scripts/nlp_example.py

# coding=utf-8
# Copyright 2021 The HuggingFace Inc. 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.

import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup

from accelerate import Accelerator, DistributedType
from accelerate.utils import set_seed


def get_dataloaders(accelerator: Accelerator, batch_size: int = 16):
    """
    Creates a set of `DataLoader`s for the `glue` dataset,
    using "bert-base-cased" as the tokenizer.

    Args:
        accelerator (`Accelerator`):
            An `Accelerator` object
        batch_size (`int`, *optional*):
            The batch size for the train and validation DataLoaders.
    """
    tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
    datasets = load_dataset("glue", "mrpc")

    def tokenize_function(examples):
        # max_length=None => use the model max length (it's actually the default)
        outputs = tokenizer(examples["sentence1"], examples["sentence2"], truncation=True, max_length=None)
        return outputs

    # Apply the method we just defined to all the examples in all the splits of the dataset
    # starting with the main process first:
    with accelerator.main_process_first():
        tokenized_datasets = datasets.map(
            tokenize_function,
            batched=True,
            remove_columns=["idx", "sentence1", "sentence2"],
        )

    # We also rename the 'label' column to 'labels' which is the expected name for labels by the models of the
    # transformers library
    tokenized_datasets = tokenized_datasets.rename_column("label", "labels")

    def collate_fn(examples):
        # On TPU it's best to pad everything to the same length or training will be very slow.
        max_length = 128 if accelerator.distributed_type == DistributedType.TPU else None
        # When using mixed precision we want round multiples of 8/16
        if accelerator.mixed_precision != "no":
            pad_to_multiple_of = 8
        else:
            pad_to_multiple_of = None

        return tokenizer.pad(
            examples,
            padding="longest",
            max_length=max_length,
            pad_to_multiple_of=pad_to_multiple_of,
            return_tensors="pt",
        )

    # Instantiate dataloaders.
    train_dataloader = DataLoader(
        tokenized_datasets["train"], shuffle=True, collate_fn=collate_fn, batch_size=batch_size, drop_last=True
    )
    eval_dataloader = DataLoader(
        tokenized_datasets["validation"],
        shuffle=False,
        collate_fn=collate_fn,
        batch_size=32,
        drop_last=(accelerator.mixed_precision == "fp8"),
    )

    return train_dataloader, eval_dataloader


def training_function(config):
    # Initialize accelerator
    accelerator = Accelerator(
        mixed_precision="fp16",
        log_with="aim",
        project_dir="aim_logs"
    )
    # Sample hyper-parameters for learning rate, batch size, seed and a few other HPs
    lr = config["lr"]
    num_epochs = int(config["num_epochs"])
    seed = int(config["seed"])
    batch_size = 16 if accelerator.num_processes > 1 else 32
    config["batch_size"] = batch_size
    metric = evaluate.load("glue", "mrpc")

    set_seed(seed, device_specific=True)
    train_dataloader, eval_dataloader = get_dataloaders(accelerator, batch_size)
    model = AutoModelForSequenceClassification.from_pretrained("bert-base-cased", return_dict=True)
    lr = lr * accelerator.num_processes

    optimizer = AdamW(params=model.parameters(), lr=lr)
    lr_scheduler = get_linear_schedule_with_warmup(
        optimizer=optimizer,
        num_warmup_steps=0,
        num_training_steps=(len(train_dataloader) * num_epochs),
    )

    model, optimizer, train_dataloader, eval_dataloader, lr_scheduler = accelerator.prepare(
        model, optimizer, train_dataloader, eval_dataloader, lr_scheduler
    )

    accelerator.init_trackers(f'{accelerator.num_processes}_gpus', config)

    current_step = 0
    for epoch in range(num_epochs):
        model.train()
        total_loss = 0
        for _, batch in enumerate(train_dataloader):
            lr = lr_scheduler.get_lr()
            outputs = model(**batch)
            loss = outputs.loss
            batch_loss = accelerator.gather(loss).detach().mean().cpu().float()
            total_loss += batch_loss
            current_step += 1
            accelerator.log(
                {
                    "batch_loss":batch_loss,
                    "learning_rate":lr,
                }, 
                step=current_step, 
                log_kwargs={"aim":{"epoch":epoch}}
                )
            accelerator.backward(loss)
            optimizer.step()
            lr_scheduler.step()
            optimizer.zero_grad()
            current_step += 1

        model.eval()
        for step, batch in enumerate(eval_dataloader):
            # We could avoid this line since we set the accelerator with `device_placement=True`.
            batch.to(accelerator.device)
            with torch.no_grad():
                outputs = model(**batch)
            predictions = outputs.logits.argmax(dim=-1)
            predictions, references = accelerator.gather_for_metrics((predictions, batch["labels"]))
            metric.add_batch(
                predictions=predictions,
                references=references,
            )

        eval_metric = metric.compute()
        
        # Use accelerator.print to print only on the main process.
        accelerator.print(f"epoch {epoch}:", eval_metric)

        accelerator.log(
            {
                "accuracy": eval_metric["accuracy"],
                "f1": eval_metric["f1"],
                "train_loss": total_loss.item() / len(train_dataloader),
            },
            log_kwargs = {"aim":{"epoch":epoch}}
        )
    accelerator.end_training()


def main():
    config = {"lr": 2e-5, "num_epochs": 3, "seed": 42}
    training_function(config)


if __name__ == "__main__":
    main()