CRST
CRST copied to clipboard
Loss explodes
![image](https://user-images.githubusercontent.com/39029444/73524482-8f45a500-4450-11ea-8d0c-63dfdd3713fe.png)
When training with generated pseudo label, loss explodes. I run with hyperparameters you provided for SYNTHIA (https://github.com/yzou2/CRST/issues/14) and set my environment to required version. Can you give any advice to solve this problem?
Did you use correct label mapping function to map label to train_id? And also I resized data of SYNTHIA to 1024x2048
In (https://github.com/yzou2/CRST/issues/14), You mentioned INPUT_SIZE = '600,900'. '1024,2048' is correct parameter?
'600,900' is the size of the cropped size. 1024,2048 is the size of the image.