Wangzhen

Hello, Can I ask how long you trained to get this result?

> 可以先用sisnr预训练几步,初始化参数,然后在换成mixloss进行训练 我尝试先用sisnr预训练了几步，并使用第一次保存下的模型初始化，但是mix-loss依然为nan，请问您具体是怎么做的呢

> I had this problem too. I think the non-differentiability of sqrt at 0 is the culprit here. Adding a very small value before taking sqrt, or somehow avoid passing...

> 我觉得aishell还挺干净的,我的SiSNR差不多13左右,但效果还是不太好.我再试试其他数据集吧. 你好，请问您最后换了其他数据集测试了嘛？最后的效果怎么样？