How did you add c_current and c_context for MultiEnc-enc model?

[Yusuke196]

Thank you for the interesting work.

I would like to ask how you performed $\lambda c_{current} + (1 - \lambda) c_{context}$ to do the integration on the encoder side. Generally, the number of tokens in $c_{current}$ and $c_{context}$ should be different, so did you perform some kind of padding?

If you could include some actual codes to explain, I would also appreciate it very much.