search

akanimax / BMSG-GAN

[MSG-GAN] Any body can GAN! Highly stable and robust architecture. Requires little to no hyperparameter tuning. Pytorch Implementation
MIT License
629 stars 105 forks source link

I want to stop and resume the training continuously. #31

Open bemoregt opened 4 years ago

bemoregt commented 4 years ago

Hi, @markstrefford @owang @sridharmahadevan @akanimax @huangzh13

I want to stop and resume the training continuously with saved model.

My resuming commad : "$ python3 train.py --start 300" # I stop at 350 epoch

But, My Training sample is started from the scratch. --;

What's wrong to me ??

Thanks. Best, @bemoregt.

materialvision commented 4 years ago

I am also curious about this. The start flag does not seem to resume the training, only store with another number for the files... It would be great to have a resume flag.

ademus4 commented 3 years ago

Also want to be able to continue from a given point. I tried just now, after training for 100 epoch, starting at 101, it appears to have started from scratch again.

ademus4 commented 3 years ago

Ah figured this out; you can set the path to the previous weights (generator and discriminator), using the input parameters in the train.py file, lines 35 to 51