Lightning-AI/pytorch-lightning

Returning None from training_step with multi GPU DDP training

Open

#5.243 aberto em 23 de dez. de 2020

Ver no GitHub
 (26 comments) (4 reactions) (1 assignee)Python (3.233 forks)batch import
distributedfeaturehelp wantedpriority: 1

Métricas do repositório

Stars
 (26.687 stars)
Métricas de merge de PR
 (Mesclagem média 9d 15h) (3 fundiu PRs em 30d)

Description

🐛 Bug

Returning None from training_step with multi GPU DDP training freezes the training without exception

To Reproduce

Starting multi-gpu training with a None-returning training_step function.

Example training_step function:

    def training_step(self, batch, batch_idx):
        data, target = batch
        model_outputs = self.forward(images)
        loss = calc_loss(model_outputs, target)

        if torch.isnan(loss) or random.random() < .05:
            return None

        return loss

Example trainer:

 trainer = Trainer(
    gpus=2,
    distributed_backend="ddp"
)

Expected behavior

To continue training with skipping the current batch as pointed out at here.

Environment

No specific environment is needed to reproduce this bug.

Additional context

This issue was mentioned here: #4956 but not with specifics.

Note: By the time this issue being investigated, a help for a workaround would be great!

Guia do colaborador