Closed donghyunli closed 6 months ago
Hi, thanks for your interest.
Before conducting experiments, it is recommended to ensure that the learning rate is linear to the batch size. The chosen learning rate is based on our current hardware situation, which includes four graphics cards.
Regarding your second question, there is some concern about the exact difference between the accuracy of your reproduced gesture and our reported gap. If the difference is less than 1%, it could be due to numerical errors resulting from different hardware and software environments. It may be worth considering using a newer version of CUDA or PyTorch.
Thank you for the answer. I. re-check the learning rate and version of CUDA or PyTorch.
Hello,i have the same question,can you tell me what is your current learning rate and version of CUDA or PyTorch?
Hello, I have a question about the accuracy of CIFAR10-dvs and DVS128gesture datasets. I've tried many different seeds with given configs (For CIFAR10-dvs, I used spikingjelly, not .pt files), but the accuracy is lower than your proposed results. The accuracy of CIFAR10-dvs and gesture is around 77% and under 99% respectively.
So, I just wonder if there are other configs or settings like augmentation when you got the proposed accuracy.
Thank you.