search

AAnoosheh / ToDayGAN

http://arxiv.org/abs/1809.09767
BSD 2-Clause "Simplified" License
172 stars 32 forks source link

Training on Google Colab #27

Closed iambankaratharva closed 3 years ago

iambankaratharva commented 3 years ago

Want to know where should I change the path such that it corresponds to my drive in the code. I've already tested car images on pre-trained model by changing the paths but am not able to figure out where i should make changes in path for saving the trained model?

AAnoosheh commented 3 years ago

Hi, I have very little experience with Colab, and have never done anything relating to custom paths there.

But this is where saving is done in my code:
https://github.com/AAnoosheh/ToDayGAN/blob/12de3af4f209cd227bdc54160fc3164f97f6732d/models/base_model.py#L42

... and that in turn calls this function to save each network individually in pieces:
https://github.com/AAnoosheh/ToDayGAN/blob/12de3af4f209cd227bdc54160fc3164f97f6732d/models/networks.py#L333

iambankaratharva commented 3 years ago

!python /content/drive/MyDrive/ToDayGAN/train.py --continue_train --which_epoch 75 --name cars --checkpoints_dir /content/drive/MyDrive/ToDayGAN/checkpoints --dataroot /content/drive/MyDrive/ToDayGAN/datasets/d2n --n_domains 2 --niter 100 --niter_decay 0 --save_epoch_freq 25