starstarb

Results 5 comments of starstarb

郑老师,没有修改过您的transformer模型,我最近重新训练了一遍您的模型swin Transformer脚本,训练模型没有问题,测试的时候还是报这个错误,训练脚本和测试脚本也没有作修改,可是依然报这个错误,调试了很久还是没有结果。 问题: Traceback (most recent call last): File "test.py", line 325, in gallery_feature = extract_feature(model, dataloaders['gallery']) File "test.py", line 227, in extract_feature ff += outputs TypeError: unsupported operand type(s)...

6. outputs = outputs[0] outputs.unsqueeze(dim = 0) 我再次对outpus进行降维,然后升维,意思是取第一个维度进行测试,代码能跑通,但是 Rank@1:0.936234 Rank@5:0.971990 Rank@10:0.985101 mAP:0.694360 精度比Resnet提升了3个点,感觉还是有问题

7. #------------------------------------------------------------------------------ outputs = torch.tensor([item.cpu().detach().numpy() for item in outputs]).cuda() outputs = outputs[1] outputs.unsqueeze(dim = 0) 第二个维度测试,精度依然一样,Rank@1:0.936234 Rank@5:0.971990 Rank@10:0.985101 mAP:0.694360 比您提出的低了10个点,我训练时仅修改了batchsize = 100,感觉还是有问题

and when I remove the ```use_tanh``` argument of all, I meet this bug ``` return [val for val in binary_codes_list if val.shape[1] in self.eval_dims] AttributeError: 'dict' object has no attribute...