What's the difference between training and evaluation other than the training data set path? Why do gradient updates occur during evaluation ? Should I retrain the evaluation with a pre-trained model ? And Where does the evaluation index in the paper come from？
What's the difference between training and evaluation other than the training data set path? Why do gradient updates occur during evaluation ? Should I retrain the evaluation with a pre-trained model ? And Where does the evaluation index in the paper come from？