about the query_pos and query_sine_embed

[Pujin823]

hey, thanks for ur wonderful work. but here i got some problem when i read the code,

1. what is the difference between the query_pos and query_sine_embed. it seems that query_sine_embed is the pos embedding vector, but in your code, the first decoder layer, q = q_content + q_pos. Why
2. what is the function of the hyperparameter of self.keep_query_pos? in the original paper, your key insight seems to keep content query and pos query apart so that they can compute attention respectively I would appreciated it if could give me a insight.