Inconsistencies between paper and code

[sojiro0]

Hello,

thank you for open sourcing your work, it is very interesting! I am reading the paper and the code and i find that there are a few inconsistencies. In the paper you say and show two encoders but in the code i see three. There is an additional encoder for lr RGB images. Accordingly the query in the paper and in the code does not match with the one in the code having more arguments than what is stated in the paper. Also, in the code the BICUBIC interpolation has RMSE of 4.28, 7.14, 11.58 for x4, x8 and x16 respectively. In the code there is a commend stating 7.13. I am trying to replicate this for x4 and i am getting around 23. Is there code that will help replicate the ones as in the paper?

Thank you