TensorSpeech / TensorFlowTTS

:stuck_out_tongue_closed_eyes: TensorFlowTTS: Real-Time State-of-the-art Speech Synthesis for Tensorflow 2 (supported including English, French, Korean, Chinese, German and Easy to adapt for other languages)
https://tensorspeech.github.io/TensorFlowTTS/
Apache License 2.0
3.85k stars 815 forks source link

About eval data curve #438

Closed unparalleled-ysj closed 3 years ago

unparalleled-ysj commented 3 years ago

@dathudeptrai Hi, I have a question. Why does the training data loss curve look steady downward when training the fastspeech2 v2 model, but the eval data loss curve oscillates so much?

截屏2020-12-22 下午5 12 38 截屏2020-12-22 下午5 12 56
dathudeptrai commented 3 years ago

@dathudeptrai Hi, I have a question. Why does the training data loss curve look steady downward when training the fastspeech2 v2 model, but the eval data loss curve oscillates so much?

截屏2020-12-22 下午5 12 38 截屏2020-12-22 下午5 12 56

that's ok. You can see that the vertical value of eval loss curves is very small that is why you see it oscillates so much :D.