where is the setting about position embedding

[liwenssss]

hello, thx for your nice work. I want to know how do you set position embedding? Do you just follow the setting of BERT?

[kevinlin311tw]

There are two related codes for the position encoding/embedding.

First of all, when preparing the input tokens, we concatenate a reference template mesh and the image features for position encoding. You can find relevant code here

Secondly, inside the transformer encoder module, we set the position embedding following the conventional BERT. Relevant code can be found here