代码运用到自己的模型

[Tangzhaotz]

怎么把这种方法运用到自己的训练中？仅仅多训练12个epochs吗？

[hyz-xmaster]

Yes, but you need use cyclical learning rates for the extra training and then average these 12 checkpoints.

[Tangzhaotz]

where the code for this part?

[hyz-xmaster]

Please see the Getting started.

[Tangzhaotz]

ok，thank you

[Mobu59]

Yes, but you need use cyclical learning rates for the extra training and then average these 12 checkpoints.

hello ,I use the get_swa_model.py and CosineAnnealingLR（initial learning rate 1e-3,eta_min=1e-4）to train my best model for 12 extra epochs,and then get avg_model.But when I use the avg_model to test, map decreased...