# kagami语音模型训练参数



duration of dataset：96min

epoch: 1424

batch_size :24

segments = 638

step:38400

Losses vs step

![截屏2023-04-15 18.48.09](./README.assets/%E6%88%AA%E5%B1%8F2023-04-15%2018.48.09.png)