints will be written.",
    )
    parser.add_argument(
        "--task_name",
        default="xnli",
        type=str,
        help="The task name",
    )

    # Other parameters
    parser.add_argument(
        "--config_name",
        default="",
        type=str,
        help="Pretrained config name or path if not the same as model_name",
    )
    parser.add_argument(
        "--tokenizer_name",
        default="",
        type=str,
        help="Pretrained tokenizer name or path if not the same as model_name",
    )
    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 tokenization. Sequences longer "
             "than this will be truncated, sequences shorter will be padded.",
    )
    parser.add_argument(
        "--do_train", default=True, help="Whether to run training."
    )
    parser.add_argument(
        "--do_eval", action="store_true", help="Whether to run eval on the test set."
    )
    parser.add_argument(
        "--do_predict", default=True, help="Whether to run prediction."
    )
    parser.add_argument(
        "--do_predict_dev", action="store_true", help="Whether to run prediction."
    )
    parser.add_argument(
        "--init_checkpoint",
        type=str,
        default=None,
        help="initial checkpoint for predicting the dev set",
    )
    parser.add_argument(
        "--evaluate_during_training",
        action="store_true",
        help="Rul evaluation during training at each logging step.",
    )
    parser.add_argument(
        "--do_lower_case",
        action="store_true",
        help="Set this flag if you are using an uncased model.",
    )
    parser.add_argument(
        "--few_shot", default=-1, type=int, help="num of few-shot exampes"
    )
    parser.add_argument(
        "--train_split", type=str, default="train", help="split of training set"
    )
    parser.add_argument(
        "--test_split", type=str, default="test", help="split of training set"
    )

    parser.add_argument(
        "--per_gpu_train_batch_size",
        default=32,
        type=int,
        help="Batch size per GPU/CPU for training.",
    )
    parser.add_argument(
        "--per_gpu_eval_batch_size",
        default=32,
        type=int,
        help="Batch size per GPU/CPU for evaluation.",
    )
    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(
        "--learning_rate",
        default=2e-5,
        type=float,
        help="The initial learning rate for Adam.",
    )
    parser.add_argument(
        "--weight_decay", default=0.0, type=float, help="Weight deay if we apply some."
    )
    parser.add_argument(
        "--adam_epsilon", default=1e-8, type=float, help="Epsilon for Adam optimizer."
    )
    parser.add_argument(
        "--max_grad_norm", default=1.0, type=float, help="Max gradient norm."
    )
    parser.add_argument(
        "--num_train_epochs",
        default=10.0,
        type=float,
        help="Total number of training epochs to perform.",
    )
    parser.add_argument(
        "--max_steps",
        default=-1,
        type=int,
        help="If > 0: set total number of training steps to perform. Override num_train_epochs.",
    )
    parser.add_argument(
        "--warmup_steps", default=0, type=int, help="Linear warmup over warmup_steps."
    )

    parser.add_argument(
        "--logging_steps", type=int, default=50, help="Log every X updates steps."
    )
    parser.add_argument("--log_file", default="train", type=str, help="log file")
    parser.add_argument(
        "--save_steps",
        type=int,
        default=2000,
        help="Save checkpoint every X updates steps.",
    )
    parser.add_argument(
        "--eval_all_checkpoints",
        default=True,
        help="Evaluate all checkpoints starting with the same prefix as model_name ending and ending with step number",
    )
    parser.add_argument(
        "--no_cuda", action="store_true", help="Avoid using CUDA when available"
    )
    parser.add_argument(
        "--overwrite_output_dir",
        default=True,
        help="Overwrite the content of the output directory",
    )
    parser.add_argument(
        "--overwrite_cache",
        action="store_true",
        help="Overwrite the cached training and evaluation sets",
    )
    parser.add_argument(
        "--seed", type=int, default=42, help="random seed for initialization"
    )

    parser.add_argument(
        "--fp16",
        action="store_true",
        help="Whether to use 16-bit (mixed) precision (through NVIDIA apex) instead of 32-bit",
    )
    parser.add_argument(
        "--eval_test_set",
        action="store_true",
        help="Whether to evaluate test set durinng training",
    )
    parser.add_argument(
        "--fp16_opt_level",
        type=str,
        default="O1",
        help="For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
             "See details at https://nvidia.github.io/apex/amp.html",
    )
    parser.add_argument(
        "--save_only_best_checkpoint",
        default=True,
        help="save only the best checkpoint",
    )
    parser.add_argument(
        "--local_rank",
        type=int,
        default=-1,
        help="For distributed training: local_rank",
    )
    parser.add_argument(
        "--server_ip", type=str, default="", help="For distant debugging."
    )
    parser.add_argument(
        "--server_port", type=str, default="", help="For distant debugging."
    )

    ######################################
    parser.add_argument(
        "--syntactic_layers",
        type=str, default='0,1,2,3,4,5,6,7,8,9,10,11',
        help="comma separated layer indices for syntax fusion",
    )
    parser.add_argument(
        "--num_syntactic_heads",
        default=1, type=int,
        help="Number of syntactic heads",
    )
    parser.add_argument(
        "--use_syntax",
        type=bool,
        default=True,
        help="Whether to use syntax-based modeling",
    )
    parser.add_argument(
        "--freeze_word_embeddings",
        action="store_true",
        help="Whether to freeze word embeddings",
    )
    parser.add_argument(
        "--freeze_gat",
        action="store_true",
        help="Freeze the graph attention networks",
    )
    parser.add_argument(
        "--use_dependency_tag",
        type=bool,
        default=False,
        help="Whether to use dependency tag in structure modeling",
    )
    parser.add_argument(
        "--use_pos_tag",
        type=bool,
        default=True,
        help="Whether to use pos tags in structure modeling",
    )
    parser.add_argument(
        "--use_structural_loss",
        type=bool,
        default=True,
        help="Whether to use structural loss along with task loss",
    )
    parser.add_argument(
        "--struct_loss_coeff",
        default=1.0, type=float,
        help="Multiplying factor for the structural loss",
    )
    parser.add_argument(
        "--max_syntactic_distance",
        default=4, type=int,
        help="Max distance to consider during graph attention",
    )
    parser.add_argument(
        "--num_gat_layer",
        default=4, type=int,
        help="Number of layers in Graph Attention Networks (GAT)",
    )
    parser.add_argument(
        "--num_gat_head",
        default=4, type=int,
        help="Number of attention heads in Graph Attention Networks (GAT)",
    )
    parser.add_argument(
        "--pretrained_gat",
        type=str, default=None,
        help="Path of the pretrained GAT model",
    )
    parser.add_argument(
        "--batch_normalize",
        action="store_true",
        help="Apply batch normalization to <s> representation",
    )
    parser.add_argument(
        "--swap_pairs",
        action="store_true",
        help="Swap the input sentence pairs.",
    )

    args = parser.parse_args()

    args.log_file = "train.log"

    if not os.path.exists(args.output_dir):
        os.makedirs(args.output_dir, exist_ok=True)

    if (
            os.path.exists(args.output_dir)
            and os.listdir(args.output_dir)
            and args.do_train
            and not args.overwrite_output_dir
    ):
        raise ValueError(
            "Output directory ({}) already exists and is not empty. Use --overwrite_output_dir to overcome.".format(
                args.output_dir
            )
        )

    logging.basicConfig(
        handlers=[
            logging.FileHandler(os.path.join(args.output_dir, args.log_file)),
            logging.StreamHandler(),
        ],
        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,
    )
    logging.info("Input args: %r" % args)

    # Setup distant debugging if needed
    if args.server_ip and args.server_port:
        import ptvsd

        print("Waiting for debugger attach")
        ptvsd.enable_attach(
            address=(args.server_ip, args.server_port), redirect_output=True
        )
        ptvsd.wait_for_attach()

    # Setup CUDA, GPU & distributed training
    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"
        )
        args.n_gpu = torch.cuda.device_count()
    else:  # Initializes the distributed backend which sychronizes nodes/GPUs
        torch.cuda.set_device(args.local_rank)
        device = torch.device("cuda", args.local_rank)
        torch.distributed.init_process_group(backend="nccl")
        args.n_gpu = 1
    args.device = device

    # Setup logging
    logger.warning(
        "Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s",
        args.local_rank,
        device,
        args.n_gpu,
        bool(args.local_rank != -1),
        args.fp16,
    )

    # Set seed
    set_seed(args)

    # Prepare dataset
    if args.task_name not in PROCESSORS:
        raise ValueError("Task not found: %s" % (args.task_name))
    processor = PROCESSORS[args.task_name](args.covert_rate)
    args.output_mode = "classification"
    label_list = processor.get_labels()
    num_labels = len(label_list)

    # Load pretrained model and tokenizer
    # Make sure only the first process in distributed training loads model & vocab
    if args.local_rank not in [-1, 0]:
        torch.distributed.barrier()

    args.model_type = args.model_type.lower()
    config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
    config = config_class.from_pretrained(
        args.config_name if args.config_name else args.model_name_or_path,
        num_labels=num_labels,
        finetuning_task=args.task_name,
        cache_dir=args.cache_dir if args.cache_dir else None,
    )

    ####################################
    config.dep_tag_vocab_size = len(DEPTAG_SYMBOLS) + NUM_SPECIAL_TOKENS
    config.pos_tag_vocab_size = len(POS_SYMBOLS) + NUM_SPECIAL_TOKENS
    config.use_dependency_tag = args.use_dependency_tag
    config.use_pos_tag = args.use_pos_tag
    config.use_structural_loss = args.use_structural_loss
    config.struct_loss_coeff = args.struct_loss_coeff
    config.num_syntactic_heads = args.num_syntactic_heads
    config.syntactic_layers = args.syntactic_layers
    config.max_syntactic_distance = args.max_syntactic_distance
    config.use_syntax = args.use_syntax
    config.batch_normalize = args.batch_normalize
    config.num_gat_layer = args.num_gat_layer
    config.num_gat_head = args.num_gat_head
    ####################################

    logger.info("config = {}".format(config))

    tokenizer = tokenizer_class.from_pretrained(
        args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
        do_lower_case=args.do_lower_case,
        cache_dir=args.cache_dir if args.cache_dir else None,
    )

    lang2id = config.lang2id if args.model_type == "xlm" else None
    logger.info("lang2id = {}".format(lang2id))

    # Make sure only the first process in distributed training loads model & vocab
    if args.local_rank == 0:
        torch.distributed.barrier()
    logger.info("Training/evaluation parameters %s", args)

    print("###############  Epoch： ",args.num_train_epochs)

    #"""
    # Training
    if args.do_train:
        if args.init_checkpoint:
            logger.info("loading from folder {}".format(args.init_checkpoint))
            model = model_class.from_pretrained(
                args.init_checkpoint,
                config=config,
                cache_dir=args.init_checkpoint,
            )
        else:
            logger.info(
                "loading from existing model {}".format(args.model_name_or_path)
            )
            model = model_class.from_pretrained(
                args.model_name_or_path,
                from_tf=bool(".ckpt" in args.model_name_or_path),
                config=config,
                cache_dir=args.cache_dir if args.cache_dir else None,
            )

        if args.freeze_word_embeddings:
            if args.model_type == 'bert':
                model.bert.embeddings.word_embeddings.weight.requires_grad = False

        if args.use_syntax:
            if args.pretrained_gat:
                state_dict = torch.load(
                    args.pretrained_gat, map_location=lambda storage, loc: storage
                )
                model.bert.encoder.gat_model.load_state_dict(state_dict)
                logger.info(" the gat model states are loaded from %s", args.pretrained_gat)

            if args.freeze_gat:
                for p in model.bert.encoder.gat_model.parameters():
                    p.requires_grad = False

        model.to(args.device)
        train_dataset = load_and_cache_examples(
            args,
            args.task_name,
            tokenizer,
            split=args.train_split,
            language=args.train_language,
            lang2id=lang2id,
            evaluate=False,
            few_shot=args.few_shot
        )
        global_step, tr_loss, best_score, best_checkpoint = train(
            args, train_dataset, model, tokenizer, lang2id
        )
        logger.info(" global_step = %s, average loss = %s", global_step, tr_loss)
        logger.info(
            " best checkpoint = {}, best score = {}".format(best_checkpoint, best_score)
        )

    # Saving best-practices: if you use defaults names for the model, you can reload it using from_pretrained()
    if args.do_train and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
        print("#####################################################################################################")
        print("#####################################################################################################")
        print("###  do_train")
        # Create output directory if needed
        if not os.path.exists(args.output_dir) and args.local_rank in [-1, 0]:
            os.makedirs(args.output_dir)

        logger.info("Saving model checkpoint to %s", args.output_dir)
        # Save a trained model, configuration and tokenizer using `save_pretrained()`.
        # They can then be reloaded using `from_pretrained()`
        model_to_save = (
            model.module if hasattr(model, "module") else model
        )  # Take care of distributed/parallel training
        model_to_save.save_pretrained(args.output_dir)
        tokenizer.save_pretrained(args.output_dir)

        # Good practice: save your training arguments together with the trained model
        torch.save(args, os.path.join(args.output_dir, "training_args.bin"))

        # Load a trained model and vocabulary that you have fine-tuned
        model = model_class.from_pretrained(args.output_dir)
        tokenizer = tokenizer_class.from_pretrained(args.output_dir)
        model.to(args.device)
    #"""
    # Evaluation
    results = {}
    if args.init_checkpoint:
        best_checkpoint = args.init_checkpoint
    elif os.path.exists(os.path.join(args.output_dir, "checkpoint-best")):
        best_checkpoint = os.path.join(args.output_dir, "checkpoint-best")
    else:
        best_checkpoint = args.output_dir

    best_score = 0
    if args.do_eval and args.local_rank in [-1, 0]:
        print("#####################################################################################################")
        print("#####################################################################################################")
        print("###  do_eval")
        tokenizer = tokenizer_class.from_pretrained(
            args.output_dir, do_lower_case=args.do_lower_case
        )
        checkpoints = [args.output_dir]
        if args.eval_all_checkpoints:
            checkpoints = list(
                os.path.dirname(c)
                for c in sorted(
                    glob.glob(args.output_dir + "/**/" + WEIGHTS_NAME, recursive=True)
                )
            )
            logging.getLogger("transformers.modeling_utils").setLevel(
                logging.WARN
            )  # Reduce logging
        logger.info("Evaluate the following checkpoints: %s", checkpoints)
        for checkpoint in checkpoints:
            global_step = checkpoint.split("-")[-1] if len(checkpoints) > 1 else ""
            prefix = (
                checkpoint.split("/")[-1] if checkpoint.find("checkpoint") != -1 else ""
            )

            model = model_class.from_pretrained(checkpoint)
            model.to(args.device)
            result = evaluate(
                args,
                model,
                tokenizer,
                split="dev",
                language=args.train_language,
                lang2id=lang2id,
                prefix=prefix,
            )
            if result["acc"] > best_score:
                best_checkpoint = checkpoint
                best_score = result["acc"]
            result = dict((k + "_{}".format(global_step), v) for k, v in result.items())
            results.update(result)

        output_eval_file = os.path.join(args.output_dir, "eval_results")
        with open(output_eval_file, "w") as writer:
            for key, value in results.items():
                writer.write("{} = {}\n".format(key, value))
            writer.write(
                "Best checkpoint is {}, best accuracy is {}".format(
                    best_checkpoint, best_score
                )
            )
            logger.info(
                "Best checkpoint is {}, best accuracy is {}".format(
                    best_checkpoint, best_score
                )
            )

    # Prediction
    if args.do_predict and args.local_rank in [-1, 0]:
        print("#####################################################################################################")
        print("#####################################################################################################")
        print("###  do_predict")
        tokenizer = tokenizer_class.from_pretrained(
            args.model_name_or_path if args.model_name_or_path else best_checkpoint,
            do_lower_case=args.do_lower_case,
        )
        model = model_class.from_pretrained(best_checkpoint)
        model.to(args.device)
        output_predict_file = os.path.join(
            args.output_dir, args.test_split + "_results.txt"
        )
        total = total_correct = 0.0
        with open(output_predict_file, "a") as writer:
            writer.write(
                "======= Predict using the model from {} for {}:\n".format(
                    best_checkpoint, args.test_split
                )
            )
            for language in args.predict_languages.split(","):
                output_file = os.path.join(
                    args.output_dir, "test-{}.tsv".format(language)
                )
                result = evaluate(
                    args,
                    model,
                    tokenizer,
                    split=args.test_split,
                    language=language,
                    lang2id=lang2id,
                    prefix="best_checkpoint",
                    output_file=output_file,
                    label_list=label_list,
                    output_only_prediction=False,
                )
                # writer.write('{}={}\n'.format(language, result['acc']))
                logger.info("{}={}".format(language, result["acc"]))
                writer.write("=====================\nlanguage={}\n".format(language))
                for key in sorted(result.keys()):
                    writer.write("{} = {}\n".format(key, result[key]))
                total += result["num"]
                total_correct += result["correct"]
            writer.write("=====================\n")
            writer.write("total={}\n".format(total_correct / total))

    if args.do_predict_dev:
        tokenizer = tokenizer_class.from_pretrained(
            args.model_name_or_path if args.model_name_or_path else best_checkpoint,
            do_lower_case=args.do_lower_case,
        )
        model = model_class.from_pretrained(args.init_checkpoint)
        model.to(args.device)
        output_predict_file = os.path.join(args.output_dir, "dev_results")
        total = total_correct = 0.0
        with open(output_predict_file, "w") as writer:
            writer.write(
                "======= Predict using the model from {}:\n".format(
                    args.init_checkpoint
                )
            )
            for language in args.predict_languages.split(","):
                output_file = os.path.join(
                    args.output_dir, "dev-{}.tsv".format(language)
                )
                result = evaluate(
                    args,
                    model,
                    tokenizer,
                    split="dev",
                    language=language,
                    lang2id=lang2id,
                    prefix="best_checkpoint",
                    output_file=output_file,
                    label_list=label_list,
                )
                # writer.write('{}={}\n'.format(language, result['acc']))
                writer.write("=====================\nlanguage={}\n".format(language))
                for key in sorted(result.keys()):
                    writer.write("{} = {}\n".format(key, result[key]))
                total += result["num"]
                total_correct += result["correct"]
            writer.write("=====================\n")
            writer.write("total={}\n".format(total_correct / total))

    return result


if __name__ == "__main__":
    main()