The results of testing on the pre-trained model about ditto

[Steve-Tod]

Hi, thank you for pointing it out! I found I uploaded the wrong model. Now I updated the link for Ditto_s2m, and it should give performance close to the ones reported in the paper.

[buptmjj]

Hi, thank you very much for your reply. We have obtained the correct results through your latest pre-trained model. But when we test with our own training model, there seem to be some errors, such as cd_mobile. During the training, we changed the batch_size from 8 to 4 because of GPU. We are trying to find out the reason, but we also hope to get your help. Thank you very much.

[Steve-Tod]

Hmm, did you half the learning rate when changing the batch size?

[buptmjj]

Thank you very much for your reminder. I seem to know where the problem is!