I uncommented the following code that calculates the validation loss.
However, when training the model the validation loss stays constant and doesn't move a bit. I trained the model on the COCO Dataset and on my Custom Dataset and in both the validation loss stays the same.
e. g. COCO2017
e. g. Custom Dataset (I removed the scaling on the val loss)
Im using the newest commit and the pytorch version. Additionally, I newly cloned the repository when training on the COCO2017 dataset.
I uncommented the following code that calculates the validation loss.
However, when training the model the validation loss stays constant and doesn't move a bit. I trained the model on the COCO Dataset and on my Custom Dataset and in both the validation loss stays the same.
e. g. COCO2017
e. g. Custom Dataset (I removed the scaling on the val loss)
Im using the newest commit and the pytorch version. Additionally, I newly cloned the repository when training on the COCO2017 dataset.
Any help is appreciated.