Location Embedding of End token in sequential prediction module.

[anilbatra2185]

Hi @LuoweiZhou ,

Thanks for sharing the code of an interesting work. I am re-implementing the code in Pytorch, as I am unable to install luatorch due to some older libraries.

Query about Location Embedding: As the flaten output of MaxPool is 200 which will be used for location embedding. However, there is only one additional token is used i.e. Start embedding token. I could not find how to identify the end token or stop the sequential prediction.

Appreciate you response!