Function train_resnet

python/paddle/distributed/launch/plugins/test.py:57–96

()

def train_resnet():
    fleet.init(is_collective=True)

    resnet = ResNet(BottleneckBlock, 18, num_classes=class_dim)
    optimizer = optimizer_setting(parameter_list=resnet.parameters())
    optimizer = fleet.distributed_optimizer(optimizer)
    resnet = fleet.distributed_model(resnet)

    dataset = RandomDataset(batch_num * batch_size)
    train_loader = DataLoader(
        dataset,
        batch_size=batch_size,
        shuffle=True,
        drop_last=True,
        num_workers=2,
    )

    print("Distributed training start...")
    for eop in range(epoch):
        resnet.train()

        for batch_id, data in enumerate(train_loader()):
            img, label = data
            label.stop_gradient = True

            out = resnet(img)
            loss = paddle.nn.functional.cross_entropy(input=out, label=label)
            avg_loss = paddle.mean(x=loss)
            acc_top1 = paddle.metric.accuracy(input=out, label=label, k=1)
            acc_top5 = paddle.metric.accuracy(input=out, label=label, k=5)

            avg_loss.backward()
            optimizer.step()
            resnet.clear_gradients()

            print(
                f"[Epoch {eop}, batch {batch_id}] loss: {avg_loss:.5f}, acc1: {acc_top1:.5f}, acc5: {acc_top5:.5f}"
            )

    print("Distributed training completed")


if __name__ == '__main__':