Is the pre-trained model in the repo already fine tuned on FER+?

[shv07]

Hi,

Thank you for providing the codes and the pre-trained models.

Is the pre-trained model supplied in the repo fine tuned on FER+ dataset or just trained on AffectNet? You have mentioned in paper that a good performance is achieved on FER+ after its fine tuning.

It would be great if you could provide the performance of the trained model supplied in the repo on the datasets you tested with.

Thanks,