jaywalnut310 / vits

VITS: Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech
https://jaywalnut310.github.io/vits-demo/index.html
MIT License
6.48k stars 1.21k forks source link

negative loss #164

Open JunhaoHuang0615 opened 11 months ago

JunhaoHuang0615 commented 11 months ago

Got negative loss:

INFO:feng_pai:Saving model and optimizer state at iteration 10 to ../drive/MyDrive/feng_pai/G_6000.pth INFO:feng_pai:Saving model and optimizer state at iteration 10 to ../drive/MyDrive/feng_pai/D_6000.pth INFO:feng_pai:====> Epoch: 10 INFO:feng_pai:Train Epoch: 11 [5%] INFO:feng_pai:[3.00002384185791, 1.4992263317108154, 1.4684797861264087e-05, 0.2452947050333023, 1.7586746215820312, -0.5710883140563965, 6200, 0.00019975014057813518] INFO:feng_pai:Train Epoch: 11 [37%] INFO:feng_pai:[3.0000157356262207, 1.4988996982574463, 1.599153392817243e-06, 0.2704768180847168, 1.5593206882476807, -0.07069837301969528, 6400, 0.00019975014057813518] INFO:feng_pai:Train Epoch: 11 [70%] INFO:feng_pai:[3.000070571899414, 1.4969582557678223, 1.2474944242057973e-06, 0.23094238340854645, 1.8005414009094238, -0.605614423751831, 6600, 0.00019975014057813518]

Could any one tell me these Infos are normal or not? i am traning with Mutiple speakers...

Thank you so mush!