Skip to main content
. 2021 Apr 28;22(9):4632. doi: 10.3390/ijms22094632

Figure 2.

Figure 2

The effect of the use of different sets of transfer learning hyperparameters. Models F1–F6 correspond to the model finetuning with learning rate: 5 × 102, batch size: 32; learning rate: 5 × 10−2, batch size: 64; learning rate: 5 × 10−2, batch size: 128; learning rate: 5 × 10−3, batch size: 32; learning rate: 5 × 10−3, batch size: 64; learning rate: 5 × 10−3, batch size: 128.