Congratulations on the SNN you designed for defeating SwinTransformer with the same parameters! Looking forward to your code and pre-trained ckpts.

ZhangHanN1 / SGLFormer

Spiking Global-Local Fusion Transformer

MIT License

14 stars 1 forks source link

Congratulations on the SNN you designed for defeating SwinTransformer with the same parameters! Looking forward to your code and pre-trained ckpts. #1

Open jkhu29 opened 7 months ago

jkhu29 commented 7 months ago

At the same time, I would like to kindly inquire about the training details, such as the dataset, pre-training, optimizer, and training time. Additionally, I am interested in the reasons behind the superior performance of this SNN network architecture compared to ANN. Thank you for your valuable contribution to the field of SNN!

ZhangHanN1 commented 7 months ago

Thank you for your congratulations. I am currently organizing the codes for SGLFormer and will open source pre-trained ckpts and training parameter settings together.

jkhu29 commented 7 months ago

LFTI :)

ldkong1205 commented 6 months ago

Hi @ZhangHanN1, thanks for sharing your work! Any update for the code release?