Open chienyiwang opened 2 years ago
I accidentally set "no_ndc" to "False" in the previous experiments... After setting back "no_ndc" to "True", the results are getting much better now.
If possible, could you elaborate more on the choice of "no_ndc"? Is it heavily dependent on the input format or just empirical? Thanks a lot!
I am not able to get the results with the same quality as the paper either.
I accidentally set "no_ndc" to "False" in the previous experiments... After setting back "no_ndc" to "True", the results are getting much better now.
![]()
I get similar results even if no_ndc is true.
Same here. Trained with person_1's data and config. Training psnr gradually reaches 32 while validation psnr plateaus at 20. Would really appreciate if author can help here!
Hi, Thanks for the great work and kind release of the implementation. I followed the modification in Issue#16 and could train the model with your default .yml file. However, the validation and testing results are not good visually (using ckpt at epoch 500k/600k/...). Most of them are blurry and some of them even only contains background. The training losses are decreasing, but the coarse loss of validation set is not. Could you please provide some suggestions about how to tune the parameters or fix the model training? Thanks a lot!
The validation results and loss curves from tensorboard snapshots are attached below:
![]()
Hi!How did you get these charts?
Same here. Trained with person_1's data and config. Training psnr gradually reaches 32 while validation psnr plateaus at 20. Would really appreciate if author can help here!
Hello, I have also encountered the same problem. Have you resolved it
Hi, Thanks for the great work and kind release of the implementation. I followed the modification in Issue#16 and could train the model with your default .yml file. However, the validation and testing results are not good visually (using ckpt at epoch 500k/600k/...). Most of them are blurry and some of them even only contains background. The training losses are decreasing, but the coarse loss of validation set is not. Could you please provide some suggestions about how to tune the parameters or fix the model training? Thanks a lot!
The validation results and loss curves from tensorboard snapshots are attached below:
![train_loss](https://user-images.githubusercontent.com/15913177/152935585-171b7dda-3066-40e4-b5d0-406f7b4c67e2.PNG)