search

eeyhsong / EEG-Conformer

EEG Transformer 2.0. i. Convolutional Transformer for EEG Decoding. ii. Novel visualization - Class Activation Topography.
GNU General Public License v3.0
389 stars 55 forks source link

About the size of the model on a different dataset #19

Open zhyliu-devon opened 1 year ago

zhyliu-devon commented 1 year ago

Hi there, I read the paper and was excited to see a method that is lightweight and efficient. However, when I was trying on my own data set, the size of the model surged up to 600 Gb! Do you have any ideas why? Is it because my dataset comes with 128 channels and 2500 sampling points? I am not that familiar with how the code works so I wonder if you can help me :)

eeyhsong commented 1 year ago

hello @Joscelin-666, Is that 600G on GPU? You may use larger kernel of convolution module to reduce the scale of the data and also capture better features for Transformer module.

zhyliu-devon commented 1 year ago

Yes that's 600 G in my GPU. Thank you for the suggestion!