- This topic has 4 replies, 2 voices, and was last updated 8 years ago by .
Viewing 2 reply threads
Viewing 2 reply threads
- You must be logged in to reply to this topic.
› Forums › Speech Synthesis › Merlin › Epoch number
When I train acoustic model(epoch=20, learning rate=0.003, validation set is around 7% of my training data), the plot is like this (see attached picture). It seems that the error for training set keeps decreasing significantly after each epoch, whereas that for validation set becomes steady. Is it necessary to increase the number of epochs?
You are not using warmup_epoch? why?
At the moment, the graph shows that the training may lead to over fitting if you extend beyond 20 epochs.
Sorry I forgot to mention that. I used the default setting for warmup_epoch (20). What’s the difference between training_epochs and warmup_epoch?
I did not realize the problem of over fitting. Thanks for pointing that out and I will not change the number of epochs then.
warmup_epoch is usually set at 10. The learning rate remains same as the value you have set until warmup_epoch (i.e., 10) and then the learning rate becomes half by every epoch thereafter.
If warmup_epoch is used, then the number of epochs can be set any value — and then the network training stops after certain number of epochs, if validation error is not improving.
Thanks!
Some forums are only available if you are logged in. Searching will only return results from those forums if you log in.
Copyright © 2024 · Balance Child Theme on Genesis Framework · WordPress · Log in