Why should the length of the vqcode for data processing be multiplied by four?

Thank you for providing such a great open source code!

I want to know why the vqcode is multiplied by four in line 58 of the [convert_imagepair_cc512.py](https://github.com/FoundationVision/Liquid/blob/main/data_process/convert_imagepair_cc512.py) file.