Hello, I'm pretty curious about the retraining strategy you used in Table 4. (I understand you use the fine-tuning setting by default). For retraining, do you use the same training strategy with DeiT? or the one you presented in Table 3 with 300 epochs?
Hello, I'm pretty curious about the retraining strategy you used in Table 4. (I understand you use the fine-tuning setting by default). For retraining, do you use the same training strategy with DeiT? or the one you presented in Table 3 with 300 epochs?