search

lovelyqian / AMeFu-Net

Repository for the paper "Depth Guided Adaptive Meta-Fusion Network for Few-shot Video Recognition"
MIT License
45 stars 3 forks source link

当服务器有4张卡的时候就训练不了了，准确来说是96/GPU数要是16的倍数。因为代码默认跑满四张卡，tensor的size不符合后面的处理。 #2

Open Li-Gui-233 opened 3 years ago

Li-Gui-233 commented 3 years ago

可以将network_train_meta_learning.py文件的第70行代码修改为self.mymodel = torch.nn.DataParallel(self.mymodel, device_ids=[0, 1, 2])

lovelyqian commented 3 years ago

Thanks，u are right!