|
def loss_spatials(self, outputs, targets, indices, num_masks, layer_id, extra): |
As I can not find this loss function in mask2former or maskformer repo,should I remove this loss function if I do not need language encoder part and make x-decoder a generic segment model?
X-Decoder/modeling/modules/criterion.py
Line 469 in 165f8a6
As I can not find this loss function in mask2former or maskformer repo,should I remove this loss function if I do not need language encoder part and make x-decoder a generic segment model?