Zhengyang Geng

Hi! I appreciate your interest! I am not sure what the ``depth'' here refers to. Does this stand for depth estimation? If so, you can also train your depth estimation...

Update: The bug is reported on 1080Ti with build-in GPU estimator. When I run the code on TPUs, everything is fine. Hope this issue could help you!

Hi @polo5 , In addition to @jerrybai1995 , 1. Yes. For MDEO-Tiny, different inexact gradients can work properly. But for the DEQ-transformer, it would be much harder to **directly** employ...

Hi @BraveDrXuTF , Thank you for your interest in this work! Regarding your questions, 1. This codebase does not contain the naming of "fixed point." The discussion presented in blog...

Hi Andreas @BurgerAndreas , Thanks a lot for your interest! I think a quick fix is to enable `self.deq` to be in the `train` mode while other components of the...

Hi @dorazhiyuyang , Thank you for your interest in DEQs! Could you please elaborate on your problems? In addition, we will release a library and a model zoo for DEQs...