scripts/train_detection.py [224:237]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
def train(model, optimizer, scheduler, dataloader, meters, **varargs):
    model.train()
    dataloader.batch_sampler.set_epoch(varargs["epoch"])
    optimizer.zero_grad()
    global_step = varargs["global_step"]
    loss_weights = varargs["loss_weights"]

    data_time_meter = AverageMeter((), meters["loss"].momentum)
    batch_time_meter = AverageMeter((), meters["loss"].momentum)

    data_time = time.time()
    for it, batch in enumerate(dataloader):
        # Upload batch
        batch = {k: batch[k].cuda(device=varargs["device"], non_blocking=True) for k in NETWORK_INPUTS}
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



scripts/train_panoptic.py [270:283]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
def train(model, optimizer, scheduler, dataloader, meters, **varargs):
    model.train()
    dataloader.batch_sampler.set_epoch(varargs["epoch"])
    optimizer.zero_grad()
    global_step = varargs["global_step"]
    loss_weights = varargs["loss_weights"]

    data_time_meter = AverageMeter((), meters["loss"].momentum)
    batch_time_meter = AverageMeter((), meters["loss"].momentum)

    data_time = time.time()
    for it, batch in enumerate(dataloader):
        # Upload batch
        batch = {k: batch[k].cuda(device=varargs["device"], non_blocking=True) for k in NETWORK_INPUTS}
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -