Some doubts about training process

[ZhuangLii]

Hi, it is an interesting and enlightening work. But i have some doubts, first, why don't you just set the learning rate of block 4,3,2 to zero to intently train the block 1. It may be a substitute of rolling back block 2,3,4. second, why do not join triplet loss in the training process.
TAHNKS for your time to read this message.

[youngminPIL]

Thank you for your interest in my research.

1. It seems to be a meaningful attempt. I thought it was difficult to ignore the continuity between each layer in the learning of the deep network, so I gave a learning rate to every layer.
2. I didn't use triplet loss because it converges slowly and sampling method is tricky.

[ZhuangLii]

Thank you for your interest in my research.

1. It seems to be a meaningful attempt. I thought it was difficult to ignore the continuity between each layer in the learning of the deep network, so I gave a learning rate to every layer.
2. I didn't use triplet loss because it converges slowly and sampling method is tricky.

Thanks for your reply, I will close the issue