Hi,
Thank you for your excellent work! I have a quick question regarding train.py (lines 73–89). The dataloader appears to return two synthetic images, but only mgt is used, while mgt2 seems not... I couldn’t find a related training scheme in the paper. Could you please clarify the purpose of mgt2?
Many thanks!
Hi,
Thank you for your excellent work! I have a quick question regarding train.py (lines 73–89). The dataloader appears to return two synthetic images, but only mgt is used, while mgt2 seems not... I couldn’t find a related training scheme in the paper. Could you please clarify the purpose of mgt2?
Many thanks!