search

rfelixmg / frwgan-eccv18

Code for model presented on our paper accepted on European Conference on Computer Vision 2018.
57 stars 23 forks source link

Saving Checkpoints #11

Closed naumanbashir70 closed 4 years ago

naumanbashir70 commented 4 years ago

Hi @rfelixmg ,

Thank you for sharing your valuable code. I run src/sota/awa1/train_gan.sh and I am getting error while saving checkpoints. Could you please guide me in solving this. I am using Google Colab error I will appreciate.

rfelixmg commented 4 years ago

@naumanbashir70 you can just comment the line which is performing symlink, and do it manually before running the next batch. You can also copy the weights, instead.

naumanbashir70 commented 4 years ago

Great!!

Thanks

rfelixmg commented 4 years ago

@naumanbashir70 let me know how it goes, and I can close this issue.

Cheers

naumanbashir70 commented 4 years ago

Sure, I will let you know.

naumanbashir70 commented 4 years ago

@rfelixmg

Yes, I commented the line performing symlink and it is working fine now. There was no need of manually copying the weights. The way you have arranged your code and navigation between them....Impressive!!!

A question and now you can close this issue.

Why there is need of symlink in you code, If I don't consider the last_epoch and other various folders/file saved by symlink like knn.h5, will it affect the performance?

Thanks,

rfelixmg commented 4 years ago

First, tks for your appreciation for the code. I've worked to make it easier and intuitive 😊

Answering your question, I believe it was due to previous version that i've use only the last epoch, and also to make it easier for running manual experiments with the last epoch only. It does not affect the performance. Note that if you want to do experiments in different data sets, you can change the frequency which the checkpoints are made, so you will consider different epochs. Cheers, Rafa