Continue training of existing model

Mukosame / Zooming-Slow-Mo-CVPR-2020

Fast and Accurate One-Stage Space-Time Video Super-Resolution (accepted in CVPR 2020)

GNU General Public License v3.0

908 stars 165 forks source link

Continue training of existing model #41

Closed mpriessner closed 3 years ago

mpriessner commented 3 years ago

Hello Mukosame,

Congratulations to this amazing paper. I just tried it and it is really good! Is there a way where I can continue training on an existing pretrained model? I couldn't find this functionality in your code (but maybe I missed it).

Otherwise, is there a way to easily adapt the code that this is also possible?

mpriessner commented 3 years ago

I think I found it now. I will test it soon. Sorry for the confusion!