loss is nan

[xiaoli123456789]

Hello,author!I'm really interested in your project..But I encountered some problems in the training.During my training,my loss value appeared int the form of nan.Can you give me some suggestions or tips?Looking forward to your reply.
my error:

[NormXU]

@xiaoli123456789

The image link is dead. Could you please re-upload it?

I suspect the problem of CrossEntropyLoss Loss turning to NaN comes from a mismatch in class labels when certain labels are set to -1. Please verify if the label IDs assigned to each instance precisely match the label list.

[xiaoli123456789]

Thank you very much for your reply.I uploaded th picture again.

[NormXU]

@xiaoli123456789 Please check your dataset and its assigned labels

[xiaoli123456789]

@NormXU
I'm a bit unclear what data_root refers to ,and what files are included in it.Do you mean the files in label_root when you say assigned labels?Only these two dataset files were used during training process.

Did an error occur while processing the datase?

[xiaoli123456789]

@NormXU
Hello,author,I encountered many problems during the training process.I'm not sure if it's possible to check the author's training weight to see the effect.If you find it inconvenient,I'm sorry.Thank you for your response to me all along.