loss=nan

[uchiha99999]

Why did I lose=nan after training for a while?

My input and output channels are 30, and the sequence length is 10

[QQYYHH]

I had the same problem. I solved it by removing the weight_norm from the code, but I don't know why it worked

[uchiha99999]

I had the same problem. I solved it by removing the weight_norm from the code, but I don't know why it worked
Me too, but I solved it by replacing Relu with Sigmoid.