作者你好。我利用GAT模型，基本在前30个epoch就能跑出bestmodel，之后出现严重的过拟合情况，SF会达到上千。04-15 19:34:14 current lr: [0.001]
04-15 19:34:16 Epoch: 26 , train Loss: 255.3702 MSE: 255.3702RMSE: 15.9803 SF: 165088.39596586.0 examples/sec 0.04 sec/batch
04-15 19:34:17 Epoch: 26 train-Loss: 274.0050 train-MSE: 274.0049-RMSE: 16.5531-SF: 241326.9732 Cost 2.6 sec
04-15 19:34:17 Epoch: 26 val-Loss: 234.8861 val-MSE: 234.8861-RMSE: 15.3260-SF: 358.5778 Cost 0.0 sec 请问这是正常的吗，还是我出现了一些错误，或者需要调模型的参数呢？
作者你好。我利用GAT模型，基本在前30个epoch就能跑出bestmodel，之后出现严重的过拟合情况，SF会达到上千。04-15 19:34:14 current lr: [0.001]
04-15 19:34:16 Epoch: 26 , train Loss: 255.3702 MSE: 255.3702RMSE: 15.9803 SF: 165088.39596586.0 examples/sec 0.04 sec/batch
04-15 19:34:17 Epoch: 26 train-Loss: 274.0050 train-MSE: 274.0049-RMSE: 16.5531-SF: 241326.9732 Cost 2.6 sec
04-15 19:34:17 Epoch: 26 val-Loss: 234.8861 val-MSE: 234.8861-RMSE: 15.3260-SF: 358.5778 Cost 0.0 sec 请问这是正常的吗，还是我出现了一些错误，或者需要调模型的参数呢？