run_summarization.py

#!/usr/bin/env python
# coding=utf-8
# Copyright 2021 The HuggingFace Team. 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.
"""
Fine-tuning the library models for sequence to sequence.
"""
# You can also adapt this script on your own sequence to sequence task. Pointers for this are left as comments.

import json
import logging
import os
import re
import sys
from dataclasses import dataclass, field
from pathlib import Path
from typing import Any, Dict, List, Optional

import datasets
import evaluate
import nltk  # Here to have a nice missing dependency error message early on
import numpy as np
import pandas as pd
import transformers
from datasets import load_dataset
from filelock import FileLock
from omegaconf import OmegaConf
from transformers import (
    AutoConfig,
    AutoModelForSeq2SeqLM,
    AutoTokenizer,
    DataCollatorForSeq2Seq,
    HfArgumentParser,
    MBart50Tokenizer,
    MBart50TokenizerFast,
    MBartTokenizer,
    MBartTokenizerFast,
    Seq2SeqTrainer,
    Seq2SeqTrainingArguments,
    set_seed,
)
from transformers.trainer_utils import get_last_checkpoint
from transformers.utils import check_min_version, is_offline_mode, send_example_telemetry
from transformers.utils.versions import require_version


# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.27.0.dev0")

require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/summarization/requirements.txt")

logger = logging.getLogger(__name__)

try:
    nltk.data.find("tokenizers/punkt")
except (LookupError, OSError):
    if is_offline_mode():
        raise LookupError(
            "Offline mode: run this script without TRANSFORMERS_OFFLINE first to download nltk data files"
        )
    with FileLock(".lock") as lock:
        nltk.download("punkt", quiet=True)

# A list of all multilingual tokenizer which require lang attribute.
MULTILINGUAL_TOKENIZERS = [MBartTokenizer, MBartTokenizerFast, MBart50Tokenizer, MBart50TokenizerFast]

# Constants for the MEDIQA-Chat @ ACL-ClinicalNLP challenge task
GENHX = "GENHX"
# A list of valid task names for the MEDIQA-Chat @ ACL-ClinicalNLP
TASK_A = "A"
TASK_B = "B"
TASK_C = "C"
TASKS = [TASK_A, TASK_B, TASK_C]
# These are all related to the output files
ID_COL = "ID"
ENCOUNTER_ID_COL = "encounter_id"
TEST_ID = "TestID"
SYSTEM_OUTPUT_1 = "SystemOutput1"
SYSTEM_OUTPUT_2 = "SystemOutput2"
SYSTEM_OUTPUT = "SystemOutput"
TEAM_NAME = "wanglab"


@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"}
    )
    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 to store the pretrained models downloaded from huggingface.co"},
    )
    use_fast_tokenizer: bool = field(
        default=True,
        metadata={"help": "Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."},
    )
    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 `huggingface-cli login` (necessary to use this script "
                "with private models)."
            )
        },
    )
    resize_position_embeddings: Optional[bool] = field(
        default=None,
        metadata={
            "help": (
                "Whether to automatically resize the position embeddings if `max_source_length` exceeds "
                "the model's position embeddings."
            )
        },
    )


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

    lang: Optional[str] = field(default=None, metadata={"help": "Language id for summarization."})

    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)."}
    )
    text_column: Optional[str] = field(
        default=None,
        metadata={"help": "The name of the column in the datasets containing the full texts (for summarization)."},
    )
    summary_column: Optional[str] = field(
        default=None,
        metadata={"help": "The name of the column in the datasets containing the summaries (for summarization)."},
    )
    train_file: Optional[str] = field(
        default=None, metadata={"help": "The input training data file (a jsonlines or csv file)."}
    )
    validation_file: Optional[str] = field(
        default=None,
        metadata={
            "help": (
                "An optional input evaluation data file to evaluate the metrics (rouge) on (a jsonlines or csv file)."
            )
        },
    )
    test_file: Optional[str] = field(
        default=None,
        metadata={
            "help": "An optional input test data file to evaluate the metrics (rouge) on (a jsonlines or csv 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_source_length: Optional[int] = field(
        default=1024,
        metadata={
            "help": (
                "The maximum total input sequence length after tokenization. Sequences longer "
                "than this will be truncated, sequences shorter will be padded."
            )
        },
    )
    max_target_length: Optional[int] = field(
        default=128,
        metadata={
            "help": (
                "The maximum total sequence length for target text after tokenization. Sequences longer "
                "than this will be truncated, sequences shorter will be padded."
            )
        },
    )
    val_max_target_length: Optional[int] = field(
        default=None,
        metadata={
            "help": (
                "The maximum total sequence length for validation target text after tokenization. Sequences longer "
                "than this will be truncated, sequences shorter will be padded. Will default to `max_target_length`."
                "This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
                "during ``evaluate`` and ``predict``."
            )
        },
    )
    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."
            )
        },
    )
    num_beams: Optional[int] = field(
        default=None,
        metadata={
            "help": (
                "Number of beams to use for evaluation. This argument will be passed to ``model.generate``, "
                "which is used during ``evaluate`` and ``predict``."
            )
        },
    )
    ignore_pad_token_for_loss: bool = field(
        default=True,
        metadata={
            "help": "Whether to ignore the tokens corresponding to padded labels in the loss computation or not."
        },
    )
    source_prefix: Optional[str] = field(
        default="", metadata={"help": "A prefix to add before every source text (useful for T5 models)."}
    )
    source_suffix: Optional[str] = field(default="", metadata={"help": "A suffix to add after every source text."})

    forced_bos_token: Optional[str] = field(
        default=None,
        metadata={
            "help": (
                "The token to force as the first generated token after the decoder_start_token_id."
                "Useful for multilingual models like mBART where the first generated token"
                "needs to be the target language token (Usually it is the target language token)"
            )
        },
    )
    task: str = field(
        default="A", metadata={"help": "Which challenge task to train or evaluate on. Should be one of A, B, or C."}
    )
    run: str = field(
        default="1", metadata={"help": "Which challenge run to produce predictions for. Should be one of 1, 2, or 3."}
    )
    bertscore_model_type: str = field(
        default="microsoft/deberta-large-mnli",
        metadata={
            "help": (
                "Which model to use for BERTScore. This is a good default that balances metric performance and model"
                " size. For options, see:"
                " https://docs.google.com/spreadsheets/d/1RKOVpselB98Nnh_EOC4A2BYn8_201tmPODpNWu4w7xI/edit?usp=sharing"
            )
        },
    )
    bleurt_checkpoint: str = field(
        default="BLEURT-20-D12",
        metadata={
            "help": (
                "Which checkpoint to use for BLEURT. This is a good default that balances metric performance and"
                " model size. For options, see:"
                " https://github.com/google-research/bleurt/blob/master/checkpoints.md"
            )
        },
    )

    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, or test file.")
        else:
            if self.train_file is not None:
                extension = self.train_file.split(".")[-1]
                assert extension in ["csv", "json"], "`train_file` should be a csv or a json file."
            if self.validation_file is not None:
                extension = self.validation_file.split(".")[-1]
                assert extension in ["csv", "json"], "`validation_file` should be a csv or a json file."
            if self.test_file is not None:
                extension = self.test_file.split(".")[-1]
                assert extension in ["csv", "json"], "`test_file` should be a csv or a json file."
        if self.val_max_target_length is None:
            self.val_max_target_length = self.max_target_length
        if self.task not in TASKS:
            raise ValueError(f"Task should be one of {TASKS}. Got: {self.task}.")


summarization_name_mapping = {
    "amazon_reviews_multi": ("review_body", "review_title"),
    "big_patent": ("description", "abstract"),
    "cnn_dailymail": ("article", "highlights"),
    "orange_sum": ("text", "summary"),
    "pn_summary": ("article", "summary"),
    "psc": ("extract_text", "summary_text"),
    "samsum": ("dialogue", "summary"),
    "thaisum": ("body", "summary"),
    "xglue": ("news_body", "news_title"),
    "xsum": ("document", "summary"),
    "wiki_summary": ("article", "highlights"),
    "multi_news": ("document", "summary"),
}


def sanitize_text(text: str, lowercase: bool = False) -> str:
    """Cleans text by removing whitespace, newlines and tabs and (optionally) lowercasing."""
    sanitized_text = " ".join(text.strip().split())
    sanitized_text = sanitized_text.lower() if lowercase else sanitized_text
    return sanitized_text


def parse_omega_conf() -> Dict[str, Any]:
    """A helper function to parse the command line arguments and YAML config files.

    This allows you to call the script with one or more YAML config files and CLI args.

    Example:

    python train.py config1.yml config2.yml --learning_rate=0.0001 --train_batch_size=32

    Arguments in config1.yml will overridden arguments in config2.yml, which will be overridden by any CLI args.
    """
    # Assume anything that ends in ".yml" is a YAML file to be parsed, and everything else is cli args
    cli_args = [arg for arg in sys.argv[1:][:] if not arg.endswith(".yml")]
    yml_confs = [OmegaConf.load(arg) for arg in sys.argv[1:][:] if arg.endswith(".yml")]
    cli_conf = OmegaConf.from_dotlist(cli_args)
    # Merge the YAML configs in the order they were given, with the cli args taking precedence
    conf = OmegaConf.merge(*yml_confs, cli_conf)
    # HuggingFace expects a vanilla python dict, so perform the conversion here
    return OmegaConf.to_object(conf)


def get_global_attention_mask(input_ids: List[List[int]], token_ids: List[int]) -> List[List[int]]:
    """Returns a corresponding global attention mask for `input_ids`, which is 1 for any tokens in
    `token_ids` (indicating the model should attend to those tokens) and 0 elsewhere (indicating the
    model should not attend to those tokens).

    # Parameters

    input_ids : `List[List[str]]`
        The input ids that will be provided to a model during the forward pass.
    token_ids : `List[List[str]]`
        The token ids that should be globally attended to.
    """
    # TODO (John): Ideally this would be vectorized
    global_attention_mask = [[1 if token_id in token_ids else 0 for token_id in batch] for batch in input_ids]
    return global_attention_mask


def main():
    # See all possible arguments in src/transformers/training_args.py
    # or by passing the --help flag to this script.
    # We now keep distinct sets of args, for a cleaner separation of concerns.

    parser = HfArgumentParser((ModelArguments, DataTrainingArguments, Seq2SeqTrainingArguments))
    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]))
    # Our unique parsing strategy (which depends on OmegaConf) exists here
    elif any(argv.endswith(".yml") for argv in sys.argv[1:]):
        conf = parse_omega_conf()
        model_args, data_args, training_args = parser.parse_dict(conf)
        output_dir = Path(training_args.output_dir)
        output_dir.mkdir(parents=True, exist_ok=True)
        (output_dir / "omega_conf.json").write_text(json.dumps(conf, indent=2))
    else:
        model_args, data_args, training_args = parser.parse_args_into_dataclasses()

    # Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
    # information sent is the one passed as arguments along with your Python/PyTorch versions.
    send_example_telemetry("run_summarization", model_args, data_args)

    # 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)],
    )

    if training_args.should_log:
        # The default of training_args.log_level is passive, so we set log level at info here to have that default.
        transformers.utils.logging.set_verbosity_info()

    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}")

    if data_args.source_prefix is None and model_args.model_name_or_path in [
        "t5-small",
        "t5-base",
        "t5-large",
        "t5-3b",
        "t5-11b",
    ]:
        logger.warning(
            "You're running a t5 model but didn't provide a source prefix, which is the expected, e.g. with "
            "`--source_prefix 'summarize: ' `"
        )

    # 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 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 first column for the full texts and the second column for the
    # summaries (unless you specify column names for this with the `text_column` and `summary_column` arguments).
    #
    # 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,
            use_auth_token=True if model_args.use_auth_token else None,
        )
    else:
        data_files = {}
        if data_args.train_file is not None:
            data_files["train"] = data_args.train_file
            extension = data_args.train_file.split(".")[-1]
        if data_args.validation_file is not None:
            data_files["validation"] = data_args.validation_file
            extension = data_args.validation_file.split(".")[-1]
        if data_args.test_file is not None:
            data_files["test"] = data_args.test_file
            extension = data_args.test_file.split(".")[-1]
        raw_datasets = load_dataset(
            extension,
            data_files=data_files,
            cache_dir=model_args.cache_dir,
            use_auth_token=True if model_args.use_auth_token else None,
        )
    # 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.

    # 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,
        cache_dir=model_args.cache_dir,
        revision=model_args.model_revision,
        use_auth_token=True if model_args.use_auth_token else None,
    )
    tokenizer = AutoTokenizer.from_pretrained(
        model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path,
        cache_dir=model_args.cache_dir,
        use_fast=model_args.use_fast_tokenizer,
        revision=model_args.model_revision,
        use_auth_token=True if model_args.use_auth_token else None,
    )
    model = AutoModelForSeq2SeqLM.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,
    )

    # from peft import get_peft_model, LoraConfig, TaskType
    # peft_config = LoraConfig(
    #     task_type=TaskType.SEQ_2_SEQ_LM, inference_mode=False, r=8, lora_alpha=32, lora_dropout=0.1
    # )
    # model = get_peft_model(model, peft_config)
    # logger.info(f"Using LoraConfig: {peft_config}")
    # model.print_trainable_parameters()

    # We resize the embeddings only when necessary to avoid index errors. If you are creating a model from scratch
    # on a small vocab and want a smaller embedding size, remove this test.
    embedding_size = model.get_input_embeddings().weight.shape[0]
    if len(tokenizer) > embedding_size:
        model.resize_token_embeddings(len(tokenizer))

    if model.config.decoder_start_token_id is None and isinstance(tokenizer, (MBartTokenizer, MBartTokenizerFast)):
        if isinstance(tokenizer, MBartTokenizer):
            model.config.decoder_start_token_id = tokenizer.lang_code_to_id[data_args.lang]
        else:
            model.config.decoder_start_token_id = tokenizer.convert_tokens_to_ids(data_args.lang)

    if model.config.decoder_start_token_id is None:
        raise ValueError("Make sure that `config.decoder_start_token_id` is correctly defined")

    if (
        hasattr(model.config, "max_position_embeddings")
        and model.config.max_position_embeddings < data_args.max_source_length
    ):
        if model_args.resize_position_embeddings is None:
            logger.warning(
                "Increasing the model's number of position embedding vectors from"
                f" {model.config.max_position_embeddings} to {data_args.max_source_length}."
            )
            model.resize_position_embeddings(data_args.max_source_length)
        elif model_args.resize_position_embeddings:
            model.resize_position_embeddings(data_args.max_source_length)
        else:
            raise ValueError(
                f"`--max_source_length` is set to {data_args.max_source_length}, but the model only has"
                f" {model.config.max_position_embeddings} position encodings. Consider either reducing"
                f" `--max_source_length` to {model.config.max_position_embeddings} or to automatically resize the"
                " model's position encodings by passing `--resize_position_embeddings`."
            )

    prefix = data_args.source_prefix if data_args.source_prefix is not None else ""
    suffix = data_args.source_suffix if data_args.source_suffix is not None else ""

    # Preprocessing the datasets.
    # We need to tokenize inputs and targets.
    if training_args.do_train:
        if "train" not in raw_datasets:
            raise ValueError("--do_train requires a train dataset")
        column_names = raw_datasets["train"].column_names
    elif training_args.do_eval:
        if "validation" not in raw_datasets:
            raise ValueError("--do_eval requires a validation dataset")
        column_names = raw_datasets["validation"].column_names
    elif training_args.do_predict:
        if "test" not in raw_datasets:
            raise ValueError("--do_predict requires a test dataset")
        column_names = raw_datasets["test"].column_names
    else:
        logger.info("There is nothing to do. Please pass `do_train`, `do_eval` and/or `do_predict`.")
        return

    if isinstance(tokenizer, tuple(MULTILINGUAL_TOKENIZERS)):
        assert (
            data_args.lang is not None
        ), f"{tokenizer.__class__.__name__} is a multilingual tokenizer which requires --lang argument"

        tokenizer.src_lang = data_args.lang
        tokenizer.tgt_lang = data_args.lang

        # For multilingual translation models like mBART-50 and M2M100 we need to force the target language token
        # as the first generated token. We ask the user to explicitly provide this as --forced_bos_token argument.
        forced_bos_token_id = (
            tokenizer.lang_code_to_id[data_args.forced_bos_token] if data_args.forced_bos_token is not None else None
        )
        model.config.forced_bos_token_id = forced_bos_token_id

    # Get the column names for input/target.
    dataset_columns = summarization_name_mapping.get(data_args.dataset_name, None)
    if data_args.text_column is None:
        text_column = dataset_columns[0] if dataset_columns is not None else column_names[0]
    else:
        text_column = data_args.text_column
        if text_column not in column_names:
            raise ValueError(
                f"--text_column' value '{data_args.text_column}' needs to be one of: {', '.join(column_names)}"
            )
    if data_args.summary_column is None:
        summary_column = dataset_columns[1] if dataset_columns is not None else column_names[1]
    else:
        summary_column = data_args.summary_column
        if summary_column not in column_names:
            raise ValueError(
                f"--summary_column' value '{data_args.summary_column}' needs to be one of: {', '.join(column_names)}"
            )

    # Temporarily set max_target_length for training.
    max_target_length = data_args.max_target_length
    padding = "max_length" if data_args.pad_to_max_length else False

    if training_args.label_smoothing_factor > 0 and not hasattr(model, "prepare_decoder_input_ids_from_labels"):
        logger.warning(
            "label_smoothing is enabled but the `prepare_decoder_input_ids_from_labels` method is not defined for"
            f"`{model.__class__.__name__}`. This will lead to loss being calculated twice and will take up more memory"
        )

    def preprocess_function(examples):
        # remove pairs where at least one record is None

        inputs, targets = [], []
        for i in range(len(examples[text_column])):
            if examples[text_column][i] and examples[summary_column][i]:
                inputs.append(examples[text_column][i])

                # Preprocess the targets, so the model must generate the section header before the section text
                if data_args.task == TASK_A and training_args.do_train:
                    target = (
                        f'Section header: {examples["section_header"][i]} Section text: {examples[summary_column][i]}'
                    )
                elif data_args.task == TASK_B:
                    target = examples[summary_column][i].replace("CC:", "CHIEF COMPLAINT")
                    target = target.replace("HPI:", "HISTORY OF PRESENT ILLNESS")
                else:
                    target = examples[summary_column][i]

                targets.append(target)

        inputs = [f"{prefix} {inp} {suffix}".strip() for inp in inputs]

        # Do some basic cleanup on the source and target texts
        inputs = [sanitize_text(inp) for inp in inputs]
        targets = [sanitize_text(tgt) for tgt in targets]

        model_inputs = tokenizer(inputs, max_length=data_args.max_source_length, padding=padding, truncation=True)

        # Add global attention to start tokens (for models that use it)
        model_inputs["global_attention_mask"] = get_global_attention_mask(model_inputs["input_ids"], token_ids=[0])

        # Tokenize targets with the `text_target` keyword argument
        labels = tokenizer(text_target=targets, max_length=max_target_length, padding=padding, truncation=True)

        # If we are padding here, replace all tokenizer.pad_token_id in the labels by -100 when we want to ignore
        # padding in the loss.
        if padding == "max_length" and data_args.ignore_pad_token_for_loss:
            labels["input_ids"] = [
                [(l if l != tokenizer.pad_token_id else -100) for l in label] for label in labels["input_ids"]
            ]

        model_inputs["labels"] = labels["input_ids"]
        return model_inputs

    if training_args.do_train:
        train_dataset = raw_datasets["train"]
        if data_args.max_train_samples is not None:
            max_train_samples = min(len(train_dataset), data_args.max_train_samples)
            train_dataset = train_dataset.select(range(max_train_samples))
        with training_args.main_process_first(desc="train dataset map pre-processing"):
            train_dataset = train_dataset.map(
                preprocess_function,
                batched=True,
                num_proc=data_args.preprocessing_num_workers,
                remove_columns=column_names,
                load_from_cache_file=not data_args.overwrite_cache,
                desc="Running tokenizer on train dataset",
            )

    if training_args.do_eval:
        max_target_length = data_args.val_max_target_length
        eval_dataset = raw_datasets["validation"]
        if data_args.max_eval_samples is not None:
            max_eval_samples = min(len(eval_dataset), data_args.max_eval_samples)
            eval_dataset = eval_dataset.select(range(max_eval_samples))
        with training_args.main_process_first(desc="validation dataset map pre-processing"):
            eval_dataset = eval_dataset.map(
                preprocess_function,
                batched=True,
                num_proc=data_args.preprocessing_num_workers,
                remove_columns=column_names,
                load_from_cache_file=not data_args.overwrite_cache,
                desc="Running tokenizer on validation dataset",
            )

    if training_args.do_predict:
        max_target_length = data_args.val_max_target_length
        predict_dataset = raw_datasets["test"]
        if data_args.max_predict_samples is not None:
            max_predict_samples = min(len(predict_dataset), data_args.max_predict_samples)
            predict_dataset = predict_dataset.select(range(max_predict_samples))
        with training_args.main_process_first(desc="prediction dataset map pre-processing"):
            predict_dataset = predict_dataset.map(
                preprocess_function,
                batched=True,
                num_proc=data_args.preprocessing_num_workers,
                remove_columns=column_names,
                load_from_cache_file=not data_args.overwrite_cache,
                desc="Running tokenizer on prediction dataset",
            )

    # Data collator
    label_pad_token_id = -100 if data_args.ignore_pad_token_for_loss else tokenizer.pad_token_id
    data_collator = DataCollatorForSeq2Seq(
        tokenizer,
        model=model,
        label_pad_token_id=label_pad_token_id,
        pad_to_multiple_of=8 if training_args.fp16 else None,
    )

    # download_configs are necessary to load certain metrics offline.
    # See: https://github.com/huggingface/evaluate/issues/428
    if data_args.task == TASK_A:
        exact_match = evaluate.load(
            "exact_match",
            download_config=datasets.DownloadConfig(
                cache_dir=model_args.cache_dir, local_files_only=True, use_etag=False
            )
            if is_offline_mode()
            else None,
            cache_dir=model_args.cache_dir,
        )
    rouge = evaluate.load(
        "rouge",
        download_config=datasets.DownloadConfig(cache_dir=model_args.cache_dir, local_files_only=True, use_etag=False)
        if is_offline_mode()
        else None,
        cache_dir=model_args.cache_dir,
    )
    bertscore = None
    if data_args.bertscore_model_type:
        bertscore = evaluate.load(
            "bertscore",
            download_config=datasets.DownloadConfig(
                cache_dir=model_args.cache_dir, local_files_only=True, use_etag=False
            )
            if is_offline_mode()
            else None,
            cache_dir=model_args.cache_dir,
        )
    bleurt = None
    if data_args.bleurt_checkpoint:
        bleurt = evaluate.load(
            "bleurt",
            data_args.bleurt_checkpoint,
            # Don't ask me why, but BLEURT needs a different download_config than the other metrics
            download_config=datasets.DownloadConfig(use_etag=False) if is_offline_mode() else None,
            cache_dir=model_args.cache_dir,
        )

    def postprocess_text(preds, labels):
        preds = [sanitize_text(pred) for pred in preds]
        labels = [sanitize_text(label) for label in labels]

        # rougeLSum expects newline after each sentence
        preds = ["\n".join(nltk.sent_tokenize(pred)) for pred in preds]
        labels = ["\n".join(nltk.sent_tokenize(label)) for label in labels]

        return preds, labels

    def extract_header_and_text(texts):
        """Extracts section header and section text predictions from model outputs and targets."""
        section_headers, section_texts = [], []
        for text in texts:
            # Extract from the model predictions and the labels the section headers and the section texts
            section_header = re.findall(r"(?:Section header:)(.*)(?:Section text)", text, re.DOTALL)
            section_text = re.findall(r"(?:Section text:)(.*)", text, re.DOTALL)
            # It is possible the mdoel does not format its outputs as expected. In this case, take the section
            # header to be GENHX (the most likely section header) and the section text to be the whole text.
            section_header = section_header[0].strip() if section_header else GENHX
            section_text = section_text[0].strip() if section_text else text.strip()
            section_headers.append(section_header)
            section_texts.append(section_text)
        return section_texts, section_headers

    def compute_metrics(eval_preds):
        preds, labels = eval_preds
        if isinstance(preds, tuple):
            preds = preds[0]
        decoded_preds = tokenizer.batch_decode(preds, skip_special_tokens=True)
        if data_args.ignore_pad_token_for_loss:
            # Replace -100 in the labels as we can't decode them.
            labels = np.where(labels != -100, labels, tokenizer.pad_token_id)
        decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)

        # Some simple post-processing
        decoded_preds, decoded_labels = postprocess_text(decoded_preds, decoded_labels)

        result = {}

        # For TaskA, we need to extract the header and text from the predictions and labels
        if data_args.task == TASK_A:
            decoded_preds, header_preds = extract_header_and_text(decoded_preds)
            decoded_labels, header_labels = extract_header_and_text(decoded_labels)

            # Compute section header metrics
            result.update(exact_match.compute(predictions=header_preds, references=header_labels))

        # Compute section text metrics...

        # ROUGE
        rouge_results = rouge.compute(predictions=decoded_preds, references=decoded_labels, use_stemmer=True)
        result.update(rouge_results)

        # Compute the arithmetic mean of ROUGE-1, ROUGE-2 and ROUGE-L following: https://arxiv.org/abs/2110.08499
        result["rouge_avg"] = np.mean([result["rouge1"], result["rouge2"], result["rougeL"]]).item()

        # BERTScore
        if bertscore is not None:
            bertscore_result = bertscore.compute(
                predictions=decoded_preds,
                references=decoded_labels,
                batch_size=training_args.per_device_eval_batch_size * 4,
                device=training_args.device,
                # These are mostly based on the recommendations in https://github.com/Tiiiger/bert_score
                model_type=data_args.bertscore_model_type,
                lang="en",
                rescale_with_baseline=True,
                use_fast_tokenizer=True,
            )
            result.update(
                {
                    "bertscore_p": np.mean(bertscore_result["precision"]).item(),
                    "bertscore_r": np.mean(bertscore_result["recall"]).item(),
                    "bertscore_f1": np.mean(bertscore_result["f1"]).item(),
                }
            )

        # BLEURT
        if bleurt is not None:
            bleurt_result = bleurt.compute(predictions=decoded_preds, references=decoded_labels)
            result.update({"bleurt": np.mean(bleurt_result["scores"]).item()})

        # Compute an ensemble score for the generations
        # Use ROUGE-1, following the challenge task evaluation
        result["ensemble_gen_score"] = (
            sum([result["rouge1"], result.get("bertscore_f1", 0), result.get("bleurt", 0)]) / 3
        )
        # If this is Task A, also compute an ensemble score that includes the exact match score
        if data_args.task == TASK_A and "exact_match" in result:
            result["ensemble_score"] = np.mean([result["ensemble_gen_score"], result["exact_match"]]).item()

        result = {k: round(v * 100, 4) for k, v in result.items()}

        # Add length of generated and reference summaries
        generated_lens = [np.count_nonzero(pred != tokenizer.pad_token_id) for pred in preds]
        reference_lens = [np.count_nonzero(label != tokenizer.pad_token_id) for label in labels]
        result["mean_generated_len"] = np.mean(generated_lens)
        result["mean_reference_len"] = np.mean(reference_lens)

        return result

    # Initialize our Trainer
    trainer = Seq2SeqTrainer(
        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 if training_args.predict_with_generate else None,
    )

    # 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)
        trainer.save_model()  # Saves the tokenizer too for easy upload

        metrics = train_result.metrics
        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
    results = {}
    max_length = (
        training_args.generation_max_length
        if training_args.generation_max_length is not None
        else data_args.val_max_target_length
    )
    num_beams = data_args.num_beams if data_args.num_beams is not None else training_args.generation_num_beams
    if training_args.do_eval:
        logger.info("*** Evaluate ***")
        metrics = trainer.evaluate(max_length=max_length, num_beams=num_beams, metric_key_prefix="eval")
        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)

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

        predict_results = trainer.predict(
            predict_dataset, metric_key_prefix="predict", max_length=max_length, num_beams=num_beams
        )
        metrics = predict_results.metrics
        max_predict_samples = (
            data_args.max_predict_samples if data_args.max_predict_samples is not None else len(predict_dataset)
        )
        metrics["predict_samples"] = min(max_predict_samples, len(predict_dataset))

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

        if trainer.is_world_process_zero():
            if training_args.predict_with_generate:
                predictions = tokenizer.batch_decode(
                    predict_results.predictions, skip_special_tokens=True, clean_up_tokenization_spaces=True
                )
                # Predictions may contain "\n", which causes this file to contain an incorrect number of lines.
                # Pointed this out to HF maintainers here: https://github.com/huggingface/transformers/issues/18992
                # but they weren't interested in fixing it.
                predictions = [sanitize_text(pred) for pred in predictions]
                output_prediction_file = os.path.join(training_args.output_dir, "generated_predictions.txt")
                with open(output_prediction_file, "w") as writer:
                    writer.write("\n".join(predictions))

                # Output predictions to a file expected by the challenge task
                # See: https://github.com/abachaa/MEDIQA-Chat-2023#submission-instructions for details on format
                if data_args.task == TASK_A:
                    text_preds, header_preds = extract_header_and_text(predictions)
                    ct_output = {
                        TEST_ID: raw_datasets["test"][ID_COL],
                        SYSTEM_OUTPUT_1: header_preds,
                        SYSTEM_OUTPUT_2: text_preds,
                    }
                else:
                    ct_output = {TEST_ID: raw_datasets["test"][ENCOUNTER_ID_COL], SYSTEM_OUTPUT: predictions}
                ct_fn = f"task{data_args.task.upper().strip()}_{TEAM_NAME}_run{data_args.run}.csv"
                ct_fp = os.path.join(training_args.output_dir, ct_fn)
                pd.DataFrame.from_dict(ct_output).to_csv(ct_fp, index=False)

    kwargs = {"finetuned_from": model_args.model_name_or_path, "tasks": "summarization"}
    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 data_args.lang is not None:
        kwargs["language"] = data_args.lang

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

    return results


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


if __name__ == "__main__":
    main()