main.py

# coding=utf-8
# Modification of the original run_glue.py script of Hugging Face's Transformers for 
# News-Frame detection task. Copyright 2020 of the authors of Multi-label and 
# Multilingual News Framing Analysis.
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
# Copyright (c) 2018, NVIDIA CORPORATION.  All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.


print(""" Finetuning the library models for sequence classification on GLUE (Bert, XLM, XLNet, RoBERTa).""")
print(""" Modification of the original run_glue.py script of Hugging Face's Transformers for News-Frame detection task.""")

import argparse
import glob
import logging
import os
import random
import sys
import pandas as pd
from scipy.special import expit, softmax
from sklearn import metrics
import matplotlib.pyplot as plt
import numpy as np
import pickle

import torch
from torch import nn
from torch.nn import BCEWithLogitsLoss
import torch.nn.functional as F

from torch.utils.data import (DataLoader, RandomSampler, SequentialSampler,
                              TensorDataset)
from torch.utils.data.distributed import DistributedSampler
# from tensorboardX import SummaryWriter
from tqdm import tqdm, trange

from transformers import (WEIGHTS_NAME, BertPreTrainedModel, BertConfig, BertModel,
                                  BertForSequenceClassification, BertTokenizer,
                                  RobertaConfig,
                                  RobertaForSequenceClassification,
                                  RobertaTokenizer,
                                  XLMConfig, XLMForSequenceClassification,
                                  XLMTokenizer, XLNetConfig,
                                  XLNetForSequenceClassification,
                                  XLNetTokenizer,
                                  DistilBertConfig,
                                  DistilBertForSequenceClassification,
                                  DistilBertTokenizer)

from transformers import AdamW, get_linear_schedule_with_warmup
from transformers import DataProcessor
from transformers import glue_compute_metrics as compute_metrics
from transformers import glue_output_modes as output_modes
output_modes["frame"] = "classification"
from transformers import glue_processors as processors

from losses import *

logger = logging.getLogger(__name__)

ALL_MODELS = sum((tuple(conf.pretrained_config_archive_map.keys()) for conf in (BertConfig, XLNetConfig, XLMConfig, 
                                                                                RobertaConfig, DistilBertConfig)), ())

losses = {
    "focal": FocalLoss,
    "focal2": FocalLoss2,
    "focal3": FocalLoss3,
    "bce": BCELoss,
    "softmax-focal": SoftmaxFocalLoss,
    "softmax": SoftmaxLoss,
    "softmax-weighted":SoftmaxWeightedLoss,
    "normalized-log-softmax": NormalizedLogSoftmaxLoss,
    "log-normalized-softmax": LogNormalizedSoftmaxLoss
}

class BertForMultiLabelSequenceClassification(BertPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels

        self.bert = BertModel(config)
        self.dropout = nn.Dropout(config.hidden_dropout_prob)
        self.classifier = nn.Linear(config.hidden_size, self.config.num_labels)
        self.init_weights()
        self.inverse_normed_freqs = None
        self.normed_freqs = None
        self.loss = None
        self.loss_name = None
        
    # used for weighted focal loss 1 and 2
    def set_inverse_normed_freqs(self, args, inverse_normed_freqs):
        self.inverse_normed_freqs = inverse_normed_freqs.to(args.device)
        
    # used for weighted focal loss 3
    def set_normed_freqs(self, args, normed_freqs):
        self.normed_freqs = normed_freqs.to(args.device)
    
    def set_loss(self, loss_name):
        print(loss_name)
        self.loss_name = loss_name
        self.loss = losses[loss_name]

    def forward(self, input_ids, attention_mask=None, token_type_ids=None,
                position_ids=None, head_mask=None, labels=None):
        outputs = self.bert(input_ids,
                            attention_mask=attention_mask,
                            token_type_ids=token_type_ids,
                            position_ids=position_ids, 
                            head_mask=head_mask)
        pooled_output = outputs[1]
        pooled_output = self.dropout(pooled_output)
        logits = self.classifier(pooled_output)
        outputs = (logits,) + outputs[2:]  # add hidden states and attention if they are here

        if labels is not None:
            if self.loss_name == "focal3":
                loss = self.loss(logits, labels, self.normed_freqs) # normed freqs are weights_0
            else:
                loss = self.loss(logits, labels, self.inverse_normed_freqs) 
#             loss_fct = nn.BCEWithLogitsLoss()
#             loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1, self.num_labels).float())
            outputs = (loss,) + outputs
        return outputs  # (loss), logits, (hidden_states), (attentions)



MODEL_CLASSES = {
    'bert': (BertConfig, BertForSequenceClassification, BertTokenizer),
    'bertmultilabel': (BertConfig, BertForMultiLabelSequenceClassification, BertTokenizer),
    'xlnet': (XLNetConfig, XLNetForSequenceClassification, XLNetTokenizer),
    'xlm': (XLMConfig, XLMForSequenceClassification, XLMTokenizer),
    'roberta': (RobertaConfig, RobertaForSequenceClassification, RobertaTokenizer),
    'distilbert': (DistilBertConfig, DistilBertForSequenceClassification, DistilBertTokenizer)
}


class InputExample(object):
    """
    A single training/test example for simple sequence classification.
    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.
    """
    def __init__(self, guid, text_a, text_b=None, labels=None):
        self.guid = guid
        self.text_a = text_a
        self.text_b = text_b
        self.labels = labels

    def __repr__(self):
        return str(self.to_json_string())

    def to_dict(self):
        """Serializes this instance to a Python dictionary."""
        output = copy.deepcopy(self.__dict__)
        return output

    def to_json_string(self):
        """Serializes this instance to a JSON string."""
        return json.dumps(self.to_dict(), indent=2, sort_keys=True) + "\n"


class InputFeatures(object):
    """
    A single set of features of data.
    Args:
        input_ids: Indices of input sequence tokens in the vocabulary.
        attention_mask: Mask to avoid performing attention on padding token indices.
            Mask values selected in ``[0, 1]``:
            Usually  ``1`` for tokens that are NOT MASKED, ``0`` for MASKED (padded) tokens.
        token_type_ids: Segment token indices to indicate first and second portions of the inputs.
        label: Label corresponding to the input
    """

    def __init__(self, input_ids, attention_mask, token_type_ids, labels):
        self.input_ids = input_ids
        self.attention_mask = attention_mask
        self.token_type_ids = token_type_ids
        self.labels = labels
        
class MultiLabelTextProcessor(DataProcessor):

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

    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 ["1","2","3","4","5","6","7","8","9"]
    
    # get the inverse normed frequencies of the training set
    def get_inverse_normed_freqs(self, data_dir):
        lines = self._read_tsv(os.path.join(data_dir, "train.tsv"))
        freqs = [0] * 9
        for line in lines:
            labels = line[3:]
            for (i,label) in enumerate(labels):
                if float(label):
                    freqs[i] = freqs[i]+1
        inverse_freqs = [1./f for f in freqs]
        total_inverse_freqs = sum(inverse_freqs)
        inverse_normed_freqs = [i/total_inverse_freqs for i in inverse_freqs]
        return torch.FloatTensor(inverse_normed_freqs)
    
    # get the normed frequencies of the training set which will serve as weights_0 for focal3
    def get_normed_freqs(self, data_dir):
        lines = self._read_tsv(os.path.join(data_dir, "train.tsv"))
        freqs = [0] * 9
        for k,line in enumerate(lines):
            labels = line[3:]
            for (i,label) in enumerate(labels):
                if float(label):
                    freqs[i] = freqs[i]+1
        total_labels = (k+1) * (i+1) #train size * 9
#         print("total_labels ",total_labels)
#         print("freqs ", freqs)
        normed_freqs = [f/total_labels for f in freqs]
#         print("normed_freqs ",normed_freqs)
        return torch.FloatTensor(normed_freqs)
    


    def _create_examples(self, lines, set_type):
        """Creates examples for the training and dev sets."""
        examples = []
        for (i, line) in enumerate(lines):
            guid = "%s-%s" % (set_type, i)
#             logger.info("i:%d, %s" % (i,line))
            text_a = line[1]
            labels = line[3:]
            examples.append(
                InputExample(guid=guid, text_a=text_a, text_b=None, labels=labels))
        return examples

processors["frame"] = MultiLabelTextProcessor


def convert_examples_to_features(examples, tokenizer,
                                      max_length=512,
                                      task=None,
                                      label_list=None,
                                      output_mode=None,
                                      pad_on_left=False,
                                      pad_token=0,
                                      pad_token_segment_id=0,
                                      mask_padding_with_zero=True):
    """
    Loads a data file into a list of ``InputFeatures``
    Args:
        examples: List of ``InputExamples`` or ``tf.data.Dataset`` containing the examples.
        tokenizer: Instance of a tokenizer that will tokenize the examples
        max_length: Maximum example length
        task: GLUE task
        label_list: List of labels. Can be obtained from the processor using the ``processor.get_labels()`` method
        output_mode: String indicating the output mode. Either ``regression`` or ``classification``
        pad_on_left: If set to ``True``, the examples will be padded on the left rather than on the right (default)
        pad_token: Padding token
        pad_token_segment_id: The segment ID for the padding token (It is usually 0, but can vary such as for XLNet where it is 4)
        mask_padding_with_zero: If set to ``True``, the attention mask will be filled by ``1`` for actual values
            and by ``0`` for padded values. If set to ``False``, inverts it (``1`` for padded values, ``0`` for
            actual values)
    Returns:
        If the ``examples`` input is a ``tf.data.Dataset``, will return a ``tf.data.Dataset``
        containing the task-specific features. If the input is a list of ``InputExamples``, will return
        a list of task-specific ``InputFeatures`` which can be fed to the model.
    """
    
    features = []
    for (ex_index, example) in enumerate(examples):
        if ex_index % 10000 == 0:
            logger.info("Writing example %d" % (ex_index))

        inputs = tokenizer.encode_plus(
            example.text_a,
            example.text_b,
            add_special_tokens=True,
            max_length=max_length,
        )
        input_ids, token_type_ids = inputs["input_ids"], inputs["token_type_ids"]

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

        # Zero-pad up to the sequence length.
        padding_length = max_length - len(input_ids)
        if pad_on_left:
            input_ids = ([pad_token] * padding_length) + input_ids
            attention_mask = ([0 if mask_padding_with_zero else 1] * padding_length) + attention_mask
            token_type_ids = ([pad_token_segment_id] * padding_length) + token_type_ids
        else:
            input_ids = input_ids + ([pad_token] * padding_length)
            attention_mask = attention_mask + ([0 if mask_padding_with_zero else 1] * padding_length)
            token_type_ids = token_type_ids + ([pad_token_segment_id] * padding_length)

        assert len(input_ids) == max_length, "Error with input length {} vs {}".format(len(input_ids), max_length)
        assert len(attention_mask) == max_length, "Error with input length {} vs {}".format(len(attention_mask), max_length)
        assert len(token_type_ids) == max_length, "Error with input length {} vs {}".format(len(token_type_ids), max_length)

#         if output_mode == "classification":
#             label = label_map[example.label]
#         elif output_mode == "regression":
#             label = f`loat(example.label)
#         else:
#             raise KeyError(output_mode)

#         labels_ids = []
#         for label in example.labels:
#             labels_ids.append(float(label))
#         logger.info("LABELS: %s" % (example.labels))    
        labels = [float(i) for i in example.labels]

        if ex_index < 5:
            logger.info("*** Example ***")
            logger.info("guid: %s" % (example.guid))
            logger.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
            logger.info("attention_mask: %s" % " ".join([str(x) for x in attention_mask]))
            logger.info("token_type_ids: %s" % " ".join([str(x) for x in token_type_ids]))
            logger.info("label: %s" % (example.labels))

        features.append(
                InputFeatures(input_ids=input_ids,
                              attention_mask=attention_mask,
                              token_type_ids=token_type_ids,
                              labels=labels))

    return features

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):
    """ 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)

    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)
    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
    tr_loss, logging_loss = 0.0, 0.0
    model.zero_grad()
    train_iterator = trange(int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0])
    set_seed(args)  # Added here for reproductibility (even between python 2 and 3)
    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):
            model.train()
            batch = tuple(t.to(args.device) for t in batch)
            inputs = {'input_ids':      batch[0],
                      'attention_mask': batch[1],
                      'labels':         batch[3]}
            if args.model_type != 'distilbert':
                inputs['token_type_ids'] = batch[2] if args.model_type in ['bert', 'xlnet'] else None  # XLM, DistilBERT and RoBERTa don't use segment_ids
            outputs = model(**inputs)
            loss = outputs[0]  # model outputs are always tuple in transformers (see doc)

            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()
                torch.nn.utils.clip_grad_norm_(amp.master_params(optimizer), args.max_grad_norm)
            else:
                loss.backward()
                torch.nn.utils.clip_grad_norm_(model.parameters(), args.max_grad_norm)

            tr_loss += loss.item()
            if (step + 1) % args.gradient_accumulation_steps == 0:
                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
                    if args.local_rank == -1 and args.evaluate_during_training:  # Only evaluate when single GPU otherwise metrics may not average well
                        results = evaluate(args, model, tokenizer)
#                         for key, value in results.items():
#                             tb_writer.add_scalar('eval_{}'.format(key), value, global_step)
#                     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

                if args.local_rank in [-1, 0] and args.save_steps > 0 and global_step % args.save_steps == 0:
                    # 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)
                    torch.save(args, os.path.join(output_dir, 'training_args.bin'))
                    logger.info("Saving model checkpoint 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


def plot_roc(fpr, tpr, roc_auc, output_dir=False):
    plt.figure()
    lw = 2
    for i in range(len(fpr)):
        plt.plot(fpr[i], tpr[i], #color='darkorange',
                 lw=lw, label='Frame %d (area = %0.2f)' % (int(i+1), roc_auc[i]))
    plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')
    plt.xlim([0.0, 1.0])
    plt.ylim([0.0, 1.05])
    plt.xlabel('False Positive Rate')
    plt.ylabel('True Positive Rate')
    plt.title('Receiver operating characteristic example')
    plt.legend(loc="lower right")
    if output_dir:
        plt.savefig(os.path.join(output_dir, "auc.png"))
    plt.show()
    
    
def evaluate(args, model, tokenizer, prefix=""):
    # Loop to handle MNLI double evaluation (matched, mis-matched)
    eval_task_names = ("mnli", "mnli-mm") if args.task_name == "mnli" else (args.task_name,)
    eval_outputs_dirs = (args.output_dir, args.output_dir + '-MM') if args.task_name == "mnli" else (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, 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) if args.local_rank == -1 else DistributedSampler(eval_dataset)
        eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)

        # Eval!
        logger.info("***** Running evaluation {} *****".format(prefix))
        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
        for batch in tqdm(eval_dataloader, desc="Evaluating"):
            model.eval()
            batch = tuple(t.to(args.device) for t in batch)

            with torch.no_grad():
                inputs = {'input_ids':      batch[0],
                          'attention_mask': batch[1],
                          'labels':         batch[3]}
                if args.model_type != 'distilbert':
                    inputs['token_type_ids'] = batch[2] if args.model_type in ['bert', 'xlnet'] else None  # XLM, DistilBERT and RoBERTa don't use segment_ids
                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()
            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)

        eval_loss = eval_loss / nb_eval_steps
#         if args.output_mode == "classification":
#             preds = np.argmax(preds, axis=1)
#         elif args.output_mode == "regression":
#             preds = np.squeeze(preds)

        if args.loss in ["focal","focal2","focal3","bce"]:
            preds = expit(preds)
        else:
            preds = softmax(preds, axis=1)
        
        print("preds shape: ", preds.shape)
        print("out label ids: ", out_label_ids.shape)
        print(preds[:5])
        print(out_label_ids[:5])

        return preds
#         output_eval_file = os.path.join(eval_output_dir, "eval_results.txt")
#         with open(output_eval_file, "w") as writer:
#             logger.info("***** Eval results {} *****".format(prefix))
#             for key in sorted(results.keys()):
#                 logger.info("  %s = %s", key, str(results[key]))
#                 writer.write("%s = %s\n" % (key, str(results[key])))

#     return results


def load_and_cache_examples(args, task, tokenizer, evaluate=False):
    if args.local_rank not in [-1, 0] and not evaluate:
        torch.distributed.barrier()  # Make sure only the first process in distributed training process the dataset, and the others will use the cache

    processor = processors[task]()
    output_mode = output_modes[task]
    # Load data features from cache or dataset file
    cached_features_file = os.path.join(args.data_dir, 'cached_{}_{}_{}_{}'.format(
        'dev' if evaluate else 'train',
        list(filter(None, args.model_name_or_path.split('/'))).pop(),
        str(args.max_seq_length),
        str(task)))
    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 task in ['mnli', 'mnli-mm'] and args.model_type in ['roberta']:
            # HACK(label indices are swapped in RoBERTa pretrained model)
            label_list[1], label_list[2] = label_list[2], label_list[1] 
        examples = processor.get_dev_examples(args.data_dir) if evaluate else processor.get_train_examples(args.data_dir)
        features = convert_examples_to_features(examples,
                                                tokenizer,
                                                label_list=label_list,
                                                max_length=args.max_seq_length,
                                                output_mode=output_mode,
                                                pad_on_left=bool(args.model_type in ['xlnet']),                 # pad on the left for xlnet
                                                pad_token=tokenizer.convert_tokens_to_ids([tokenizer.pad_token])[0],
                                                pad_token_segment_id=4 if args.model_type in ['xlnet'] else 0,
        )
#         if args.local_rank in [-1, 0]:
#             logger.info("Saving features into cached file %s", cached_features_file)
#             torch.save(features, cached_features_file)

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

    # Convert to Tensors and build dataset
    all_input_ids = torch.tensor([f.input_ids for f in features], dtype=torch.long)
    all_attention_mask = torch.tensor([f.attention_mask for f in features], dtype=torch.long)
    all_token_type_ids = torch.tensor([f.token_type_ids for f in features], dtype=torch.long)
    if output_mode == "classification":
        all_labels = torch.tensor([f.labels for f in features], dtype=torch.long) #list of list happens ?
    elif output_mode == "regression":
        all_labels = torch.tensor([f.labels for f in features], dtype=torch.float)

    dataset = TensorDataset(all_input_ids, all_attention_mask, all_token_type_ids, all_labels)
    return dataset


def main():
    parser = argparse.ArgumentParser()

    ## Required parameters
    parser.add_argument("--data_dir", default=None, type=str, required=True,
                        help="The input data dir. Should contain the .tsv files (or other data files) for the task.")
    parser.add_argument("--model_type", default=None, type=str, required=True,
                        help="Model type selected in the list: " + ", ".join(MODEL_CLASSES.keys()))
    parser.add_argument("--model_name_or_path", default=None, type=str, required=True,
                        help="Path to pre-trained model or shortcut name selected in the list: " + ", ".join(ALL_MODELS))
    parser.add_argument("--task_name", default=None, type=str, required=True,
                        help="The name of the task to train selected in the list: " + ", ".join(processors.keys()))
    parser.add_argument("--output_dir", default=None, type=str, required=True,
                        help="The output directory where the model predictions and checkpoints will be written.")
    parser.add_argument("--loss", default=None, type=str, required=True,
                        help="The loss option selected in the list: " + ", ".join(losses.keys()))
    parser.add_argument("--exp_name", default=None, type=str, required=True,
                        help="Name of the experiment to save the evaluated predictions.")

    ## 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", action='store_true',
                        help="Whether to run training.")
    parser.add_argument("--do_eval", action='store_true',
                        help="Whether to run eval on the dev set.")
    parser.add_argument("--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("--set_weights", action='store_true',
                        help="Set appropriate weights for weighted loss based on the loss.")
    parser.add_argument("--zero_shot", action='store_true',
                        help="If zero-shot, model is loaded from model_name_or_path for evaluation.")

    parser.add_argument("--per_gpu_train_batch_size", default=8, type=int,
                        help="Batch size per GPU/CPU for training.")
    parser.add_argument("--per_gpu_eval_batch_size", default=8, 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=5e-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=3.0, 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('--save_steps', type=int, default=500,
                        help="Save checkpoint every X updates steps.")
    parser.add_argument("--eval_all_checkpoints", action='store_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', action='store_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('--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("--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("--eval_specific_checkpoint", default="", type=str,
                        help="Evaluate the specified checkpoint under model path.")
    
    args = parser.parse_args()
    print(args)

    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))

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

    # 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 will take care of sychronizing 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
    logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s -   %(message)s',
                        datefmt = '%m/%d/%Y %H:%M:%S',
                        level = logging.INFO if args.local_rank in [-1, 0] else logging.WARN)
    logger.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 GLUE task
    args.task_name = args.task_name.lower()
    if args.task_name not in processors:
        raise ValueError("Task not found: %s" % (args.task_name))
    processor = processors[args.task_name]()
    args.output_mode = output_modes[args.task_name]
    label_list = processor.get_labels()
    num_labels = len(label_list)

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

    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)
    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)
    model = model_class.from_pretrained(args.model_name_or_path, 
                                        from_tf=bool('.ckpt' in args.model_name_or_path),
                                        config=config)

    model.set_loss(args.loss)
    
    
    if args.local_rank == 0:
        torch.distributed.barrier()  # Make sure only the first process in distributed training will download model & vocab

    

    logger.info("Training/evaluation parameters %s", args)
    
    
    
    if args.loss == "focal3":
        normed_freqs = processor.get_normed_freqs(args.data_dir) if args.set_weights else torch.ones(num_labels)
        model.set_normed_freqs(args, normed_freqs)
    else:
        # Set inverse normed freqs if needed for weighted loss
        inverse_normed_freqs = processor.get_inverse_normed_freqs(args.data_dir) if args.set_weights else torch.ones(num_labels)
        model.set_inverse_normed_freqs(args, inverse_normed_freqs)

    model.to(args.device)
    # Training
    if args.do_train:
        train_dataset = load_and_cache_examples(args, args.task_name, tokenizer, evaluate=False)
        global_step, tr_loss = train(args, train_dataset, model, tokenizer) #here
        logger.info(" global_step = %s, average loss = %s", global_step, tr_loss)


    # 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):
        # 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, 
                                                    do_lower_case=args.do_lower_case)
        model.to(args.device)


    # Evaluation
    results = {}
    if args.do_eval and args.local_rank in [-1, 0]:
        
        model_directory = args.model_name_or_path if args.zero_shot else args.output_dir
        
        tokenizer = tokenizer_class.from_pretrained(model_directory, do_lower_case=args.do_lower_case) 
        checkpoints = [model_directory]
        if args.eval_all_checkpoints:
            checkpoints = list(os.path.dirname(c) for c in sorted(glob.glob(model_directory + '/**/' + WEIGHTS_NAME, recursive=True)))
            logging.getLogger("transformers.modeling_utils").setLevel(logging.WARN)  # Reduce logging
        
        if args.eval_specific_checkpoint:
            checkpoints = [os.path.join(model_directory, args.eval_specific_checkpoint)] + checkpoints
        logger.info("Evaluate the following checkpoints: %s", checkpoints)
        for checkpoint in checkpoints:
            global_step = checkpoint.split('-')[-1] if len(checkpoints) > 1 else ""
            model = model_class.from_pretrained(checkpoint)
            model.set_loss(args.loss)
            model.set_inverse_normed_freqs(args, torch.ones(num_labels))
            model.set_normed_freqs(args, torch.ones(num_labels)) #you could rather set a default for loss
            model.to(args.device)
            result = evaluate(args, model, tokenizer, prefix=global_step)
            
            # write the predictions array (num_samples) x (num_frames) into pickle file
            filename = os.path.join(args.output_dir, "predictions.pkl")
            with open(filename, "wb") as f:
                pickle.dump(result, f)
                f.close()
            # assuming we're only evaluating a single checkpoint
            print("Saved ", os.path.join(args.output_dir, "predictions.pkl"))
            return(result)

if __name__ == "__main__":
    main()