【SegRNN 源码理解】验证集和测试集
print("Epoch: {} cost time: {}".format(epoch + 1, time.time() - epoch_time))
train_loss = np.average(train_loss)
vali_loss = self.vali(vali_data, vali_loader, criterion)
test_loss = self.vali(test_data, test_loader, criterion)问题，验证集到底是什么？验证集和测试集不是一样的嘛，调用的函数都是一样的
- 训练集：用于模型拟合的数据样本，模型通过训练集学习数据的模式。23
- 验证集：在模型训练过程中单独留出的样本集，用于调整模型的超参数和初步评估模型的能力。
- 测试集：用于评估最终模型的泛化能力，但不能用于调参或选择特征等算法相关的选择。
模型训练
Use GPU: cuda:0
>>>>>>>start training : illness_60_24_SegRNN_custom_ftM_sl60_pl24_dm512_dr0.0_rtgru_dwpmf_sl12_mae_test_0>>>>>>>>>>>>>>>>>>>>>>>>>>
train 593
val 74
test 170
Epoch: 1 cost time: 207.93248915672302
Epoch: 1, Steps: 37 | Train Loss: 0.6410040 Vali Loss: 0.6463686 Test Loss: 1.3848118
Validation loss decreased (inf --> 0.646369). Saving model ...
Updating learning rate to 0.001
Epoch: 2 cost time: 62.998446226119995
Epoch: 2, Steps: 37 | Train Loss: 0.4989109 Vali Loss: 0.3734880 Test Loss: 1.0756294
Validation loss decreased (0.646369 --> 0.373488). Saving model ...
Updating learning rate to 0.001
Epoch: 3 cost time: 9.611003160476685
Epoch: 3, Steps: 37 | Train Loss: 0.3889044 Vali Loss: 0.3518658 Test Loss: 1.0003682
Validation loss decreased (0.373488 --> 0.351866). Saving model ...
Updating learning rate to 0.001
Epoch: 4 cost time: 9.890604734420776
Epoch: 4, Steps: 37 | Train Loss: 0.3497405 Vali Loss: 0.4116149 Test Loss: 1.0889241
EarlyStopping counter: 1 out of 10
Updating learning rate to 0.0008
Epoch: 5 cost time: 11.188161849975586
Epoch: 5, Steps: 37 | Train Loss: 0.3310404 Vali Loss: 0.3023584 Test Loss: 0.8752079
Validation loss decreased (0.351866 --> 0.302358). Saving model ...
Updating learning rate to 0.0006400000000000002
Epoch: 6 cost time: 10.03389573097229
Epoch: 6, Steps: 37 | Train Loss: 0.2898096 Vali Loss: 0.3139473 Test Loss: 0.8356396
EarlyStopping counter: 1 out of 10
Updating learning rate to 0.0005120000000000001
Epoch: 7 cost time: 9.94393253326416
Epoch: 7, Steps: 37 | Train Loss: 0.2752131 Vali Loss: 0.2802873 Test Loss: 0.8298662
Validation loss decreased (0.302358 --> 0.280287). Saving model ...
Updating learning rate to 0.0004096000000000001
Epoch: 8 cost time: 47.79346227645874
Epoch: 8, Steps: 37 | Train Loss: 0.2621685 Vali Loss: 0.2933615 Test Loss: 0.7996481
EarlyStopping counter: 1 out of 10
Updating learning rate to 0.0003276800000000001
Epoch: 9 cost time: 13.392791509628296
Epoch: 9, Steps: 37 | Train Loss: 0.2522859 Vali Loss: 0.2907325 Test Loss: 0.7990233
EarlyStopping counter: 2 out of 10
Updating learning rate to 0.0002621440000000001
Epoch: 10 cost time: 13.887414693832397
Epoch: 10, Steps: 37 | Train Loss: 0.2448694 Vali Loss: 0.2579939 Test Loss: 0.7900772
Validation loss decreased (0.280287 --> 0.257994). Saving model ...
Updating learning rate to 0.0002097152000000001
Epoch: 11 cost time: 10.111633777618408
Epoch: 11, Steps: 37 | Train Loss: 0.2389001 Vali Loss: 0.2839487 Test Loss: 0.7887061
EarlyStopping counter: 1 out of 10
Updating learning rate to 0.0001677721600000001
Epoch: 12 cost time: 10.194474697113037
Epoch: 12, Steps: 37 | Train Loss: 0.2347271 Vali Loss: 0.2807631 Test Loss: 0.7819831
EarlyStopping counter: 2 out of 10
Updating learning rate to 0.00013421772800000008
Epoch: 13 cost time: 10.330410718917847
Epoch: 13, Steps: 37 | Train Loss: 0.2332735 Vali Loss: 0.2688113 Test Loss: 0.7899975
EarlyStopping counter: 3 out of 10
Updating learning rate to 0.00010737418240000006
Epoch: 14 cost time: 10.05665922164917
Epoch: 14, Steps: 37 | Train Loss: 0.2266295 Vali Loss: 0.2714401 Test Loss: 0.7808742
EarlyStopping counter: 4 out of 10
Updating learning rate to 8.589934592000005e-05
Epoch: 15 cost time: 9.99810791015625
Epoch: 15, Steps: 37 | Train Loss: 0.2243598 Vali Loss: 0.2903035 Test Loss: 0.7775844
EarlyStopping counter: 5 out of 10
Updating learning rate to 6.871947673600005e-05
Epoch: 16 cost time: 10.282510757446289
Epoch: 16, Steps: 37 | Train Loss: 0.2225707 Vali Loss: 0.2826172 Test Loss: 0.7801220
EarlyStopping counter: 6 out of 10
Updating learning rate to 5.497558138880004e-05
Epoch: 17 cost time: 10.422674179077148
Epoch: 17, Steps: 37 | Train Loss: 0.2194050 Vali Loss: 0.2822956 Test Loss: 0.7810600
EarlyStopping counter: 7 out of 10
Updating learning rate to 4.398046511104004e-05
Epoch: 18 cost time: 10.491711378097534
Epoch: 18, Steps: 37 | Train Loss: 0.2180462 Vali Loss: 0.2709097 Test Loss: 0.7783265
EarlyStopping counter: 8 out of 10
Updating learning rate to 3.518437208883203e-05
Epoch: 19 cost time: 10.142646074295044
Epoch: 19, Steps: 37 | Train Loss: 0.2161210 Vali Loss: 0.2623108 Test Loss: 0.7727545
EarlyStopping counter: 9 out of 10
Updating learning rate to 2.8147497671065623e-05
Epoch: 20 cost time: 9.860526323318481
Epoch: 20, Steps: 37 | Train Loss: 0.2158552 Vali Loss: 0.2736951 Test Loss: 0.7759588
EarlyStopping counter: 10 out of 10
Early stopping
>>>>>>>testing : illness_60_24_SegRNN_custom_ftM_sl60_pl24_dm512_dr0.0_rtgru_dwpmf_sl12_mae_test_0<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
