-
我使用您们提供的源码跑cifar和mnist时,使用resnet50的addernet版要比torchvision里自带的resnet50慢很多(用5000个train samples和1000个test samples时,resnet50是6s一轮,adder_resnet50()居然是两分钟一轮,请问这是为什么呢?如果GPU是专门优化的话,我用cpu跑的时候速度依然慢得多,问题出在哪里呢?
-
Hi, Author.
I have noticed that the experiments shown in the paper say that "Specifically, with full-precision
floating-point (FP32) ShiftAddNet even surpasses both the multiplication-based ConvNet …
-
您好
我们参考 https://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html
利用自己的数据集对网络进行微调,我们遇到了困难
普通的resnet通过fc层输出in_featrues,进行微调训练,但是addernet中resnet的fc是conv2d,我不知道如何得到fc层的输入,或者说avgpool的输出。
…
-
您好,冒昧打扰。 之前拜读了您团队的Pretrained ImageProcessingTransformer, 受益匪浅。在尝试复现您的论文,但是在Dataloader和Datasets的写过程中遇到了问题。不知如何裁剪图片, 如何对数据进行预处理。论文中提到Overlap10Pix, 但是边缘部分如何处理!
若能分享解惑,不甚感激。
感谢感谢,期待回复。
-
hi,
How does transfer learning train its own data set to modify pre-training parameters?
thanks
-
您好!想请问下开源代码设置的学习率是否与论文中提到的学习率是不同的?
一个是余弦衰减,一个是自适应学习率?
麻烦解答下,谢谢啊
-
I replace adder.adder2d with torch.nn.Conv2d, and replace torch.cdist with [my_cdist](https://github.com/pytorch/pytorch/issues/15253). train my network, the speed of the new model slower than the old…
-
Hi,
We recently noticed that you update the comparison with ShiftAddNet in your readme.
However, the chosen accuracy of ShiftAddNet (i.e., 85.10%) is **wrongly** from our adaption experiments, …
-
Hi,
Thanks for this excellenct work!
I cannot find the project page (detailed codes) of AdderSR: Towards Energy Efficient Image Super-Resolution
-
I got an issue that CUDA out of memory although I've changed any batch size such as 1, 2, 4, 8, 16.
Any one can help me, please?