Analysis of loss value

[Vitusyc]

Hello, I recently ran the implementation with your code, but I found that the loss value is increasing as the number of iterations increases.According to the theory, the loss value should be reduced. At the same time, I understand what you mean by this code: when the predicted tag is the same as the actual tag, the weight of the sample will be reduced. Could you consider increasing the weight of the sample when the predicted tag is different from the actual tag?I hope to receive your reply