ZikangZhou/HiVT

Question Regarding to Temporal Encoder

SwagJ opened this issue · 1 comments

SwagJ commented

Hi @ZikangZhou,

There is one question regarding to the temporal encoder layer. When computing the attention in temporal encoder layer, all the agent within the batch is calculated instead of batch-wise attention, I was wondering fusing such batched information? Thank you in advance.

Best,

The temporal encoder performs attention across different time steps within the same agent, so there isn't any information fusion across different agents.