ogroth / tf-gqn

Tensorflow implementation of Neural Scene Representation and Rendering
Apache License 2.0
188 stars 36 forks source link

It stays in local minimum and never goes down #36

Closed wenbingz closed 5 years ago

wenbingz commented 5 years ago

Just like the question raised by the last person, in the step of 200k, the loss goes down to 6930 and never decrease! However, the visualization shows that the model can not reach the expected results and only a mess. I repeated this train process again and the same question happened again. Can anyone help me to repeat the results? Or share your experience. Thanks a lot!

ogroth commented 5 years ago

@wenbingz Please refer to this issue.