[Python] Why are the results so different between my training rounds？？？ [closed]

In my training process, I usually set 5 trials and 400 epochs, which is also the setting of other people's work. However, when I was training, the results I got for each trial key were very different. For example, the AUC of the first trial was around 0.7. The AUC obtained by the next trial was around 0.6 and even got worse with more training. This results in a very poor average result. Trying to fix the random seed, increase the dropout value, adjust the learning rate, etc., did not fully solve this problem. I want to know if you have encountered this situation? I really don't know what to do ......T-T

Trying to fix the random seed, increase the dropout value, adjust the learning rate, etc., did not fully solve this problem. I want to know if you have encountered this situation? I really don't know what to do ......T-T

Continue reading...