hfailab / clip-gen Goto Github PK

Thanks for your great work.

I noticed that you utilize "nn.TransformerEncoderLayer" when constructing "conditional transformer". Since it is used to predict the next token index, I am wondering whether the decoder of transformer is more appropriate for the construction of your conditional transformer? or what's the reason that you don't adopt "nn.TransformerdecoderLayer" ?

Because of the structure of "nn.TransformerEncoderLayer" is simpler or more concise than that of "nn.TransformerDEcoderLayer" ?

Hi, if I want to train the model, I need the hfai package. May I ask if the hfai package cannot be downloaded to the local, can it only be run in your side?

Hi, in the following codes of "TransformerMapper", afther concatenation, why not adding learnable positional embedding?

Hello, Would you like to share the FFR data of coco ?

Hello, I reported an error when running demo, how to solve it?
python demo.py --text "A city bus driving on the city street" --out "bus.jpg"

Hi,

The links provided here seem invalid https://github.com/HFAiLab/clip-gen/blob/main/README_en.md#demo?