Hello,
This might be a silly question, but I am getting started with CNN and computer vision and had a small doubt.
I am currently working with the Traffic sign detection dataset and am using a GPU to run my own CNN model. Earlier the epochs weren’t taking much time and used to run quickly (~2 min/epoch) but when I run the same model architecture now, the epochs seem to take much time (~25 min/epoch).
Can anyone please provide some high level information about why the epochs are taking up so much time as compared to the previous runs, when the model and data loading configuration is exactly the same?
Thanks.