Personalized Attention Details

[WillPowellUk]

Dear @akaneUS and @yh0903 ,

Thank you for providing your code alongside your very interesting paper - it is unfortunately quite rare to have code available alongside a publication so this makes a change.

I was just going through your network and I cannot find the PA (personalized attention) framework. I.e. the implementation of a personalized SAB (self-attention block). Unless I have missed it, please could you provide more details on how this is implemented?

Also, could I ask how the 60 steps x 1 min segments are fed into the dataloader?

Cheers,
Will