4fde867a550b532a6bbc6ea82db69a0d902862f9,train.py,,,#,22

Before Change

        // Determine approximate time left for epoch
        epoch_batches_left = len(dataloader) - (batch_i + 1)
        time_left = datetime.timedelta(seconds=epoch_batches_left * (time.time() - prev_time))
        prev_time = time.time()
        print(f"---- ETA {time_left}")

After Change

optimizer = torch.optim.Adam(filter(lambda p: p.requires_grad, model.parameters()))

for epoch in range(opt.epochs):
    start_time = time.time()
    for batch_i, (_, imgs, targets) in enumerate(dataloader):

        batches_done = len(dataloader) * epoch + batch_i

        imgs = Variable(imgs.to(device))
        targets = Variable(targets.to(device), requires_grad=False)

        optimizer.zero_grad()

        loss = model(imgs, targets)

        loss.backward()
        optimizer.step()

        // ----------------
        //   Log progress
        // ----------------

        print("\n---- [Epoch %d/%d, Batch %d/%d] ----" % (epoch, opt.epochs, batch_i, len(dataloader)))
        for i in range(4):
            print(
                "[%s] [loss %f, x %f, y %f, w %f, h %f, conf %f, cls %f, cls_acc: %.2f%%, recall: %.5f, precision: %.5f]"
                % (
                    "Total" if i + 1 == 4 else "YOLO Layer %d" % (i + 1),
                    model.losses[i]["loss"],
                    model.losses[i]["x"],
                    model.losses[i]["y"],
                    model.losses[i]["w"],
                    model.losses[i]["h"],
                    model.losses[i]["conf"],
                    model.losses[i]["cls"],
                    model.losses[i]["cls_acc"],
                    model.losses[i]["recall"],
                    model.losses[i]["precision"],
                )
            )

            // Tensorboard logging
            for name, loss in model.losses[i].items():
                loss_name = f"{name}_total" if i + 1 == 4 else f"{name}_{i+1}"
                logger.scalar_summary(loss_name, loss, batches_done)

        // Determine approximate time left for epoch
        epoch_batches_left = len(dataloader) - (batch_i + 1)
        time_left = datetime.timedelta(seconds=epoch_batches_left * (time.time() - start_time) / (batch_i + 1))
        print(f"---- ETA {time_left}")

        model.seen += imgs.size(0)

Instances