run_eqa.py

#!/usr/bin/env python
# coding=utf-8
# Copyright 2020 The HuggingFace Team All rights reserved.
# Modified by Wonjin Yoon for "Sequence Tagging for Biomedical Extractive Question Answering"
#
# 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.
"""
Fine-tuning the library models for token classification.
"""
# You can also adapt this script on your own token classification task and datasets. Pointers for this are left as
# comments.

import logging
import os
import sys
from dataclasses import dataclass, field
from typing import Optional
import numpy as np

from tqdm.auto import tqdm

import datasets
from datasets import ClassLabel, load_dataset, load_metric

import transformers
from transformers import (
    AutoConfig,
    AutoTokenizer,
    DataCollatorForTokenClassification,
    HfArgumentParser,
    PreTrainedTokenizerFast,
    Trainer,
    TrainingArguments,
    set_seed,
)
from transformers.trainer_utils import get_last_checkpoint
from transformers.utils import check_min_version
from transformers.utils.versions import require_version

from seqtagqa import (
    BertForLinearSeqTagQA,
)


# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.9.2") # Our codes are tested using '4.9.2' version

require_version("datasets>=1.8.0", "Our codes are tested using '1.18.2' version of datasets")

logger = logging.getLogger(__name__)


@dataclass
class ModelArguments:
    """
    Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
    """

    model_name_or_path: str = field(
        metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
    )
    model_struct: str = field(
        metadata={"help": "Structure of output layer"}
    )
    config_name: Optional[str] = field(
        default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
    )
    tokenizer_name: Optional[str] = field(
        default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
    )
    cache_dir: Optional[str] = field(
        default=None,
        metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"},
    )
    model_revision: str = field(
        default="main",
        metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
    )
    use_auth_token: bool = field(
        default=False,
        metadata={
            "help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
            "with private models)."
        },
    )


@dataclass
class DataTrainingArguments:
    """
    Arguments pertaining to what data we are going to input our model for training and eval.
    """

    task_name: Optional[str] = field(default="ner", metadata={"help": "The name of the task (ner, pos...)."})
    dataset_name: Optional[str] = field(
        default=None, metadata={"help": "The name of the dataset to use (via the datasets library)."}
    )
    dataset_config_name: Optional[str] = field(
        default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
    )
    train_file: Optional[str] = field(
        default=None, metadata={"help": "The input training data file (a csv or JSON file)."}
    )
    validation_file: Optional[str] = field(
        default=None,
        metadata={"help": "An optional input evaluation data file to evaluate on (a csv or JSON file)."},
    )
    test_file: Optional[str] = field(
        default=None,
        metadata={"help": "An optional input test data file to predict on (a csv or JSON file)."},
    )
    text_column_name: Optional[str] = field(
        default=None, metadata={"help": "The column name of text to input in the file (a csv or JSON file)."}
    )
    label_column_name: Optional[str] = field(
        default=None, metadata={"help": "The column name of label to input in the file (a csv or JSON file)."}
    )
    overwrite_cache: bool = field(
        default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
    )
    preprocessing_num_workers: Optional[int] = field(
        default=None,
        metadata={"help": "The number of processes to use for the preprocessing."},
    )
    max_seq_length: int = field(
        default=None,
        metadata={
            "help": "The maximum total input sequence length after tokenization. If set, sequences longer "
            "than this will be truncated, sequences shorter will be padded."
        },
    )
    pad_to_max_length: bool = field(
        default=False,
        metadata={
            "help": "Whether to pad all samples to model maximum sentence length. "
            "If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
            "efficient on GPU but very bad for TPU."
        },
    )
    max_train_samples: Optional[int] = field(
        default=None,
        metadata={
            "help": "For debugging purposes or quicker training, truncate the number of training examples to this "
            "value if set."
        },
    )
    max_eval_samples: Optional[int] = field(
        default=None,
        metadata={
            "help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
            "value if set."
        },
    )
    max_predict_samples: Optional[int] = field(
        default=None,
        metadata={
            "help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
            "value if set."
        },
    )
    label_all_tokens: bool = field(
        default=False,
        metadata={
            "help": "Whether to put the label for one word on all tokens of generated by that word or just on the "
            "one (in which case the other tokens will have a padding index)."
        },
    )
    return_entity_level_metrics: bool = field(
        default=False,
        metadata={"help": "Whether to return all the entity levels during evaluation or just the overall ones."},
    )
    keep_question_tokens: bool = field(
        default=False,
        metadata={"help": "Whether to mask question tokens or label them with O tag."},
    )

    def __post_init__(self):
        if self.dataset_name is None and self.train_file is None and self.validation_file is None and self.test_file is None:
            raise ValueError("Need either a dataset name or a training/validation/test file.")
        else:
            if self.train_file is not None:
                extension = self.train_file.split(".")[-1]
                assert extension in ["tsv", "csv", "json"], "`train_file` should be a csv, tsv, or a json file."
            if self.validation_file is not None:
                extension = self.validation_file.split(".")[-1]
                assert extension in ["tsv", "csv", "json"], "`validation_file` should be a csv, tsv, or a json file."
            if self.test_file is not None:
                extension = self.test_file.split(".")[-1]
                assert extension in ["tsv", "csv", "json"], "`test_file` should be a csv, tsv, or a json file."
        self.task_name = self.task_name.lower()


def main():
    # See all possible arguments in huggingface-transformers : src/transformers/training_args.py
    # or by passing the --help flag to this script.

    parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
    if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
        # If we pass only one argument to the script and it's the path to a json file,
        # let's parse it to get our arguments.
        model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
    else:
        model_args, data_args, training_args = parser.parse_args_into_dataclasses()

    # Setup logging
    logging.basicConfig(
        format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
        datefmt="%m/%d/%Y %H:%M:%S",
        handlers=[logging.StreamHandler(sys.stdout)],
    )

    log_level = training_args.get_process_log_level()
    logger.setLevel(log_level)
    datasets.utils.logging.set_verbosity(log_level)
    transformers.utils.logging.set_verbosity(log_level)
    transformers.utils.logging.enable_default_handler()
    transformers.utils.logging.enable_explicit_format()

    # Log on each process the small summary:
    logger.warning(
        f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
        + f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
    )
    logger.info(f"Training/evaluation parameters {training_args}")

    # Detecting last checkpoint.
    last_checkpoint = None
    if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
        last_checkpoint = get_last_checkpoint(training_args.output_dir)
        if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
            raise ValueError(
                f"Output directory ({training_args.output_dir}) already exists and is not empty. "
                "Use --overwrite_output_dir to overcome."
            )
        elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
            logger.info(
                f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
                "the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
            )

    # Set seed before initializing model.
    set_seed(training_args.seed)

    # Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
    # or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
    # (the dataset will be downloaded automatically from the datasets Hub).
    #
    # For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
    # 'text' is found. You can easily tweak this behavior (see below).
    #
    # In distributed training, the load_dataset function guarantee that only one local process can concurrently
    # download the dataset.
    if data_args.dataset_name is not None:
        # Downloading and loading a dataset from the hub.
        raw_datasets = load_dataset(
            data_args.dataset_name, data_args.dataset_config_name, cache_dir=model_args.cache_dir
        )
    else:
        data_files = {}
        if data_args.train_file is not None:
            data_files["train"] = data_args.train_file
        if data_args.validation_file is not None:
            data_files["validation"] = data_args.validation_file
        if data_args.test_file is not None:
            data_files["test"] = data_args.test_file
        extension = list(data_files.values())[0].split(".")[-1]
        # for extension == "tsv", tsv/tsv.py https://huggingface.co/docs/datasets/add_dataset.html
        raw_datasets = load_dataset(extension, data_files=data_files, cache_dir=model_args.cache_dir)
    # See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
    # https://huggingface.co/docs/datasets/loading_datasets.html.

    if training_args.do_train:
        column_names = raw_datasets["train"].column_names
        features = raw_datasets["train"].features
    elif training_args.do_eval:
        column_names = raw_datasets["validation"].column_names
        features = raw_datasets["validation"].features
    elif training_args.do_predict:
        column_names = raw_datasets["test"].column_names
        features = raw_datasets["test"].features
    else:
        raise ValueError("Select at least one of train, eval or test")

    if data_args.label_column_name is not None:
        label_column_name = data_args.label_column_name
    elif f"{data_args.task_name}_tags" in column_names:
        label_column_name = f"{data_args.task_name}_tags"
    else:
        label_column_name = "labels"#"answer_labels"

    IGNORE_LABELS = ["[SEP]", "X", "[CLS]", "[PAD]"]

    # In the event the labels are not a `Sequence[ClassLabel]`, we will need to go through the dataset to get the
    # unique labels.
    def get_label_list(labels):
        unique_labels = set()
        for label in labels:
            unique_labels = unique_labels | set(label)
        label_list = list(unique_labels)
        label_list.sort()
        return label_list

    if isinstance(features[label_column_name].feature, ClassLabel):
        label_list = features[label_column_name].feature.names
        # No need to convert the labels since they are already ints.
        label_to_id = {i: i for i in range(len(label_list))}
        #label_to_id = {label: str(idx) for idx, label in enumerate(label_list)}
    else:
        label_list = get_label_list(raw_datasets["train"][label_column_name])
        label_to_id = {l: i for i, l in enumerate(label_list)}
        #label_to_id = {label: str(idx) for idx, label in enumerate(label_list)}
    num_labels = len(label_list)

    # Map that sends B-Xxx label to its I-Xxx counterpart
    b_to_i_label = []
    for idx, label in enumerate(label_list):
        if label.startswith("B-") and label.replace("B-", "I-") in label_list:
            b_to_i_label.append(label_list.index(label.replace("B-", "I-")))
        else:
            b_to_i_label.append(idx)

    # Load pretrained model and tokenizer
    #
    # Distributed training:
    # The .from_pretrained methods guarantee that only one local process can concurrently
    # download model & vocab.
    config = AutoConfig.from_pretrained(
        model_args.config_name if model_args.config_name else model_args.model_name_or_path,
        num_labels=num_labels,
        label2id=label_to_id,
        id2label={i: l for l, i in label_to_id.items()},
        finetuning_task=data_args.task_name,
        cache_dir=model_args.cache_dir,
        revision=model_args.model_revision,
        use_auth_token=True if model_args.use_auth_token else None,
    )

    tokenizer_name_or_path = model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path
    if config.model_type in {"gpt2", "roberta"}:
        tokenizer = AutoTokenizer.from_pretrained(
            tokenizer_name_or_path,
            cache_dir=model_args.cache_dir,
            use_fast=True,
            revision=model_args.model_revision,
            use_auth_token=True if model_args.use_auth_token else None,
            add_prefix_space=True,
        )
    else:
        tokenizer = AutoTokenizer.from_pretrained(
            tokenizer_name_or_path,
            cache_dir=model_args.cache_dir,
            use_fast=True,
            revision=model_args.model_revision,
            use_auth_token=True if model_args.use_auth_token else None,
        )

    if model_args.model_struct.lower() == "linear":
        model = BertForLinearSeqTagQA.from_pretrained(
            model_args.model_name_or_path,
            from_tf=bool(".ckpt" in model_args.model_name_or_path),
            config=config,
            cache_dir=model_args.cache_dir,
            revision=model_args.model_revision,
            use_auth_token=True if model_args.use_auth_token else None,
        )
    else:
        raise NotImplementedError

    # Tokenizer check: this script requires a fast tokenizer.
    if not isinstance(tokenizer, PreTrainedTokenizerFast):
        raise ValueError(
            "This example script only works for models that have a fast tokenizer. Checkout the big table of models "
            "at https://huggingface.co/transformers/index.html#supported-frameworks to find the model types that meet this "
            "requirement"
        )

    # Preprocessing the dataset
    # Padding strategy
    padding = "max_length" if data_args.pad_to_max_length else False

   # Tokenize all texts and align the labels with them.
    def tokenize_and_align_labels_pair(examples):
        tokenized_inputs = tokenizer(
            examples["question"], examples["context_tokens"],
            padding=padding,
            truncation="only_second",
            max_length=data_args.max_seq_length,
            # We use this argument because the texts in our dataset are lists of words (with a label for each word).
            is_split_into_words=True,
        )
        labels = []
        for i, label in enumerate(examples["answer_labels"]): # TODO : change this part to include question queries
            word_ids = tokenized_inputs.word_ids(batch_index=i)
            token_type_ids = tokenized_inputs.token_type_ids[i]
            assert len(word_ids) == len(token_type_ids)
            
            previous_word_idx = None
            label_ids = []
            for word_idx, token_type_idx in zip(word_ids, token_type_ids):
                # for keep_question_tokens==False, tag (label) for questions are ignore for loss calculation. 
                if not(data_args.keep_question_tokens) and token_type_idx == 0:
                    label_ids.append(-100)
                # Special tokens have a word id that is None. We set the label to -100 so they are automatically
                # ignored in the loss function.
                elif word_idx is None:
                    label_ids.append(-100)
                # We set the label for the first token of each word.
                elif word_idx != previous_word_idx:
                    if label[word_idx] in [label_list.index(ignore_label) for ignore_label in IGNORE_LABELS]: 
                        label_ids.append(-100)
                    else:
                        label_ids.append(label_to_id[label[word_idx]])
                # For the other tokens in a word, we set the label to either the current label or -100, depending on
                # the label_all_tokens flag.
                else:
                    if data_args.label_all_tokens:
                        label_ids.append(b_to_i_label[label_to_id[label[word_idx]]])
                    else:
                        label_ids.append(-100)
                previous_word_idx = word_idx

            labels.append(label_ids)
            
        tokenized_inputs["labels"] = labels

        return tokenized_inputs

    if training_args.do_train:
        if "train" not in raw_datasets:
            raise ValueError("--do_train requires a train dataset")
        train_dataset = raw_datasets["train"]
        if data_args.max_train_samples is not None:
            train_dataset = train_dataset.select(range(data_args.max_train_samples))
        with training_args.main_process_first(desc="train dataset map pre-processing"):
            train_dataset = train_dataset.map(
                tokenize_and_align_labels_pair,
                batched=True,
                num_proc=data_args.preprocessing_num_workers,
                load_from_cache_file=not data_args.overwrite_cache,
                desc="Running tokenizer on train dataset",
            )

    if training_args.do_eval:
        if "validation" not in raw_datasets:
            raise ValueError("--do_eval requires a validation dataset")
        eval_dataset = raw_datasets["validation"]
        if data_args.max_eval_samples is not None:
            eval_dataset = eval_dataset.select(range(data_args.max_eval_samples))
        with training_args.main_process_first(desc="validation dataset map pre-processing"):
            eval_dataset = eval_dataset.map(
                tokenize_and_align_labels_pair,
                batched=True,
                num_proc=data_args.preprocessing_num_workers,
                load_from_cache_file=not data_args.overwrite_cache,
                desc="Running tokenizer on validation dataset",
            )

    if training_args.do_predict:
        if "test" not in raw_datasets:
            raise ValueError("--do_predict requires a test dataset")
        predict_dataset = raw_datasets["test"]
        if data_args.max_predict_samples is not None:
            predict_dataset = predict_dataset.select(range(data_args.max_predict_samples))
        with training_args.main_process_first(desc="prediction dataset map pre-processing"):
            predict_dataset = predict_dataset.map(
                tokenize_and_align_labels_pair,
                batched=True,
                num_proc=data_args.preprocessing_num_workers,
                load_from_cache_file=not data_args.overwrite_cache,
                desc="Running tokenizer on prediction dataset",
            )

    # for sanity checking
    logger.info(f"#### Printing pre-processed samples, each from train_dataset and predict_dataset.")
    show_sample = [train_dataset] if training_args.do_train else []
    if training_args.do_predict:
        show_sample.append(predict_dataset)

    for sample_data_type in show_sample:
        for key, values in sample_data_type[0].items():
            logger.info(f"{key} ({len(values)}): {values}")
    logger.debug("labels\ttoken_type_ids input_ids")
    for values in  zip(
        sample_data_type[0]["labels"],
        sample_data_type[0]["token_type_ids"],
        sample_data_type[0]["input_ids"],
        ):
        logger.debug("\t".join([str(ele) for ele in values]) + " : "+tokenizer.decode(values[2]))

    # Data collator
    data_collator = DataCollatorForTokenClassification(tokenizer, pad_to_multiple_of=8 if training_args.fp16 else None)

    # Metrics
    metric = load_metric("seqeval", cache_dir=training_args.output_dir)

    def compute_metrics(p, predictions_are_clean=False):
        """
        p (:obj:`Tuple` of size:2):
            A tuple (predictions, labels). 
            predictions and labels should have (batch_size, seq_len, num_labe) and (batch_size, seq_len) shape, respectively.
        predictions_are_clean (:obj:`Boolean`):
            True if predictions is post-processed, both Argmax AND Ignore_Idx removal (cleaning). 
            In this case, each element of p should have (batch_size, seq_len) shape
        """
        predictions, labels = p
        if not predictions_are_clean:
            predictions = np.argmax(predictions, axis=2)

        if predictions_are_clean:
            clean_labels = [
                [l for l in label if l != -100]
                for label in labels
            ]
        else:
            clean_labels=labels

        # Remove ignored index (special tokens)
        true_predictions = [
            [label_list[p] for (p, l) in zip(prediction, label) if l != -100]
            for prediction, label in zip(predictions, clean_labels)
        ]
        true_labels = [
            [label_list[l] for (p, l) in zip(prediction, label) if l != -100]
            for prediction, label in zip(predictions, clean_labels)
        ]

        results = metric.compute(predictions=true_predictions, references=true_labels)
        if data_args.return_entity_level_metrics:
            # Unpack nested dictionaries
            final_results = {}
            for key, value in results.items():
                if isinstance(value, dict):
                    for n, v in value.items():
                        final_results[f"{key}_{n}"] = v
                else:
                    final_results[key] = value
            return final_results
        else:
            return {
                "precision": results["overall_precision"],
                "recall": results["overall_recall"],
                "f1": results["overall_f1"],
                "accuracy": results["overall_accuracy"],
            }

    # Initialize our Trainer
    trainer = Trainer(
        model=model,
        args=training_args,
        train_dataset=train_dataset if training_args.do_train else None,
        eval_dataset=eval_dataset if training_args.do_eval else None,
        tokenizer=tokenizer,
        data_collator=data_collator,
        compute_metrics=compute_metrics,
    )

    # Training
    if training_args.do_train:
        checkpoint = None
        if training_args.resume_from_checkpoint is not None:
            checkpoint = training_args.resume_from_checkpoint
        elif last_checkpoint is not None:
            checkpoint = last_checkpoint
        train_result = trainer.train(resume_from_checkpoint=checkpoint)
        metrics = train_result.metrics
        trainer.save_model()  # Saves the tokenizer too for easy upload

        max_train_samples = (
            data_args.max_train_samples if data_args.max_train_samples is not None else len(train_dataset)
        )
        metrics["train_samples"] = min(max_train_samples, len(train_dataset))

        trainer.log_metrics("train", metrics)
        trainer.save_metrics("train", metrics)
        trainer.save_state()

    # Evaluation
    if training_args.do_eval:
        logger.info("*** Evaluate ***")

        metrics = trainer.evaluate()

        max_eval_samples = data_args.max_eval_samples if data_args.max_eval_samples is not None else len(eval_dataset)
        metrics["eval_samples"] = min(max_eval_samples, len(eval_dataset))

        trainer.log_metrics("eval", metrics)
        trainer.save_metrics("eval", metrics)

    # Predict
    if training_args.do_predict:
        logger.info("*** Predict ***")

        if model_args.model_struct.lower() == "linear":
            predictions_logit, labels, metrics = trainer.predict(predict_dataset, metric_key_prefix="predict")
            predictions = np.argmax(predictions_logit, axis=2)

            # Remove ignored index (question tokens)
            if data_args.keep_question_tokens:
                # we need to remove question tokens
                logger.info("*** Removing question tokens in output file (predictions.txt) ***")
                num_q_tokens = []
                test_dataloader = trainer.get_test_dataloader(predict_dataset)
                for step, batch in tqdm(enumerate(test_dataloader), total=len(test_dataloader)):
                    device_batch = {key: values.to(model.device) for key,values in batch.items()}
                    tti_one = (device_batch["attention_mask"] & (device_batch["token_type_ids"]==0)).to("cpu").detach().numpy()
                    num_q_tokens.extend(np.sum(tti_one, axis=1))
                predictions = [prediction[num_q:] for prediction, num_q in zip(predictions, num_q_tokens)]
                predictions_prob = [prediction[num_q:] for prediction, num_q in zip(predictions_logit, num_q_tokens)] # TODO : softmax function
                labels = [label[num_q:] for label, num_q in zip(labels, num_q_tokens)]

            # Remove ignored index (special tokens)
            true_predictions = [
                [label_list[p] for (p, l) in zip(prediction, label) if l != -100]
                for prediction, label in zip(predictions, labels)
            ]
        else:
            raise ValueError("Wrong model_args.model_struct!")
        
        predictions_prob = predictions_logit # TODO : softmax function

        # Remove ignored index (special tokens)
        true_predictions_prob = [
            [p for (p, l) in zip(prediction, label) if l != -100]
            for prediction, label in zip(predictions_prob, labels)
        ]

        trainer.log_metrics("predict", metrics)
        trainer.save_metrics("predict", metrics)

        # Save predictions
        output_predictions_file = os.path.join(training_args.output_dir, "predictions.txt")
        output_predictions_prob_file = os.path.join(training_args.output_dir, "predictions_prob.txt")
        if trainer.is_world_process_zero():
            with open(output_predictions_file, "w") as writer:
                for prediction, unique_id in zip(true_predictions, predict_dataset.data['unique_id']):
                    writer.write(" ".join([str(unique_id)] + prediction) + "\n")

            if len(true_predictions_prob) > 10000:
                logger.info("Saving of predictions_prob.txt omitted:: it takes long time to save long prediction.")
            else:
                with open(output_predictions_prob_file, "w") as writer:
                    for prediction, unique_id in zip(true_predictions_prob, predict_dataset.data['unique_id']):
                        if str(unique_id) == "":
                            assert prediction == [], f"prediction:{prediction}, unique_id:{unique_id}"
                            continue
                        writer.write("UNIQUEID " + str(unique_id) + "\n")
                        for token_prob in prediction: # not including [SEP], [CLS], X = questions
                            writer.write("[" + " ".join([str(ele).replace("\n", " ").replace("  "," ") for ele in token_prob]) + "]\n")

        logger.info(f"*** Predictions written to {output_predictions_file} ***")

    kwargs = {"finetuned_from": model_args.model_name_or_path, "tasks": "token-classification"}
    if data_args.dataset_name is not None:
        kwargs["dataset_tags"] = data_args.dataset_name
        if data_args.dataset_config_name is not None:
            kwargs["dataset_args"] = data_args.dataset_config_name
            kwargs["dataset"] = f"{data_args.dataset_name} {data_args.dataset_config_name}"
        else:
            kwargs["dataset"] = data_args.dataset_name

    if training_args.push_to_hub:
        trainer.push_to_hub(**kwargs)
    else:
        trainer.create_model_card(**kwargs)


def _mp_fn(index):
    # For xla_spawn (TPUs)
    main()


if __name__ == "__main__":
    main()