Yimeng Shan

Thank you very much, it has been very helpful to me. My current goal is to carefully read the LEMS file and find a way to reproduce the synapses, neurons,...

From the perspective of rate coding, we can consider input data. For SNN, we generally understand it as encoding the data first, and then inputting the encoded result into the...

I'm very sorry, I didn't elaborate enough on the issue. I am using the Ubuntu18.04 operating system and did not run build.sh. Instead, I ran run.sh(As you mentioned in readme,...

From the perspective of a single step mode, we will reset the neurons in the entire network after T (timesteps) inputs of a certain sample, rather than performing a reset...

I think a simple logic can solve your problem: when the second sample is inputted, the residual membrane potential of each neuron is meaningless, which is equivalent to noise for...

The function of ``` function.reset_net(net) ``` is to reset the membrane potential of each neuron to zero. The function of ``` optimizer.zero_grad() ``` is to reset the gradient to zero....

Perhaps my words caused your misunderstanding. Below, I will give you an example based on the specific implementation (still using single step mode as an example): ``` for img, label...

https://spikingjelly.readthedocs.io/zh-cn/latest/activation_based/monitor.html 可以通过spikingjelly的monitor对每层IF神经元的Spike进行监视以获取spike数量或者firing rate. 某一层的神经元数量就是输入特征图的元素数量(C * H * W)，但要注意时间维度对神经元层的作用方式是重复输入，而不是每一个time都重新创建一层尺寸为[C, H, W]的神经元. 所以可以通过输出神经元前一层的特征图尺寸，再通过上述方法计算得到神经元数量.

最好提供下你的代码，是在某个epoch更改了输入的tensor尺寸吗？

I greatly appreciate your help; this has been very beneficial to me.