def train()

in main_moco.py [0:0]

• 35 lines of code
• 8 McCabe index (conditional complexity)

def train(train_loader, model, optimizer, scaler, summary_writer, epoch, args):
    batch_time = AverageMeter('Time', ':6.3f')
    data_time = AverageMeter('Data', ':6.3f')
    learning_rates = AverageMeter('LR', ':.4e')
    losses = AverageMeter('Loss', ':.4e')
    progress = ProgressMeter(
        len(train_loader),
        [batch_time, data_time, learning_rates, losses],
        prefix="Epoch: [{}]".format(epoch))

    # switch to train mode
    model.train()

    end = time.time()
    iters_per_epoch = len(train_loader)
    moco_m = args.moco_m
    for i, (images, _) in enumerate(train_loader):
        # measure data loading time
        data_time.update(time.time() - end)

        # adjust learning rate and momentum coefficient per iteration
        lr = adjust_learning_rate(optimizer, epoch + i / iters_per_epoch, args)
        learning_rates.update(lr)
        if args.moco_m_cos:
            moco_m = adjust_moco_momentum(epoch + i / iters_per_epoch, args)

        if args.gpu is not None:
            images[0] = images[0].cuda(args.gpu, non_blocking=True)
            images[1] = images[1].cuda(args.gpu, non_blocking=True)

        # compute output
        with torch.cuda.amp.autocast(True):
            loss = model(images[0], images[1], moco_m)

        losses.update(loss.item(), images[0].size(0))
        if args.rank == 0:
            summary_writer.add_scalar("loss", loss.item(), epoch * iters_per_epoch + i)

        # compute gradient and do SGD step
        optimizer.zero_grad()
        scaler.scale(loss).backward()
        scaler.step(optimizer)
        scaler.update()

        # measure elapsed time
        batch_time.update(time.time() - end)
        end = time.time()

        if i % args.print_freq == 0:
            progress.display(i)