train models with multicrops

[lxysl]

I wonder why can directly inference the model with a list? In main_swav.py line 284

for it, inputs in enumerate(train_loader):
    embedding, output = model(inputs)

Doesn't the multicrop dataloader yield a list of tensors with different sized image crop?

    def __getitem__(self, index):
        path, _ = self.samples[index]
        image = self.loader(path)
        multi_crops = list(map(lambda trans: trans(image), self.trans))
        if self.return_index:
            return index, multi_crops
        return multi_crops