search

AFeng-x / SMT

This is an official implementation for "Scale-Aware Modulation Meet Transformer".
https://arxiv.org/abs/2307.08579
MIT License
186 stars 15 forks source link

DW_Conv #5

Open Jinchen2028 opened 1 year ago

Jinchen2028 commented 1 year ago

Thank you very much for sharing your great work, in the paper I noticed that DW_Conv were used in the SAM module. If I use vanilla convolution instead of DW_Conv, will it have a performance improvement in addition to growing parameters?

AFeng-x commented 1 year ago

Thank you for your interest! We initially attempted to use vanilla convolution, but we found that it led to a larger number of parameters in the module. As a result, we ultimately opted for depthwise convolution. Theoretically, vanilla convolution has stronger modeling capabilities than depthwise convolution and can potentially improve accuracy. However, we are uncertain about the extent of this improvement in practice. Therefore, we suggest that you run an experiment to see for yourself, and we welcome you to share the results with us. Thank you!

Jinchen2028 commented 1 year ago

Thank you so much for your answer！