Xuerui Qiu

Results 14 comments of Xuerui Qiu

Thank you for paying attention to our work. This is simple to implement. Here's a step-by-step refinement: 1. "The output of a spiking neuron is binary: it's either 0 or...

Thank you for paying attention to our work. I re-tested the training file and modified the code of main.py. You can try to re-test it.

感谢您关注我们的工作,我这边重新测试了一下权重文件,没有问题, github上开源的、CODE/CIFAR/models/MS_ResNet.py和权重有的地方有出入,我重新上传了新的网络文件可以重新测试一下

![image](https://github.com/bollossom/GAC/assets/87843847/638bde73-a327-4d57-a639-1db5b8153967)

感谢您关注我们的工作,GAC是一种即插即用的编码模块,可以与任意的架构相结合, 同时GAC理论上来说是可以和spikeformer架构结合的, 我这边还没有做过相关实验,但相信结合起来会有一定的性能提升。

Thank you for paying attention to our work. I re-tested the training file and modified the code of main.py. You can try to re-test it.

![image](https://github.com/bollossom/GAC/assets/87843847/3d26f9a5-dbc4-4282-9af4-bac44099901a)

Thank you for paying attention to our work. Next I will answer your questions point to point. 1,2 Please recheck whether you have turned on data enhancements such as cutmix...

感谢您关注我们的工作。这很容易实现。以下是逐步细化: 脉冲神经元的输出是二进制的:它要么是 0,要么是 1。 计算每个脉冲神经元之后的特征图的平均值。例如使用 x.mean()。 接下来,将这些平均值相加以计算SNN中每一层的发放率。 计算得到发放率之后根据文章中的公式11来计算功耗。

使用openai api key辅助测试出来发现qwen2_5_vl还是比官方测出来的低3个点: ``` ----------------------------------- ------------------- ------------------- split dev validation Overall 0.4866666666666667 0.5066666666666667 Accounting 0.6 0.4 Agriculture 0.4 0.5666666666666667 Architecture_and_Engineering 0.0 0.23333333333333334 Art 0.4 0.5666666666666667 Art_Theory 0.8 0.7333333333333333 Basic_Medical_Science 0.8...