search

vanhuyz / CycleGAN-TensorFlow

An implementation of CycleGan using TensorFlow
MIT License
1.19k stars 436 forks source link

so,I can't set the training step by myself,right? #77

Open yt605155624 opened 6 years ago

yt605155624 commented 6 years ago

Could you plese tell me when will the training step stops? I can't set the steps myself, will it stops ubtil the loss lower than some number,can you tell me what the number is?

mcbuehler commented 5 years ago

In model.CycleGAN#optimize you can see that the training stops after a fixed number of steps (100K steps).

Of course, you can modify this number. You also might have to adjust the criterion for the learning rate decay.