How to update simnet using loss_ce_supervised?

[pengzhenjia]

How to update simnet using loss_ce_supervised? It seems that loss_ce_supervised is generated by model_ and has nothing to do with simnet.

update simnet parameters

feat, logits = model_(inputs) logits = de_interleave(logits, 2*args.mu+1) logits_l = logits[:batch_l[0]] loss_ce_supervised = F.cross_entropy(logits_l, targets_l)

optimizer_simnet.zero_grad() loss_ce_supervised.backward() optimizer_simnet.step()