after loading a pretrained torch lightning model, I assume that I should run model.eval() and it turns off the gradiants and dropouts, still, Unexpectedly, the progress bar is shown and some epochs start to proceed. Does anyone have any idea what’s the problem?
model.eval() doesn’t do anything of this as all of this is controlled by the trainer.
The only thing model.eval() does is change the behavior for specific layers like dropout or batchnorm.
2 Likes