Why the FID results I got is different from the results in your paper?

mit-han-lab / gan-compression

[CVPR 2020] GAN Compression: Efficient Architectures for Interactive Conditional GANs

Other

1.1k stars 148 forks source link

Why the FID results I got is different from the results in your paper? #21

Closed zhuzm0902 closed 4 years ago

zhuzm0902 commented 4 years ago

Hi, I follow the instructions in Pix2pix Model Compression "bash scripts/pix2pix/edges2shoes-r/train_mobile.sh", but the model I got only got 28.79 on FID score which is different from the results in your paper. Is there anything I forgot to do?

lmxyy commented 4 years ago

Could you please provide the best FID you've got in your log? FID is sometimes unstable, I think you could set up a new run with a different seed.

zhuzm0902 commented 4 years ago

The results in the log.txt is: " End of epoch 20 / 20 Time Taken: 2441.05 sec

(Evaluate epoch: 20, iters: 239901, time: 83.268) fid: 28.786 fid-mean: 28.266 fid-best: 27.372

Saving the model at the end of epoch 20, iters 239901 learning rate = 0.0000000 " I will run again with a different seed to see the results. Thank you!

lmxyy commented 4 years ago

Hello, did you get the same result? If you did, I would close this issue.