Writing Reusable and Reproducible Pipelines for Training Neural Networks

Andrey Lukyanenko

Senior DS, Careem

About me

  • ~4 years as ERP-system consultant
  • self-study for switching career
  • DS since 2017
  • Lead a medical chatbot project
  • Lead an R&D CV team
  • Senior DS in anti-fraud team

Content

  • Styles of writing training code
  • Reusable pipeline: why do it and how to start
  • Functionality of training pipeline

Styles of writing code

Training pipeline

Reasons for writing pipeline

  • Writing everything from scratch takes time and can have errors
  • You have repeatable pieces of code anyway
  • Better understanding how the things work
  • Standardization among the team

My pipeline

My pipeline: core ideas

  • Replaceable modules
  • Hydra/OmegaConf for configuration files
  • Values in configuration files can be changed in CLI
  • Logging and reproducibility

My pipeline

My pipeline

    def configure_optimizers(self):
        optimizer = load_obj(self.cfg.optimizer.class_name)(self.model.parameters(),
        **self.cfg.optimizer.params)
        scheduler = load_obj(self.cfg.scheduler.class_name)(optimizer,
        **self.cfg.scheduler.params)

        return (
            [optimizer],
            [{'scheduler': scheduler,
            'interval': self.cfg.scheduler.step,
            'monitor': self.cfg.scheduler.monitor}],
        )

My pipeline

>>> python train.py
>>> python train.py optimizer=sgd
>>> python train.py model=efficientnet_model
>>> python train.py model.encoder.params.arch=resnet34
>>> python train.py datamodule.fold_n=0,1,2 -m
@hydra.main(config_path='conf', config_name='config')
def run_model(cfg: DictConfig) -> None:
    os.makedirs('logs', exist_ok=True)
    print(cfg.pretty())
    if cfg.general.log_code:
        save_useful_info()
    run(cfg)


if __name__ == '__main__':
    run_model()

Training loop

def training_step(self, batch, *args, **kwargs):  # type: ignore
    image = batch['image']
    logits = self(image)

    target = batch['target']
    shuffled_target = batch.get('shuffled_target')
    lam = batch.get('lam')
    if shuffled_target is not None:
        loss = self.loss(logits, (target, shuffled_target, lam)).view(1)
    else:
        loss = self.loss(logits, target)
    self.log('train_loss', loss,
             on_step=True, on_epoch=True, prog_bar=True, logger=True)

    for metric in self.metrics:
        score = self.metrics[metric](logits, target)
        self.log(f'train_{metric}', score,
		 on_step=True, on_epoch=True, prog_bar=True, logger=True)
    return loss

Reproducibility

def set_seed(seed: int = 42) -> None:
    np.random.seed(seed)
    random.seed(seed)
    os.environ['PYTHONHASHSEED'] = str(seed)
    torch.backends.cudnn.benchmark = False
    torch.backends.cudnn.deterministic = True
    torch.manual_seed(seed)
    torch.cuda.manual_seed(seed)
    torch.cuda.manual_seed_all(seed)

Experiment tracking

Changing hyperparameters

>>> python train.py optimizer=sgd
>>> python train.py trainer.gpus=2

Basic functionality

  • Easy to modify for a similar problem
  • Make predictions
  • Make predictions without pipeline
  • Changing isn't very complicated

Useful functionality

  • Configs, configs everywhere
  • Templates of everything
  • Training on folds and hyperparameter optimization
  • Training with stages
  • Using pipeline for a variety of tasks
  • Sharable code and documentation
  • Various cool tricks

Links

Contacts

Writing reusable and reproducible pipelines for training neural networks

By Andrey Lukyanenko

Writing reusable and reproducible pipelines for training neural networks

  • 170

More from Andrey Lukyanenko