search

mlangguth89 / downscaling_benchmark

6 stars 0 forks source link

Resume training of Harris WGAN #37

Closed mlangguth89 closed 1 month ago

mlangguth89 commented 1 month ago

Currently, saving the traind model does not include the optimizer state, so that resuming training from a checkpointed model is impossible. In this branch, this issue will solved by explicitly saving the optimizer states of the generator- and critic-models. Additionally, a load_checkpoint-method will be integrated as well as a corresponding parsing argument to main_train.py.

mlangguth89 commented 1 month ago

This issue-branch was closed with merging #38.