Reconstruction id embedding loss after detach about dg-net HOT 7 CLOSED

In the file of 'configs/latest.yaml', recon_s_w and recon_f_w are set 0, this is strange. Are the two reconstruction loss not used during training?

from dg-net.

@shuxjweb in the trainer you can check that at each iteration after several thousands iters the weights of these losses are increased up to a certain weight

from dg-net.

Hi @juliendenize ,
Sorry for the late response.

Yes. We detached the id embedding to prevent the gradient to E_appearance, which may compromise the re-id performance. The gradient of loss_gen_recon_f_a mainly works for optimising the decoder.

from dg-net.

Hi @shuxjweb

Sorry for the late reply. Actually, we use the losses in one warm-up manner., as @juliendenize mentioned.

from dg-net.

@layumi thank you for your answer. Indeed, I tried to remove the detach function and the GAN collapsed, so I get why you did that and it is truely inspiring. However, I don't see how the gradient of loss_gen_recon_f_a would optimize the decoder because both f_a_recon and f_a are detached.

from dg-net.

Thanks. @juliendenize Yes. You are right.
The gradient of loss_gen_recon_f_a would not optimize the decoder, due to the detach. I think I just left the api for ablation study two years ago.

from dg-net.

Thank you for your answers and your work

from dg-net.